Serveur d'exploration sur les pandémies grippales

Attention, ce site est en cours de développement !
Attention, site généré par des moyens informatiques à partir de corpus bruts.
Les informations ne sont donc pas validées.

2006 EUROPEAN SOCIETY FOR MEDICAL DECISION MAKING MEETING

Identifieur interne : 002538 ( Istex/Corpus ); précédent : 002537; suivant : 002539

2006 EUROPEAN SOCIETY FOR MEDICAL DECISION MAKING MEETING

Auteurs :

Source :

RBID : ISTEX:525CB0AB84383F0EC9BA6467590F85311F2844B5

English descriptors


Url:
DOI: 10.1177/0272989X080270040104

Links to Exploration step

ISTEX:525CB0AB84383F0EC9BA6467590F85311F2844B5

Le document en format XML

<record>
<TEI wicri:istexFullTextTei="biblStruct">
<teiHeader>
<fileDesc>
<titleStmt>
<title xml:lang="en">2006 EUROPEAN SOCIETY FOR MEDICAL DECISION MAKING MEETING</title>
</titleStmt>
<publicationStmt>
<idno type="wicri:source">ISTEX</idno>
<idno type="RBID">ISTEX:525CB0AB84383F0EC9BA6467590F85311F2844B5</idno>
<date when="2007" year="2007">2007</date>
<idno type="doi">10.1177/0272989X080270040104</idno>
<idno type="url">https://api.istex.fr/ark:/67375/M70-DPPD2M40-0/fulltext.pdf</idno>
<idno type="wicri:Area/Istex/Corpus">002538</idno>
<idno type="wicri:explorRef" wicri:stream="Istex" wicri:step="Corpus" wicri:corpus="ISTEX">002538</idno>
</publicationStmt>
<sourceDesc>
<biblStruct>
<analytic>
<title level="a" type="main" xml:lang="en">2006 EUROPEAN SOCIETY FOR MEDICAL DECISION MAKING MEETING</title>
</analytic>
<monogr></monogr>
<series>
<title level="j">Medical Decision Making</title>
<idno type="ISSN">0272-989X</idno>
<idno type="eISSN">1552-681X</idno>
<imprint>
<publisher>Sage Publications</publisher>
<pubPlace>Sage CA: Los Angeles, CA</pubPlace>
<date type="published" when="2007-07">2007-07</date>
<biblScope unit="volume">27</biblScope>
<biblScope unit="issue">4</biblScope>
<biblScope unit="page" from="-3">e-3</biblScope>
<biblScope unit="page" to="-4">e-4</biblScope>
</imprint>
<idno type="ISSN">0272-989X</idno>
</series>
</biblStruct>
</sourceDesc>
<seriesStmt>
<idno type="ISSN">0272-989X</idno>
</seriesStmt>
</fileDesc>
<profileDesc>
<textClass>
<keywords scheme="Teeft" xml:lang="en">
<term>2department</term>
<term>2imperial college</term>
<term>2university</term>
<term>Abstracts conclusions</term>
<term>Abstracts methods</term>
<term>Abstracts purpose</term>
<term>Abstracts results</term>
<term>Actual decision</term>
<term>Acute infrarenal</term>
<term>Additional language versions</term>
<term>Alternative approach</term>
<term>Alternative screening strategies</term>
<term>Amsterdam</term>
<term>Analogue</term>
<term>Anchor state</term>
<term>Anesthetist</term>
<term>Annual health care expenditures</term>
<term>Antihypertensive medication</term>
<term>Appropriate diagnoses</term>
<term>Aqlq scores</term>
<term>Artery disease</term>
<term>Asthma</term>
<term>Asthma control</term>
<term>Asthma quality</term>
<term>Aston university</term>
<term>Asymmetry</term>
<term>Atrial fibrillation</term>
<term>Australia purpose</term>
<term>Autonomous decision</term>
<term>Available evidence</term>
<term>Barthel index</term>
<term>Baseline</term>
<term>Bayesian</term>
<term>Bayesian approach</term>
<term>Better asthma control</term>
<term>Better decisions</term>
<term>Bias</term>
<term>Bias term</term>
<term>Biennial european meeting</term>
<term>Biosource corporation</term>
<term>Birth weight</term>
<term>Blood pressure</term>
<term>Bootstrap approach</term>
<term>Breast cancer</term>
<term>Budget impact analysis</term>
<term>Canada purpose</term>
<term>Cancer</term>
<term>Cancer screening</term>
<term>Cardiovascular</term>
<term>Cardiovascular disease</term>
<term>Cardiovascular events</term>
<term>Carlo</term>
<term>Case study</term>
<term>Cdss</term>
<term>Center rotterdam</term>
<term>Centre</term>
<term>Cervical screening states</term>
<term>Characteristic curve</term>
<term>Chemotherapy</term>
<term>Child characteristics</term>
<term>Childhood asthma</term>
<term>Chlamydia</term>
<term>Chlamydia prevalence</term>
<term>Chlamydia screening</term>
<term>Chlamydia screening studies</term>
<term>Chlamydial infection</term>
<term>Choice experiments</term>
<term>Chronic disease management</term>
<term>Chronic diseases</term>
<term>Chronic illness</term>
<term>Chronic kidney disease</term>
<term>Clinical decisions</term>
<term>Clinical guidelines</term>
<term>Clinical outcome</term>
<term>Clinical outcomes</term>
<term>Clinical practice</term>
<term>Clinical prediction rule</term>
<term>Clinical stage</term>
<term>Clinical studies</term>
<term>Clinical trial data</term>
<term>Clinical trials</term>
<term>Clinical validity</term>
<term>Clinician</term>
<term>Coefficient</term>
<term>Cognitive</term>
<term>Cognitive process model</term>
<term>Cohort</term>
<term>Cohort model</term>
<term>Colorectal cancer</term>
<term>Common comparator</term>
<term>Common diseases</term>
<term>Competitive athletes</term>
<term>Complex task</term>
<term>Confidence intervals</term>
<term>Confidentiality</term>
<term>Confounder selection</term>
<term>Congestive heart failure</term>
<term>Consultation time</term>
<term>Content analysis</term>
<term>Context independence</term>
<term>Control group</term>
<term>Convergent validity</term>
<term>Coronary artery disease</term>
<term>Cost effectiveness</term>
<term>Cost estimates</term>
<term>Costeffectiveness analysis</term>
<term>Critical event</term>
<term>Cue</term>
<term>Current practice</term>
<term>Data collection</term>
<term>Database</term>
<term>Decision analysis</term>
<term>Decision makers</term>
<term>Decision paths</term>
<term>Decision process</term>
<term>Decision rules</term>
<term>Decision strategies</term>
<term>Decision support system</term>
<term>Decision tree</term>
<term>Decision trees</term>
<term>Decisionmaking</term>
<term>Denmark purpose</term>
<term>Depressed patients</term>
<term>Depressive patients</term>
<term>Diabetes</term>
<term>Diabetes management</term>
<term>Diagnostic accuracy</term>
<term>Diagnostic costs</term>
<term>Diagnostic decision</term>
<term>Diagnostic odds ratio</term>
<term>Diagnostic performance</term>
<term>Different approaches</term>
<term>Different causes</term>
<term>Different interview formats</term>
<term>Different results</term>
<term>Different risk formats</term>
<term>Different screening strategies</term>
<term>Different strategies</term>
<term>Different techniques</term>
<term>Differential diagnoses</term>
<term>Discrete event simulation</term>
<term>Discriminative ability</term>
<term>Disease management programs</term>
<term>Disease prognosis</term>
<term>Disease progression</term>
<term>Drug users</term>
<term>Dutch tariff</term>
<term>Dynamic model</term>
<term>East lansing</term>
<term>Ecological criterion</term>
<term>Economic appraisal</term>
<term>Economic appraisals</term>
<term>Economic data</term>
<term>Economic evaluation</term>
<term>Economic evaluations</term>
<term>Economics unit</term>
<term>Ectopic pregnancy</term>
<term>Edmunds</term>
<term>Effect size</term>
<term>Electronic databases</term>
<term>Elicitation</term>
<term>Emergency department</term>
<term>Emergency room visits</term>
<term>Endometrial cancer</term>
<term>Endoscopy</term>
<term>Endovascular</term>
<term>Endovascular repair</term>
<term>Eq5d</term>
<term>Eq5d domains</term>
<term>Eq5d tariffs</term>
<term>Erasmus</term>
<term>Eskd</term>
<term>Eskd patients</term>
<term>Ethical decision</term>
<term>Ethical principles</term>
<term>Everyday life</term>
<term>Exfoliation glaucoma</term>
<term>Expert panel</term>
<term>Expert policy</term>
<term>Extensive decision</term>
<term>Family history</term>
<term>Final model</term>
<term>First questionnaire</term>
<term>Focus group interviews</term>
<term>Framing effects</term>
<term>Framing treatment benefits</term>
<term>France purpose</term>
<term>Freiburg</term>
<term>Full information</term>
<term>Further information</term>
<term>Further research</term>
<term>General population</term>
<term>General practitioners</term>
<term>Genetic algorithms</term>
<term>Genetic counselling</term>
<term>Genetic testing</term>
<term>Glaucoma</term>
<term>Good location</term>
<term>Groningen</term>
<term>Group differences</term>
<term>Guideline</term>
<term>Hall austria purpose</term>
<term>Health care</term>
<term>Health care expenditures</term>
<term>Health economics</term>
<term>Health economics facility</term>
<term>Health outcomes</term>
<term>Health professionals</term>
<term>Health protection agency</term>
<term>Health risk information</term>
<term>Health sciences</term>
<term>Health sector</term>
<term>Health services</term>
<term>Health state</term>
<term>Health state descriptions</term>
<term>Health state valuation techniques</term>
<term>Health state valuations</term>
<term>Health states</term>
<term>Health technology assessment</term>
<term>Herd immunity</term>
<term>Heterogeneity</term>
<term>Home haemodialysis</term>
<term>Hormone therapy</term>
<term>Hospitalisation</term>
<term>Hrhpv</term>
<term>Hrhpv knowledge</term>
<term>Illicit drug</term>
<term>Illness management</term>
<term>Immunisation</term>
<term>Important role</term>
<term>Inadequate treatment</term>
<term>Incident data</term>
<term>Inclusion criteria</term>
<term>Incremental</term>
<term>Incremental cost</term>
<term>Incremental costs</term>
<term>Incremental ratio</term>
<term>Indirect costs</term>
<term>Indirect effects</term>
<term>Individualized quality</term>
<term>Indolent cancer</term>
<term>Influence diagram</term>
<term>Influenza</term>
<term>Influenza vaccination</term>
<term>Information requirements</term>
<term>Information search</term>
<term>Information visualisation</term>
<term>Insurance premiums</term>
<term>International health</term>
<term>International survey</term>
<term>Internet panel</term>
<term>Intervention group</term>
<term>Iqwig</term>
<term>James university hospital</term>
<term>Judgement</term>
<term>Judgement performance</term>
<term>Kidney check australia taskforce</term>
<term>Kidney disease</term>
<term>Kidney health australia</term>
<term>Kidney transplant</term>
<term>Kingdom purpose</term>
<term>Knowledge errors</term>
<term>Language therapists</term>
<term>Language versions</term>
<term>Large number</term>
<term>Large rearrangements</term>
<term>Latanoprost</term>
<term>Leiden</term>
<term>Leiden university</term>
<term>Life extension</term>
<term>Life utilities</term>
<term>Life weights</term>
<term>Life year</term>
<term>Life years</term>
<term>Little evidence</term>
<term>Local search methods</term>
<term>Logistic regression</term>
<term>Lower levels</term>
<term>Lung cancer</term>
<term>Maastricht university</term>
<term>Managerial decisions</term>
<term>Manchester triage system</term>
<term>Markov</term>
<term>Markov approach</term>
<term>Markov model</term>
<term>Markov models</term>
<term>Mathematical model</term>
<term>Maximum number</term>
<term>Medical decision</term>
<term>Medical decisions</term>
<term>Medical informatics</term>
<term>Medical literature</term>
<term>Medical technology assessment</term>
<term>Medication</term>
<term>Mental health professionals</term>
<term>Methodology</term>
<term>Model results</term>
<term>Model stewart stewart passmore</term>
<term>Modeling</term>
<term>Modeling uncertainty</term>
<term>Modelling</term>
<term>Monte carlo sampling</term>
<term>Morbidity</term>
<term>More accuracy</term>
<term>Mortality rates</term>
<term>Multiple objectives</term>
<term>Multivariable model</term>
<term>Mutation</term>
<term>National patient safety agency</term>
<term>Netherlands</term>
<term>Netherlands purpose</term>
<term>Norway</term>
<term>Norway purpose</term>
<term>Norwegian knowledge centre</term>
<term>Norwich university hospital</term>
<term>Novice</term>
<term>Nurse performance</term>
<term>Obese patients</term>
<term>Occupational health</term>
<term>Older patients</term>
<term>Older people</term>
<term>Oncologist</term>
<term>Open angle glaucoma</term>
<term>Open repair</term>
<term>Open surgery</term>
<term>Opportunistic</term>
<term>Opportunistic chlamydia screening</term>
<term>Opportunistic screening</term>
<term>Option</term>
<term>Organisational benefits</term>
<term>Other health professionals</term>
<term>Other measures</term>
<term>Outcome measures</term>
<term>Ovarian</term>
<term>Ovarian cancer</term>
<term>Ovarian cancer patients</term>
<term>Overall survival</term>
<term>Paediatric emergency care</term>
<term>Pandemic</term>
<term>Parameter values</term>
<term>Partner notification</term>
<term>Passive immunisation</term>
<term>Patient consultations</term>
<term>Patient involvement</term>
<term>Patient outcomes</term>
<term>Patient participation</term>
<term>Patient preferences</term>
<term>Patient safety</term>
<term>Payer perspective</term>
<term>Performance instructions</term>
<term>Perinatal period</term>
<term>Peritoneal dialysis</term>
<term>Person trade</term>
<term>Personal risk</term>
<term>Pharmaceutical product</term>
<term>Pharmaceutical research network</term>
<term>Pictorial representation</term>
<term>Plot asymmetry</term>
<term>Pneumococcal</term>
<term>Pneumococcal conjugate vaccine</term>
<term>Pneumococcal disease</term>
<term>Pneumonia</term>
<term>Policy makers</term>
<term>Population screening</term>
<term>Positive aspects</term>
<term>Positive family history</term>
<term>Positive prediction value</term>
<term>Possible reasons</term>
<term>Practitioner</term>
<term>Preference</term>
<term>Preferred information format</term>
<term>Pregnant women</term>
<term>Preliminary analyses</term>
<term>Preliminary analysis</term>
<term>Preliminary results</term>
<term>Prenatal</term>
<term>Prenatal screening</term>
<term>Present study</term>
<term>Prevalence</term>
<term>Preventive behaviour</term>
<term>Previous pandemics</term>
<term>Primary care</term>
<term>Prince alfred hospital</term>
<term>Prison reception</term>
<term>Prison vaccination</term>
<term>Process information</term>
<term>Production rules</term>
<term>Programme</term>
<term>Progression</term>
<term>Prolabels database</term>
<term>Prostaglandin analogs</term>
<term>Prostate</term>
<term>Prostate cancer</term>
<term>Proxy outcome measures</term>
<term>Psychological distress</term>
<term>Psychometric properties</term>
<term>Psychometric validation</term>
<term>Public health</term>
<term>Public health interventions</term>
<term>Publication bias</term>
<term>Qaly</term>
<term>Qalys</term>
<term>Radboud university nijmegen</term>
<term>Radiation dose</term>
<term>Radiographic pneumonia</term>
<term>Random sample</term>
<term>Randomised</term>
<term>Rasch analysis</term>
<term>Reasoning biases</term>
<term>Reference case</term>
<term>Regression analysis</term>
<term>Regression model</term>
<term>Relative risk reduction</term>
<term>Relative treatment effects</term>
<term>Reliability</term>
<term>Renal</term>
<term>Renal replacement therapy</term>
<term>Respiratory morbidity</term>
<term>Respiratory tract infections</term>
<term>Respiratory tract symptoms</term>
<term>Results show</term>
<term>Rheumatoid arthritis</term>
<term>Risk assessment</term>
<term>Risk factors</term>
<term>Risk information</term>
<term>Risk perception</term>
<term>Risk status</term>
<term>Rotavirus</term>
<term>Rotterdam</term>
<term>Routine vaccination</term>
<term>Same comparison</term>
<term>Scenario</term>
<term>Screening</term>
<term>Screening program</term>
<term>Screening strategies</term>
<term>Screening test</term>
<term>Seasonal month</term>
<term>Sensitivity analyses</term>
<term>Sensitivity analysis</term>
<term>Sequela</term>
<term>Severe side effects</term>
<term>Severe uncertainty</term>
<term>Sheffield</term>
<term>Side effects</term>
<term>Significant correlations</term>
<term>Significant impact</term>
<term>Significant reductions</term>
<term>Significant variation</term>
<term>Similar analyses</term>
<term>Simulation</term>
<term>Situation awareness</term>
<term>Smaller studies</term>
<term>Social decision</term>
<term>Social medicine</term>
<term>Societal perspective</term>
<term>Socioeconomic status</term>
<term>Southern denmark</term>
<term>Stage kidney disease</term>
<term>Standard approach</term>
<term>Standard gamble</term>
<term>Statistical modelling</term>
<term>Structural equation modelling</term>
<term>Subjective norm</term>
<term>Substitute preferences</term>
<term>Support decision</term>
<term>Support patients</term>
<term>Survival studies</term>
<term>Survival trials</term>
<term>Switzerland purpose</term>
<term>Symptom</term>
<term>Syndrome</term>
<term>Systematic review</term>
<term>Systematic reviews</term>
<term>Test characteristics</term>
<term>Test evaluation program</term>
<term>Test recommendation</term>
<term>Thematic content analysis</term>
<term>Time horizon</term>
<term>Time points</term>
<term>Time trade</term>
<term>Timolol maleate</term>
<term>Total elimination</term>
<term>Trading decision</term>
<term>Transition probabilities</term>
<term>Transmission dynamics</term>
<term>Transplant</term>
<term>Transplant registry</term>
<term>Treatment adherence</term>
<term>Treatment choices</term>
<term>Treatment decision</term>
<term>Treatment decisionmaking</term>
<term>Treatment effect</term>
<term>Treatment effects</term>
<term>Treatment options</term>
<term>Treatment questionnaire</term>
<term>Trial conditions</term>
<term>Trial participation</term>
<term>Unidimensional scale</term>
<term>Unit costs</term>
<term>University hospital</term>
<term>Unrealistic optimism</term>
<term>Unrealistic pessimism</term>
<term>Urinary symptoms</term>
<term>Useful tool</term>
<term>Usual activities</term>
<term>Utility estimates</term>
<term>Utility scores</term>
<term>Utility values</term>
<term>Utility weights</term>
<term>Vaccination</term>
<term>Vaccine</term>
<term>Validation</term>
<term>Variance</term>
<term>Verbal screening</term>
<term>Visual acuity</term>
<term>Visual analogue scale</term>
<term>Website</term>
<term>Weight change</term>
<term>Weighting</term>
<term>Wider economy</term>
<term>Young people</term>
<term>Zealand dialysis</term>
</keywords>
</textClass>
<langUsage>
<language ident="en">en</language>
</langUsage>
</profileDesc>
</teiHeader>
</TEI>
<istex>
<corpusName>sage</corpusName>
<keywords>
<teeft>
<json:string>kingdom purpose</json:string>
<json:string>modelling</json:string>
<json:string>medical decision</json:string>
<json:string>health states</json:string>
<json:string>chlamydia</json:string>
<json:string>rotterdam</json:string>
<json:string>netherlands purpose</json:string>
<json:string>markov</json:string>
<json:string>publication bias</json:string>
<json:string>public health</json:string>
<json:string>guideline</json:string>
<json:string>qalys</json:string>
<json:string>2university</json:string>
<json:string>prenatal</json:string>
<json:string>anesthetist</json:string>
<json:string>cdss</json:string>
<json:string>asthma</json:string>
<json:string>clinical trials</json:string>
<json:string>cohort</json:string>
<json:string>programme</json:string>
<json:string>freiburg</json:string>
<json:string>hospitalisation</json:string>
<json:string>latanoprost</json:string>
<json:string>life years</json:string>
<json:string>sheffield</json:string>
<json:string>patient participation</json:string>
<json:string>database</json:string>
<json:string>health protection agency</json:string>
<json:string>clinical outcome</json:string>
<json:string>heterogeneity</json:string>
<json:string>website</json:string>
<json:string>vaccine</json:string>
<json:string>patient preferences</json:string>
<json:string>hrhpv</json:string>
<json:string>immunisation</json:string>
<json:string>rotavirus</json:string>
<json:string>morbidity</json:string>
<json:string>ovarian</json:string>
<json:string>groningen</json:string>
<json:string>2department</json:string>
<json:string>clinician</json:string>
<json:string>screening</json:string>
<json:string>endoscopy</json:string>
<json:string>elicitation</json:string>
<json:string>prostate cancer</json:string>
<json:string>bayesian</json:string>
<json:string>edmunds</json:string>
<json:string>oncologist</json:string>
<json:string>pneumococcal</json:string>
<json:string>influence diagram</json:string>
<json:string>risk factors</json:string>
<json:string>treatment options</json:string>
<json:string>eq5d</json:string>
<json:string>decisionmaking</json:string>
<json:string>netherlands</json:string>
<json:string>side effects</json:string>
<json:string>mutation</json:string>
<json:string>chronic diseases</json:string>
<json:string>qaly</json:string>
<json:string>young people</json:string>
<json:string>randomised</json:string>
<json:string>time trade</json:string>
<json:string>transition probabilities</json:string>
<json:string>cue</json:string>
<json:string>eskd</json:string>
<json:string>prenatal screening</json:string>
<json:string>iqwig</json:string>
<json:string>primary care</json:string>
<json:string>present study</json:string>
<json:string>plot asymmetry</json:string>
<json:string>sequela</json:string>
<json:string>endovascular</json:string>
<json:string>chemotherapy</json:string>
<json:string>general practitioners</json:string>
<json:string>patient involvement</json:string>
<json:string>vaccination</json:string>
<json:string>glaucoma</json:string>
<json:string>leiden</json:string>
<json:string>scenario</json:string>
<json:string>incremental</json:string>
<json:string>validation</json:string>
<json:string>renal</json:string>
<json:string>radiation dose</json:string>
<json:string>visual analogue scale</json:string>
<json:string>screening program</json:string>
<json:string>general population</json:string>
<json:string>passive immunisation</json:string>
<json:string>unrealistic optimism</json:string>
<json:string>cost effectiveness</json:string>
<json:string>herd immunity</json:string>
<json:string>risk information</json:string>
<json:string>judgement</json:string>
<json:string>erasmus</json:string>
<json:string>baseline</json:string>
<json:string>different techniques</json:string>
<json:string>endovascular repair</json:string>
<json:string>breast cancer</json:string>
<json:string>outcome measures</json:string>
<json:string>decision process</json:string>
<json:string>ovarian cancer</json:string>
<json:string>health sciences</json:string>
<json:string>leiden university</json:string>
<json:string>patient outcomes</json:string>
<json:string>university hospital</json:string>
<json:string>abstracts methods</json:string>
<json:string>health technology assessment</json:string>
<json:string>health risk information</json:string>
<json:string>different causes</json:string>
<json:string>australia purpose</json:string>
<json:string>health economics</json:string>
<json:string>chlamydia screening</json:string>
<json:string>health care</json:string>
<json:string>sensitivity analyses</json:string>
<json:string>person trade</json:string>
<json:string>variance</json:string>
<json:string>cardiovascular</json:string>
<json:string>modeling</json:string>
<json:string>confidentiality</json:string>
<json:string>asymmetry</json:string>
<json:string>centre</json:string>
<json:string>pregnant women</json:string>
<json:string>medical informatics</json:string>
<json:string>diagnostic odds ratio</json:string>
<json:string>systematic review</json:string>
<json:string>economic evaluations</json:string>
<json:string>cognitive process model</json:string>
<json:string>test recommendation</json:string>
<json:string>economics unit</json:string>
<json:string>emergency room visits</json:string>
<json:string>urinary symptoms</json:string>
<json:string>incremental ratio</json:string>
<json:string>peritoneal dialysis</json:string>
<json:string>regression analysis</json:string>
<json:string>health professionals</json:string>
<json:string>case study</json:string>
<json:string>decision paths</json:string>
<json:string>large number</json:string>
<json:string>trial participation</json:string>
<json:string>markov model</json:string>
<json:string>current practice</json:string>
<json:string>genetic counselling</json:string>
<json:string>control group</json:string>
<json:string>depressed patients</json:string>
<json:string>common comparator</json:string>
<json:string>indirect costs</json:string>
<json:string>weight change</json:string>
<json:string>disease prognosis</json:string>
<json:string>screening strategies</json:string>
<json:string>positive prediction value</json:string>
<json:string>reliability</json:string>
<json:string>methodology</json:string>
<json:string>simulation</json:string>
<json:string>prostate</json:string>
<json:string>norway</json:string>
<json:string>analogue</json:string>
<json:string>opportunistic</json:string>
<json:string>weighting</json:string>
<json:string>bias term</json:string>
<json:string>economic evaluation</json:string>
<json:string>james university hospital</json:string>
<json:string>cardiovascular disease</json:string>
<json:string>center rotterdam</json:string>
<json:string>survival trials</json:string>
<json:string>population screening</json:string>
<json:string>decision tree</json:string>
<json:string>diagnostic accuracy</json:string>
<json:string>data collection</json:string>
<json:string>information requirements</json:string>
<json:string>maximum number</json:string>
<json:string>test evaluation program</json:string>
<json:string>psychometric properties</json:string>
<json:string>kidney health australia</json:string>
<json:string>health outcomes</json:string>
<json:string>occupational health</json:string>
<json:string>obese patients</json:string>
<json:string>norwegian knowledge centre</json:string>
<json:string>pictorial representation</json:string>
<json:string>health services</json:string>
<json:string>random sample</json:string>
<json:string>clinical decisions</json:string>
<json:string>clinical prediction rule</json:string>
<json:string>respiratory tract symptoms</json:string>
<json:string>discriminative ability</json:string>
<json:string>multivariable model</json:string>
<json:string>abstracts results</json:string>
<json:string>support decision</json:string>
<json:string>genetic testing</json:string>
<json:string>everyday life</json:string>
<json:string>anchor state</json:string>
<json:string>context independence</json:string>
<json:string>cervical screening states</json:string>
<json:string>health state valuations</json:string>
<json:string>timolol maleate</json:string>
<json:string>asthma control</json:string>
<json:string>actual decision</json:string>
<json:string>treatment adherence</json:string>
<json:string>clinical trial data</json:string>
<json:string>utility values</json:string>
<json:string>ectopic pregnancy</json:string>
<json:string>inadequate treatment</json:string>
<json:string>older people</json:string>
<json:string>subjective norm</json:string>
<json:string>clinical practice</json:string>
<json:string>paediatric emergency care</json:string>
<json:string>opportunistic screening</json:string>
<json:string>acute infrarenal</json:string>
<json:string>life weights</json:string>
<json:string>abstracts conclusions</json:string>
<json:string>framing effects</json:string>
<json:string>personal risk</json:string>
<json:string>intervention group</json:string>
<json:string>discrete event simulation</json:string>
<json:string>depressive patients</json:string>
<json:string>open repair</json:string>
<json:string>eq5d tariffs</json:string>
<json:string>health state valuation techniques</json:string>
<json:string>public health interventions</json:string>
<json:string>clinical outcomes</json:string>
<json:string>different results</json:string>
<json:string>diagnostic performance</json:string>
<json:string>positive aspects</json:string>
<json:string>pharmaceutical product</json:string>
<json:string>ethical decision</json:string>
<json:string>time horizon</json:string>
<json:string>decision rules</json:string>
<json:string>switzerland purpose</json:string>
<json:string>further research</json:string>
<json:string>available evidence</json:string>
<json:string>incremental cost</json:string>
<json:string>chronic illness</json:string>
<json:string>indolent cancer</json:string>
<json:string>coronary artery disease</json:string>
<json:string>expert panel</json:string>
<json:string>consultation time</json:string>
<json:string>treatment decision</json:string>
<json:string>congestive heart failure</json:string>
<json:string>patient safety</json:string>
<json:string>prison reception</json:string>
<json:string>prolabels database</json:string>
<json:string>decision trees</json:string>
<json:string>influenza vaccination</json:string>
<json:string>chlamydia prevalence</json:string>
<json:string>model results</json:string>
<json:string>cohort model</json:string>
<json:string>time points</json:string>
<json:string>social medicine</json:string>
<json:string>smaller studies</json:string>
<json:string>markov models</json:string>
<json:string>treatment effect</json:string>
<json:string>open surgery</json:string>
<json:string>substitute preferences</json:string>
<json:string>medication</json:string>
<json:string>transplant</json:string>
<json:string>influenza</json:string>
<json:string>preference</json:string>
<json:string>progression</json:string>
<json:string>practitioner</json:string>
<json:string>syndrome</json:string>
<json:string>cognitive</json:string>
<json:string>novice</json:string>
<json:string>amsterdam</json:string>
<json:string>biennial european meeting</json:string>
<json:string>cardiovascular events</json:string>
<json:string>risk perception</json:string>
<json:string>indirect effects</json:string>
<json:string>norwich university hospital</json:string>
<json:string>bayesian approach</json:string>
<json:string>unrealistic pessimism</json:string>
<json:string>national patient safety agency</json:string>
<json:string>preferred information format</json:string>
<json:string>regression model</json:string>
<json:string>effect size</json:string>
<json:string>mathematical model</json:string>
<json:string>significant correlations</json:string>
<json:string>unit costs</json:string>
<json:string>routine vaccination</json:string>
<json:string>significant reductions</json:string>
<json:string>cancer screening</json:string>
<json:string>better asthma control</json:string>
<json:string>markov approach</json:string>
<json:string>psychological distress</json:string>
<json:string>focus group interviews</json:string>
<json:string>survival studies</json:string>
<json:string>monte carlo sampling</json:string>
<json:string>results show</json:string>
<json:string>decision analysis</json:string>
<json:string>different strategies</json:string>
<json:string>diagnostic costs</json:string>
<json:string>different risk formats</json:string>
<json:string>maastricht university</json:string>
<json:string>health state</json:string>
<json:string>autonomous decision</json:string>
<json:string>parameter values</json:string>
<json:string>renal replacement therapy</json:string>
<json:string>stage kidney disease</json:string>
<json:string>partner notification</json:string>
<json:string>zealand dialysis</json:string>
<json:string>screening test</json:string>
<json:string>transplant registry</json:string>
<json:string>international health</json:string>
<json:string>kidney check australia taskforce</json:string>
<json:string>policy makers</json:string>
<json:string>prince alfred hospital</json:string>
<json:string>positive family history</json:string>
<json:string>illness management</json:string>
<json:string>eskd patients</json:string>
<json:string>severe side effects</json:string>
<json:string>diagnostic decision</json:string>
<json:string>international survey</json:string>
<json:string>home haemodialysis</json:string>
<json:string>nurse performance</json:string>
<json:string>extensive decision</json:string>
<json:string>complex task</json:string>
<json:string>electronic databases</json:string>
<json:string>inclusion criteria</json:string>
<json:string>other health professionals</json:string>
<json:string>decision support system</json:string>
<json:string>mental health professionals</json:string>
<json:string>production rules</json:string>
<json:string>radiographic pneumonia</json:string>
<json:string>respiratory tract infections</json:string>
<json:string>aston university</json:string>
<json:string>content analysis</json:string>
<json:string>characteristic curve</json:string>
<json:string>organisational benefits</json:string>
<json:string>little evidence</json:string>
<json:string>ecological criterion</json:string>
<json:string>support patients</json:string>
<json:string>family history</json:string>
<json:string>large rearrangements</json:string>
<json:string>logistic regression</json:string>
<json:string>colorectal cancer</json:string>
<json:string>endometrial cancer</json:string>
<json:string>risk status</json:string>
<json:string>test characteristics</json:string>
<json:string>significant variation</json:string>
<json:string>standard gamble</json:string>
<json:string>risk assessment</json:string>
<json:string>expert policy</json:string>
<json:string>utility scores</json:string>
<json:string>possible reasons</json:string>
<json:string>judgement performance</json:string>
<json:string>common diseases</json:string>
<json:string>health state descriptions</json:string>
<json:string>biosource corporation</json:string>
<json:string>alternative approach</json:string>
<json:string>patient consultations</json:string>
<json:string>multiple objectives</json:string>
<json:string>radboud university nijmegen</json:string>
<json:string>critical event</json:string>
<json:string>social decision</json:string>
<json:string>life utilities</json:string>
<json:string>health sector</json:string>
<json:string>trading decision</json:string>
<json:string>wider economy</json:string>
<json:string>hrhpv knowledge</json:string>
<json:string>knowledge errors</json:string>
<json:string>more accuracy</json:string>
<json:string>societal perspective</json:string>
<json:string>canada purpose</json:string>
<json:string>final model</json:string>
<json:string>preventive behaviour</json:string>
<json:string>process information</json:string>
<json:string>atrial fibrillation</json:string>
<json:string>trial conditions</json:string>
<json:string>systematic reviews</json:string>
<json:string>disease management programs</json:string>
<json:string>relative treatment effects</json:string>
<json:string>same comparison</json:string>
<json:string>situation awareness</json:string>
<json:string>previous pandemics</json:string>
<json:string>other measures</json:string>
<json:string>chronic kidney disease</json:string>
<json:string>hall austria purpose</json:string>
<json:string>chlamydia screening studies</json:string>
<json:string>medical technology assessment</json:string>
<json:string>modeling uncertainty</json:string>
<json:string>framing treatment benefits</json:string>
<json:string>differential diagnoses</json:string>
<json:string>further information</json:string>
<json:string>first questionnaire</json:string>
<json:string>east lansing</json:string>
<json:string>overall survival</json:string>
<json:string>treatment decisionmaking</json:string>
<json:string>blood pressure</json:string>
<json:string>lung cancer</json:string>
<json:string>group differences</json:string>
<json:string>eq5d domains</json:string>
<json:string>norway purpose</json:string>
<json:string>dutch tariff</json:string>
<json:string>usual activities</json:string>
<json:string>internet panel</json:string>
<json:string>utility estimates</json:string>
<json:string>socioeconomic status</json:string>
<json:string>mortality rates</json:string>
<json:string>artery disease</json:string>
<json:string>total elimination</json:string>
<json:string>similar analyses</json:string>
<json:string>convergent validity</json:string>
<json:string>utility weights</json:string>
<json:string>perinatal period</json:string>
<json:string>managerial decisions</json:string>
<json:string>important role</json:string>
<json:string>individualized quality</json:string>
<json:string>rheumatoid arthritis</json:string>
<json:string>decision makers</json:string>
<json:string>annual health care expenditures</json:string>
<json:string>different approaches</json:string>
<json:string>insurance premiums</json:string>
<json:string>dynamic model</json:string>
<json:string>health care expenditures</json:string>
<json:string>medical decisions</json:string>
<json:string>information search</json:string>
<json:string>budget impact analysis</json:string>
<json:string>birth weight</json:string>
<json:string>hormone therapy</json:string>
<json:string>manchester triage system</json:string>
<json:string>childhood asthma</json:string>
<json:string>severe uncertainty</json:string>
<json:string>southern denmark</json:string>
<json:string>open angle glaucoma</json:string>
<json:string>alternative screening strategies</json:string>
<json:string>chronic disease management</json:string>
<json:string>decision strategies</json:string>
<json:string>denmark purpose</json:string>
<json:string>diabetes management</json:string>
<json:string>emergency department</json:string>
<json:string>clinical stage</json:string>
<json:string>ovarian cancer patients</json:string>
<json:string>clinical guidelines</json:string>
<json:string>useful tool</json:string>
<json:string>rasch analysis</json:string>
<json:string>unidimensional scale</json:string>
<json:string>barthel index</json:string>
<json:string>proxy outcome measures</json:string>
<json:string>health economics facility</json:string>
<json:string>treatment choices</json:string>
<json:string>kidney transplant</json:string>
<json:string>kidney disease</json:string>
<json:string>better decisions</json:string>
<json:string>incident data</json:string>
<json:string>information visualisation</json:string>
<json:string>statistical modelling</json:string>
<json:string>economic appraisal</json:string>
<json:string>standard approach</json:string>
<json:string>economic appraisals</json:string>
<json:string>confidence intervals</json:string>
<json:string>economic data</json:string>
<json:string>bootstrap approach</json:string>
<json:string>local search methods</json:string>
<json:string>costeffectiveness analysis</json:string>
<json:string>confounder selection</json:string>
<json:string>genetic algorithms</json:string>
<json:string>preliminary analyses</json:string>
<json:string>relative risk reduction</json:string>
<json:string>life extension</json:string>
<json:string>structural equation modelling</json:string>
<json:string>chlamydial infection</json:string>
<json:string>pneumococcal conjugate vaccine</json:string>
<json:string>thematic content analysis</json:string>
<json:string>ethical principles</json:string>
<json:string>reasoning biases</json:string>
<json:string>preliminary analysis</json:string>
<json:string>abstracts purpose</json:string>
<json:string>language therapists</json:string>
<json:string>appropriate diagnoses</json:string>
<json:string>antihypertensive medication</json:string>
<json:string>older patients</json:string>
<json:string>treatment effects</json:string>
<json:string>different interview formats</json:string>
<json:string>lower levels</json:string>
<json:string>model stewart stewart passmore</json:string>
<json:string>pharmaceutical research network</json:string>
<json:string>exfoliation glaucoma</json:string>
<json:string>medical literature</json:string>
<json:string>visual acuity</json:string>
<json:string>payer perspective</json:string>
<json:string>incremental costs</json:string>
<json:string>prison vaccination</json:string>
<json:string>respiratory morbidity</json:string>
<json:string>reference case</json:string>
<json:string>child characteristics</json:string>
<json:string>seasonal month</json:string>
<json:string>prostaglandin analogs</json:string>
<json:string>life year</json:string>
<json:string>opportunistic chlamydia screening</json:string>
<json:string>different screening strategies</json:string>
<json:string>disease progression</json:string>
<json:string>cost estimates</json:string>
<json:string>verbal screening</json:string>
<json:string>illicit drug</json:string>
<json:string>competitive athletes</json:string>
<json:string>2imperial college</json:string>
<json:string>preliminary results</json:string>
<json:string>pneumococcal disease</json:string>
<json:string>france purpose</json:string>
<json:string>treatment questionnaire</json:string>
<json:string>language versions</json:string>
<json:string>additional language versions</json:string>
<json:string>drug users</json:string>
<json:string>psychometric validation</json:string>
<json:string>good location</json:string>
<json:string>clinical validity</json:string>
<json:string>transmission dynamics</json:string>
<json:string>clinical studies</json:string>
<json:string>asthma quality</json:string>
<json:string>aqlq scores</json:string>
<json:string>sensitivity analysis</json:string>
<json:string>performance instructions</json:string>
<json:string>choice experiments</json:string>
<json:string>full information</json:string>
<json:string>significant impact</json:string>
<json:string>carlo</json:string>
<json:string>diabetes</json:string>
<json:string>bias</json:string>
<json:string>symptom</json:string>
<json:string>coefficient</json:string>
<json:string>option</json:string>
<json:string>pandemic</json:string>
<json:string>pneumonia</json:string>
<json:string>prevalence</json:string>
<json:string>cancer</json:string>
</teeft>
</keywords>
<articleId>
<json:string>10.1177_0272989X080270040104</json:string>
</articleId>
<arkIstex>ark:/67375/M70-DPPD2M40-0</arkIstex>
<language>
<json:string>eng</json:string>
</language>
<originalGenre>
<json:string>other</json:string>
</originalGenre>
<qualityIndicators>
<score>7.012</score>
<pdfWordCount>28632</pdfWordCount>
<pdfCharCount>188264</pdfCharCount>
<pdfVersion>1.3</pdfVersion>
<pdfPageCount>31</pdfPageCount>
<pdfPageSize>594 x 783 pts</pdfPageSize>
<refBibsNative>false</refBibsNative>
<abstractWordCount>1</abstractWordCount>
<abstractCharCount>0</abstractCharCount>
<keywordCount>0</keywordCount>
</qualityIndicators>
<title>2006 EUROPEAN SOCIETY FOR MEDICAL DECISION MAKING MEETING</title>
<genre>
<json:string>other</json:string>
</genre>
<host>
<title>Medical Decision Making</title>
<language>
<json:string>unknown</json:string>
</language>
<issn>
<json:string>0272-989X</json:string>
</issn>
<eissn>
<json:string>1552-681X</json:string>
</eissn>
<publisherId>
<json:string>MDM</json:string>
</publisherId>
<volume>27</volume>
<issue>4</issue>
<pages>
<first>e-3</first>
<last>e-4</last>
</pages>
<genre>
<json:string>journal</json:string>
</genre>
</host>
<namedEntities>
<unitex>
<date>
<json:string>2005</json:string>
<json:string>1957</json:string>
<json:string>2001</json:string>
<json:string>2017</json:string>
<json:string>2006</json:string>
<json:string>1000</json:string>
<json:string>1998</json:string>
<json:string>1016</json:string>
<json:string>between January 1998 and November 2005</json:string>
<json:string>the 20th century</json:string>
<json:string>2002</json:string>
<json:string>2007</json:string>
<json:string>2003</json:string>
<json:string>1589</json:string>
<json:string>1995</json:string>
<json:string>from August 2003 to November 2004</json:string>
<json:string>between January 1990 and December 2005</json:string>
<json:string>2015</json:string>
<json:string>2004</json:string>
</date>
<geogName></geogName>
<orgName>
<json:string>Haga Hospital, Juliana Children</json:string>
<json:string>Technion–Israel Institute of Technology, Haifa, Israel</json:string>
<json:string>University Hospital of Freiburg, Freiburg, Germany</json:string>
<json:string>GERMAN INSTITUTE FOR QUALITY AND EFFICIENCY IN HEALTH CARE</json:string>
<json:string>Switzerland Purpose</json:string>
<json:string>Norway, Sweden, Denmark, and the United Kingdom</json:string>
<json:string>University of Leeds</json:string>
<json:string>Department of Industrial Engineering and Management</json:string>
<json:string>Institute for Public Health, Medical Decision Making and Health</json:string>
<json:string>Chile, Santiago, Chile</json:string>
<json:string>Women’s Hospital, Boston, MA</json:string>
<json:string>Department of Health Systems Management, Ben-Gurion University of the Negev, Beer-Sheva, Israel</json:string>
<json:string>University of Basel</json:string>
<json:string>Rikshospitalet University Hospital, Oslo</json:string>
<json:string>University of Groningen</json:string>
<json:string>UK</json:string>
<json:string>C Department of Health Sciences, University of York, York, United Kingdom ABSTRACTS E</json:string>
<json:string>Department of Public Health, Rotterdam, the Netherlands</json:string>
<json:string>T National Patient</json:string>
<json:string>UK Jit</json:string>
<json:string>Cardiff University</json:string>
<json:string>Health Protection Agency, London, United Kingdom</json:string>
<json:string>Mueller E Analytica International</json:string>
<json:string>UMCG</json:string>
<json:string>University Medical Center Groningen</json:string>
<json:string>University of Oklahoma</json:string>
<json:string>Max Planck Institute of Human Development, Berlin, Germany</json:string>
<json:string>United Kingdom MODELLING THE SPREAD OF PANDEMIC INFLUENZA AND OPTIONS FOR CONTROL Edmunds WJ, Vynnycky E, Cooper BS, Pitman RJ, and Gay NJ Purpose</json:string>
<json:string>Division of Hepatology</json:string>
<json:string>Medical Center, Rotterdam</json:string>
<json:string>Sophia Children’s Hospital, Rotterdam, the Netherlands</json:string>
<json:string>Department of Nursing</json:string>
<json:string>Department of Obstetrics and Gynecology</json:string>
<json:string>Department of Ambulatory Care and Prevention</json:string>
<json:string>United Kingdom ABSTRACTS Purpose</json:string>
<json:string>Medical Center</json:string>
<json:string>Bonsel GJ Academic Medical Center</json:string>
<json:string>Department of Radiation Oncology, Radboud University Nijmegen Medical Center, Nijmegen, the Netherlands</json:string>
<json:string>James University Hospital, Leeds, United Kingdom Purpose</json:string>
<json:string>UK Electoral Register</json:string>
<json:string>Department of Public Health, Medical Decision Making and Health</json:string>
<json:string>Division of Applied Cachexia Research</json:string>
<json:string>Intervention Research, Medical College of Wisconsin, Milwaukee, WI</json:string>
<json:string>Institute of Health Management and Health</json:string>
<json:string>Norway, Sweden, and Denmark</json:string>
<json:string>Haukeland University Hospital, Bergen</json:string>
<json:string>Department of Medical Decision Making</json:string>
<json:string>Kidney Check Australia Taskforce, Kidney Health Australia, Sydney, Australia Purpose</json:string>
<json:string>Department of Medical Technology Assessment, Radboud University Nijmegen Medical Center, Nijmegen, the Netherlands</json:string>
<json:string>Institute for Research</json:string>
<json:string>University of Hertfordshire</json:string>
<json:string>University of Health Sciences</json:string>
<json:string>JB Institute of Public Health, University of Southern Denmark, Odense, Denmark Purpose</json:string>
<json:string>University of Freiburg</json:string>
<json:string>University of Stirling</json:string>
<json:string>EW Department of Public Health, Erasmus</json:string>
<json:string>Department of Obstetrics and Gynaecology</json:string>
<json:string>Methodist Hospital Research Institute and University of Houston, Houston, TX Purpose</json:string>
<json:string>EMGO-Institute, VU Medical Centre, Amsterdam, the Netherlands</json:string>
<json:string>IN NORWAY, SWEDEN, AND DENMARK USING A DECISION</json:string>
<json:string>Netherlands Background</json:string>
<json:string>Dana-Farber Cancer Institute, Boston, MA</json:string>
<json:string>University of Queensland</json:string>
<json:string>University of Berne</json:string>
<json:string>UNITED KINGDOM USING A DECISIONANALYTIC HEALTH ECONOMIC MODEL Stewart WC</json:string>
<json:string>University of Exeter</json:string>
<json:string>Germany and the USA</json:string>
<json:string>UK, Dutch and US</json:string>
<json:string>Department of Health Care Studies/Medical Sociology, Maastricht University, Maastricht, the Netherlands</json:string>
<json:string>Genetics Research Group, King</json:string>
<json:string>Department of Medical Informatics, Rotterdam, the Netherlands Purpose</json:string>
<json:string>Division of Gastroenterology</json:string>
<json:string>University of Oslo</json:string>
<json:string>Department of Human Communication Sciences, University of Sheffield, Sheffield, United Kingdom</json:string>
<json:string>University of Calgary</json:string>
<json:string>Department of Public Health, Center for Medical Decision Making, Rotterdam, the Netherlands</json:string>
<json:string>Mapi Research Institute, Lyon, France</json:string>
<json:string>Institute for Technology Assessment and Department</json:string>
<json:string>Charite University</json:string>
<json:string>Switzerland Methods</json:string>
<json:string>Norwegian Institute of Public Health, Oslo</json:string>
<json:string>Birmingham University</json:string>
<json:string>FM Department of Clinical Pharmacology, University of Groningen, Groningen, the Netherlands Purpose</json:string>
<json:string>Netherlands Purpose</json:string>
<json:string>Imperial University</json:string>
<json:string>George Institute for International Health, Sydney, Australia</json:string>
<json:string>Flinders University</json:string>
<json:string>United Kingdom Purpose</json:string>
<json:string>University of Southampton</json:string>
<json:string>Department of Social Medicine, University of Bristol, Bristol, United Kingdom Purpose</json:string>
<json:string>University of Warwick</json:string>
<json:string>Michigan State University</json:string>
<json:string>University of Colorado</json:string>
<json:string>Department of Health Sciences, University of York, York, United Kingdom</json:string>
<json:string>US English, and Dutch</json:string>
<json:string>Australia Purpose</json:string>
<json:string>Belgium Purpose</json:string>
<json:string>University of Applied Sciences Northwestern Switzerland, Department of Applied Psychology, Olten, Switzerland Purpose</json:string>
<json:string>IQWIG</json:string>
<json:string>Pfizer Inc.</json:string>
<json:string>Department of Radiology, Leiden University Medical Center, Leiden, the Netherlands This</json:string>
<json:string>University of Glasgow</json:string>
<json:string>Health Sciences Center, Denver</json:string>
<json:string>Aston University</json:string>
<json:string>University of Southampton, United Kingdom</json:string>
<json:string>Health Protection Agency, London</json:string>
<json:string>United BioSource Corporation, London, United Kingdom</json:string>
<json:string>Norwich University Hospital, Norwich, United Kingdom Purpose</json:string>
<json:string>UK NHS</json:string>
<json:string>Department of Medical Decision Making, LUMC, Leiden</json:string>
<json:string>Germany, Switzerland</json:string>
<json:string>University of Munich</json:string>
<json:string>Erasmus University</json:string>
<json:string>Helse Øst Health Services Research Unit, Akershus University Hospital, Oslo, Norway Purpose</json:string>
<json:string>University Hospital of Munich, Munich, Germany</json:string>
<json:string>University Medical Center, Groningen</json:string>
<json:string>Department of Clinical Genetics, Amsterdam, the Netherlands Purpose</json:string>
<json:string>Myriad Genetic Laboratories Inc., Salt Lake City</json:string>
<json:string>Environmental and Enteric Diseases Department, Health Protection Agency, London, United Kingdom Purpose</json:string>
<json:string>Leiden University</json:string>
<json:string>Bergemann R Analytica International</json:string>
<json:string>Wessex Institute for Health Research and Development</json:string>
<json:string>University Hospital of Berlin, Berlin, Germany</json:string>
<json:string>L Department of Public and Occupational Health</json:string>
<json:string>Germany Purpose</json:string>
<json:string>Arnhems Radiotherapeutic Institute, Arnhem, the Netherlands Purpose</json:string>
<json:string>Maastricht University</json:string>
<json:string>United Kingdom Objectives</json:string>
<json:string>C Department of Public Health and Epidemiology</json:string>
<json:string>Department of Public Health, Erasmus</json:string>
<json:string>Institute of Social Medicine, Epidemiology and Health</json:string>
<json:string>SPECIFIC PREFERENCEBASED INSTRUMENT</json:string>
<json:string>University of Fribourg, Fribourg</json:string>
<json:string>University of Pittsburgh School of Medicine, Pittsburgh</json:string>
<json:string>Linköping University</json:string>
<json:string>Humboldt University</json:string>
<json:string>Medical Center, Leiden</json:string>
<json:string>Pharmaceutical Research Network LLC, Charleston, SC</json:string>
<json:string>Caro Research Institute, Concord, MA</json:string>
<json:string>Department of Public and Occupational Health</json:string>
<json:string>Department of Medical Biometry, Charité University Medical Center, Berlin, Germany</json:string>
<json:string>Sweden and Norway</json:string>
<json:string>Brunel University</json:string>
<json:string>Sweden Purpose</json:string>
<json:string>A Department of Medical Decision Making, Leiden University Medical Center, Leiden, the Netherlands Purpose</json:string>
<json:string>Related Research, Sheffield, University of Sheffield, United Kingdom Purpose</json:string>
<json:string>Norway Purpose</json:string>
<json:string>Pediatrics Department, Soroka University Medical Center, Beer-Sheva, Israel</json:string>
<json:string>Department of Gastroenterology and Hepatology</json:string>
<json:string>University of Sheffield</json:string>
<json:string>Mary’s Hospital, Manchester, United Kingdom</json:string>
<json:string>THE EFFECT OF CLINICAL DECISION SUPPORT SYSTEMS</json:string>
<json:string>Durham University</json:string>
<json:string>Sheffield University</json:string>
<json:string>Norway, Denmark, Sweden</json:string>
<json:string>University of Marburg</json:string>
<json:string>Australia and New Zealand Dialysis and Transplant</json:string>
<json:string>Department of Geography, Tel Aviv University, Tel Aviv, Israel Purpose</json:string>
<json:string>United Kingdom Methods</json:string>
<json:string>Jonge P United BioSource Corporation, London</json:string>
<json:string>Health Protection Agency</json:string>
<json:string>Canada Purpose</json:string>
<json:string>EFFECTIVENESS OF SCREENING YOUNG COMPETITIVE ATHLETES FOR HYPERTROPHIC CARDIOMYOPATHY Brodtkorb TH Center for Medical Technology Assessment</json:string>
<json:string>Decision Sciences and Clinical Systems Modeling</json:string>
<json:string>France Purpose</json:string>
<json:string>Charité University</json:string>
<json:string>Germany Objectives</json:string>
<json:string>University of Bristol</json:string>
<json:string>Department of Medical Decision Making/Surgery, Leiden University Medical Center, Leiden</json:string>
<json:string>Open University</json:string>
<json:string>Health Economics Research Unit, University of Aberdeen, Aberdeen, United Kingdom</json:string>
<json:string>University Medical Center Rotterdam, Rotterdam</json:string>
<json:string>Sophia Children’s Hospital, Department of Paediatrics, Rotterdam, the Netherlands</json:string>
<json:string>Mapi Research Trust, Lyon, France Purpose</json:string>
<json:string>Norway and Sweden</json:string>
<json:string>Department of Psychology, University of Sheffield, Sheffield, United Kingdom E</json:string>
<json:string>Ben-Gurion University</json:string>
<json:string>National Patient</json:string>
<json:string>Health Services Research Unit, University of Aberdeen, Aberdeen, United Kingdom Purpose</json:string>
<json:string>Department of Quantitative Health Sciences, Cleveland Clinic Foundation, Cleveland, OH Purpose</json:string>
<json:string>Geographic Information Systems</json:string>
<json:string>University of Newcastle</json:string>
<json:string>Medical Center, Berlin, Germany</json:string>
<json:string>EC-EURATOM</json:string>
</orgName>
<orgName_funder>
<json:string>EC-EURATOM</json:string>
</orgName_funder>
<orgName_provider></orgName_provider>
<persName>
<json:string>Studies</json:string>
<json:string>Van Asperen</json:string>
<json:string>Our</json:string>
<json:string>Findings</json:string>
<json:string>René Dubos</json:string>
<json:string>Milton Keynes</json:string>
<json:string>Results</json:string>
<json:string>Alfred Hospital</json:string>
<json:string>Van den Hout</json:string>
<json:string>Component</json:string>
<json:string>K. Geleijns</json:string>
<json:string>La Pitié-Salpêtrière</json:string>
<json:string>Kretzschmar</json:string>
<json:string>London School</json:string>
<json:string>The</json:string>
<json:string>Van Hoe</json:string>
<json:string>Van Vugt</json:string>
<json:string>Section</json:string>
<json:string>Analyses</json:string>
<json:string>Safety</json:string>
<json:string>People</json:string>
<json:string>To</json:string>
<json:string>Respondents</json:string>
<json:string>Charles Nicolle</json:string>
<json:string>WALES Sutton</json:string>
<json:string>Models</json:string>
</persName>
<placeName>
<json:string>DELPHI</json:string>
<json:string>Maastricht</json:string>
<json:string>Groningen</json:string>
<json:string>Aberdeen</json:string>
<json:string>Leiden</json:string>
<json:string>Delphi</json:string>
<json:string>Calgary</json:string>
<json:string>Berlin</json:string>
<json:string>Brisbane</json:string>
<json:string>Birmingham</json:string>
<json:string>Paris</json:string>
<json:string>Bristol</json:string>
<json:string>Southampton</json:string>
<json:string>Marburg</json:string>
<json:string>VAS</json:string>
<json:string>Hamm</json:string>
<json:string>Germany</json:string>
<json:string>Key West</json:string>
<json:string>Munich</json:string>
<json:string>NY</json:string>
<json:string>Australia</json:string>
<json:string>Lansing</json:string>
<json:string>US</json:string>
<json:string>Austria</json:string>
<json:string>NICE</json:string>
<json:string>Canada</json:string>
<json:string>Oslo</json:string>
<json:string>Specificity</json:string>
<json:string>Cardiff</json:string>
<json:string>London</json:string>
<json:string>The Hague</json:string>
<json:string>Boston</json:string>
<json:string>Exeter</json:string>
<json:string>Norfolk</json:string>
<json:string>Europe</json:string>
<json:string>Rouen</json:string>
<json:string>Sheffield</json:string>
<json:string>Leeds</json:string>
<json:string>Manchester</json:string>
<json:string>Basel</json:string>
<json:string>Amsterdam</json:string>
<json:string>Glasgow</json:string>
<json:string>Denmark</json:string>
<json:string>Rotterdam</json:string>
<json:string>Coventry</json:string>
<json:string>United Kingdom</json:string>
<json:string>Adelaide</json:string>
<json:string>New Molecular</json:string>
<json:string>York</json:string>
<json:string>Norman</json:string>
<json:string>MA</json:string>
<json:string>Denver</json:string>
<json:string>France</json:string>
<json:string>OK</json:string>
<json:string>Plymouth</json:string>
<json:string>Sweden</json:string>
<json:string>England</json:string>
<json:string>Oxford</json:string>
<json:string>Lyon</json:string>
<json:string>Belgium</json:string>
<json:string>Stirling</json:string>
</placeName>
<ref_url></ref_url>
<ref_bibl>
<json:string>Huber, Wider, and Huber, 1997</json:string>
<json:string>Flesch, 1948</json:string>
<json:string>July 2006</json:string>
<json:string>Roberts et al.</json:string>
</ref_bibl>
<bibl></bibl>
</unitex>
</namedEntities>
<ark>
<json:string>ark:/67375/M70-DPPD2M40-0</json:string>
</ark>
<categories>
<wos>
<json:string>1 - science</json:string>
<json:string>2 - medical informatics</json:string>
<json:string>2 - health care sciences & services</json:string>
</wos>
<scienceMetrix>
<json:string>1 - health sciences</json:string>
<json:string>2 - public health & health services</json:string>
<json:string>3 - health policy & services</json:string>
</scienceMetrix>
<scopus>
<json:string>1 - Health Sciences</json:string>
<json:string>2 - Medicine</json:string>
<json:string>3 - Health Policy</json:string>
</scopus>
</categories>
<publicationDate>2007</publicationDate>
<copyrightDate>2007</copyrightDate>
<doi>
<json:string>10.1177/0272989X080270040104</json:string>
</doi>
<id>525CB0AB84383F0EC9BA6467590F85311F2844B5</id>
<score>1</score>
<fulltext>
<json:item>
<extension>pdf</extension>
<original>true</original>
<mimetype>application/pdf</mimetype>
<uri>https://api.istex.fr/ark:/67375/M70-DPPD2M40-0/fulltext.pdf</uri>
</json:item>
<json:item>
<extension>zip</extension>
<original>false</original>
<mimetype>application/zip</mimetype>
<uri>https://api.istex.fr/ark:/67375/M70-DPPD2M40-0/bundle.zip</uri>
</json:item>
<istex:fulltextTEI uri="https://api.istex.fr/ark:/67375/M70-DPPD2M40-0/fulltext.tei">
<teiHeader>
<fileDesc>
<titleStmt>
<title level="a" type="main" xml:lang="en">2006 EUROPEAN SOCIETY FOR MEDICAL DECISION MAKING MEETING</title>
</titleStmt>
<publicationStmt>
<authority>ISTEX</authority>
<publisher scheme="https://scientific-publisher.data.istex.fr">Sage Publications</publisher>
<pubPlace>Sage CA: Los Angeles, CA</pubPlace>
<availability>
<licence>
<p>sage</p>
</licence>
</availability>
<p scheme="https://loaded-corpus.data.istex.fr/ark:/67375/XBH-0J1N7DQT-B"></p>
<date>2007</date>
</publicationStmt>
<notesStmt>
<note type="other" scheme="https://content-type.data.istex.fr/ark:/67375/XTP-7474895G-0">other</note>
<note type="journal" scheme="https://publication-type.data.istex.fr/ark:/67375/JMC-0GLKJH51-B">journal</note>
</notesStmt>
<sourceDesc>
<biblStruct type="inbook">
<analytic>
<title level="a" type="main" xml:lang="en">2006 EUROPEAN SOCIETY FOR MEDICAL DECISION MAKING MEETING</title>
<idno type="istex">525CB0AB84383F0EC9BA6467590F85311F2844B5</idno>
<idno type="ark">ark:/67375/M70-DPPD2M40-0</idno>
<idno type="DOI">10.1177/0272989X080270040104</idno>
<idno type="article-id">10.1177_0272989X080270040104</idno>
</analytic>
<monogr>
<title level="j">Medical Decision Making</title>
<idno type="pISSN">0272-989X</idno>
<idno type="eISSN">1552-681X</idno>
<idno type="publisher-id">MDM</idno>
<idno type="PublisherID-hwp">spmdm</idno>
<imprint>
<publisher>Sage Publications</publisher>
<pubPlace>Sage CA: Los Angeles, CA</pubPlace>
<date type="published" when="2007-07"></date>
<biblScope unit="volume">27</biblScope>
<biblScope unit="issue">4</biblScope>
<biblScope unit="page" from="-3">e-3</biblScope>
<biblScope unit="page" to="-4">e-4</biblScope>
</imprint>
</monogr>
</biblStruct>
</sourceDesc>
</fileDesc>
<profileDesc>
<creation>
<date>2007</date>
</creation>
<langUsage>
<language ident="en">en</language>
</langUsage>
</profileDesc>
<revisionDesc>
<change when="2007-07">Published</change>
</revisionDesc>
</teiHeader>
</istex:fulltextTEI>
<json:item>
<extension>txt</extension>
<original>false</original>
<mimetype>text/plain</mimetype>
<uri>https://api.istex.fr/ark:/67375/M70-DPPD2M40-0/fulltext.txt</uri>
</json:item>
</fulltext>
<metadata>
<istex:metadataXml wicri:clean="corpus sage not found" wicri:toSee="no header">
<istex:xmlDeclaration>version="1.0" encoding="UTF-8"</istex:xmlDeclaration>
<istex:docType PUBLIC="-//NLM//DTD Journal Publishing DTD v2.3 20070202//EN" URI="journalpublishing.dtd" name="istex:docType"></istex:docType>
<istex:document>
<article article-type="other" dtd-version="2.3" xml:lang="EN">
<front>
<journal-meta>
<journal-id journal-id-type="hwp">spmdm</journal-id>
<journal-id journal-id-type="publisher-id">MDM</journal-id>
<journal-title>Medical Decision Making</journal-title>
<issn pub-type="ppub">0272-989X</issn>
<publisher>
<publisher-name>Sage Publications</publisher-name>
<publisher-loc>Sage CA: Los Angeles, CA</publisher-loc>
</publisher>
</journal-meta>
<article-meta>
<article-id pub-id-type="doi">10.1177/0272989X080270040104</article-id>
<article-id pub-id-type="publisher-id">10.1177_0272989X080270040104</article-id>
<article-categories>
<subj-group subj-group-type="heading">
<subject>Articles</subject>
</subj-group>
</article-categories>
<title-group>
<article-title>2006 EUROPEAN SOCIETY FOR MEDICAL DECISION MAKING MEETING</article-title>
</title-group>
<pub-date pub-type="ppub">
<month>7</month>
<year>2007</year>
</pub-date>
<volume>27</volume>
<issue>4</issue>
<fpage>e-3</fpage>
<lpage>e-4</lpage>
<custom-meta-wrap>
<custom-meta xlink:type="simple">
<meta-name>sagemeta-type</meta-name>
<meta-value>Other</meta-value>
</custom-meta>
<custom-meta xlink:type="simple">
<meta-name>cover-date</meta-name>
<meta-value>July/August 2007</meta-value>
</custom-meta>
<custom-meta xlink:type="simple">
<meta-name>search-text</meta-name>
<meta-value>2006 EUROPEAN SOCIETY FOR MEDICAL DECISION MAKING MEETING, BIRMINGHAM, UK, JUNE 2006 MEDICAL DECISION MAKING/JUL_AUG 2007 E1 PARALLEL I 1 DISCRETE EVENT SIMULATION (DES): A COMPREHENSIVE AND FLEXIBLE TOOL FOR ESTIMATING THE LONG-TERM ECONOMIC AND HEALTH OUTCOMES IN OBESE PATIENTS AT HIGH CARDIOMETABOLIC RISK Caro J,1 Getsios D,1 Möller J,1 Lavaud V,2 and McEwan P3 1Caro Research Institute, Concord, MA; 2Sanofi-aventis, Paris, France; 3Cardiff Research Consortium, Cardiff, United Kingdom Purpose: To develop a comprehensive and flexible model for estimating long-term economic and health impact of managing obese patients at cardiometabolic risk. Methods: DES was used to model comprehensively a range of cardiometabolic disease processes over time. A simulated population of individuals can be created by assigning each patient specific characteristics (e.g. demographics, smoking status, lipid profile, blood pressure, FPG, HbA1c, medical history and medications) drawn from patient-level data collected during randomized control trials (RCT) examining over 6600 obese patients. During the simulation, each patient_s cardiovascular risk was estimated in terms of developing diabetes, diabetes-related microvascular complications, cardiovascular events and mortality. Quality-adjusted life-years, event rates and costs were also calculated. Results: The simulation occupies a 3.1 meg file; 100 replications of 10,000 patients take 13.6 minutes. To test the model, 2 diet regimens were simulated: sustained diet with ongoing benefits versus short-term diet where benefits are lost after 2 years. The effect of diet on risk factors was based on results from analysis of the RCT. A sustained diet increases life expectancy by 1.1 years; 13.2 cardiovascular events are avoided per 100 patients. Validation against source studies showed close concordance. Probabilistic sensitivity analysis and cost effectiveness analysis calculation can easily be produced. Conclusion: DES represents the course of the disease naturally avoiding mutually exclusive branches, states, fixed cycles, memory assumptions and other limitations of Markov models. The technique should be considered when carrying out economic evaluations which are used to inform policy makers particularly in case of complex disease processes. 2 IMPACT OF CHLAMYDIA SCREENING IN ENGLAND: RESULTS FROM A MATHEMATICAL MODEL Turner KME,* Adams EJ, Edmunds WJ, Baster K, Emmett L, and LaMontagne DS* On behalf of the Chlamydia Recall Study Advisory Group, Centre for Infections, Health Protection Agency, London, United Kingdom (*formerly of the Health Protection Agency) Purpose: There is a high prevalence of sexually transmitted chlamydial infection among certain groups in England. Many of those infected are asymptomatic and will not seek treatment, and untreated infection may lead to sequelae. We explore the potential impact of screening on Chlamydia prevalence using a mathematical model. Methods: A dynamic individual-based stochastic model was developed to simulate the effects of different Chlamydia screening strategies in England. Parameter values were obtained directly, or estimated by fitting the model to UK data on Chlamydia prevalence, sexual behaviour and healthcare attendance. Results: Screening young women annually reduced the overall Chlamydia prevalence by 2/3 after 5 years, given 85% annual healthcare attendance and 50% test acceptance. Reductions are seen in targeted age groups, but prevalence also decreases in those unscreened. If partner notification (PN) is maintained at the current level (approx 50%), then the reduction in men is comparable to that in women, without screening men. The main limiting factor to the success of the screening programme was the frequency of attendance at healthcare settings. Conclusions: Chlamydia screening is likely to reduce Chlamydia prevalence, providing that PN activities are maintained and that the emphasis on screening young, newly sexually active women is continued. Screening men and women may increase the speed and magnitude of the reduction in prevalence, but improving PN efficacy would have a similar effect. Further information is needed on the healthcare attendance patterns to determine whether active recall of Chlamydia positives would be more effective than opportunistic re-screening. 3 INCORPORATING DIRECT AND INDIRECT EVIDENCE USING BAYESIAN METHODS: AN APPLIED CASE STUDY IN OVARIAN CANCER Griffin SC,1 Bojke L,2 Main C,3 and Palmer S,4 1Centre for Health Economics, University of York; Centre for Reviews and Dissemination, University of York, York, United Kingdom Purpose: To demonstrate the application of a Bayesian mixed treatment comparison (MTC) model to synthesise data from clinical trials to inform decisions based on all relevant evidence. Methods: The value of an MTC model is demonstrated using a probabilistic decision analytic model developed to assess the cost-effectiveness of second-line chemotherapy in ovarian cancer. Three clinical trials were found that each made a different pairwise comparison of 3 treatments of interest in the overall patient population. As no common comparator existed between the 3 trials, an MTC model was used to assess the combined weight of evidence on survival from all 3 trials simultaneously. This analysis was compared to an alternative approach that combined 2 of the trials to make the same comparison of all 3 treatments using a common comparator, and an informal approach that did not synthesise the available evidence. Results: By including all 3 trials using an MTC model, the credible intervals around estimated overall survival were reduced compared to making the same comparison using only 2 trials and a common comparator. However, the survival estimates from the MTC model result in greater uncertainty around the optimal treatment strategy at a cost-effectiveness threshold of £30,000 per quality-adjusted life-year. Conclusions: MTC models can be used to combine more data than would typically be included in a traditional meta-analysis that relies on a common comparator. They can formally quantify the combined uncertainty from all available evidence, and can be conducted using the same analytical approaches as standard meta-analyses. Abstracts SMDM 10TH BIENNIAL EUROPEAN MEETING PARALLEL ABSTRACTS E2 _ MEDICAL DECISION MAKING/JUL_AUG 2007 4 DYNAMIC MODELLING OF DISEASE PROGNOSIS: RECOVERY OF INAPPROPRIATELY IGNORED CLINICAL INFORMATION Kievit J1 and Jacobi CE2 1Department of Medical Decision Making/Surgery, Leiden University Medical Center, Leiden; 2 Department of Medical Decision Making, Leiden University Medical Center, Leiden, the Netherlands Purpose: Disease prognosis is traditionally estimated at the time of diagnosis and/or first treatment. This approach ignores the fact that during the course of disease, the occurrence or absence of specific events and health states may strongly affect survival. As a result, doctors and patients may base their expectations and make their choices on outdated information. Methods: We developed a dynamic approach to disease prognosis, by combining traditional Kaplan Meyer estimates with the Markov approach that has been used extensively in decision analysis. We tested this on 634 patients who were treated for papillary thyroid cancer over 30 years. Initial estimates of prognosis were based on initial patient and tumour characteristics. Dynamic changes in prognosis were quantified on the basis of the absence or occurrence of cancer recurrence over time. Results: The initial disease-specific 10-year survival was 92% (100% for T1, 74% for T4 cancers). This overall survival increased to over 99% in those patients who remained cancer free for 6 years or more, and decreased to 42% for those who developed a local recurrence within two years. The health state _early local recurrence_ most strongly decreased disease specific survival, followed by _any distant metastasis_ and _regional recurrence_ in order of decreasing harm. Conclusion: The notion of dynamic prognostic modelling, analogous to the Markov approach in decision analysis, turns out to be feasible. It recovers clinically highly relevant prognostic information, which would go unnoticed in traditional approaches to disease prognosis. 5 THE IMPACT OF PATIENT PARTICIPATION ON ADHERENCE AND CLINICAL OUTCOME IN PRIMARY CARE OF DEPRESSION Simon D,1 Loh A,1 Leonhart R,2 Wills CE,3 and Härter M1 1University Hospital of Freiburg, Freiburg, Germany; 2University of Freiburg, Freiburg, Germany; 3Michigan State University, East Lansing, MI Purpose: Patient participation in shared treatment decisionmaking is hypothesized to improve depression treatment adherence and clinical outcomes in depressed patients. The study aim was to evaluate the impact of patient participation on these factors via structural equation modelling. Methods: A survey was realized with 30 general practitioners and 207 depressed patients, at initial consultation and six to eight weeks later. General practitioners documented their clinical practice, and patients completed questionnaires including Brief-PHQ for depression and clinical outcome, patient participation (Man-Son-Hing- Scale), and visual analogue scales for treatment adherence. Results: Correlation analysis showed significant correlations between patient participation and adherence (patient rating r = .36, p .01, physician rating r = .21, p .05) and between patient participation and clinical outcome (r = .26, p .01). Structural equation modelling revealed that sixty percent of the variance in clinical outcome was attributable to patient adherence (beta = .41) and baseline depression severity (beta = .65). Depression severity predicted clinical outcome but not patient participation. Participation predicted adherence (beta = .39) but did not directly affect clinical outcome. Adherence was explainable by physician- (beta = .57) and patient-reported treatment adherence (beta = .66). Conclusions: In a specific pathway via adherence improving patient participation in decision-making can foster improved clinical outcomes. The research findings reveal the significance of patient participation as a key factor to address for improving treatment adherence and clinical outcome. Quality improvement strategies for depression treatment should emphasize patient participation. Keywords: depression shared decision making; primary care; adherence; clinical outcome 6 PATIENT INVOLVEMENT IN THE CHOICE OF RADIATION DOSE FOR PROSTATE CANCER van Tol-Geerdink JJ, Stalmeier PFM, Huizenga H, van Lin ENJT, Schimmel EC, van Daal WAJ, and Leer JW Nijmegen, the Netherlands Purpose: Cancer treatment with a higher radiation dose leads to a higher probability of (disease-free) survival but also to a higher risk of severe side effects. Our questions are 1) do patients with prostate cancer want to be involved in the treatment choice and if so, 2) do they choose the higher dose, that is, the more aggressive treatment? Methods: In an interview, the effects of two alternative treatment options (radiation with 70 Gy or 74 Gy) were explained to 150 patients with prostate cancer (T1-3N0M0). The patients were asked whether they wanted to leave the decision to the physician or choose their own treatment, and if so, which treatment. Subsequently, the patient_s choice was carried out. Results: Even in this older patient population, most of the patients (79%) wanted to choose their radiation dose. This desire to choose was associated with hopefulness. A majority of these patients (75%) chose the lower dose (70 Gy). This choice was consistent with their personal medical risk status and the importance weights they assigned to possible outcomes. Conclusion: Once informed, most patients want to make their own choice of the radiation dose and most prefer the less aggressive treatment out of these two active treatment options. In conclusion, patients with T1-3N0M0 prostate cancer should be informed and involved in the choice of treatment. Secondly, since many patients find quality-of-life aspects more important than cure, the general focus on cure may need to be reconsidered. 7 DOCTORS_ AND PATIENTS_ PREFERENCES FOR PARTICIPATION AND TREATMENT Stalmeier PFM,1,2 van Tol-Geerdink JJ,1 van Lin ENJT,1 Schimmel EC,3 Huizenga H,1 van Daal WAJ,1 and Leer JW,1 1Department of Radiation Oncology, Radboud University Nijmegen Medical Center, Nijmegen, the Netherlands; 2Department of Medical Technology Assessment, Radboud University Nijmegen Medical Center, Nijmegen, the Netherlands; 3Arnhems Radiotherapeutic Institute, Arnhem, the Netherlands Purpose: Physicians hold opinions about unvoiced patient preferences. Such opinions are called substitute preferences. Objective: To study agreement between substitute and patient preferences for participation in decision making and treatment planning. Methods: One hundred and fifty patients with prostate cancer (T1 T3N0M0) facing radiotherapy. Patients chose between two radiation doses, involving a trade-off between (disease free) survival and side effects. A decision aid conveyed the information. Subsequently, patient preferences for participation (whether or not they wanted to choose a radiation dose) were obtained. The chosen radiation dose was truly administered. Radiotherapists gave substitute judgments for patient preferences for participation and radiation dose, and in addition stated their own preferred plan for the patient. ABSTRACTS Results: Favouring participation were 79% of the patient preferences, and 66% of the substitute preferences; agreement was poor (64%, kappa = 0.13, p = 0.11), and was better in more hopeful patients (OR = 5, p = 0.003), and in more experienced physicians (OR = 1.10, p = 0.03). Favouring the less toxic dose were 71% of the patient preferences and 51% of the substitute preferences; agreement was 70% (kappa = 0.2, p = 0.03). Physicians_ own preference indicated a less toxic treatment in only 23% of the patients. Discussion: Physicians slightly underestimated patients_ desire to choose their own treatment, and strongly underestimated patients_ preferences for a less toxic treatment. Physicians cannot reliably guess patients_ participation and treatment preferences. Therefore, in order to meet standards for informed consent and patient autonomy, patient preferences should be explicitly asked. 9 COMMUNICATION OF EFFECTIVENESS OF INTERVENTION FOR CHRONIC DISEASES: WHAT SINGLE FORMAT RESULTS IN THE _BEST_ DECISIONS? Kristiansen IS, Dahl, R, Gyrd-Hansen D, Nexoe, and Nielsen JB Institute of Public Health, University of Southern Denmark, Odense, Denmark Purpose: The effectiveness of interventions for chronic diseases can be explained in terms of absolute (ARR, NNT) or relative risk reduction (RRR), or life extension (LE). We aimed at exploring which single format yields decisions that are closest to those of the fully informed individual. Methods: A random sample of 4,000 non-institutionalised individuals aged 40-59 were invited to a face-to-face interview and asked about preferences for a hypothetical preventive drug. In total, 1,491 (37%) interviews were completed. Respondents were randomised to 24 different interview formats of which 16 (n = 1,169) are presented here. Here, respondents were randomly allocated to receive first one primary piece of information (ARR, NNT, RRR or LE) and then indicate preferences for the drug on a 1-10 scale (1 = definitely no; 10 = definitely yes). Subsequently they were given _full information_ (all four pieces + pictorial representation) and asked again to indicate preferences. Results: On the 1-10 scale, the mean change in preference from primary to _full information_ was 0.12 for ARR (n = 294) (p = 0.66), _0.14 for NNT (n = 288) (p = 0.73), _0.98 for RRR (n = 292) (p = 0.02), and 0.43 for LE (n = 295) (p .0001). In regression analysis, RRR as primary information was associated with significant change in preference while ARR and NNT were not (LE reference), nor were age, sex, income or numeracy significant. Conclusion: While ARR and NNT as single information format seem to result in decisions that are close to those of the _fully informed_ individual, RRR may result in too optimistic views and LE possibly too pessimistic. 10 COMMUNICATION OF EFFECTIVENESS OF INTERVENTION FOR CHRONIC DISEASES: WHAT FORMAT DO LAY-PERSONS PREFER? Kristiansen IS, Dahl R, Gyrd-Hansen D, Nexoe J, and Nielsen JB Institute of Public Health, University of Southern Denmark, Odense, Denmark Purpose: The effectiveness of interventions for chronic diseases can be explained in terms of absolute (NNT) or relative risk reduction (RRR), life extension (LE) or pictorial representation (PR) with population crowd figures representing the number of successes and failures. We aimed at eliciting lay-persons_ preferences for information format and comprehensibility of each of these formats. Methods: A random sample of 4,000 non-institutionalised individuals aged 40-59 in an urban area were invited to a face-to-face interview and asked about preferences for a hypothetical preventive drug. In total, 1,491 (37%) interviews were completed. Respondents were randomised to 24 different interview formats of which 16 (n = 1,169) are presented here. All respondents were presented with all four information formats. Results: While 27 persons were uncertain about their preferred information format, 106 (9%) preferred NNT, 436 (37%) RRR, 149 (13%) LE and 451 (39%) PR. The proportions that found the information very easy to understand were 72% (NNT), 67% (RRR), 61% (LE) and 82% (PR). There was no clear association between numeracy and preferred information format or comprehensibility. Conclusion: Lay-persons prefer RRR and pictorial representation as expression of effectiveness of interventions for chronic diseases, but the self-reported comprehension is better for pictorial representation. 11 PRESENTING HEALTH RISK INFORMATION IN DIFFERENT FORMATS: THE EFFECTS ON PARTICIPANTS_ COGNITIVE AND EMOTIONAL EVALUATION Timmermans DRM, Vermey K, and Henneman L Department of Public and Occupational Health, VU University Medical Center, Amsterdam, the Netherlands Purpose: Effective communication of treatment risks is important to enable patients to make informed decisions. This study aimed to determine the effects of different risk formats on participants_ evaluation and interpretation of risk information. Methods: Participants (N = 101) were recruited among students of the Vrije Universiteit Amsterdam and were asked to evaluate four cases. They were asked to imagine that a good friend of theirs was at risk for (1) giving birth to a child with Down syndrome; (2) developing cardiovascular disease in the forthcoming 10 years; (3) developing breast cancer; and (4) developing colon cancer when having a positive family history. Risk information was presented in one of three risk formats (percentages, natural frequencies, or icons). Risk format was varied between subjects. Results: Analysis showed that risk information presented in percentages was evaluated as smallest but also easiest to understand compared with risk information presented in other formats (p .05). Risk information presented in icons was evaluated as most serious, most frightening and most worrisome, while risk information presented in percentages was evaluated as least serious, worrisome and frightening (p .001). Mixed results were found on the dimension complexity. An effect was found on decision making, that is, what they would advise their friend to do, but this was not clear cut. Conclusion: This study showed that different risk formats have different effects on participants_ cognitive and emotional evaluation of the information and on their choice. Doctors should therefore be careful in choosing the format in which they present treatment risks. 12 SUPPORTING PATIENT INVOLVEMENT IN ILLNESS MANAGEMENT: AN EVALUATION OF INFORMATION PROVIDED BY UK RENAL UNITS Winterbottom A,1 Bekker HL,1 Conner M,1 and Mooney A2 1University of Leeds, Leeds; 2St. James University Hospital, Leeds, United Kingdom Purpose: Patient involvement in illness management is a service priority. Provision of accurate information is fundamental to facilitating patient involvement. Patients with end stage renal ABSTRACTS ABSTRACTS E3 E4 _ MEDICAL DECISION MAKING/JUL_AUG 2007 failure (ESRF) receive information from the NHS, charities and pharmaceutical companies. There is little evidence of the quality or effectiveness of information that facilitates involvement across services. This study assesses the consistency and quality of information provided to support patients_ renal replacement therapy choices. Methods: Questionnaire audit of information provided by UK renal services; quality assessment of written information about renal replacement treatment options in terms of: readability; adequacy of information about treatment options; theoretical techniques enhancing evaluation of information. Results: 67/100 renal units returned questionnaires; services provide patients with information over and above routine consultations: written information (91%); videos (54%); workshops (70%), specialist nurse home visits (81%). There was variation across services by amount and type of information provided. 31/47 provided written information about renal replacement therapies; Flesch readability scores were low (mean 48; SD 10.8) (Flesch, 1948), description of treatment options varied across services, few included techniques known to assist patients_ processing of information. Conclusions: These data suggest a chaotic pattern of information provision across the UK for patients with ESRF. Further, it is unclear if services are providing comprehensive information in a way that enables patients to evaluate the information in accord with their own beliefs and preferences. Evidence is required to explore how patients use this information and what types of information enable patients to become involved in the management of their treatments for ESRF. 13 EXPLAINING HETEROGENEITY AND ASSESSING PUBLICATION BIAS IN META-ANALYSIS AND META-REGRESSION OF THE EFFICACY OF DISEASE MANAGEMENT PROGRAMS IN CONGESTIVE HEART FAILURE Göhler A,1 Worrell SS,1 Gazelle SG,1 and Siebert U2 1Harvard Medical School, Boston, MA; 2Institute for Public Health, Medical Decision Making and Health Technology Assessment, UMIT, Hall i.T., Austria Purpose: Several meta-analyses have been performed to investigate the efficacy of disease management programs (DMP) in congestive heart failure. However, none of them formally assessed heterogeneity and publication bias. Based on these analyses, we sought (1) to compare the adequacy of different effect measures, (2) to identify and explain heterogeneity, and (3) to assess potential publication bias. Methods: Our analyses were based on 36 RCTs with a total of 8,341 patients. L_Abbé plot was used to display different effect estimates and their overall effects. Heterogeneity was assessed visually producing Galbraith plots and quantitatively performing meta-regression and subgroup analyses. Different baseline risks were assessed using bivariate models. To investigate publication bias, we used Begg_s funnel plots. Quantitative testing for publication bias was conducted using Begg_s test, the regression test of Egger, and the _trim and fill_ methods of Duval and Tweedie. Results: The l_Abbé plot did not indicate a preferable effect estimate. However, for the overall mortality reduction a multiplicative approach showed less heterogeneity among studies compared to an additive. Factors explaining heterogeneity between studies included severity of disease, age, medication, baseline risk, country, duration of follow-up, and mode of post-discharge contact. Although indicated by the funnel plot, no statistically significant publication bias was detected and the _trim and fill_ methods produced an unadjusted estimate. Conclusions: Substantial part of heterogeneity could be explained by a few relevant covariates. As in all meta-analyses, results may be affected by publication bias. However, we could not find statistical evidence for a publication bias. 14 MULTIDETECTOR CT ANGIOGRAPHY FOR ASSESSMENT OF CORONARY ARTERY DISEASE: A SYSTEMATIC REVIEW OF DIAGNOSTIC PERFORMANCE Heijenbrok-Kal MH,1 Vanhoenacker PK,2 Van Heste R,2 Decramer I,2,3 Van Hoe LR,2 Wijns W,2 and Hunink MGM1 1Erasmus MC_University Medical Center, Rotterdam, the Netherlands; 2OLV Ziekenhuis, Aalst, Belgium; 3Cardiovascular Center, Aalst, Belgium Purpose: To review the literature on the diagnostic performance of multi-detector computed tomography angiography (MDCTA), with conventional coronary angiography (CA) as reference standard for assessment of symptomatic coronary artery disease. Methods: A Pubmed and manual search of the literature published between January 1998 and November 2005 on MD-CTA compared with CA in patients with symptomatic coronary artery disease was performed. Diagnostic performance was expressed in summary estimates of sensitivity and specificity and the diagnostic odds ratio. Random-effects regression models were used to compare the diagnostic odds ratio of 4-, 16- and 64-slice-CTA and the proportion of non-assessable segments was evaluated. Results: Forty-six studies were included: 17 studies on 4-slice- CTA, 25 on 16-slice CTA and 4 on 64-slice-CTA. The pooled sensitivity and specificity for detecting >50% stenosis per segment were 0.90 (95% CI, 0.83-0.97) and 0.96 (95% CI, 0.95-0.98) for 64- slice-CT, 0.83 (95% CI, 0.76-0.90) and 0.96 (95% CI, 0.95-098) for 16-slice-CT, and 0.84 (95% CI, 0.80-0.88) and 0.93 (95% CI, 0.90- 0.95) for 4-slice-CT, respectively. Regression analysis showed that the diagnostic odds ratio significantly improved with the new generations of MD-CTA-scanners (64- and 16- versus 4-slice-CT), adjusted for exclusion of non-assessable segments, and contrast concentration used. Simultaneously, the non-assessable proportion of coronary segments significantly decreased with the new generations of MD-CTA-scanners, adjusted for heart rate, betablocker use, age, and prevalence of significant disease. Conclusion: With the new generations of MD-CTA-scanners the diagnostic performance for assessment of coronary artery disease has significantly improved, while the non-assessable proportion of segments decreased. 15 VALIDITY OF THE MANCHESTER TRIAGE SYSTEM IN PAEDIATRIC EMERGENCY CARE Roukema J,1 Steyerberg EW,2 Meurs A,3 Ruige M,3 var van der Lei J,4 and Moll HA,1 1Erasmus MC/Sophia Children_s Hospital, Department of Paediatrics, Rotterdam, the Netherlands; 2Erasmus MC/Center for Medical Decision Making, Department of Public Health, Rotterdam, the Netherlands; 3Haga Hospital, Juliana Children_s Hospital, The Hague, the Netherlands; 4Erasmus MC/Department of Medical Informatics, Rotterdam, the Netherlands Purpose: Hospital emergency departments are increasingly visited by patients with non-urgent problems or who are not referred by their general practitioner, leading to overcrowded waiting rooms and long waiting times. The Manchester Triage System (MTS) has been developed to categorize patients by _urgency of care._ In the MTS all patients are assigned one of five priority levels, ranging from emergent to non-urgent. Objective: To assess the validity of the MTS in paediatric emergency care. Methods: Patients were eligible if they attended the emergency department of a large inner-city teaching hospital in the period from August 2003 to November 2004 and were under 16 years. A representative sample of 1,065 patients was drawn from 18,469 eligible patients. The originally assigned MTS urgency level was compared with a predefined reference standard for true urgency. ABSTRACTS The reference standard was based on objective information on vital parameters, resource use and final diagnosis. Sensitivity, specificity, and percentage over- and undertriage of the MTS were calculated. Results: Undertriage occurred in 15%, of which 96% by only one urgency category lower than the reference standard. The sensitivity of the MTS to detect emergent/very-urgent cases was 63%. Overtriage occurred in 40%, mostly in the lower MTS-categories. In 36% of these cases, the MTS classified at least two urgency categories higher than the reference standard. Conclusions: The MTS lacks sensitivity and specificity in paediatric emergency care. Flowchart-specific modifications of the MTS should be considered for paediatric emergency care to reduce overtriage, whilst maintaining sensitivity in the highest urgency categories. 16 RELATIONSHIP BETWEEN THE EUROQOL-5D AND BARTHEL INDEX: EXAMINING THE USE OF PROXY OUTCOME MEASURES FOR OLDER PEOPLE Kaambwa BC, Bryan S, and Barton PM Health Economics Facility, Health Services Management Centre, University of Birmingham, Birmingham, United Kingdom Purpose: Intermediate care (IC) of older people is a key component of UK government health policy. Evidence on outcomes associated with IC is scarce. In many instances, health-related quality of life (HRQoL) outcome measures are not available mainly because older people are physically or mentally not able to (self-) report their HRQoL. The resulting missing outcome values may lead to biased statistical results. Proxy outcome measures may help but their suitability as proxies needs to be tested and verified. The aim of this study was to examine the direct relationship between a conventional clinical scale of functional status that is suited for proxy-assessment (Barthel Index [BI]) and a measure of HRQoL (EuroQoL-5D). Methods: We considered outcome measures from a total of 1589 IC patients participating in the national evaluation of IC for older people in the UK. To test the relationship between the BI and the EuroQoL-5D, we used jackknife regression, correlation coefficients, multiple comparison tests (Dwass, Steel, Critchlow- Flinger tests) and multiple imputation methods. Results: We established a plausible and significant relationship between the measures. There was evidence of the ability of the BI to predict EuroQoL-5D scores. This relationship was valid for all IC types and settings as well as across all diagnostic groups in the sample. Conclusions: There is considerable documentation supporting the construct validity of using the EuroQoL-5D in geriatric care, and these results add to the evidence that BI scores may be used as proxy for HRQoL in case of missing values on EuroQoL-5D. 17 WOMEN_S DECISION MAKING ABOUT CLINICAL TRIAL PARTICIPATION DURING TREATMENT FOR BREAST AND OVARIAN CANCER Abhyankar P,1 Bekker HL,1 Latchford G,1 and Velikova G2 1University of Leeds, Leeds; 2St. James University Hospital, Leeds, United Kingdom Purpose: During treatment for breast and ovarian cancer, some women are invited to participate in clinical trials. Some women need to make informed choices simultaneously about trial (non) participation and treatment options. Although studies have described peoples_ reasons for trial (non) participation, few have investigated how women_s beliefs about cancer treatment impact on trial participation. This study explores the association between beliefs about cancer and treatment, and decision making about trial participation. Methods: 9 women with breast and 9 with ovarian cancer who were interviewed retrospectively to their decision about trial participation. Interviews were transcribed and coded using thematic content analysis. Women completed questionnaires after the interview assessing: Anxiety (6-item STAI); Beliefs about illness (BIPQ); Beliefs about treatment (BMQ). Results: Preliminary analysis of interviews revealed several themes that informed trial participation choices including: beliefs about cancer treatments; perceptions of risk about treatment effects; and dynamic nature of illness and choices. At interview, women_s level of anxiety were normal (M = 38), they reported pessimistic views of disease timeline (M = 7.3), moderate levels of personal control (M = 4.9), high levels of treatment control (M = 8.5) and high concerns with illness (M = 7.3). Conclusions: Findings suggest women_s choices about trial participation are influenced by past experiences and beliefs about cancer and treatment options, rather than trial information. Further, these beliefs affect the way women process trial information. If interventions are to be designed to encourage involvement in clinical trials, further research needs to understand how best to support women when faced with treatment and trial decisions simultaneously. 18 UNWILLINGNESS TO VACCINATE: A PROBLEM OF UNREALISTIC OPTIMISM? Fischer K University of Applied Sciences Northwestern Switzerland, Department of Applied Psychology, Olten, Switzerland Purpose: Vaccinating is one of the most effective ways protecting people from serious diseases. Nevertheless, the willingness to vaccinate is decreasing in many industrial countries. In three studies we investigated the occurrence of unrealistic optimism in the risk perception of getting ill without vaccination and unrealistic pessimism concerning side effects of vaccination. Methods: Using questionnaires, we asked for (1) Ss_ willingness to vaccinate, (2) judgments of their personal risk for specific diseases without vaccination, (3) their personal risk for side effects, (4) risks for disease for a normal adult person of the same age (other_s risk). Furthermore, our Ss had to compare directly their own risks with other_s risks. Results: Indeed, Ss underestimated their personal risk of getting ill without vaccination compared to the population risk (unrealistic optimism). This result occurred in direct comparative judgements as well as when unrealistic optimism was derived from probability judgements. Unexpected, we did not find an unrealistic pessimism in the perception of the risk of side effects. In contrast, there was an unrealistic optimism too. Though underestimating their probability, the fear of severe side effects is the reason most frequently given by our Ss against vaccination. Conclusions: Unwillingness to vaccinate seems to be due to an unrealistic optimism concerning diseases without vaccination and to the fear of side effects. We discuss these findings and their implications for interventions designed to influence risk perception. 19 ADOLESCENTS_ VIEWS ON ASTHMA CONTROL AND SELFMANAGEMENT: INTERNET-BASED SELF-MANAGEMENT OFFERS AN OPPORTUNITY TO ACHIEVE BETTER ASTHMA CONTROL van der Meer V,1 van Stel HF,1 Otten W,1 Detmar SB,2 and Sont JK1 1Department of Medical Decision Making, LUMC, Leiden; 2TNO Prevention and Health, Leiden, the Netherlands Purpose: Guided asthma self-management programmes including information, monitoring, regular medical review and a written ABSTRACTS ABSTRACTS E5 E6 _ MEDICAL DECISION MAKING/JUL_AUG 2007 action plans are effective in reducing the burden of asthma. We explored the barriers and supportive factors for (ICT-based) selfmanagement in adolescents with well and poorly controlled asthma. Methods: 97 adolescents (12-17 yr) who had mild to moderate persistent asthma (> 3 mo inhaled corticosteroid usage in the past year) were asked to monitor asthma control on a designated website (weekly Asthma Control Questionnaire [ACQ]; daily lung function registrations via website or short-message service [SMS]). After 4 weeks, 35 patients participated in focus group interviews stratified on sex, age and asthma control (ACQ, Asthma Therapy Assessment Questionnaire). Focus group interviews were audiotaped, fully transcribed and coded with respect to the components of asthma self-management. Results: All patients expressed limited self-efficacy on acute situations, avoidance of triggers and medication use. Patients with poor asthma control accepted symptoms as a part of everyday life. They expressed a need for education, and more often reported monitoring and getting instant feedback by ICT when their asthma deteriorated as useful than patients with well controlled asthma. The former group was willing to incorporate ICT-based selfmanagement on a long-term base into their everyday life. Conclusion: Asthmatic adolescents have limited self-efficacy expectations. In contrast to patients with well controlled asthma, patients with poorly controlled asthma accepted that symptoms were a part of everyday life, but held positive views to a guided ICT-based self-managament programme in order to achieve better asthma control. 20 DOES OFFERING PRENATAL SCREENING AFFECT ATTITUDES? Kleinveld JH,1 Timmermans DRM,1 van den Berg M,1 van Eijk JTh,2 van Vugt JMG,3 and van der Wal G1 1Institute for Research in Extramural Medicine, Department of Public and Occupational Health, VU University Medical Center, Amsterdam, the Netherlands; 2Department of Health Care Studies/Medical Sociology, Maastricht University, Maastricht, the Netherlands; 3Department of Obstetrics and Gynaecology, VU University Medical Center, Amsterdam, the Netherlands Purpose: In health care, making informed decisions is highly valued and seen as a reflection of autonomous decision making. Informed decisions are decisions that are based on sufficient knowledge and are in congruence with the decision maker_s attitudes. However, attitudes are often not stable over time and may change due to the decision. The present paper aims to give more insight into how attitudes change when deciding about whether or not to have prenatal screening done. Methods: Based on randomisation, 1500 women were offered the nuchal translucency measurement or triple test. Questionnaires were filled in before prenatal screening was offered, after the offer and after the test result was known or at a comparable point in time. Attitudes towards testing their unborn child were measured using a self-developed Attitude Scale. Results: Attitudes of women with a neutral attitude about prenatal screening at baseline changed in the direction of their choice: they became more positive if they intended to accept the screening test and became more negative if they intended to decline the test. Attitudes of women with a positive or negative attitude at baseline who intended to act opposite their attitude, changed in the direction of their intended choice. Regarding postdecisional processes, attitudes of women who were not certain of their intended decision changed slightly in the direction of their executed choice. The instability of women_s attitudes before and after making the decision regarding prenatal screening questions the concept of informed decision making as a reflection of autonomous decision making. 21 A COGNITIVE PROCESS MODEL OF PERIOPERATIVE DECISION MAKING IN ANESTHESIA Pott CM1 and Ballast A2 1University of Groningen, Groningen, the Netherlands; 2University Medical Center, Groningen, the Netherlands Purpose: Common models for decision making are not suitable to determine information requirements of anesthetists during critical periods in perioperative tasks. We developed a cognitive process model for diagnostic decision making under uncertainty and time pressure. Methods: To investigate the diagnostic decision making of anesthetists, we conducted an international survey in 2002. Based on these results (246 returned forms from 29 different countries), we developed a cognitive process model for decision making. The model approximates anesthetists_ internal conceptualization of their working environment. Results: Our model distinguishes three different problem states of the patient requiring decision making by the anesthetist: a familiar state, an urgent state, or an unidentified state. The anesthetist chooses a causative therapy for familiar and therefore well-diagnosed problems. For urgent but incompletely identified problems a symptomatic therapy is appropriate. In case a problem cannot be categorized, extensive decision making processes by the anesthetist or other experts are necessary before therapy can be given. Conclusions: Decision making in anesthesia is a complex task made by highly trained experts. To handle the complexity, anesthetists frequently use _satisficing_ heuristic for diagnosing. If the state of the patient is familiar to the anesthetist, usually no extensive decision making processes have to take place. In urgent situations, decision making heuristics like _treat first what kills first_ are used. However, for uncategorized patient problems, special human abilities of decision making are essential. Using the model, we developed a decision support system supporting the information requirements of anesthetists in the three different situations. 22 A METHOD FOR ELICITING KNOWLEDGE STRUCTURES USED BY MENTAL HEALTH PROFESSIONALS WHEN MAKING RISK JUDGEMENTS Buckingham CD1 and Adams AE2 1Aston University, Birmingham, United Kingdom; 2University of Warwick, Coventry, United Kingdom Purpose: To describe a method that elicits a consensual riskassessment knowledge structure from multiple experts. Methods: Semi-structured interviews were conducted with 46 multi-disciplinary mental health professionals and content analysis was applied to the transcripts. A new method was developed for recording the results, using open-source mind-mapping software called Freemind. Each interview was coded as a mind map and then integrated into a single map encapsulating the combined knowledge of experts. Nodes (concepts) in the combined mind map were linked to the individual experts who mentioned them and then to their interview transcripts, providing a full audit trail for the map_s construction. Analysis of maps was facilitated by their storage in Freemind as XML, a format that structures data. Results: The combined mind map represented a hierarchical knowledge structure with 455 unique concepts (many occurring in several places) and 840 low-level cues. Cues (eg, patient lives alone) are linked via increasingly abstract intervening concepts (eg, living arrangements, social context) to top-level risk concepts (eg, suicide). The varying numbers of experts supporting concepts are quantified, highlighting parts with greatest consensus, and exposing different emphases of concepts and cues, depending on their context (eg, 14 experts associated living alone with suicide risk but only 1 with risk of harm to others). ABSTRACTS Conclusions: Content analysis using mind maps and XML provides a powerful tool for understanding consensual knowledge structures. It can inform construction of decision support systems that better integrate with and disseminate clinical expertise. 23 TRAINING DECISION MAKING IN NOVICES: A WEBSITE TO DEVELOP SKILLS IN REFERRAL PRIORITISATION Harries P,1 Gilhooly K,2 Weiss D,3 Tomlinson C,4 and Harries C5 1Brunel University London; 2University of Hertfordshire, Hatfield,3 4Clinical Sciences Centre/Imperial College Microarray Centre, London; 5University College Hospital, London, United Kingdom Purpose: Demand for occupational therapy services within the community mental health setting far outweighs service availability; occupational therapists must therefore be skilled in the prioritisation of those referrals that they receive. This paper will present the development of a website that will be used to train novices in the decision-making skill of referral prioritisation. Methods: Initially 40 expert occupational therapy decisions on 120 referrals were modeled using judgement analysis. A training tool was then developed that comprised of a pretraining set of referrals, a training programme based on the expert policy, and a posttraining set of referrals. This was tested with 37 novices to see if it had the capacity to develop the novices_ policies. It has now been further developed into an interactive website which is being piloted at Brunel University where approximately 450 novice occupational therapists are educated. The website takes the user through a training package and then provides feedback on any skill development. Results: Prior to training, the correlation between the judgments of the novices and the expert policy was 0.23. After training, it rose to 0.70. It was also found that novices_ CWS ratios (a measure of expertise based on inconsistency and discrimination data) rose for 29 of the 36 students. Conclusions: The tool appears to have the capacity to develop novices_ decision-making skills. Once the pilot stage is complete, the website will be offered nationally for its use in the educational curriculum of undergraduate occupational therapists. 24 A SOCIAL JUDGEMENT ANALYSIS OF BRITISH, DUTCH, CANADIAN AND AUSTRALIAN ACUTE CARE NURSES_ INTUITIVE RISK ASSESSMENT Thompson C Department of Health Sciences, University of York, York, United Kingdom Purpose: Simple algorithmic guidelines for assessing risk of a critical event in patients (such as the Modified Early Warning System or MEWS) have known performance qualities. However, many nurses prefer to override these and make use of their own internal, intuitive, resources as they believe their performance to be better. This study examines the following: _ judgement performance of nurses against an ecological criterion with known properties (sensitivity/specificity/PPV), _ whether critical care (contact with sicker patients) and length of experience makes a difference to performance, and _ whether the clinical variables that should be given most weight in assessing risk of a critical event are the same variables that are given most weight. Methods: A double system social judgement lens model study of 200 critical and acute care nurses from the Netherlands, UK, Canada and Australia. The ecological criterion was a patient_s risk score. The patient profiles (social judgement scenarios) were extracted from the records of 350 acute care patients in the UK. Results: Analysis is still ongoing, but preliminary findings include the following: _ more experienced nurses are not necessarily better than junior nurses at assessing risk; _ cue forms and decision thresholds are not sensitive to education, experience or work environment; _ significant variation in judgements can be attributed to significant variation in the ways that people handle information; and _ receiving training in risk assessment does not increase performance. Conclusions: The number of preventable adverse events involving risk assessment may be due in part to the variability in judgement performance of nurses in acute care. PARALLEL III 25 EVALUATION FOR PUBLIC HEALTH INTERVENTIONS: DECISION RULES WITH MULTIPLE OBJECTIVES AND MULTIPLE CONSTRAINTS Claxton K University of York, York, United Kingdom Purpose: To demonstrate that a social decision making approach to evaluation can be generalised to interventions which have multiple objectives and impact on multiple constraints within and beyond the health sector. Background: Public health interventions and national policies which will have an impact across public sector budgets and the wider economy poses the question whether the existing approach to the evaluation of health technologies within the health sector are sufficient. Methods: Social decision making or _welfarist_ approach to evaluation will be are examined. We identify the generalisations of existing decision rules which are required. Results: Current decision rules in CEA and a welfarist analysis both fail to fully address the allocation problem posed by public health interventions. The information requirements of a mathematical programming solution to this problem make it impractical. We propose a simple compensation test for interventions based on the shadow prices of budget constraints and the net benefits falling on different sectors. We show that it is not necessary to pay compensation for each decision if the net compensation required is accounted for over a budget period and informs the marginal changes in subsequent allocations between sectors. Conclusions: A generalisation of decision rules to multiple sectors is required based on compensation valued in a way which is consistent with the existing allocation between the public sector(s) and the wider economy. A _welfarist_ societal perspective is not sufficient; rather, a multiple perspective evaluation which accounts for costs and effects falling on each sector is required. 26 EVALUATION: STATIC OR DYNAMIC? DOES IT MATTER HOW WE MODEL THE COST-EFFECTIVENESS OF CHLAMYDIA SCREENING? Roberts T,1 Barton P,1 Robinson S,1 Bryan S,1 and Low N2 1University of Birmingham, Birmingham, United Kingdom; 2University of Berne, Berne, Switzerland Purpose: The appropriateness of different modelling approaches used in economic evaluations of infectious diseases such as ABSTRACTS ABSTRACTS E7 E8 _ MEDICAL DECISION MAKING/JUL_AUG 2007 Chlamydia is becoming increasingly controversial. Our recent review showed the majority of economic evaluations on Chlamydia screening used decision analytic models, referred to as _static_ because they assume a constant force of infection (Roberts et al. Sexually transmitted infections. In press). Very few studies had used an appropriate transmission _dynamic_ model. The aim of this paper is to explore the extent of the difference in results using the different modeling approaches. Methods: We compared the data required and the results of the economic evaluation carried out as part of the Chlamydia Screening Studies (ClaSS) project in the UK which evaluated population screening using a discrete event simulation approach, with the data required and results of a static analysis using as far as possible the same empirical data collected as part of the study. Results: The results of using the two alternative modelling approaches were very different. It was not possible to show a systematic difference in any particular direction. The results of the static modelling approach were very sensitive to small changes in the estimates for the probability of experiencing complications associated with Chlamydia. Conclusion: This is a controversial issue as some experts argue that static models give biased results of the impact of screening; others believe that a static approach is adequate. However, use of an inadequate model risks presenting misleading results to policy makers. The implications will be discussed in July 2006. 27 HIV TRANSMISSION MODELING UNCERTAINTY: A COMPARISON OF 4 MODELS Pinkerton SD1 and Benotsch EG2 1Center for AIDS Intervention Research, Medical College of Wisconsin, Milwaukee, WI; 2University of Colorado at Denver Health Sciences Center, Denver, CO Purpose: Cost-effectiveness analyses often rely on probabilistic models to estimate health-related outcomes. The choice of a specific model can introduce _modeling uncertainty_ into the analyses. This important source of uncertainty seldom is addressed in published cost-effectiveness modeling analyses. To assess the impact of modeling uncertainty on the results of HIV prevention CEAs that rely on mathematical models of transmission, we compared 4 different modeling procedures for estimating the annual number of HIV infections acquired by gay/bisexual men vacationing in Key West, Florida. Methods: Using sexual behavior data collected from 97 vacationing men, we compared: a simple linear model of HIV transmission; a Bernoulli model in which sex acts were evenly distributed among sex partners; and two non-even distribution models (a Monte Carlo simulation and exhaustive enumeration of possible distributions). Results: The 4 modeling approaches produced similar estimates of the number of infections acquired by men vacationing in Key West. The _true_ number was found to lie between 195.7 and 206.7. The 4 HIV transmission models generated estimates ranging from 201.1 to 202.0. At worst, the maximum error was no larger than 3.6% (7.0 infections). Conclusions: Our findings have implications for both modeling and empirical data collection. Two of the 4 methods examined in this study are computationally complex, one is intermediate, and one is quite simple. The results of the analysis suggest that little precision is sacrificed through the use of the simpler models. These models also have fewer data requirements than the more complex models, substantially simplifying data collection. 28 IMPACT OF AVOIDING DIFFERENT CAUSES OF DEATH Wisløff T,1 Waaler HT,1 and Kristiansen IS2 1Norwegian Knowledge Centre for the Health Services, Oslo, Norway; 2University of Oslo, Oslo, Norway Purpose: Cause-specific mortality is usually presented in terms of cause-specific mortality rates. The purpose of this work is to present it in terms of life years gained by eliminating different causes of death. Methods: We modelled life expectancy in the Norwegian population on the basis of cause-, age- and sex-specific mortality rates from Statistics Norway. The model encompasses 17 groups of diagnoses, and 9 subgroups (specific diagnoses). The potential life-year gains from total elimination of different causes of death were estimated by subtracting cause-specific mortality risks from the totals. We also compared these results with similar analyses that were conducted about 30 years ago. Results: The potential life-year gain of total elimination of the different causes of death are summarised in the table. 2002 1973_1974 Cause of Death Men Women Men Women Neoplasms 3.26 3.24 2.41 2.62 Mental and 0.47 0.27 0.07 0.03 behavioural disorders Diseases of the 4.33 3.79 7.39 6.38 circulatory system Diseases of the 0.78 0.85 0.70 0.82 respiratory system Certain conditions 0.11 0.14 0.50 0.33 originating in the perinatal period Violent deaths 1.26 0.57 1.78 0.62 Conclusions: During the past 30 years, there has been a shift in the importance of different causes of death away from diseases in the circulatory system and conditions originating in the perinatal period to Neoplasms and mental disorders. The modelling results are based on the assumption that the diseases and their causes are independent. In reality this assumption may be violated, and the results consequently biased. 29 STABILITY OF QUASI-REALISTIC RISKY LAY MEDICAL AND MANAGERIAL DECISIONS BASED ON RESTRICTED INFORMATION Huber OW University of Fribourg, Fribourg, Switzerland Purpose: When choosing an alternative, lay decision makers often do not know if they will find useful additional information in the future. In lay medical decisions this problem is especially relevant as usually multiple information sources are available. The paper investigates the stability of quasi-realistic risky lay medical and managerial decisions based on restricted information search if later a possibility to continue information acquisition arises. Theories following Festingers (1964) assumption of postdecisional directional processing with the goal of representational coherence predict decision stability as subjects should not actively destabilize their representation by searching possibly disconfirming information. However, Heckhausens (1989) theory of action predicts a seamless continuation of the decision process as long as execution of an alternative has not been started. ABSTRACTS Methods: An experiment (90 subjects) uses the Method of Active Information Search (Huber, Wider, and Huber, 1997) to present one medical and two managerial tasks: after a short description of the decision situation, subjects can ask questions and receive matching answers. Subjects have to decide after search of two items of information either unexpectedly or previously announced, in a control condition information search is unlimited. Results: In contrary to the control condition, in the experimental conditions a great majority continues information search (up to 90%) and decision revisions are frequent (26%). There is no difference between tasks. Conclusions: The results disconfirm strong automatic directional processing, pointing towards an unbiased continuation of the decision process. Thus, in situations as described above decisions cannot be relied on as definitive. 30 DEFINING LOCALITIES OF INADEQUATE TREATMENT FOR CHILDHOOD ASTHMA: A GIS APPROACH Peled R,1 Reuveni H,1,2 Pliskin JS,1,3 Benenson I,4 Hatna E,4 and Tal A2 1Department of Health Systems Management, Ben-Gurion University of the Negev, Beer-Sheva, Israel; 2Pediatrics Department, Soroka University Medical Center, Beer-Sheva, Israel; 3Department of Industrial Engineering and Management, Ben-Gurion University of the Negev, Beer-Sheva, Israel; 4Department of Geography, Tel Aviv University, Tel Aviv, Israel Purpose: The use of Geographic Information Systems (GIS) has great potential in decision making analysis. Asthma is a chronic disease associated with substantial morbidity, mortality, and health care use. There is an increasing prevalence of asthma morbidity and mortality despite the availability of effective treatment. Objective: To develop an efficient tool for quality assurance and decision making analysis of chronic disease management. Databases: Administrative claims data of the largest HMO in Israel: drug dispensing registry, demographic data, Emergency Room visits, and hospitalization data bases. Methods: We created a list of six markers for inadequate pharmaceutical treatment of childhood asthma from the Israeli clinical guidelines. We used this list to search the drug dispensing registry to identify asthmatic children who received inadequate treatment and to assess their health care utilization and bad outcomes: emergency room visits and hospitalizations. Using GIS we located the clinics with a high percentage of children for whom the treatment provided was not in adherence with the clinical guidelines. Results: 81% of the children were found to have at least one marker for inadequate treatment; 17.5% were found to have more than one marker. Children with markers were found to have statistically significant higher rates of Emergency Room visits, hospitalizations and longer length of stay in hospital compared with children without markers. The maps show in a robust way which clinics provided treatment not in accord with the clinical guidelines. Those clinics have high rates of Emergency Room visits, hospitalizations and length of stay. GIS can be a useful tool in chronic disease management and decision making analysis. 31 WEIGHT CHANGE AS RISK FACTOR FOR CONGESTIVE HEART FAILURE: DEVELOPMENT OF A PROTOCOL AND PRELIMINARY ANALYSIS BASED ON THE FRAMINGHAM OFFSPRING COHORT Göhler A,1 Siebert U,2 and Anker SD3 1Harvard Medical School, Boston, MA; 2Institute for Public Health, Medical Decision Making and Health Technology Assessment, UMIT, Hall i.T., Austria; 3Division of Applied Cachexia Research, Charité Campus Virchow-Klinikum, Berlin, Germany Purpose: We aimed to assess the effect of weight change on the risk of CHF or death in a general population. Methods: We developed a protocol to investigate the effect of weight change on CHF and mortality in the Framingham Offspring study. We defined weight change as percentage increase (+5%) or decrease (_5%) in the body-mass index between two exams: Our endpoints will be CHF or death (primary), CHF with deaths censored and overall mortality. For univariate analysis and confounder selection, we will use Kaplan-Meier curves (KMC). After confounder selection, we will perform multivariate analyses using Cox proportional hazard models. Results: Considering individual cohort member follow-up, we identified 4547 cancer-free subjects (48% male) with baseline values determined at enrollment with follow-up over a mean period of 14 years. 411 subjects developed CHF or death and 91 CHF. In a preliminary univariate analysis, KMC showed statistically significant different outcomes for the 3 groups _+5%,_ __5%,_ and _stable_ for CHF or death (p .0001) and CHF (p = 0.0003). Compared to stable weight, the crude hazard ratio (95% confidence interval) for __5%_ was 1.97 (1.58-2.45) for CHF or death and 2.61 (1.68-4.04) for CHF; that for _+5%_ was 0.59 (0.47-0.75) for CHF or death and 0.65 (0.39-1.06) for CHF. Conclusion: Preliminary analyses suggested a univariate statistical association between weight loss and the incidence of congestive heart failure and death in a general population. Further research is needed to identify relevant confounders, to control for them and to assess the results in sensitivity analyses. 32 ETHICAL DECISION MAKING IN MEDICINE: AN APPLICATION OF THE ANALYTIC HIERARCHY PROCESS Wilson KM University of Queensland, Brisbane, Australia Purpose: This study developed a measure of individuals_ weightings of the medical ethical principles of beneficence, non-maleficence, autonomy, justice, confidentiality and veracity. The study also investigated the extent to which these weightings were able to predict ethical decision making outcomes (judgments and decisions) in medical ethical scenarios. Methods: The Analytical Hierarchy Process (AHP) was employed as the methodological tool for the ranking of the principles. This is a measure where participants choose between pairs of statements that conflict and provide a rating of the importance of their preference in each case. Weightings are then derived for the principles for each individual based on the pair wise choices and ratings. One hundred and ninety medical, psychology and business students were tested on this measure. Participants were also asked to make judgments and decisions in a number of medical scenarios. Results: The AHP is able to measure differences in the global importance of ethical principles, and overall the principle of nonmaleficence is the most important in relation to the other principles (for all types of students). However, interestingly these weightings were unrelated to ethical judgments and decisions in specific medical case scenarios. Conclusions: These results first demonstrate the need to empirically test important concepts in medical ethics and, second, they highlight the role of ethical principles in medical decision making. Future research directions are discussed in relation to the measurement of ethical decision making outcomes in medicine. 33 REASONING BIASES IN CLINICAL DECISION MAKING IN SPEECH AND LANGUAGE THERAPY Hoben KG,1 Varley RA,1 and May J2 1Department of Human Communication Sciences, University of Sheffield, Sheffield, United Kingdom; 2Department of Psychology, University of Sheffield, Sheffield, United Kingdom ABSTRACTS ABSTRACTS E9 E10 _ MEDICAL DECISION MAKING/JUL_AUG 2007 Purpose: We describe a project that compares the diagnostic reasoning skills of a group of expert speech and language therapists with those of a group of novice speech and language therapists, to test the hypothesis that experts _leap to conclusions_ and demonstrate reasoning biases despite generally reaching appropriate diagnoses, while novices may reason more normatively but lack the domain knowledge to reach appropriate diagnoses. Methods: Novice and expert therapists were asked to diagnose child and adult speech and language therapy cases presented in a written format, whilst _thinking aloud._ The case information was constructed so that later information is not necessarily concordant with the diagnoses suggested by initial information. Responses were audio recorded and analysed for reasoning patterns and biases. Results: Experts have a large case base to draw on, well-organised knowledge about their domain, are more confident in their diagnoses and therefore are more likely to show confirmation bias. Novices, by contrast, have a limited case base and knowledge and seek but then do not always appreciate the relevance of the disconfirmatory evidence. Conclusions: Both novice and expert therapists show biases in reasoning, as predicted by reasoning theory. Experts_ tendency to seek confirmatory and to ignore disconfirming evidence could lead to misdiagnosis, especially in unusual cases with surface similarities to more common disorders. 34 CARDIOVASCULAR RISK MANAGEMENT IN DIABETES: WHICH INFORMATION CUES TRIGGER THE INITIATION OF MEDICATION? Voorham J, Denig P, and Haaijer-Ruskamp FM Department of Clinical Pharmacology, University of Groningen, Groningen, the Netherlands Purpose: To obtain insight in the use of risk factors (blood pressure, cholesterol) for the decision of general practitioners to start cardiovascular treatment in diabetes patients. Methods: Included were 970 patients with type II diabetes managed by 33 general practitioners. Data from 2003-2004 were collected from electronic medical records. The influence of the risk factors level (value of last observation), trend (value change), and intensity (number of observations above target before start of treatment) on the decision to start antihypertensive or cholesterol lowering treatment was assessed with multivariate logistic regression, adjusting for age, gender, coronary heart disease (CHD) and diabetes duration. Results: Of 327 patients not using antihypertensive medication in 2003, 81 initiated treatment in 2004. Of 642 patients not using cholesterol lowering medication in 2003, 281 started in 2004. Initiation of antihypertensive medication was influenced by the last blood pressure level (10 mm Hg systolic OR 1.33, CI 1.07- 1.65; 5 mm Hg diastolic OR 1.54, CI 1.25-1.89) and intensity (OR 1.28, CI 1.06-1.54), but not by trend or cholesterol information. Initiation was higher in males, patients with CHD and short diabetes duration. Initiation of cholesterol lowering medication was influenced by the level of total cholesterol (1 mmol/L OR 1.44, CI 1.10-1.87), but not by intensity, trend, gender, CHD or blood pressure information. Medication was initiated less in older patients. Conclusions: Decisions were mainly guided by the last value of a single risk factor. Trends in risk factors were of limited predictive value, and intensity was only relevant for blood pressure levels. 35 ONCOLOGISTS_ CHOICE REGARDING THE USE OF PHARMACODIAGNOSTIC TESTS FOR CANCER TREATMENT SELECTION: A COMPARATIVE STUDY Wegwarth O,1 Gigerenzer G,2 Kurzenhäuser S,3 and Scholl W1 1Humboldt University, Berlin, Germany; 2Max Planck Institute of Human Development, Berlin, Germany; 3University of Basel, Basel, Switzerland Purpose: Upcoming pharmacodiagnostic tests offer the opportunity to better tailor cancer treatment decisions to individual patient needs. However, they put oncologists in the position of having to deal with a new technology, which often comes with its own specific risks, simply because its effects are less well known than those of established procedures. Little is known about how oncologists will handle this situation. This study is a first attempt to examine which factors impact oncologists_ decisions regarding the employment of such tests. Methods: For the preliminary study, 19 focused interviews (total) were carried out to identify relevant attributes as well as applied strategies of oncologists_ decision-making in Germany and the USA. The attributes and decision-making strategies were then tested with 164 discrete choice experiments (total). Results: Results indicate dominant preferences for the highest level of the attribute _test recommendation._ Lower levels of _test recommendation,_ high _risk of under treatment_ by using the test and increasing _test costs_ show negative effects on the choice. However, a rise in _severity of side effects of the treatment_ increases willingness to accept a lower level of _test recommendation,_ especially in German oncologists. U.S. oncologists are more likely to opt for a test in general. Conclusions: Although the substantial effect of _test recommendation_ underlines the importance of such guidelines on oncologists_ choice, their willingness to accept lower levels of recommendation under particular circumstances points strongly towards the need to emphasize the risks of non-evidence based test applications. 37 MODELLING THE IMPACT OF PRISON VACCINATION ON HEPATITIS B TRANSMISSION IN THE INJECTING DRUG USER POPULATION OF ENGLAND AND WALES Sutton AJ,1,2,3 Gay NJ,1 Edmunds WJ,1 and Gill ON1 1Health Protection Agency, London; 2Imperial College, London; 3University of Warwick, Coventry, United Kingdom Purpose: Injecting drug users (IDUs) are a major sub-group both at risk from and responsible for the transmission of hepatitis B (HBV) in England and Wales. As over 60% of the IDU population will have been imprisoned by the age of 30 years, prison may provide a good location in which to offer HBV vaccination. The work here considers the impact of prison vaccination on the transmission of HBV in the IDU population of England and Wales. Methods: A deterministic, compartmental, mathematical model was used to describe the transmission dynamics of HBV in the IDU population in England and Wales. Heterogeneities associated with how the injecting career length of IDUs may contribute towards the transmission dynamics of HBV were also included. Key model parameters and assumptions were subject to sensitivity analysis. Results: The base case model (that assumes that 50% of prison receptions are vaccinated by 2006) predicts that the number of acute cases of HBV in IDUs might be reduced by almost 80% in 12 years, and the HBV prevalence (IDUs ever infected by HBV) may be reduced from approximately 18% in 2002 to 7% in 2015. Conclusions: The model presented here demonstrates that HBV vaccination on prison reception can have a significant impact on the prevalence and incidence of HBV in the IDU population over time. A greater expansion of the prison vaccination programme in England and Wales will have an impact on the transmission of HBV in the IDU population. ABSTRACTS 38 ADJUSTING FOR PUBLICATION BIAS IN ECONOMIC DECISION MODELS Moreno SG,1 Sutton AJ,1 Ades A,2 Cooper NJ,1 Peters JL,1 and Abrams KR1 1Department of Health Sciences, University of Leicester, Leicester, United Kingdom; 2Department of Social Medicine, University of Bristol, Bristol, United Kingdom Purpose: Publication Bias (PB) is known to be a threat to the validity of meta-analysis that may consequently lead to incorrect decisions in Health Policy. A new Bayesian approach that adjusts for PB is proposed and results compared with existing methods. The motivation for this work is to be able to adjust for PB within a comprehensive economic decision modelling framework. Methods: A Bayesian approach to meta-analysis on the log-odds scale is implemented in WinBUGS. The PB/small-study-effect is estimated by measuring the extent to which smaller studies are giving results which are different from larger ones. The treatment effect is split into a true effect and a bias term using a semi-parametric regression model. Data from the _famous_ magnesium for myocardial infarction meta-analysis is analysed and the results contrasted with those derived from existing PB adjustment meta-analysis methods. Results: The bias associated with each study depends on any systematic association between effect size and study precision. Thus, if all studies have approximately the same effect, the bias term is negligible. However, if smaller studies have larger effect sizes, then inclusion of the bias term will mean these smaller studies will be downweighted with respect to the weights they would be given in a standard meta-analysis model and their influence on the pooled effect size diminished. Conclusions: This approach to adjusting for PB seems promising, and further work is under way to validate it using simulation. Once this is done, it will be applied within an economic decision making framework. 40 ESTIMATING THE (COST-) EFFECTIVENESS OF INFLUENZA VACCINATION IN CHILDREN Edmunds WJ, Pitman RJ, Vynnycky E, Sidiqui R, and Gay NJ Modelling and Economics Unit, Health Protection Agency, London, United Kingdom Purpose: Influenza vaccination of children has been recommended in the US. The potential (cost-) effectiveness of preschool vaccination strategies was assessed. Methods: The age-specific burden of influenza in terms of GP consultations, hospitalisations and deaths was estimated by regressing non-specific outcomes (such as consultations for acute respiratory disease) against laboratory reports for various pathogens (including influenza). A transmission dynamic model was developed and parameterised in standard ways. The model assumed that immunity to epidemic influenza waned over time due to antigenic drift. The average period of immunity was estimated by comparing the model to data. Unit costs per GP consultation and hospitalisation (NHS perspective) were taken from standard sources. Results: The burden of influenza in England and Wales was estimated to be large. Roughly 20% of the population may be infected every year resulting in an estimated 800,000 GP consultations, 19,000 hospitalisations and 18,000 deaths. Routine vaccination of pre-school children at 60% effective coverage is expected to have a large long-term impact on the incidence of influenza. The targeted age groups receive the largest reduction in incidence, though significant reductions would be expected in older age groups as well. Vaccination is estimated to be highly cost-effective, but only if there are significant reductions in disease in older age groups. Conclusions: Pre-school influenza vaccination has the potential to bring significant benefits to vaccinated children and the community as a whole. However, many key parameters are poorly estimated. The indirect effects of vaccination are important drivers of the cost-effectiveness of the programme. 41 HOW TO MEASURE THE SHORT-TERM BURDEN OF CANCER SCREENING AND SURVEILLANCE: UPPER GASTROINTESTINAL ENDOSCOPY IN PATIENTS WITH BARRETT_S ESOPHAGUS Kruijshaar ME,1 Siersema PD,2 Kerkhof M,2 Steyerberg EW,1 and Essink-Bot ML1 1Department of Public Health, Erasmus MC, University Medical Center Rotterdam, Rotterdam, the Netherlands; 2Department of Gastroenterology and Hepatology, Erasmus MC, University Medical Center Rotterdam, Rotterdam, the Netherlands Purpose: Cancer screening and surveillance tests can cause anxiety, pain and discomfort in patients, and hence result in shortlasting episodes of diminished health-related quality of life. Furthermore, the number of patients experiencing such episodes is much larger than the number experiencing health benefits of the procedure. The question is how this short-term burden can be measured and included in cost-effectiveness analyses of cancer screening/surveillance. We propose a common framework to measure the short-term burden. We apply it to surveillance by upper gastrointestinal endoscopy of patients with Barrett_s esophagus (BE), who are at increased risk of developing oesophageal adenocarcinoma. Methods: Four variables were assessed: 1) pain and burden experienced during endoscopy, 2) physical symptoms, 3) psychological distress (Hospital Anxiety and Depression scale and Impact of Event Scale), and 4) willingness-to-pay. A total of 180 BE patients filled out questionnaires one week before, on the day of, one week after and one month after endoscopy. Results: Of all patients, only 20% experienced pain from the endoscopy. However, 75% reported it to be burdensome. The procedure did not cause symptoms. Distress levels were increased prior to the endoscopy, and patients were willing to travel several hours extra for a non-invasive test. Conclusion: Upper endoscopy is burdensome for patients and causes moderate psychological distress. Recommendations for endoscopic surveillance should take the burden and distress of upper GI endoscopy into account. It is important to discuss methods to include the short-term burden of cancer surveillance in cost-effectiveness analyses. 42 EXPECTED VALUE OF SAMPLE INFORMATION IN SURVIVAL TRIALS USING BAYESIAN APPROXIMATION WITH THE WEIBULL PROPORTIONAL HAZARDS MODEL Brennan A1 and Kharroubi SA2 1University of Sheffield, Sheffield, United Kingdom; 2University of York, York, United Kingdom Purpose: Sample sizing for survival trials specifies a clinically significant proportional hazard, which if obtained, implies the adoption of the new therapy. Such adoption decisions are often evaluated using health economic decision models. We develop methods to formally integrate decision modelling with sample size for survival studies using the expected value of sample information approach. Methods: The method requires characterization of prior uncertainty around the proportional hazard, together with a model of survival, quality of life benefits and economic costs. EVSI computation uses Monte Carlo sampling to produce new simulated data-sets with specified size and follow-up. Each simulated data-set is synthesised ABSTRACTS ABSTRACTS E11 E12 _ MEDICAL DECISION MAKING/JUL_AUG 2007 with existing prior information. Because Bayesian updating for Weibull parameters is not analytically tractable, we use a novel form of Laplace approximation to estimate the decision model outputs. A case study builds on a classic text book example. We examine 1st and 2nd order versions of the Laplace approximation formula, comparing EVSI estimates. Expected benefits from additional information are compared against a cost function for different proposed study designs. Results: The approach is more efficient than standard EVSI computation because it requires neither Markov Chain Monte Carlo to obtain posterior densities nor Monte Carlo sampling to quantify the effect of each simulated data-set on decisions. Conclusion: The results show how the formal integration of economic considerations is both feasible and potentially profound in the design of survival trials. This methodology provides a new and valuable alternative approach for design of survival studies. 43 COST-EFFECTIVENESS OF IMAGING STRATEGIES FOR DETERMINING DIAGNOSIS AND ASSESSING RESECTABILITY IN PATIENTS WITH SUSPECTED PANCREATIC CANCER Schink T,1,4 Böhmig M,2 Müller-Nordhorn J,3 Hur C,4 Koch I,2 Wernecke KD,1 and Siebert U45 1Department of Medical Biometry, Charité University Medical Center, Berlin, Germany; 2Division of Hepatology and Gastroenterology, Charité University Medical Center; 3Institute of Social Medicine, Epidemiology and Health Economics, Charite University Medical Center, Berlin, Germany; 4Institute for Technology Assessment and Department of Radiology, Massachusetts General Hospital, Harvard Medical School, Boston, MA; 5Department of Public Health, Medical Decision Making and Health Technology Assessment, UMIT_ University for Health Sciences, Medical Informatics and Technology, Hall i.T., Austria Purpose: To assess the cost-effectiveness of different strategies for diagnosis and determining resectability in patients with suspected pancreatic cancer (PC). Methods: We examined prospective data from 206 patients with suspected PC who received care at the Charité University Hospital from 08/1999-11/2001 and underwent each of the following examinations: ultrasound (US), magnetic resonance imaging (MRI), computed tomography (CT), endoscopic ultrasound (EUS), fluorodeoxyglucose- positron-emission-tomography (PET), and endoscopicretrograde- cholangio-pancreaticography (ERCP). Diagnostic costs were based on the German tariff for hospital services. We constructed a decision tree to predict diagnostic costs and diagnostic accuracy for both PC and resectability. We evaluated 44 strategies, based on single tests as well as combinations of two tests. For the base case analysis, we used point estimates of the conditional probabilities to perform a cohort simulation. Results: The least costly and least effective strategy was _US alone_ with 62% correctly identified patients (CIP) and mean costs of Euro 38 per patient. _CT alone_ classified 10% more patients correctly, resulting in an incremental cost-effectiveness ratio (ICER) of Euro 1,925 per CIP. Strategy _MR alone,_ correctly classifying 5% more than _CT alone,_ had an ICER of Euro 3,479 per CIP. The most effective strategy _MR followed by ERCP for positive results_ had also the largest ICER of Euro 14,467 per CIP. Conclusions: Our study suggests that using MR as first test in the evaluation of patients with suspected PC results in the maximum number of CIPs. Whether adding ERCP after positive test results is good value for the money depends on the willingness-to-pay. 44 COST-EFFECTIVENESS OF RENAL REPLACEMENT THERAPY FOR PATIENTS WITH END STAGE KIDNEY DISEASE: A MARKOV MICROSIMULATION MODEL Salkeld G,1 Howard K,1 McDonald S,2 Cass A,3 Chadban S,4 White S,3 and Craig J,1 Screening and Test Evaluation Program (STEP), School of Public Health, University of Sydney, Sydney, Australia; 2Australia and New Zealand Dialysis and Transplant Registry (ANZDATA), Adelaide, Australia; 3The George Institute for International Health and Kidney Check Australia Taskforce, Kidney Health Australia, Sydney, Australia; 4The Royal Prince Alfred Hospital and Kidney Check Australia Taskforce, Kidney Health Australia, Sydney, Australia Purpose: In developed countries, the majority of patients diagnosed with end stage kidney disease (ESKD) are offered renal replacement therapy (RRT). As the incidence of ESKD steadily increases, the impact of RRT on health resources and patient health will be substantial. The objective of this study is to estimate the cost effectiveness of increasing renal transplantation rates and increasing the proportion of ESKD patients who receive dialysis in their home rather than in hospital. Methods: A Markov simulation model of current and future patients with ESKD in Australia was developed. The model follows RRT events and outcomes in the first, then second and subsequent years of treatment. Transition probabilities were derived from a secondary data analysis of treatment and outcome patterns for a cohort of ESKD patients (in the period 1996 to 2000). A cost utility analysis of three treatment options was performed: 1) a 10% to 50% increase in the number of renal transplants, 2) increasing the rate of home dialysis (with a concomitant decrease in hospital haemodialysis), and 3) increasing the rate of peritoneal dialysis. Results: Compared to current patterns of RRT, increasing the transplant rate by between 10% to 50% by the year 2010 is dominant, producing a net discounted cost saving of $5.7 m and $25.8 m and an extra 130 and 639 QALYs, respectively. Increasing the rate of home haemodialysis and peritoneal dialysis dominates current RRT. Conclusions: Increasing renal transplant rates and home haemodialysis and peritoneal dialysis rates for patients requiring RRT will produce a clear efficiency gain. 45 THE EFFECT OF CLINICAL DECISION SUPPORT SYSTEMS ON NURSE PERFORMANCE AND PATIENT OUTCOMES: A SYSTEMATIC REVIEW Dowding D,1,2 Randall R,1 Mitchell N,1 Cullum N,1 and Thompson C1 1Department of Health Sciences, University of York, York, United Kingdom; 2Hull York Medical School, Hull and York, United Kingdom Purpose: Nurses are increasingly taking on extended roles, often supported by clinical decision support software (CDSS). The aim of this review was to evaluate the evidence for the effects of CDSS on nurse performance and patient outcomes. Methods: Search: A number of electronic databases for research published between 1967-2005, together with hand searching of relevant journals and consultation with experts. Inclusion criteria: any controlled trial, before and after study or interrupted time series study that evaluated the use of CDSS in a clinical setting by a nurse and provided measurable outcomes. Results: 6885 references were identified through the electronic search and 49 via hand searching. After examination by two reviewers 20 articles, reporting 19 studies were included in the final review. Comparisons: nurses using CDSS versus nurses not using CDSS (3 studies), nurses using CDSS with other health professionals not using CDSS (5 studies), health professionals using CDSS with health professionals not using CDSS where nurses were a subgroup of the participants (11 studies). CDSS were evaluated in a variety of clinical settings, with mixed effects on both performance ABSTRACTS and patient outcomes. Some studies suggested use of CDSS improved performance and patient outcomes, others that it had no effect, and others that their use had a detrimental effect. Conclusions: CDSS vary in the nature of the decisions they support. The impact of nurses_ use of CDSS has been evaluated in a limited number of clinical areas. However, the results do highlight areas where CDSS could be beneficial. 46 CLINICAL PREDICTION RULE FOR RADIOGRAPHIC PNEUMONIA IN CHILDREN SUSPECTED OF RESPIRATORY TRACT INFECTION Roukema J,1 Steyerberg EW,2 van der Lei J,3 and Moll HA1 1Erasmus MC, Sophia Children_s Hospital, Rotterdam, the Netherlands; 2Erasmus MC, Department of Public Health, Center for Medical Decision Making, Rotterdam, the Netherlands; 3Erasmus MC, Department of Medical Informatics, Rotterdam, the Netherlands Purpose: Respiratory tract infections are frequent in childhood. Distinguishing pneumonia from less severe respiratory tract infections is difficult, and guidelines to diagnose/exclude pneumonia were found unreliable. A clinical prediction rule for radiographic pneumonia in children presenting with fever and respiratory tract symptoms was developed. Methods: From July 2003-March 2005, data were collected of all patients, 1 month to 16 years, presenting with fever and respiratory tract symptoms at an emergency department in the Netherlands. A multivariable logistic regression model was developed using clinical and laboratory predictors. The discriminative ability of the model was quantified using the area under the receiver operating characteristic curve (AUC), and corrected for optimism by bootstrapping. Results: 512 patients with fever and respiratory tract symptoms were included. 49 (9.6%) were diagnosed with pneumonia. Important predictors for the presence of pneumonia were: age (OR 1.1/year), GP-referral (OR 2.1), duration of the fever (OR 1.1/day), temperature (OR 1.6/ C), coughing (OR 5.2), and tachypnea (OR 4.5). The AUC was 0.79. Using a clinical score, derived from the multivariable model, the risk of pneumonia ranged from 1%-54%. White blood cell count per 109/L (OR 1.08), and C-reactive protein per 10 mg/L (OR 1.09), contributed significantly to the model (AUC 0.84). Using a clinical + lab score, the risk of pneumonia ranged from 0%-79%. Conclusions: Scoring rules, based on patient history, physical examination, and laboratory tests, can be used to estimate the risk of pneumonia. If further validated, the scores can become useful tools to support diagnostic and therapeutic decisions. 47 PREDICTION OF MSH2 AND MLH1 MUTATIONS IN LYNCH SYNDROME: DEVELOPMENT AND VALIDATION OF A WEBBASED MODEL TO SUPPORT DECISION MAKING Balmaña J,1 Stockwell DH,2,3 Steyerberg EW,1,4 Stoffel EM,1,2,3 Deffenbaugh AM,5 and Syngal S1,2,3 1Population Sciences Division, Dana-Farber Cancer Institute, Boston, MA; 2Division of Gastroenterology, Brigham and Women_s Hospital, Boston, MA; 3Harvard Medical School, Boston, MA; 4Center for Medical Decision Making, Department of Public Health, Erasmus MC, Rotterdam, the Netherlands; 5Myriad Genetic Laboratories Inc., Salt Lake City, UT Purpose: Lynch syndrome is caused primarily by mutations in the mismatch repair genes, MLH1 or MSH2. Our goal was to develop a clinical prediction rule to estimate the likelihood of finding a MSH2/MLH1 mutation in individuals who currently undergo genetic testing. Methods: Personal and family history were obtained for 1914 unrelated probands who submitted blood samples for genetic analysis of point mutations and large rearrangements of MLH1/MSH2. A multivariable model was developed using logistic regression in an initial cohort of 898 individuals and subsequently prospectively validated in 1016 patients. Results: Overall, 14.5% (130/898) of the probands carried a pathogenic mutation (MLH1 = 6.5%, MSH2 = 8.0%) in the development cohort and 15.3% (155/1016) in the validation cohort, 42 (27%) of the latter being large rearrangements. Strong predictors of mutations included proband characteristics (presence of colorectal cancer, especially two or more separate diagnoses, or endometrial cancer) and family history (especially the number of 1st degree relatives with colorectal or endometrial cancer). Age at diagnosis was especially important for colorectal cancer. The multivariable model discriminated well at external validation, with an area under the ROC curve of 0.80 [0.76-0.84], and was developed as a web-based tool. Conclusions: Personal and family history characteristics can accurately predict the outcome of genetic testing in a large population at risk for Lynch syndrome. The web-based model provides clinicians with an objective, easy-to-use tool to estimate the likelihood of finding mutations in the MLH1/MSH2 genes and to support decision making on molecular evaluation and genetic testing. 48 DIAGNOSTIC ACCURACY OF CORONARY PRESSURE MEASUREMENT IN CORONARY ARTERY DISEASE Bornschein B,1 Klauss V,2 and Siebert U1,3 1University of Health Sciences, Medical Informatics, and Technology, Hall, Tyrol, Austria; 2University of Munich, Munich, Germany; 3Harvard Medical School, Boston, MA Purpose: To summarize the diagnostic accuracy of coronary pressure-based fractional flow reserve (FFR) in patients with suspected coronary artery disease. Methods:We performed a systematic literature search in electronic databases (MEDLINE, EMBASE, Cochrane and HTA-databases) to identify published studies that reported test characteristics (sensitivity, specificity) of FFR compared with standard functional tests, for example, stress electrocardiography, stress echocardiography, SPECT, or combinations of these. We extracted diagnostic 2 × 2-tables and performed a diagnostic meta-analysis using the inverse variance approach and performed sensitivity analyses. Results: Thirteen articles met the inclusion criteria. Three articles were excluded because results have been reported in other articles. From the remaining studies, data from 717 observational units (i.e. patient or coronary lesion) were extracted. Pooled sensitivity of FFR was 82% (95%-CI: 77%-86%), and specificity was 79% (74%- 83%). Diagnostic odds ratio was 16.5 (11.4-63.7). These estimates were robust after excluding single studies. Sensitivity analyses showed severity of disease (1-vessel vs. multi-vessel disease) as influential factor on test sensitivity (1-vessel-disease: 95% [87%- 99%] vs. 78% [73%-83%] in multi-vessel-disease). Also, definition of reference standard influenced results. Studies using SPECT as reference standard showed lower sensitivity than studies using other reference standards (73% [73%-83%] vs. 95% [87%-99%]). Specificity again was not affected. Both groups, that is, studies with multi-vessel disease patients and studies using SPECT as reference standard tended to be of younger publication date. Conclusions: FFR is a functional test for with high sensitivity and specificity for the detection of hemodynamic relevant coronary lesions. However, time trends might affect test characteristics of individual studies. 49 THE EFFECT OF THE ANCHOR STATE ON TWO-STAGE STANDARD GAMBLE DERIVED UTILITIES FOR CERVICAL CANCER SCREENING Howard K, Salkeld G, McCaffery K, and Irwig L ABSTRACTS ABSTRACTS E13 E14 _ MEDICAL DECISION MAKING/JUL_AUG 2007 Screening and Test Evaluation Program (STEP), School of Public Health, University of Sydney, Sydney, Australia Purpose: With many theories of decision making under uncertainty, including expected utility theory, the utility of a prospect is assumed to be independent of other options in the choice set (context or menu independence). Empirical studies that have shown this assumption may not be accurate. Most of the experimental work in this area has been conducted using monetary prospects. This study tests whether the assumption of context independence holds in the context of health decisions made under uncertainty. Methods: One hundred and forty three women participated in interviews about cervical screening. The interview consisted of two-stage standard gamble valuations of two cervical screening states (repeat Pap testing and HPV testing). The choice set was altered by comparing each state against perfect health and two different _worst state_ anchors (Anchor state 1 [AS1]: hospitalisation after car accident, and Anchor state 2 [AS2]: cervical cancer). Results: If the assumption of context independence holds, utility scores for each state should be identical when calculated against AS1 and AS2. The overall ranking of cervical screening states was consistent with both AS; however, 32% of respondents had nonidentical utility valuations against the anchors for at least one of the states. Final utilities generated against AS1 were significantly lower than those generated against AS2 for all cervical screening states. Possible reasons for this difference are discussed. Conclusions: These data suggest the assumption of context independence may not always be appropriate, and argues against the existence of pre-determined, immutable preferences. Valuation of a health prospect can be context specific. 50 ADDING DISEASE SPECIFIC INFORMATION TO GENERIC HEALTH STATE DESCRIPTIONS CAUSES A SHIFT IN HEALTH STATE VALUATIONS BY A PANEL OF LAY PEOPLE Haagsma JA, Janssen MF, and Bonsel GJ Academic Medical Center, Department of Social Medicine, Amsterdam, the Netherlands Purpose: The objective of this study was to explore the effects of adding disease specific information to generic health state descriptions on health state valuations by a panel of lay people. Methods: 23 different health states of common diseases were presented for valuation. Each of these health states consisted of a EuroQ016D5L-profile and were presented with and without a disease label and a formalized clinical description of the disease. A panel of lay people (n = 105), recruited from the general public, valued the health states with the Visual Analog Scale (VAS). Results: For 19 health states, mean VAS values for health states with a disease specific information were numerically lower (_more severe_) compared to those without this information (p .001), with a mean difference of 0.10 (range: _0.08 to 0.28). The difference in VAS values was greater for less severe diseases. Standard deviations did not differ between health states with and without disease specific information, indicating similar precision. Conclusions: We found that the panel of lay people valued health states with disease specific information as more severe, especially for mild diseases. This indicates that disease specific health state descriptions contain information about prominent symptoms not reflected in the generic health state, consequently causing a valid shift. 51 PATIENT AND SOCIETAL VALUES AND UTILITIES FOR CANCER RELATED FATIGUE: WHOSE PREFERENCES SHOULD COUNT? Lloyd A and vam Hanswijck de Jonge P United BioSource Corporation, London, United Kingdom Purpose: Agencies such as NICE necessitate the use of societal preferences in cost-utility analyses. However, members of society have commonly not experienced more severe health states and so may in effect construct their preferences during the elicitation task. The present study compares values (elicited by visual analogue scale, VAS) and utilities elicited by standard gamble (SG) and time trade off (TTO) for health states related to cancer related fatigue. Methods: Existing clinical trial data from patients with cancer related fatigue (FACTAn) were summarized in order to define 7 health states related to different haemoglobin levels. Participants (general public n = 85; and oncology patients who had experienced anaemia n = 26) rated the health states using VAS and SG (public) and TTO (patients). Results: Societal and patient based values (from VAS) had a similar range. The difference in societal and patient derived VAS values between 7-8 g/dL Hb and 12+ g/dL Hb was 34.3 and 40.7, respectively. In contrast, the equivalent difference in utility values was 0.13 (societal) and (0.31) patients. Conclusions: The general public recognised differences in the health states (as evidence by the VAS scores) but did not value those differences to the same extent as patients. Patients gave much lower utility scores to more severe anaemia health states. The patients (who had all experienced cancer related fatigue) arguably had more complete preferences on which to base their judgements. The study also highlights the non-equivalence of VAS values and SG/TTO utilities. 52 QUALITY OF LIFE UTILITIES FOR HEALTH STATES IN PELVIC INFLAMMATORY DISEASE Smith KJ and Roberts MS Section of Decision Sciences and Clinical Systems Modeling, University of Pittsburgh School of Medicine, Pittsburgh, PA Purpose: Quality of life utilities for health states associated with pelvic inflammatory disease (PID) have not been directly measured, but could have important implications on the costeffectiveness of interventions to prevent and manage this disease. Methods: Using Impact 3.0, we obtained, in women with and without a history of PID, visual analogue scale (VAS) and timetradeoff (TTO) valuations for PID-associated health states: ambulatory PID treatment, hospital PID treatment, ectopic pregnancy, chronic pelvic pain, and infertility. Subjects read brief scenarios describing the medical, functional, and social activity effects typically associated with each state, then gave valuations in the order above. Results: We obtained data from 47 women with and 150 women without a PID history. PID history subjects were older (31.2 vs. 27.4) and more frequently nonwhite (74 vs. 52%). Mean utilities were similar in women with and without a history of PID for ambulatory treatment (VAS 0.72 vs. 0.70, TTO 0.89 vs. 0.87) and hospital treatment (VAS 0.61 vs. 0.60, TTO 0.80 vs. 0.84). The PID history group had significantly lower scores (p .05) in VAS for ectopic pregnancy (0.56 vs. 0.63), pelvic pain (0.45 vs. 0.52), and infertility (0.51 vs. 0.66) but not for TTO, although similar magnitude of differences between groups was observed. Significant VAS differences remained when controlling for age, race, and fertility status. Conclusion: PID has substantial influence on utility, with some PID-related health states preferred less by women who have experienced PID, which could have some impact on societal- vs. patient-perspective analyses of PID interventions. 53 TOWARD AN UNDERSTANDING OF MEDICATION NONADHERENCE IN THE ELDERLY WITH MULTIPLE ILLNESSES Elliott RA,1 Ross-Degnan D,2 Adams AS,2 Safran DG,3 and Soumerai SB2 1School of Pharmacy, University of Manchester, Manchester, United Kingdom; 2Department of Ambulatory Care and Prevention, ABSTRACTS Harvard Medical School and Harvard Pilgrim Healthcare, Boston, MA; 3The Health Institute, Boston, MA Purpose: This study explored patient decision-making to identify whether, and how, people with multiple illnesses _trade_ between medicines. Methods: Twenty insured community-dwelling seniors were interviewed. Interviewees were selected by gender; income; >3 medicines and >2 morbidities. In-depth qualitative interviews covered: knowledge and beliefs about the disease and medicines; influence of prescribers, _the system,_ media and family; and _trading_ behavior. Results: 12 women and 8 men were interviewed, aged 67-90, taking 4-12 drugs, with 3-9 comorbidities. People reported trading between medicines for different diseases; medicines for the same disease and between medicines and non-medicine health related behavior. All interviewees had made at least one trading decision in the past leading to adjusting dosing, swapping, or stopping a medicine. Most would consider trading one of their medicines over another in the future. There was general resistance to taking medicines, with minimal medicine-taking preferred, particularly with mental health medicines. The most common motivators to trade between medicines were symptom control, previous experience, fear about the future, side effects, beliefs about the illness and cost of medicines. Decision-making based on one motivator was much more common than decisions using multiple motivators. There was no dominant disease, medicine or motivator. Where individuals reported more than one trading decision, they did not use the same motivator. Conclusions: Community-dwelling seniors with multiple morbidities trade between medicines and use many decision mechanisms during trading. Specific decisions are generally driven by one motivator. Within one individual, adherence to one medicine does not predict adherence to other medicines. 54 MOTIVATED ERRORS IN KNOWLEDGE ABOUT HPV: PATIENTS MAY NOT WANT TO HEAR WHAT PHYSICIANS TELL THEM Hamm RM,1 Dailey SB,2 Boci M,1 and Scheid DC1 1University of Oklahoma Health Sciences Center, Norman, OK; 2University of Oklahoma, Norman, OK Purpose: Patients should be informed or educated by physicians. To assess patient knowledge, we often measure how many factual questions they answer correctly. Missed questions, attributed to ignorance, suggest physician neglect of patient education. But patient knowledge errors could reflect motivated ignorance, optimism or denial. We measured such optimism by judging whether erroneous responses are better or worse than correct responses. Thus, to deny _You can have HPV without knowing it_ is optimistic; to endorse _HPV causes HIV/AIDS_ is pessimistic. Methods: Computerized questionnaire administered to 270 women to whom physicians had given a DNA test for High Risk Human Papillomavirus (HRHPV). HRHPV test result, demographics, and numeracy were compared to accuracy of HRHPV knowledge, optimism of knowledge errors, and reported emotional reactions (positive and negative emotions, stigma). Optimism (+1) or pessimism (_1) of errors was measured on 38 T/F or Y/N questions. Results: 58.5% of HRHPV results were positive. Mean number of knowledge questions wrong was 19 of 38. Mean proportion of wrong answers _optimistic_ was 0.39 (a function of the question set). Statistically significant relations: The more accurate the responses, the less optimistic the errors (r = _0.63); Education, income, and numeracy predicted more accuracy and less error optimism; Those with HRHPV+ test had more accuracy (but it did not affect error optimism). The HRHPV+ and those with more accurate HRHPV knowledge had stronger emotional reactions, which in turn predicted less optimism. Conclusions: The optimism of knowledge errors about HRHPV is associated with SES and emotional reaction to HRHPV test. Present study did not measure what physician actually said, so we can not distinguish whether patients guess optimistically only if they don_t know, or distort their knowledge optimistically. 55 UNDERSTANDING PRENATAL SCREENING BEHAVIOUR van den Berg M,1 Timmermans DRM,1 Knol DL,2 van Eijk JThM,3 de Smit DJ,4 van Vugt JMG,5 and van der Wal G1 1EMGO, VU University Medical Center, Amsterdam, the Netherlands; 2Department of Clinical Epidemiology and Biostatistics, VU University Medical Center, Amsterdam, the Netherlands; 3Department of Medical Sociology, University of Maastricht, Maastricht, the Netherlands; 4I BV, Abcoude, the Netherlands; 5Department of Obstetrics and Gynecology, VU University Medical Center, Amsterdam, the Netherlands Purpose: The present study aimed to increase the understanding of prenatal screening decisions, by testing a hypothesized prenatal screening behaviour model that was based on health behaviour theory (theory of planned behaviour) as well as on previous findings. Methods: 1669 pregnant women who were offered prenatal screening for Down syndrome filled in a questionnaire. The questionnaire assessed perceived risk and perceived severity of having a child with Down syndrome, perceived efficacy of the screening test, attitude toward termination, attitude toward undergoing prenatal screening, subjective norm about having the test, childrelated anxiety, and intention to test. Results: Path analysis (using LISREL 8.3) showed moderate fit of the initial model. Three modifications resulted in a final model with reasonable model fit. The final model explained 68% of the variance. Attitude toward termination of pregnancy, and response efficacy determined a woman_s attitude toward having a prenatal test. Anxiety was influenced by perceived risk and severity of having a child with Down syndrome, and by perceived test efficacy. Pregnant women with a positive attitude and women who perceived a subjective norm to undergo prenatal screening, were more intended to have the test done. Anxiety appeared to be a weak predictor of intention to test. Conclusion: Path analyses revealed that attitude and subjective norm play a central role in the prenatal screening decision making process. Moreover, perceived risk and perceived severity appeared to play a marginal role in this decision. 56 RESPONSES TO HEALTH RISK INFORMATION: RISK TYPE, CONTROLLABILITY AND THE ROLE OF THE SELF Claassen L,1 Henneman L,1 de Vet HCW,1 Marteau TM,2 and Timmermans DRM1 1EMGO-Institute, VU Medical Centre, Amsterdam, the Netherlands; 2Psychology Genetics Research Group, King_s College, London, United Kingdom Purpose: The present study was conducted to help explain differences in responses to health risk information. Providing people with health risk information may induce preventive behaviour if a person beliefs that this behaviour will reduce the risk. However, people may also adopt fatalistic attitudes towards the risk especially when the threat is associated with a genetic susceptibility. Perceptions of controllability of health risks may also be linked with the way people see themselves; the self-concept. People with a static selfconcept (SSC) consider their personal qualities as fixed and process self-relevant information in a way that is consistent with this deterministic perspective. People with a dynamic self-concept (DSC) ABSTRACTS ABSTRACTS E15 E16 _ MEDICAL DECISION MAKING/JUL_AUG 2007 understand themselves in more flexible terms. It is likely that, when faced with a health threat, people with SSC are more susceptible to feelings of fatalism and show less preventive behaviour. Methods: Subjects were asked to imagine themselves in different health scenarios (representing lifestyle risk for cardiovascular diseases [CVD], risk based on a positive family history of CVD, and genetic risk for CVD). Next they completed a 27-item selfconcept questionnaire and other psychological scales. Results: Perceptions of controllability of CVD were lowest in the genetic risk scenario. The final, 7-item SSC-DSC-questionnaire showed good psychometric properties. People with SSC, compared, to people with DSC, perceived less control and showed a stronger preference for cholesterol-lowering drugs over lifestyle changes. Conclusion: The findings suggest that the SSC-DSC-questionnaire may be used in subsequent research on differences in responses to health risk information. 57 PATIENT SELF-MANAGEMENT OF ANTICOAGULATION: COMPARISON OF WITHIN-TRIAL AND ROUTINE PRACTICE COSTS Jowett S,1 Bryan S,1 Murray E,2 McCahon D,2 and Fitzmaurice D2 1Health Economics Facility, University of Birmingham, Birmingham, United Kingdom; 2Department of Primary Care and General Practice, University of Birmingham, Birmingham, United Kingdom Purpose: Patient self-management (PSM) is a means of enhancing patient choice and autonomy. In the management of patients receiving warfarin for atrial fibrillation, regular patient monitoring is undertaken in a primary or secondary care clinic. The development of near patient testing means PSM is now feasible and the first UK trial (the SMART trial) showed self-management to be both safe and reliable. Therefore the cost implications of PSM must be considered, but resource use under trial conditions may not reflect actual use by patients carrying out PSM routinely. Methods: A cost-effectiveness analysis was conducted alongside a randomised controlled trial involving 617 patients (SMART) comparing PSM with routine care. A subset of patients (n = 78) were followed up for a further 12 months in a case-control study and a cost analysis undertaken. Results: The economic evaluation alongside the trial revealed PSM had a higher mean health care cost per patient (£417 vs £122), with little difference in clinical outcomes or quality of life over 12 months. In the follow-up study, whilst PSM again had a higher mean cost per patient (£305 vs £118), the difference was smaller. Conclusions: The reduction might not change the decision in this case, but such a significant reduction in costs under non-trial conditions should alert both researchers and decision makers to the importance of assessing the impact of trial conditions on the results of evaluations. This paper explores the extent to which this problem is likely to arise in other contexts. 58 MODELLING THE-COST EFFECTIVENESS OF EPO FOR CANCER TREATMENT_INDUCED ANAEMIA Yao G, Raftery J, Miner A, Wilson J, and Hyde C University of Birmingham, Birmingham, United Kingdom; University of Southampton, United Kingdom; London School of Hygiene Southampton, Tropical Medicine, London, United Kingdom Purpose: Anaemia is a complication associated with chemotherapy. Whilst the use of erythropoietin (EPO) for the treatment of anaemia has been shown to decrease the risk of blood transfusion and increase haemoglobin (HB), its cost-effectiveness remains uncertain. To evaluate the cost-effectiveness of EPO for the treatment of anaemia associated with chemotherapy compared with best standard care from the NHS perspective. Methods: A simulation model, incorporating survival, quality adjusted life years (QALYs) and costs, was developed. Patients were characterized by their baseline HB levels. Input values were derived from systematic reviews. The model has a 3-year time frame. The base case was 6 months chemotherapy treatment followed by an offchemotherapy period. Results: The incremental cost-effectiveness ratio (ICER) was £119,247 per QALY gained with no survival gain to EPO. The ICER decreased to £31,119 when the most favourable survival gain was assumed. Results were sensitive to the length of timeframe and survival benefit of the treatment. Conclusions: EPO is effective in improving red blood cell transfusion requirements. Its impact on survival remains highly uncertain. If there is no impact on survival, it seems highly unlikely that EPO is a cost effective use of health care resources. SMDM 10TH BIENNIAL EUROPEAN MEETING PLENARY ABSTRACTS PLENARY 1 MODELLING TREATMENT EFFECTS IN DISCRETE-TIME MARKOV DATA FROM CLINICAL TRIALS OF ASTHMA Price M, Welton N, and Ades T HSRC and University of Bristol, Bristol, United Kingdom Purpose: To develop a structured approach to compare the relative treatment effects between the different arms of randomised controlled trials (RCTs) where data are reported in the form of a discrete-time Markov Model. The methods are then extended to allow a multiple treatment comparison (MTC) meta-analysis. Methods: We analysed data taken from five RCTs comparing eight different treatments for asthma. Here data are collected for subjects moving between four discrete health states with nine possible transitions. The state transition rates are modelled using log-linear regressions, and these transition rates are converted to transition probabilities using Kolmogorovs_ Forward Equations. Bayesian inferential techniques are used to generate posterior distributions for these transition probabilities. We sample from these distributions using Markov Chain Monte Carlo simulation techniques via WinBUGS and WBDiff. We considered modelling differing numbers of transitions, structured and unstructured baselines and numerous plausible treatment structures including: no treatment effect, all forward transitions only, all backward transitions only, transitions to treatment failure only, transitions from successful treatment only, and combinations of the above. Results: The preferred (best fitting) model included an unstructured (saturated) baseline and a structured treatment effect. The treatment model found to have the best fit (lowest Dic) was the all backwards (health improvement) transitions only structure. Conclusions: Modelling transition rates, rather than probabilities, allows the estimation of relative treatment effects in a generalisable form, Meta-analysis with MTC can then readily be performed on the data. PLENARY 2 MODELLING THE SPREAD OF PANDEMIC INFLUENZA AND OPTIONS FOR CONTROL Edmunds WJ, Vynnycky E, Cooper BS, Pitman RJ, and Gay NJ Modelling and Economics Unit, Health Protection Agency, London, United Kingdom Purpose: The emergence and spread of avian influenza has heightened concern that the next pandemic may be imminent. Methods: A range of transmission dynamic models were developed to address different aspects of control. The models were parameterised based on analyses of data from the previous pandemics of the 20th century (notably from 1957 due to the relative abundance of data on this pandemic). ABSTRACTS Results: Data from previous pandemics suggest that the main wave may last around 10-12 weeks and that 20%-40% of the population will become ill. The number and patterns of deaths was highly variable. Models suggest that a pandemic strain will spread very rapidly around the world, and that travel restrictions and entrance screening will have very little impact. Antivirals can help reduce the height of the epidemic peak, and reduce the overall clinical attack rate somewhat. Estimates of their impact on hospitalisations and deaths are highly speculative due to a lack of data. Pre-first wave vaccination programmes are likely to be more effective than vaccination programmes aimed at protecting people after the first wave, partly as there may only be one wave. Vaccination of children may be more effective than targeting other groups. School closure may have an impact (particularly in children), but there is a lack of any evidence for effectiveness of other measures to increase _social distance._ Conclusions: Models can and are being used to guide planning and decision-making regarding the next influenza pandemic. PLENARY 3 COST-EFFECTIVENESS OF SCREENING FOR THE EARLY DIAGNOSIS AND TREATMENT OF RISK FACTORS FOR CHRONIC KIDNEY DISEASE: A MARKOV MONTE CARLO MICROSIMULATION USING MARCK-E (MODEL FOR THE ASSESSMENT OF RISK FACTORS FOR CHRONIC KIDNEY DISEASE_ECONOMICS) Howard K,1 Salkeld G,1 White S,2 Chadban S,3,5 Craig J,1 Cass A,2,5 McDonald S,4 and Perkovic V2 1Screening and Test Evaluation Program (STEP), School of Public Health, University of Sydney, Sydney, Australia; 2The George Institute for International Health, Sydney, Australia; 3The Royal Prince Alfred Hospital, Sydney, Australia; 4Australia and New Zealand Dialysis and Transplant Registry (ANZDATA), Flinders University, Adelaide, Australia; 5Kidney Check Australia Taskforce, Kidney Health Australia, Sydney, Australia Purpose: Worldwide, increasing numbers of patients affected by chronic kidney disease (CKD) and subsequent end-stage kidney disease (ESKD) are leading to substantial health care costs. Diabetes, hypertension and proteinuria are key risk factors for CKD. Cost effectiveness of treatment for patients known to have diabetes, hypertension or proteinuria and of opportunistic screening for those risk factors in the Australian adult population is estimated. Methods:We developed a Markov Monte Carlo simulation model of development and progression of CKD and ESKD in TreeAge Pro (MARCK-E) in patients with diabetes, hypertension or proteinuria. It examined costs and effects (in LYS and QALYs) of: 1) improved management of existing patients with diabetes and/or hypertension; 2) intervention 1 plus annual opportunistic screening to detect and manage undiagnosed cases of diabetes, hypertension or proteinuria. Age-specific prevalence of risk factors and quality of life weights were calculated from the national AusDiab survey; transition probabilities, and intervention benefits were derived from the literature. Results: Median ICERs (and estimated 95% CI) for better glycaemic control in known diabetics (1), versus current practice were AUD $27,600/LYS (cost-saving to $500,000) and $30,600/QALY (cost-saving to $300,000). For screening and early treatment of diabetes, plus intervention (1) versus current practice, the ICERs were $1,200/LYS (cost-saving to $106,000) and for QALYS were costsaving (cost-saving to $120,000). Similar analyses for hypertension and proteinuria are under way. Conclusions: Analyses suggests that both better glycaemic management of existing diabetic patients and opportunistic screening for new diabetics, plus better management of existing patients may represent value for money. PLENARY 4 A META-ANALYSIS OF THE EFFECTS OF FRAMING TREATMENT BENEFITS IN DIFFERENT FORMATS Covey J Durham University, Durham, United Kingdom Purpose: To examine the effects of framing treatment benefits on the decisions of both patients and health professionals. Three formats were investigated: relative risk reductions (RR), absolute risk reductions (AR) and number needed to be treated (NN). Methods: A systematic review of the published literature was conducted. Articles were retrieved by searching a variety of databases and screened for inclusion by two reviewers. Data were extracted on characteristics of the subjects and methodologies used. Log-odds ratios were calculated to estimate the size of the framing effects. Results: A total of 25 articles were retrieved which reported on 32 unique experiments. The meta-analysis showed that treatments were evaluated more favourably when the RR frame was used rather than the AR or NN frame. However, a significant amount of heterogeneity was found between studies_the sources of which were explored using sub-group analyses and meta-regression techniques. These analyses demonstrated that framing effects were more pronounced in within-participant experiments and when the word _relative_ was not used in the RR frame. Also RR-NN framing effects were less pronounced in doctors than in other health professionals, medical students or patients. Conclusions: The published literature has consistently demonstrated that RR frames produce more favourable evaluations of treatments than AR or NN frames. However, the effects are heterogeneous and seem to be moderated by key differences between the methodologies used or types of subjects investigated. PLENARY 5 DO WOMEN SEEKING GENETIC COUNSELLING FOR HEREDITARY BREAST CANCER HAVE A PESSIMISTIC RISK PERCEPTION? Otten W,1 Van Dijk S,1 Van Asperen CJ,1 Meijers-Heijboer H,2 and Kievit J1 1Leiden University Medical Center, Leiden, the Netherlands; 2Erasmus University Medical Center, Rotterdam, the Netherlands Purpose: The purpose of the present study was to examine whether women who seek genetic counselling for hereditary breast and ovarian cancer (HBOC), assess their own risk different when they compare it to either (a) an average woman in the general population, or (b) another woman also seeking genetic counselling. In addition, we explored some medical and psychological factors related to these women_s comparative risk assessment. Methods: Before counselling, 620 women filled out a first questionnaire assessing their comparative risk estimates. Results: Results showed that these women consider their risk higher than that of a woman from the general population (i.e., a pessimistic response), but equal to the risk of another counselee. The latter comparative risk measure was also more normally distributed implying that comparing to a similar other person is a more relevant comparison. Women were more pessimistic comparing to another counselee if they had had breast cancer, and more optimistic if they had no first-degree relatives with breast cancer, perceived a low absolute risk, and had stronger dispositional optimistic tendencies. Conclusions: Overall, women who sought genetic counselling for breast cancer had unbiased comparative risk perceptions and correctly assessed their own risk status. ABSTRACTS ABSTRACTS E17 E18 _ MEDICAL DECISION MAKING/JUL_AUG 2007 PLENARY 6 THE IMPACT OF SHARED DECISION MAKING ON PATIENTS_ TREATMENT ACCEPTANCE, COMPLIANCE AND CLINICAL OUTCOME IN PRIMARY CARE OF DEPRESSION: A CLUSTER RANDOMIZED CONTROLLED TRIAL Simon D,1 Loh A,1 Wills CE,2 and Härter M1 1University Hospital of Freiburg, Freiburg, Germany; 2Michigan State University, East Lansing, MI Purpose: Depression is highly prevalent in primary care. The involvement of patients in the process of shared treatment decisionmaking is a central element to ensure high quality care consistent with patient preferences. Little is published to date on the impact of shared decision-making in primary care of depression. Methods: In a cluster randomised controlled trial evaluating the impact of a shared decision-making intervention for depression, 30 general practitioners documented their clinical management plans. 486 depressed patients rated their own participation in decisionmaking (Man-Son-Hing_s Scale), acceptance of diagnoses and treatment (5 point Likert-Scales) and mental health status (Brief PHQ). Baseline data for 227 patients were collected twice within 6 weeks. 20 physicians were randomly assigned to the intervention group and received physician training, patient information material and patient decision aids. 10 physicians were assigned to the control group performing usual care. Post-intervention data were collected again twice within 6 weeks enrolling 259 new depressive patients. Results: Patient participation (p = .000) and patients_ acceptance of treatment (p = .02) were significantly higher in the intervention group. Compliance was also higher in the intervention group but did not reach statistical significance (p = .08). Patients with higher participation rates showed more favourable clinical outcomes (p = .04). Conclusions: Depressed patients engaging in shared decisionmaking show higher acceptance of diagnosis and therapy, perform therapeutic tasks to a higher degree, and gain more clinical effects than patients not involved in decision making. These results support the usefulness of incorporating shared decisionmaking into treatment of depressed primary care patients. SMDM 10TH BIENNIAL EUROPEAN MEETING POSTER ABSTRACTS. 59 EUROQOL DATA IN TERMINALLYILL CANCER PATIENTS Van den Hout WB and Stiggelbout A Department of Medical Decision Making, Leiden University Medical Center, Leiden, the Netherlands Purpose: To investigate the alleged insensitivity of the EuroQol (EQ5D). Methods: Longitudinal EQ5D and visual analogue scale (VAS) data were obtained from two studies on palliative radiotherapy in patients with painful bone metastases (n = 1023) or non-small-cell lung cancer (n = 247). Group differences on the five EQ5D domains, three EQ5D tariffs (UK, Dutch and US), and the VAS were analyzed, by comparing the six-month area under the curve prior to death. Results: Over the last year, the UK tariff decreased by 0.68 (0.56 to 0.12), the Dutch tariff by 0.51 (0.62 to 0.11), the US tariff by 0.50 (0.68 to 0.18), and the VAS by 0.35 (0.55 to 0.20). The decrease accelerated towards death, especially due to self-care and usual activities. On all measures, men were better off than women (p = 0.047). Older patients (65+) were better off than younger patients on pain/discomfort (p .001), anxiety/depression (p = 0.046) and the Dutch tariff (p = 0.035). Longer survivors (>6 months) were better off than shorter survivors on usual activities (p = 0.014), pain/discomfort (p .001) and the EQ5D tariffs (p = 0.030). Patients in the bone metastases study were better off than patients in the lung study on the VAS (p = 0.033) but were worse off on all EQ5D domains (p = 0.007) except mobility (p = 0.435) and on the EQ5D tariffs (p .001). Conclusions: In our study the EuroQol proved sensitive to the approach of death and was able to distinguish most investigated subgroups. Moreover, it was consistently more sensitive to group differences than the VAS. 60 RECRUITMENT, PARTICIPATION AND COMPLIANCE OF AN INTERNET PANEL TO PROVIDE UTILITY ESTIMATES FOR COST UTILITY MODELLING IN HEALTH TECHNOLOGY ASSESSMENT Stein K,1 Dyer M,1 Round A,1 Milne R,2 Ratcliffe J,3 and Brazier J3 1Peninsula Technology Assessment Group, University of Exeter, Exeter, United Kingdom; 2Wessex Institute for Health Research and Development, University of Southampton, Southampton, United Kingdom; 3School of Health and Related Research, Sheffield, University of Sheffield, United Kingdom Purpose: To establish and train a panel of members of the public to provide utility estimates via the Internet to inform the specific needs of cost utility analyses being carried out as part of health technology assessments. Methods: A random sample of members of the general public in four UK cities (Exeter, Sheffield, Aberdeen and Glasgow) was obtained from the UK Electoral Register and invited to training sessions in each city. 106 health state descriptions were posted at intervals during 2004-6 and utility values obtained. Recruitment and the characteristics of the panel are described. Participation is defined as having provided any values and compliance as the proportion of potential values provided by participants. The determinants of participation and compliance are explored using univariate and multivariate (where appropriate) methods. Results: Only 5.4% of those approached expressed willingness to participate in the project. The need for training reduced recruitment further, giving an initial panel membership of 112. Recruitment was more successful in Exeter than the other cities, as was reporting of reasons for declining the initial invitation. 83 (74%) people participated in the panel. Socioeconomic and marital status were significantly associated with participation. Mean compliance was 45%. None of the potential explanatory variables were associated with compliance. Conclusions: It is feasible to establish an Internet panel for preference measurement but recruitment rates are very low. Oversampling according to ethnicity and socioeconomic status are required to ensure a demographically balanced participative panel. 61 AN INVESTIGATION OF THE CONSTRUCT VALIDITY OF HEALTH STATE VALUATION TECHNIQUES USED IN THE CALCULATION OF QUALITY ADJUSTED LIFE YEARS (QALYS) Robinson S, Bryan S, Kaambwa B, and Freeman T University of Birmingham, Birmingham, United Kingdom Purpose: Health state valuations are a key component in calculating quality adjusted life years (QALYs), which are increasingly being used in the process of priority setting and resource allocation. There are a number of different techniques which can be used to elicit valuations for health states, including the time trade off (TTO) and person trade off (PTO) techniques. Studies have suggested that using different techniques may elicit different valuations for the same health state; very few studies have explored the validity of the PTO technique. If different techniques do produce different results, then this could have an impact on the QALY and thus the resource allocation decision reached. The purpose of this paper is to consider the ABSTRACTS construct validity (i.e. association between the different techniques) of the PTO and TTO techniques. Methods: All participants were asked to value various health states using PTO and TTO. Bland and Altman test of agreement was used to explore the convergent validity of the techniques. Regression analysis was also undertaken to further explore differences in utility weights. Results: The mean TTO and PTO utility weights were similar between health states. However, the Bland and Altman plots show considerable disagreement between the techniques in terms of the absolute values they measure. The regression analysis suggested age profession and whether respondents had children had significant effects on valuations. Conclusions: The study raises concerns over the convergent validity of the health state valuation techniques and demonstrates the need for further exploration around the validity of such measures. 62 A QUALITATIVE INVESTIGATION TO EXPLORE THE VALIDITY OF THE VISUAL ANALOGUE SCALE, TIME TRADE OFF AND PERSON TRADE OFF TECHNIQUES Robinson S, Freeman T, and Bryan S University of Birmingham, Birmingham, United Kingdom Purpose: Preference elicitation techniques, such as the time trade off (TTO) and person trade off (PTO), are a key component in the calculation of quality adjusted life years, with different techniques often yielding different results. When employing these methods economists generally assume that individuals have fully formed, highly articulated preferences which can be applied to any form of decision making. However, there is suggestion that rather than having well articulated preferences, individuals use heuristics (cognitive shortcuts) to simplify the decision making process. This study attempts to gain a greater understanding of the cognitive process respondents use when undertaking a preference elicitation exercises. Methods: Semi-structured interviews were undertaken with respondents who had participated in an earlier preference elicitation exercise, using the VAS, TTO and PTO. The interviews explored a number of areas that are relevant to assessing the validity of the instrument and process of elicitation. Results: Respondents generally found PTO was the most difficult of the three exercises. They experienced more difficulty valuing the lives of others, as in PTO, rather than valuing time for themselves, that is, life years in TTO. There was evidence that some respondents were employing heuristics in order to simplify the decision making process. Conclusions: The study further reaffirmed the theory that valuing hypothetical health states is a complex task. If as the findings suggest respondents are employing decision making heuristics, then this may help explain why different preference elicitation techniques often produce different results. It also raises concerns around the validity of the instruments. 63 TEST-RETEST RELIABILITY OF TIME TRADE OFF AND PERSON TRADE OFF: A REVIEW AND EMPIRICAL INVESTIGATION Robinson S, Bryan S, and Freeman T University of Birmingham, Birmingham, United Kingdom Objectives: Economic analysis is increasingly being employed in formal resource allocation decision-making processes in health care. The consequence is that the methods being employed by economic analysts are increasingly subject to close scrutiny. One such area of methodology concerns the instruments used to elicit preferences for various health states for use in the construction of quality-adjusted life years (QALYs). The objective of this study is to explore the test-retest reliability of two techniques: time trade off (TTO) and person trade off (PTO). Methods: The study conducts a review of the literature on testretest reliability of health state valuation techniques. It also reports the results of an empirical study which analysed the test-retest reliability of both TTO and PTO valuations collected by a general population postal survey. Results: A total of 798 respondents returned questionnaires. The intra class correlation coefficients ranged from 0.32-0.84 for TTO and 0.17-0.82 for PTO, with the majority of coefficients being >0.50. The health states with low coefficients varied between techniques. The reliability coefficients varied between techniques and health states, with the TTO technique tending to produce higher coefficients. Conclusions: The reliability study reported here represents the first empirical investigation of the test-retest reliability of the PTO technique and the first reliability study to use a postal method of administration for both TTO and PTO. The study reports positive results for the reliability of the TTO technique. The reliability of the PTO technique is somewhat less clear with many individuals seeming unwilling to trade. 64 UTILITY ASSOCIATED WITH HEALTH STATES IN CHRONIC RENAL DISEASE AND END-STAGE RENAL DISEASE: RESULTS OF A SYSTEMATIC REVIEW Dale PL,1 Hutton J,1 and Elgazzar H2 1United BioSource Corporation, London, United Kingdom; 2Genzyme, Oxford, United Kingdom Purpose: To document published utilities for health states associated with CKD and ESRD, and determine the most appropriate values for use in economic models to evaluate renal treatments in the UK. Methods: A systematic review was conducted (PUBMED, NHS EED) in order to identify relevant articles published between January 1990 and December 2005. In addition, searches were conducted on websites of HTA organizations (NICE, CCOHTA, SBU) and the Harvard CEA registry. Articles were reviewed (editorials, letters and discussions were excluded) and those not containing utilities excluded. Authors were contacted when this information was unclear in the article. Results were assessed on the scientific quality of the elicitation studies; their relevance to the UK HTA environment; and on coverage of the states in the proposed model. Results: 68 articles were identified of which 32 were included. The largest numbers of studies were found in Canada (13), the US (10), and the Netherlands (4). Utilities were identified for all/most of the necessary states, but scores for individual states differed widely between studies and very few studies met our quality criteria. Only one study was judged to be directly relevant to the UK HTA environment. Conclusion: No single study provided a set of utilities meeting the criteria of quality, relevance and comprehensiveness. A set of UK-relevant utilities can be assembled from several sources but must be used with caution. Further empirical research is needed to produce more reliable utilities for economic modelling in the UK, especially in chronic renal disease. 65 THE IMPACT OF POSITIVE ASPECTS ON THE ASSIGNMENT OF UTILITIES TO HEALTH STATES De Vogel-Voogt E and Stiggelbout AM Department of Medical Decision Making, Leiden University Medical Center, Leiden, the Netherlands Purpose: In health care decision-making, health state utilities play an important role. In general, members of the public assign lower ABSTRACTS ABSTRACTS E19 E20 _ MEDICAL DECISION MAKING/JUL_AUG 2007 utilities to health states than do patients. We hypothesize that positive aspects of patients_ lives play an important role in the utilities that patients assign to their health, utilities that are therefore only moderately correlated with health-related quality of life (HRQL). Methods: Patients filled out a Visual Analogue Scale (VAS) for their quality of life and a time trade off (TTO) for their health. The Schedule for Evaluation of Individualized Quality of Life-Direct Weighting (SEIQoL-DW) was used to measure the importance of and functioning on dimensions of quality of life that are most important to the patient. Recently, we started our data collection, and by April 2006, we will have interviewed 25 patients with rheumatoid arthritis (RA) and 25 patients with breast, rectal or prostate cancer. Results: Presently, data of 10 patients with RA are available. The Pearson_s correlation between the VAS and the TTO was 0.38. Patients varied in their levels of individualized quality of life (SEIQoL-DW: Mean, 61.3; SD, 20.8). On average, 2.4 out of 5 domains concerned positive aspects (SD, 1.6). During the conference, we aim to present data about the importance of and functioning on positive domains of life, and we expect that these data may give insight in the low correlation between the VAS and the TTO. Conclusion: Our conclusion may be that when patients assign utilities to their health state, they evaluate their life instead of their health. 66 COMPARISON OF TWO BUDGET IMPACT MODELLING APPROACHES: IMPACT OF A NEW PHARMACEUTICAL PRODUCT ON ANNUAL HEALTH CARE EXPENDITURES AND THE IMPACT ON EXPENDITURES AND INSURANCE REVENUE (PREMIUMS) Walzer S, Morlotti L, and Mueller E Analytica International, Loerrach, Germany Objectives: To contribute to current discussions about budget impact modelling two different approaches for the impact of a new pharmaceutical product were analysed: firstly considering the impact on annual health care expenditures only and secondly additional inclusion of lost insurance premiums due to possible early retirement in patients with chronic diseases. Methods: The dynamic model calculates the budget impact from two different perspectives: a) the impact on health care expenditures and b) on expenditures as well as on health insurance revenues due to premiums. The latter approach could especially be useful for patients with chronic diseases who have higher probabilities of early retirement. Early retirement rates and indirect costs were derived from published data (Germany, Switzerland). Cut-off point for retirement was assumed to be 65 years. Health Care premiums were calculated based on an average premium and a mean income. Epidemiological input data were obtained from literature. Time horizon was 10 years. Results: Results in terms of reimbursement decisions of the budget impact analysis varied depending on the assumptions made for the insurance premiums, costs and early retirement rate. Sensitivity analyses revealed that in extreme cases the decision for accepting a new pharmaceutical product would probably be negative using approach a, but positive using approach b. Conclusions: Depending on the disease and population of interest in a budget impact analysis not only the health care expenditures for a health insurance have to be considered but also the revenue side for an insurance due to retirement should be included. 67 DEVELOPING OUTCOME MEASURES FOR GENETICS SERVICES: A DELPHI SURVEY Payne K,1,3 Nicholls S,1,3 McAllister M,1,2 MacLeod R,2,3 Donnai D,1,2,3 and Davies LM3 1The North West Genetics Knowledge Park, Manchester, United Kingdom; 2St. Mary_s Hospital, Manchester, United Kingdom; 3The University of Manchester, Manchester, United Kingdom Purpose: To explore which of the existing outcome measures healthcare professionals believe are most appropriate to value genetics services (GS). Methods: A Delphi survey was posted to all known consultant geneticists (n = 114) and a sample of genetic counsellors (n = 160) in hospitals (England). The survey contained 19 measures identified from a systematic review of validated outcome measures used to evaluate GS. Respondents assessed the usefulness of each outcome as a measure of patient benefit on a rating-scale (1 = strongly disagree to 7 = strongly agree). Free-text comments on each measure were collected. Descriptive statistics summarised the extent of consensus and qualitative data were analysed by the constant comparative method. Results: 131 (64 geneticists and 67 counsellors: 48% response) respondents completed the round one survey. The top three outcomes rated as _useful_ (scored 5, 6 or 7) were: ability to make informed decisions (90% of respondents); knowledge about the genetic condition (85%); satisfaction with service (84%). Rate of terminated pregnancies was rated not useful by 82% of respondents (score 1, 2 or 3). Views about other measures, such as anxiety levels (38% useful; 42% not useful), varied. The comments identified a number of emerging themes: relevance of the measure to GS; impact of good or bad news on the outcome and timing of outcome measurement. A second round is planned for February 2006. Conclusions: This study has proved a useful first step to assess the appropriateness of existing outcome measures for GS and inform the development of a core set of outcomes for evaluating GS. 68 ELICITING (MEDICAL) EXPERT OPINION ABOUT UNCERTAINTY USING PROBABILITIES Jenkinson DJ and Garthwaite PH Open University, Milton Keynes, United Kingdom Purpose: Decision making must often proceed despite uncertainty about important quantities. Bayesian statisticians argue that the only correct way to handle uncertainty is to use probability. Probabilities frequently need to be assessed when extensive data analysis is not possible (e.g., due to a lack of time or data). It then becomes important to use the intrinsic knowledge of experts to gain quantifications of the uncertainty, using their personal (subjective) probabilities. We report two experiments from a multi-centre collaboration (for the UK NHS) of statisticians, psychologists and health economists on the elicitation of such probabilities from doctors. Methods: Experiment 1 tested different approaches to eliciting uncertainty about means and variances (in particular, looking at the use of graphs). These methods were used to elicit opinions about how four variables would effect the survival time and side effects post treatment of prostate cancer patients. Experiment 2 tested different methods for eliciting correlations, eliciting opinions about the relationship between birth weight, gestation period and the mother_s age for the first-born of a pair of twins. Each experiment involved gathering a group of doctors together and them filling out a questionnaire. Results: The coherence and calibration of doctors_ opinions about uncertain quantities imply that selecting the best method with which to elicit beliefs about means, variances and correlations is far from trivial. Conclusions: These methods have wide applicability, e.g., informing clinical trials and contributing to cost effectiveness analyses. We also discuss some software we have written to elicit opinions using a graphical interface. ABSTRACTS 69 FRAUD AND PUBLICATION BIAS: THE ONLY REASONS FOR FUNNEL PLOT ASYMMETRY? Wisløff T, Myhre KI, and Norderhaug IN Norwegian Knowledge Centre for the Health Services, Oslo, Norway Purpose: We wanted to explore the reasons for funnel plot asymmetry, and the relations to publication bias. We also wanted to relate these aspects to the meta-analyses conducted alongside HTA reports from our centre. Methods: We searched Medline for literature regarding publication bias and funnel plot asymmetry. Literature was divided into whether it was concerning possible reasons for funnel plot asymmetry or more directly concerning publication bias. We also related the literature to meta-analyses performed in HTA reports from our centre. Results: Even though a funnel plot is clearly asymmetric, we cannot be sure of publication bias. We found several possible reasons for funnel plot asymmetry. They can be divided into the following four main groups; selection bias, true heterogeneity, data irregularities and artefactual. We also found articles which systematically has explored different aspects of publication bias, for instance that positive and significant results are published more often than non-positive and non-significant ones. In our own meta-analyses of passive smoking, we found obvious funnel plot asymmetry which was clearly significant (Egger publication bias statistic). The asymmetry, however, seems to disappear when the trials of lowest quality are excluded. Conclusions: Publication bias is not the only possible reason for funnel plot asymmetry. Hence it is important not to draw the conclusion that there exists publication bias if a funnel plot shows asymmetry. However, there are several examples from the literature that publication bias exists. 70 SCREENING FOR OPEN ANGLE GLAUCOMA: IS IT COSTEFFECTIVE? Hernández R1, Kannaiyan R,2 Vale L,1,2 Burr JM,2 for the OAG Screening Project Group 1Health Economics Research Unit, University of Aberdeen, Aberdeen, United Kingdom; 2Health Services Research Unit, University of Aberdeen, Aberdeen, United Kingdom Purpose: Open angle glaucoma (OAG) affects 1.2% of adults in the UK and is a leading cause of blindness in the developed world. Currently OAG cases are detected during routine sight tests performed to assess the need for spectacles. This opportunistic approach fails to identify many of those with OAG. This study compared the cost-effectiveness of different strategies for screening for OAG. Methods: Markov models were developed describing the pathways of care for a cohort of people at risk of OAG for current practice, and for two alternative screening strategies, varying in how screening was organised (i.e. invitation for a screening examination by a glaucoma trained optometrist or for a simple test assessing either visual field loss or nerve fibre layer loss together with a measurement of the intraocular pressure by a technician). The models estimated the lifetime cost and benefits of patients. All strategies allowed for the progression of glaucoma. This progression was modified according to how quickly cases of OAG were detected (and treated). Model input estimates were derived from a series of systematic reviews conducted explicitly for the purpose. Results were expressed in terms of incremental cost per case detected and incremental cost per QALY. Deterministic and probabilistic sensitivity analyses were performed. Results: Preliminary results show that screening strategies are unlikely to be cost-effective. Identifying people at greater risk, for example, siblings of people with OAG might make screening strategies worthwhile. Conclusions: Screening for OAG seems unlikely to be cost effective for the UK general population. 71 ROBUST-SATISFICING DECISIONS UNDER SEVERE UNCERTAINTY Ben-Haim Y1 and Dacso CC2 1Technion_Israel Institute of Technology, Haifa, Israel; 2Methodist Hospital Research Institute and University of Houston, Houston, TX Purpose: To test the feasibility of robust-satisficing decision strategies in patients_ management of chronic illness. Methods: Chronic illness is a severe test for patient-focused decision making. Typical decisions in chronic illness must deal with rare events of uncertain probability. This study explores the use of robust-satisficing decision strategies by patients who are faced with low-probability events, or with incompletely denied events with ambiguous uncertainties. Type II diabetes mellitus is used to illustrate the chronic illness patient_s dilemma. A webbased application is constructed to test the application of robustsatisficing decision strategies, including info-gap theory (IGT). Results: Robust-satisficing provides reliable and adequate (though possibly sub-optimal) outcomes. We show Info-Gap Theory (IGT) informing patients_ choices in diabetes management. IGT is particularly useful for decision making under the severe uncertainty facing a patient or family in choosing among mutually exclusive options. Although there exist broad-ranging epidemiologic and outcome studies for diabetes management, the patient and his/her family are always uncertain about how the data apply to them and about how satisfactory (or not) the unfamiliar potential outcomes would be. IGT approaches this problem by maximizing the robustness to uncertainty in the decision, and satisficing (rather than maximizing) the quality of the outcome. Conclusion: IGT and robust-satisficing are valid strategies for patient-focused decision making. Further research is needed to test their application in clinical practice. 72 PREDICTION OF INDOLENT PROSTATE CANCER: VALIDATION AND UPDATING OF A PROGNOSTIC NOMOGRAM TO SUPPORT DECISION MAKING ON SURGERY Steyerberg EW,1 Roobol MJ,2 Kattan MW,3 and Schröder FH2 Departments of 1Public Health and 2Urology, Erasmus MC_University Medical Center Rotterdam, Rotterdam, the Netherlands; 3Department of Quantitative Health Sciences, Cleveland Clinic Foundation, Cleveland, OH Purpose: Screening with serum prostate specific antigen (PSA) testing leads to the detection of many prostate cancers early in their natural history. Statistical models may predict the presence of _indolent cancer,_ which do not require surgical resection. We aimed to validate and update model predictions for a screening setting. Patients and Methods: We selected 247 clinical stage T1C or T2A patients from the European Randomized Study on Screening for Prostate Cancer who were treated with radical prostatectomy. We validated a nomogram, which was previously developed in a clinical setting. Predictive characteristics included serum PSA, ultrasound prostate volume, clinical stage, prostate biopsy Gleason grades 1 and 2, and length of cancer and non-cancer tissue in biopsy cores. Indolent cancer was defined as organ confined cancer _0.5 cc in volume and without poorly differentiated elements. Logistic regression was used to update the previous model and to examine the contribution of other potential predictors. Results: Overall 121 (49%) of 247 patients had indolent cancer, while the average predicted probability was around 20% (p 0.001). Effects of individual variables were similar to those found ABSTRACTS ABSTRACTS E21 E22 _ MEDICAL DECISION MAKING/JUL_AUG 2007 before, and discriminative ability was adequate (area under the receiver operating characteristic curve 0.76). An updated model was constructed which merely recalibrated the nomogram, and did not apply additional predictors. Conclusions: Prostate cancers identified in a screening setting have a substantially higher likelihood of being indolent than predicted by a previously proposed nomogram. An updated model can, however, support patients and clinicians in their decision-making on treatment options for prostate cancer. 73 A COMBINED MODEL ENHANCES ACCURACY OF CANCER DIAGNOSIS Wang D and Hyde C Department of Public Health and Epidemiology, Birmingham University, Edgbaston, Birmingham, United Kingdom Purpose: Cancer is one of the most common diseases in the world. Treatment works better the earlier it is detected. It is extremely valuable to design reliable methods of detecting early cancer so that people with cancer could be screened. We designed a combined model using artificial neural network (ANN) and random forest (RF) machine learning techniques to diagnose ovarian cancer. Methods: A total of 253 samples from 162 ovarian cancer patients and 91 normal subjects were used in this study. A set of 68 variables was identified from 15,154 variables and used to construct ANN and RF models. An independent test dataset (40 ovarian cancer patients and 23 control subjects) was randomly selected from the dataset. The remaining was randomly partitioned into 10 sub-training and 10 sub-validation sets. ANN and RF individual models were trained and validated using the subtraining and sub-validation datasets. Cancer diagnose was performed using a combined model. The performance of the model was assessed using the independent dataset. Results: The RF model gave a sensitivity of 97.5%, specificity of 82.6% and positive prediction value of 90.7%. The ANN model gave a sensitivity of 100%, specificity of 87.0%, and positive prediction value of 93.0%. The combined meta-model gave a sensitivity of 97.5%, specificity of 91.3% and positive prediction value of 95.1%. Conclusion: The combined model enhanced accuracy of cancer diagnosis, increased specificity from 82.6% to 91.3%, and positive prediction value from 93.0% to 95.1%. The proposed method can be generalised to other diseases and other types of data. 74 DEVELOPMENT AND FIRST VALIDATION OF THE SHARED DECISION-MAKING QUESTIONNAIRE (SDM-Q) Simon D,1 Schorr G,1 Wirtz M,2 Vodermaier A,3 Caspari C,3 Neuner B,4 Spies C,4 Krones T,5 Keller H,5 Edwards A,6 Loh A,1 and Härter M1 1University Hospital of Freiburg, Freiburg, Germany; 2University of Freiburg, Freiburg, Germany; 3University Hospital of Munich, Munich, Germany; 4University Hospital of Berlin, Berlin, Germany; 5University of Marburg, Marburg, Germany; 6Cardiff University, Cardiff, United Kingdom Purpose: Due to a lack of valid German instruments measuring shared decision making (SDM), a theory driven questionnaire (SDM-Q) to measure the process of SDM was developed and validated. Methods: As a theoretical basis steps of the SDM-process were defined in an expert panel. Item formulation was conducted according to the Delphi method. For the first preliminary validation on a mixed sample, Rasch analysis was used to eliminate items not fitting the underlying construct thus receiving a unidimensional scale. Results: After eliminating 4 items, the remaining 11 form a unidimensional scale with an acceptable reliability for person measures (.77) and very good reliability for item difficulties (.95). Analysis of subgroups revealed a different use of items in different conditions. Furthermore, the scale showed high ceiling effects. Conclusions: A new theory-driven instrument to measure the process of SDM has been developed and validated by use of a rigorous method revealing first promising results. Yet the ceiling effects require the addition of more discriminating items, and the different use of items in different conditions demands an in-depth analysis. While the concept of SDM is constantly receiving more attention in medical practice, its valid and reliable measurement remains challenging. Key words: shared decisionmaking; questionnaire; Rasch analysis 75 THE ASSESSMENT OF DEPRESSIVE PATIENTS_ INVOLVEMENT IN DECISION MAKING IN AUDIOTAPED PRIMARY CARE CONSULTATIONS Simon D1, Loh A,1 Hennig K,1 Hennig B,1 Härter M,1 and Elwyn G2 1University Hospital of Freiburg, Freiburg, Germany; 2Cardiff University, Cardiff, United Kingdom Purpose: In primary care of depression, treatment options such as antidepressants, counseling and psychotherapy are reasonable. Patient involvement could foster adherence and clinical outcome. However, there is a lack of empirical information about the extent to which general practitioners involve patients in decision making processes in this condition, and about the consultation time spent for distinct decision making tasks. Methods: 20 general practice consultations with depressive patients prior to a treatment decision were audio-taped and transcribed. Patient involvement in decision making was assessed with the OPTION scale and durations of decision-making stages were measured. Results: Mean duration of consultations was 16 minutes, 6 seconds. The inter-rater concordance of the OPTION-ratings was 67%, the Cohen Kappa coefficient 0.46 and the interrater intraclass coefficient (ICC) was 0.74. The means of the OPTION-items were between 0.0 and 26.9, in a scale range from 0 to 100. Total OPTION scores per consultation were between 4.2 and 34.4 (0- 100) and the mean per practitioner 14.6. Overall, 78.6% of the consultation time was spent for the step _problem definition_ (12 minutes, 42 seconds). Conclusions: Very low levels of patient involvement in medical decisions were observed in consultations about depression. Physicians used the majority of their time for the definition of the patient_s medical problem. To improve treatment decision making in this condition, general practitioners should enhance their decision-making competences and be more aware of the time spent in each decision-making stage. Key words: shared decision making; physician-patient communication; consultation time; depression 76 INVOLVING YOUNG PEOPLE WITH END STAGE RENAL FAILURE IN TREATMENT CHOICES: AN INTERVIEW STUDY Coakes JA,1 Youngson S,1 Fitzpatrick M,2 and Bekker HL1 1University of Leeds, Leeds; 2St. James University Hospital, Leeds, United Kingdom Purpose: Policy recommends young people with end stage renal failure (ESRF) participate in their treatment choices. Unlike adults, young people with ESRF are offered kidney transplant (cadaver or living-related donor) as an initial treatment option, as well as haemo and peritoneal dialysis. There is little evidence exploring young people_s views about their kidney disease, treatment options and involvement in decision making. This study describes young people_s beliefs about their disease and experiences of decision making about treatment. ABSTRACTS Methods: An interview study of young people offered a kidney transplant in a regional paediatric nephrology service (n = 36); 12 participated. Semi-structured interviews with young people in their homes. Interviews were audio-tape recorded, transcribed and analysed using thematic content analysis. Results: Current treatments were 9 cadaveric transplant, 2 LRD transplant, 1 peritoneal dialysis. Mean age 14 years; 7 males; 5 had kidney disease from birth; time since transplant 1-10 years. Participants expressed strong negative emotions when describing their illness including loss of normality, control, and personal potential, and fear about treatment failing or resulting in changes. The young people had limited understanding about the different renal replacement treatment options, rarely reported seeking information, and did not perceive there to be an active role for them in decision making about treatment. The young people felt advantages and disadvantages of cadaveric and LRD transplants were similar. Conclusion: Young peoples_ awareness of their treatment and disease is poor. Materials need to be developed to support staff and parents in engaging their children to manage their disease. 77 LEARNING FROM INCIDENTS IN RADIATION TREATMENT Cooke D,1 Dubetz M,2 Ekaette E,1 Heshmati R,2 Iftody S,2 McKimmon E,2 Powers J,2 Lee R,1 and Dunscombe P1,2 1University of Calgary, Calgary; 2Tom Baker Cancer Centre, Calgary, Alberta, Canada Purpose: The purpose is to design and implement a feedback system to enhance organizational learning from the normal incidents that occur in the radiation treatment (RT) program at a major academic cancer centre. The premise is that an improved system for processing incident information will allow better decisions to be made. Methods: The methodological approach employs a natural experiment to compare incident data, survey data and other safety metrics for a period before system implementation to similar data post implementation. There is an action research component since the researchers are interacting with RT program managers in the design of the new system. Results: The new incident learning system has been designed and documented. The system design, which has undergone independent peer review, is being implemented. A survey tool was developed and baseline climate data collected. The baseline survey data has been analyzed and documented. The results show that the safety climate in the RT program is generally positive and supports the premise that a significant number of incidents are experienced but not reported in the existing system. Conclusions: The survey results suggest that there is an opportunity to greatly increase the reporting of incidents in the RT program by implementing an incident learning system. The improved system is designed to collect incident data and process information gathered during incident investigations into better decisions for RT program improvement. 79 VISUAL INTERACTIVE MODELLING: TOOLS FOR IMPROVED UNDERSTANDING IN HEALTH TECHNOLOGY ASSESSMENT Pitt MA and Stein K The Peninsula Medical School, Universities of Exeter and Plymouth, Exeter and Plymouth, United Kingdom Purpose: Despite the growing sophistication of modelling approaches in HTA, little work to date has been directed to investigate visualisation and interactive techniques in the presentation and control of models. This presentation outlines experiments in information visualisation designed to improve the accessibility and control for a Markov model within a case study example of HTA. Methods: Experimental visualisation tools were developed for an HTA Markov model of surveillance for Barretts Oesophagus. These included critical event counters, graphical slider controls for interactive sensitivity analysis, and a model animation incorporating interactive and graphical controls to observe the effects of parameter changes during simulation trials. Results: Visual outputs and controls are demonstrated which provide enhanced user understanding, model verification and validation, sensitivity analysis, and improved presentation of outputs. These tools also provide a context for group discussion and the exploration of model dynamics. Conclusions: There is considerable scope for the application of visualisation techniques for Markov models in HTA modelling to enhance their role in decision support. In other areas, such as discrete event simulation, visual interactive approaches are accepted as central in promoting understanding and acceptance and there is a developed body of research in information visualisation which can be applied to HTA. A full investigation, however, needs a methodological frame of reference which identifies key user groups in the decision process, analyzes specific requirements and maps these to appropriate visualisation applications. Such research needs to incorporate structured evaluation to assess the potential benefits of these tools. 80 STATISTICAL MODELLING OF CLINICAL TRIAL DATA FOR ECONOMIC APPRAISAL: AN ALTERNATIVE TO THE STANDARD APPROACH? Briggs A University of Glasgow, Glasgow, United Kingdom Purpose: To explore an alternative method of analysing and presenting trial-based economic appraisals. Methods: A recent ISPOR task force has set out guidance for conducting economic appraisals alongside clinical trials. The guidance reinforces what has become an accepted _standard_ of analysing the results of trial-based economic appraisal. An alternative approach to analysing economic data collected alongside clinical trials is explored based on statistical relationships in the data. This approach is contrasted with the standard methods using a recently published example. Results: The standard approach to economic analysis yields costeffectiveness point estimates that appear to be highly cost-effective (less than £10,000/QALY). However, considerable uncertainty is evident with confidence intervals overlapping £30,000 per QALY. By contrast, the statistical modelling approach yields slightly less favourable point estimates, reflecting the cautious approach to the analysis. Nevertheless, the estimated statistical uncertainty is less and yields confidence intervals that have upper limits below £30,000/QALY. Conclusions: Standard approaches to the analysis of economic data alongside clinical trials tend to yield cost-effectiveness estimates that are highly uncertain due to the typical underpowering of trials in relation to secondary outcomes such as cost and quality of life. More formal statistical modelling of economic data may offer increased power and reduced uncertainty subject to the validity of the underlying assumptions. On the basis of the single example presented here, we can only conclude that there may be potential merit in adopting such an approach_particularly when the validity of the randomised comparison on the primary clinical outcome is retained. 81 WITHIN-TRIAL COST-EFFECTIVENESS ANALYSIS WITH CENSORED DATA: A COMPARISON OF PARAMETRIC AND BOOTSTRAP APPROACHES Epstein D Centre for Health Economics, York, United Kingdom ABSTRACTS ABSTRACTS E23 E24 _ MEDICAL DECISION MAKING/JUL_AUG 2007 Purpose: To compare a parametric and a semi-parametric approach to estimate incremental mean costs, QALYs, their variances and co-variances with censored data. Methods: The economic evaluation of clinical trial data requires the estimation of incremental mean costs and effectiveness, together with their variances and co-variances. Methods to deal with censoring of data will be needed even in trials where dropout is low, if patients enter the trial at different times but the trial is of fixed duration. The method of inverse probability weighting has been suggested to estimate mean costs and QALYs between treatments, their variances and covariance, allowing for censoring in a regression framework. However, the method assumes bi-variate normality for costs and QALYs in each time period. Bootstrapping provides an alternative, semi-parametric approach to obtain standard errors and confidence intervals without making strong distributional assumptions. This study uses the inverse probability weighting method, comparing the parametric and semi-parametric approaches using 5 year data from the Randomised Intervention Trial for unstable angina (RITA-3). Results: We obtain incremental mean costs, incremental QALYs and their variances and covariance with censored data using (i) parametric assumptions of normality and (ii) using bootstrapping. Conclusion: We find the two approaches give similar results using this dataset. The parametric approach would be expected to give better results in cases where the assumption of bivariate normality of costs and QALYs is valid, but bootstrap approach can deal with a wider range of distributional forms. Furthermore, bootstrap approach is easier to understand and to implement. 82 APPLICATION OF LOCAL SEARCH METHODS IN COSTEFFECTIVENESS ANALYSIS: THE CASE OF RHEUMATOID ARTHRITIS Barton PM, Jobanputra P, Bryan S, and Burls AJ University of Birmingham, Birmingham, United Kingdom Purpose: Rheumatoid arthritis (RA) is a chronic condition for which a large number of disease-modifying anti-rheumatic drugs (DMARDs) are available. Typically DMARDs will be stopped after a time for various reasons. Thus the appropriate long-term strategy requires the use of a sequence of DMARDs. Any possible sequence is in principle a candidate for cost-effectiveness analysis. With 11 commonly used DMARDs to consider, there are a total of nearly 40 million possible sequences of DMARDs, and over 100 million sequences if subsets are to be considered as well. It is clearly not possible to test all these sequences. Methods: Sequences of DMARDs are compared using the Birmingham Rheumatoid Arthritis Model (BRAM). The BRAM allows realistic distributions to be used for the time spent on any DMARD. The model works by generating a large number of virtual patient histories from which population mean costs and QALYs are estimated. This work extends the work presented at SMDM in 2005 and illustrates the use of local search methods from operational research, including simulated annealing and genetic algorithms, to find the optimum sequence in a computationally feasible way. The stochastic natures of the model and search algorithms are made to work together. Results: All methods gave the true optimum sequence given the assumptions and parameter values used. The efficiency of the methods depends on the stopping rules used. Conclusions: Simulated annealing and genetic algorithms are well suited to cost-effectiveness analysis in a large decision space. 83 A NEW MODEL TO ASSESS THE COST-EFFECTIVENESS OF SCREENING FOR Chlamydia trachomatis Barton PM, Roberts TE, and Robinson SM University of Birmingham, Birmingham, United Kingdom Purpose: Chlamydia trachomatis is one of the most prevalent sexually transmitted infections in developed countries. Sequelae associated with Chlamydia include reduced fertility, infertility, ectopic pregnancy and potential damage to children born to infected mothers. Our review of published economic evaluations of Chlamydia screening found that the majority used a model which does not reflect the reality of the situation. Because of the infectious nature of the condition, and the highly heterogeneous pattern of spread of the infection, only a dynamic model that considers a whole population at individual level (of a type akin to discrete event simulation) can be guaranteed to give adequate estimates of the results of applying different policies. Methods: The Chlamydia Screening Studies (ClaSS) project in the UK investigated active population screening using homecollected specimens and generated empirical data about the coverage and uptake of screening, population Chlamydia prevalence, effectiveness of partner notification, and costs of screening. This paper reports the development of a new model extending the work of Mirjam Kretzschmar and colleagues. Our model is both the first to use UK primary data to assess the effect of population screening and the first to incorporate sequelae within the dynamic model structure. Results: In most scenarios, the incremental cost-effectiveness ratio for population screening was more than £20,000 per major outcome averted. The results were highly sensitive to the parameters relating to risk of sequelae. Conclusions: It is feasible to include sequelae within an individuallevel dynamic modeling structure. More information on the risk of sequelae is needed. 84 REPRESENTING DECISION PATHS IN AN INFLUENCE DIAGRAM BY DECISION TREES: EXAMPLE OF THE BENEFITS OF CT IN ACUTE ABDOMINAL PAIN Ying-Lie O,2 J, Kievit,1 J, Sont,1 A, Meijer,2 and K. Geleijns,2 1Department of Medical Decision Making and 2Department of Radiology, Leiden University Medical Center, Leiden, the Netherlands This work was financially supported by the EC-EURATOM 6 Framework Programme (2002-2006) and forms part of the CT Safety Efficacy (Safety and Efficacy of Computed Tomography (CT): A broad perspective) project, contract FP/002388. Purpose: An influence diagram can accommodate a large number of items and the complex associations between them, for instance the differential diagnosis, tests, treatment, and outcomes. However, the decision paths are invisible. Methods: The conversion of decision paths in the influence diagram to decision trees takes several conversion steps. First, a number of suspected diagnoses in the influence diagram are ruled out by considering several laboratory tests. Then, all connected paths to these diagnoses, tests, decisions, and outcomes are traced. Node types (chance, decision, outcome) in the influence diagram translate to the same types in the decision tree, and each choice in these becomes a branch. The hard part is the ordering of these nodes, because this is not explicitly supported in an influence diagram. Therefore, ordering of _equivalent_ nodes such as tests associated to the same diagnosis must be indicated beforehand. Results: The advantages are illustrated by a case study on modelling the benefits of CT in acute abdominal pain. The restriction to inflammatory causes still leaves several possible diagnoses. The probabilities are determined by evaluation of the most recent articles. The decisions regard diagnostics and treatment (CT, ultrasound, observation, laparoscopy) according to policies and clinical findings. Conclusions: Hidden decision paths in an influence diagram can be made explicit by conversion to decision trees, thus fully utilising ABSTRACTS both the advantages of the influence diagram and the decision tree. Future work will investigate graphical tools to support the conversion and analysis of the mixed influence diagram_decision tree structure. 85 COST-EFFECTIVENESS OF THE 7-VALENT PNEUMOCOCCAL CONJUGATE VACCINE IN THE UK: COMPARISON OF COHORT AND TRANSMISSION DYNAMIC MODEL RESULTS Jit M, Edmunds WJ, Choi Y, and Melegaro A Modelling and Economics Unit, Health Protection Agency, London, United Kingdom Purpose: S. pneumoniae is responsible for several clinical conditions in children and the elderly that can lead to serious morbidity or death. A 7-valent pneumococcal conjugate vaccine was licensed in the UK in 2001 and is currently being considered for routine infant vaccination. We evaluated the cost-effectiveness of the vaccine using two approaches: a cohort model, and a transmission dynamic model that could take indirect effects (herd immunity) into account. Methods: We compared results from a cohort model we had previously developed to results from an age-structured transmission dynamic model of vaccine and nonvaccine serotypes. The models were parameterised using hospital and epidemiological data from the UK. We calculated the expected net costs and benefits (measured in terms of life-years and quality adjusted lifeyears gained) of routine vaccination of infants at the age of 2 months and 4 months, as well as a booster dose at 12 months. We also considered the health economic implications of a catch-up campaign for all children under either 2 or 5 years of age. We analysed how sensitive the results were to changes in the data. Results: The cohort model suggested that the vaccine is unlikely to be cost-effective to the health provider at its current price. However, a transmission dynamic model that took into account the effects of herd immunity in the vaccinated population showed much better cost-effectiveness. Conclusions: A transmission model may give a more realistic picture of a vaccine_s cost effectiveness, particularly when the vaccine has substantial indirect benefits. 86 THE COST-EFFECTIVENESS OF INTRODUCING ROTAVIRUS VACCINATION IN THE UK Jit M,1 Edmunds WJ,1 and Harris J2 1Modelling and Economics Unit, Health Protection Agency, London, United Kingdom; 2Environmental and Enteric Diseases Department, Health Protection Agency, London, United Kingdom Purpose: Rotavirus is the leading cause of acute gastroenteritis in children, accounting for more than half a million deaths worldwide. In the UK, there are few rotavirus deaths but a significant burden of hospitalisation. Two rotavirus vaccines have recently completed clinical trials and are being considered for routine child vaccination in several countries. We investigated the cost-effectiveness of a universal immunisation programme using either of the two vaccines in the UK. Methods: We acquired data on gastrointestinal related deaths, hospital admissions, nosocomial hospital infections, Accident and Emergency admissions and GP consultations. Using information on the seasonality of rotavirus and other gastrointestinal infections, we estimated the proportion of these cases that were attributable to rotavirus. We also acquired estimates of the number of rotavirus related calls to NHS Direct. These results were combined with the efficacy of the two vaccines discovered in clinical trials to estimate the health economic benefits of the outcomes prevented by an immunisation programme. Results: A rotavirus vaccine is unlikely to be cost saving to the UK health service or economy. Its cost-effectiveness depends primarily on the number of rotavirus-attributable deaths. If there are at least 25 such deaths a year, then a vaccine costing £20 a dose may be cost-effective. Conclusions: Routine rotavirus vaccination can reduce the substantial short-term morbidity burden due to the infection. However, it may not be cost effective unless the number of vaccine preventable deaths is substantially greater than we estimate from currently available data. 87 ENDOVASCULAR REPAIR VERSUS OPEN SURGERY IN HEMODYNAMICALLY STABLE PATIENTS WITH ACUTE ABDOMINAL AORTIC ANEURYSMS: ONE-YEAR COST-EFFECTIVENESS ANALYSIS Visser JJ, Bosch JL, Hunink MGM, van Dijk LC, Hendriks JM, and van Sambeek MRHM Erasmus MC, Rotterdam, the Netherlands Purpose: To determine the one-year cost-effectiveness for endovascular versus open repair in patients with an acute infrarenal abdominal aortic aneurysm (AAA). Methods: All consecutive patients with an acute infrarenal AAA who presented to our tertiary-care hospital between January 1, 2001, and December 31, 2004, and underwent endovascular (n = 32) or open repair (n = 35) were included in this study. Direct hospital costs of all patients were assessed using the resource utilization approach. The in-hospital costs and costs during one-year follow-up were assessed from a health care perspective. Treatment effects were expressed as one-year survival. Bootstrapping was performed to estimate precision of our results. Results: Sex (61 male, 6 female) and age (mean 72.0 years) did not differ between the treatment groups (p 5). The total costs including one-year follow-up were _21,431 versus _35,249 for patients who underwent endovascular repair and open surgical repair, respectively (p = 0.12). Total mortality including one-year follow-up was 22% (7/32) for patients who underwent endovascular repair and 26% (9/35) for patients treated with open surgery (p = 0.71). Because endovascular repair was more effective and less expensive than open repair, a situation of dominance prevails. Although, bootstrapping results suggest much uncertainty surrounding our estimates for costs and mortality. Conclusion: The one-year data from our study suggest that endovascular repair compared to open surgery is cost saving with slightly lower mortality in patients with an acute infrarenal AAA. 88 COST-EFFECTIVENESS OF LATANOPROST AND TIMOLOL MALEATE FOR THE TREATMENT OF GLAUCOMA IN SCANDINAVIA AND THE UNITED KINGDOM USING A DECISIONANALYTIC HEALTH ECONOMIC MODEL Stewart WC,1 Stewart JA,1 Passmore CL,1 and Mychaskiw M2 1Pharmaceutical Research Network LLC, Charleston, SC; 2Pfizer Inc., New York, NY Purpose: To assess the cost-effectiveness of latanoprost or timolol maleate as monotherapy in the treatment of open-angle glaucoma in Norway, Sweden, Denmark, and the United Kingdom (UK). Methods: A Markov model was constructed to perform a costeffectiveness analysis. Health states were stable and progressed glaucoma and transition probabilities for both primary open-angle and exfoliation glaucoma were derived from the medical literature. Practice patterns were obtained from surveys completed by 54 ophthalmologists geographically dispersed throughout each country. Country specific unit costs were used for medications, ABSTRACTS ABSTRACTS E25 E26 _ MEDICAL DECISION MAKING/JUL_AUG 2007 patient visits, diagnostics, and therapeutic procedures. Quality of life weights for visual acuity were 0.50-0.68 depending on extent of visual disability. The time horizon was five years. A payer perspective was adopted and costs were discounted at 3% (Scandinavia) or 3.5% (UK) per annum. Results: Latanoprost was less expensive than timolol maleate by 5.4%-6.7% (Scandinavia) and 2.1% (UK). A narrow range of effectiveness 0.003-0.01 (years to progression of glaucoma) existed between treatment cohorts. This may have resulted from model the design, which reflected that physicians ultimately control most patients_ glaucoma over five years by adding or changing therapy. The associated incremental cost-effectiveness ratios for latanoprost vs. timolol generated by the Norwegian, Swedish, Danish, and UK models, respectively, were: NOK 457,212; SEK 1,251,126; DKK 447,857 and GBP 6,087. Conclusions: Over the course of five years, latanoprost provides a cost-effective alternative to traditional timolol generics as monotherapy in Norway, Denmark, Sweden and the UK. 89 COST-EFFECTIVENESS OF LATANOPROST, TRAVOPROST, AND BIMATOPROST FOR THE TREATMENT OF GLAUCOMA IN NORWAY, SWEDEN, AND DENMARK USING A DECISION-ANALYTIC HEALTH ECONOMIC MODEL Stewart WC,1 Stewart JA,1 Passmore CL,1 and Mychaskiw M2 1Pharmaceutical Research Network LLC, Charleston, SC; 2Pfizer Inc., New York, NY Purpose: To evaluate the cost-effectiveness of latanoprost, travoprost or bimatoprost monotherapy in open-angle glaucoma in Norway, Sweden, and Denmark. Methods: A Markov decision-analytic health economic model was developed to estimate the comparative cost-effectiveness of prostaglandin analogs. Health states were stable and progressed glaucoma. Transition probabilities for both primary open-angle and exfoliation glaucoma were populated with data from published medical literature. Clinical practice patterns were derived from surveys obtained from 45 ophthalmologists dispersed throughout each country. Specific unit costs for each country were used for medications, clinic visits, diagnostics, and therapies. Quality of life weights were assigned for visual acuity from 0.50-0.68 and the quality adjusted life year (QALY) was the effectiveness metric. The time horizon was five years. All analyses were from a payer perspective and cost results (2005 currency) were discounted at 3% per annum. Results: Latanoprost dominated (i.e., costs less and is more effective than) bimatoprost and travoprost in both Norway and Sweden. Latanoprost was 3% less expensive in Sweden and Norway, and the cost of all three medicines were within 1% of each other in Denmark. In contrast, in Denmark bimatoprost dominated travoprost and was slightly less expensive than latanoprost (DKK 28.7 K versus 29.0 K). However, in Denmark, latanoprost was more effective than bimatoprost (incremental cost/efficacy ratio DKK 47,871). Effectiveness (years till progression) was within a narrow range (3.2048-3.2613 QALY) for all products in each country. Conclusions: Latanoprost provides a cost-effective alternative to other available prostaglandin analogs in Norway, Sweden, and Denmark. 90 COST-EFFECTIVENESS OF PASSIVE IMMUNISATION AGAINST RSV INCLUDING LONG-TERM EFFECTS Moll HA, Rietveld E, Groot R, and Steyerberg EW Department of Public Health, Erasmus MC, University Medical Centre, Rotterdam, the Netherlands Background: The cost-effectiveness of passive immunisation against respiratory syncytial virus (RSV) is discussed. After RSV infection, the risk of recurrent lower respiratory symptoms is increased. We aimed to assess incremental costs to prevent one hospitalisation in high-risk infants from a societal perspective, including long-term costs and effects. Methods: We performed a cost-effectiveness analysis by combining estimates of individual hospitalisation costs and monthly hospitalisation risks, with immunization costs, parental costs, and efficacy of passive immunisation, long-term respiratory morbidity and costs the 5 years after RSV infection. The reference case was a high-risk boy (gestational age 28 weeks, birth weight 1000 g and zero months at start-off the RSV season). Various sensitivity analyses were performed. Results: Cost-effectiveness of passive immunisation varied widely by child characteristics and seasonal month. For the reference case, it was most cost-effective in December at _27,227 per hospitalisation averted and _27,670 without long-term costs. Cost-effectiveness, long-term morbidity included, was not sensitive to changes in costs of long-term treatment frequency of long-term respiratory morbidity in the five years following RSV hospitalisation. Cost-effectiveness was most sensitive to changes in costs of passive immunisation. Conclusions: Cost-effectiveness of passive immunisation varied strongly by child characteristics and seasonal month. The incremental costs per hospitalisation were always high, and did not significantly decrease if costs of long-term follow-up complications are included. We recommend a restrictive immunisation policy. 91 ESTIMATING THE COST-EFFECTIVENESS OF OPPORTUNISTIC CHLAMYDIA SCREENING IN ENGLAND Adams EJ,1 Turner KME,2 and Edmunds WJ1 1Health Protection Agency, London; 2Imperial University, London (formerly of the Health Protection Agency), United Kingdom Purpose: Opportunistic Chlamydia screening is being implemented in England, but questions remain as to which populations should be targeted and the ideal rescreening interval. Cost-effectiveness analysis compares different screening strategies, in the context of limited resources. Methods: Results from a dynamic individual-based stochastic model were incorporated into a disease progression and economic model to explore the cost-effectiveness of different screening strategies. Disease progression parameter values were based on estimates from the literature. Cost estimates came from the literature, analysis of data from a Chlamydia screening pilot study, and other sources. The uncertainty of key model assumptions including disease progression and cost estimates was assessed using univariate and multivariate probabilistic sensitivity analyses. Results: Targeted screening of young women is estimated to be the most cost-effective strategy. Screening older individuals and including men increases the effectiveness but also increases the overall cost-effectiveness ratio. The model results were very sensitive to assumptions about the progression from chlamydial infection to pelvic inflammatory disease. Conclusions: Model results suggest that screening is unlikely to be cost-saving. More work is needed to estimate the results in terms of a broader outcome such as cost per QALY gained. Assumptions about intervention parameters including effective partner notification rates and screening test acceptance can be updated in the model as data is collected in the national screening programme to give better estimates of the cost-effectiveness of opportunistic screening. 92 ESTIMATING THE COST-EFFECTIVENESS OF DETECTING CASES OF CHRONIC HEPATITIS C INFECTION ON RECEPTION INTO PRISONS IN ENGLAND AND WALES Sutton AJ,1,2,3 Edmunds WJ,1 and Gill ON1 ABSTRACTS 1Health Protection Agency, London; 2Imperial College, London, United Kingdom; 3University of Warwick, Coventry, United Kingdom Purpose: In England and Wales where less than 1% of the population are injecting drug users (IDUs), 97% of HCV reports are attributed to injecting drug use. As over 60% of the IDU population will have been imprisoned by the age of 30 years, prison may provide a good location in which to offer HCV screening and treatment. The aim of this work is to examine the cost effectiveness of a number of alternative HCV screening strategies on prison reception. Methods: A decision analysis model embedded in a model of the flow of IDUs through prison was used to estimate the cost effectiveness of a number of alternative screening strategies. The model estimates the average cost of identifying a new case of HCV from the perspective of the health care provider and how these estimates may evolve over time. Results: The results suggest that administering verbal screening for a past positive HCV test and for ever having engaged in illicit drug use prior to the administering of ELISA and PCR tests can have a significant impact on the cost effectiveness of HCV screening strategies on prison reception, the discounted cost in 2017 being £2,102 per new HCV case detected compared to £3,107 when no verbal screening is employed. Conclusions: The work here demonstrates the importance of targeting those individuals that have ever engaged in illicit drug use for HCV testing in prisons, these individuals can then be targeted for future intervention measures such as treatment or monitored to prevent future transmission. 93 COST-EFFECTIVENESS OF SCREENING YOUNG COMPETITIVE ATHLETES FOR HYPERTROPHIC CARDIOMYOPATHY Brodtkorb TH Center for Medical Technology Assessment, Linköping University, Linkoping, Sweden Purpose: Most sudden deaths in athletes are due to cardiovascular disease, and hypertrophic cardiomyopathy (HCM) has been implicated as the major cause of sudden cardiac deaths in young competitive athletes. Currently there is no organized screening program for HCM in Sweden. The purpose of the present study was to analyze the cost-effectiveness of a screening program of competitive athletes at 12 years of age. Methods: A probabilistic Markov model was used to analyze the synthesized available evidence with regards to costs and effects for implementing a screening program compared to the current non screening strategy. The costs were estimated based on the Swedish procedure prize lists, the transition probabilities, and utility values were obtained from the literature. The effectiveness was expressed in QALY_s. Results: Preliminary analyses of the model indicate that there is a high probability for the screening program to be dominated by current practice, both being more costly and generating less QALY_s. Conclusion: The preliminary results of the model indicate that given the existing information on this problem it is not seen as costeffective to implement a screening program for HCM for young competitive athletes in Sweden. The main reason for these findings seems to be the large proportion of false positive the screening program would generate and by that a large loss in QALY_s due to active healthy subjects being prevented from physical activity and living with the unfounded fear of having a heart disease. 94 COST-EFFECTIVENESS OF 7-VALENT PNEUMOCOCCAL CONJUGATE VACCINE (PCV-7): IMPACT OF HERD IMMUNITY Wisløff T,1 Abrahamsen TG,2 Pedersen MK,3 Løvoll Ø,3 Bergsaker MAR,3 Møller P,4 and Kristiansen IS5 1Norwegian Knowledge Centre for the Health Services, Oslo; 2Rikshospitalet University Hospital, Oslo; 3Norwegian Institute of Public Health, Oslo; 4Haukeland University Hospital, Bergen; 5Institute of Health Management and Health Economics, University of Oslo, Oslo, Norway Background: Streptococcus pneumoniae is a leading bacterial cause of septicaemia, meningitis, pneumonia and otitis media and may cause severe sequelae and even death. A 7-valent conjugate pneumococcal vaccine (Prevenar®) has proved effective in preventing otitis and invasive pneumococcal disease (IPD) in children. A reduction in IPD has also been observed in adult age groups, possibly due to herd immunity. The aim of this study was to explore the cost-effectiveness of Prevenar® vaccination of infants in Norway accounting for herd immunity. Methods: The study was based on a Markov-model using data on risk of pneumococcal disease, efficacy of the vaccine as observed in clinical trials (adjusted for serotype differentials), costs of the vaccine and decrease in the frequency of adverse outcomes from pneumococcal disease. Results: Disregarding indirect costs, the incremental cost per life year saved using 4 Prevenar® doses was _153,000 when herd immunity was included and _311,000 when it was not, or _90,000 and _184,000 if 3 doses offer the same effectiveness as 4. Also accounting for indirect costs, the numbers would decrease markedly. The vaccine price and efficacy, and otitis incidence were also crucial factors in the analyses. Monte Carlo simulations indicate that the results were robust to uncertainty in other parameters. Conclusions: The cost-effectiveness of vaccination with pneumococcal vaccine of infants will in particular depend on the price of the vaccine, the efficacy of the vaccine, the efficacy of three versus four vaccine shots, the extent of herd immunity, the discount rate and the valuation of indirect costs. 95 MULTINATIONAL DEVELOPMENT OF A QUESTIONNAIRE ASSESSING PATIENT SATISFACTION WITH ANTI-COAGULANT TREATMENT Prins MH,1 Arnould B,2 Bhakar AL,2 Caire P,3 and Leguet P4 1Maastricht University, Maastricht, the Netherlands; 2Mapi Values, Lyon, France; 3Mapi Research Institute, Lyon, France; 4Sanofi- Synthelabo Recherche, Bagneux, France Purpose: A patient-reported questionnaire, the PACT-Q (the Perception of Anti-Coagulant Treatment Questionnaire), was developed and validated in several languages to assess both patient expectation and satisfaction with anticoagulant treatment (AC-TX). Methods: A conceptual model of expectation and satisfaction with AC-TX was designed by a multidisciplinary advisory board and used to guide patient (n = 31) and clinician (n = 17) interviews in French, US English, and Dutch. Patients had either atrial fibrillation (AF), deep venous thrombosis (DVT), or pulmonary embolism (PE). Following interviews, three PACT-Q language versions were developed simultaneously and then pilot-tested by 18 patients. Linguistic validations were performed for additional language versions. Results: Concepts were developed and classified into three groups: 1) Treatment, 2) Disease and Complications and 3) Information. After clinician and patient interviews, concepts were further refined and three language versions of the PACT-Q were created, each containing the same 27 items grouped into 4 domains: _Expectations_ (7 items), _Convenience_ (11 items), _Burden of Disease and Treatment_ (2 items), and _Anticoagulant Treatment Satisfaction_ (7 items). No item was deleted or added after pilot testing. The PACT-Q was ABSTRACTS ABSTRACTS E27 E28 _ MEDICAL DECISION MAKING/JUL_AUG 2007 divided into an _Expectation_ and a _Satisfaction_ section for separate evaluation prior to and after AC-TX respectively. Eleven additional language versions were linguistically validated. Conclusion: The Perception of Anti-Coagulant Treatment Questionnaire (PACT-Q) has been developed in 14 languages for use with AF, PE and DVT patients. Validation is currently under way to examine the psychometric properties and to further refine the PACT-Q prior to its wider use as an endpoint in clinical trials. 96 SCORING AND PSYCHOMETRIC VALIDATION OF A COMPREHENSIVE URINARY SYMPTOM CHECKLIST: THE URINARY SYMPTOM PROFILE (USP©) Arnould B,1 Bosch V,1 Benmedjahed K,1 Amarenco G,2 Coloby P,3 Grise P,4 Haab F,5 and Richard F6 1Mapi Values, Lyon, France; 2Hôpital Rothschild, Paris, France; 3Centre Hospitalier René Dubos, Pontoise, France; 4Hôpital Charles Nicolle, Rouen, France; 5Hôpital Ténon, Paris, France; 6Hôpital La Pitié-Salpêtrière, Paris, France Purpose: A unique tool, specific to urinary symptoms among men and women, was designed for use in clinical practice. This study presents the scoring and psychometric validation of the USP©. Methods: The USP© was developed by an advisory committee of four urologists and one physiotherapist. The USP© was comprehension tested by patients and clinicians. Patients suffering from urinary symptoms (n = 253) and without urinary symptoms (n = 75) completed the final questionnaire twice, seven days apart. Psychometric properties were assessed on symptom patients and included: 1) structural analysis by Principal Component Analysis (PCA) with varimax rotation, 2) internal consistency reliability determination (Cronbach_s alpha), 3) concurrent validation (Spearman correlation coefficients with the International Consultation on Incontinence Questionnaire (ICIQ-UI SF)), and 4) test-retest reliability evaluation (Intra Class Correlation (ICC) coefficients in stable patients). Clinical validity was established by comparing patient-completed micturition diary information with USP© scores. Results: PCA analysis confirmed the final USP© structure of 11 items grouped into 3 dimensions (Stress Urinary Incontinence (UI), Overactive Bladder (OAB), and Low Stream). Higher scores were indicative of higher severity. Internal consistency reliability (Cronbach_s alpha ranging from 0.69 to 0.94) and concurrent validity (ICIQ-UI SF correlations of 0.73 for Stress UI and 0.62 for OAB), were good to excellent. Test-retest reliability was excellent (ICC ranging from 0.85 to 0.91). Significant correlations between the diary information categories and certain USP© dimension scores demonstrated clinical validity. Conclusion: The USP© is valid, specifically designed for clinical use, and the first questionnaire to provide comprehensive evaluation of all urinary symptoms for men and women. 97 THE PROLABELS DATABASE: A NEW ON-LINE TOOL TO EXPLORE THE WORDING AND TYPES OF PRO INCLUDED IN APPROVED MEDICINAL PRODUCTS LABELS Caron M, Emery MP, Arnould B, and Acquadro C Mapi Research Trust, Lyon, France Purpose: Patient-reported Outcomes (PRO) are used in clinical studies to assess patients_ treatment benefit. There is an increased interest in examining PRO claims included in approved product labels. We developed the PROLabels database to summarize and report extensively PRO labels approved by the FDA or the EMEA. Methods: Efficacy results were extracted from two sources: Summary Product Characteristics (SPC) of drugs approved through the centralized procedure gathered from the EMEA website (since January 1995) and approved labels of New Molecular Entities posted on the FDA CDER website (since January 1998). Products including a PRO claim in their approved label were added to the database. Results: At the date of 12/21/2005, the database contains 121 records (57 from the FDA and 64 from the EMEA) for 91 different International Nonproprietary Names (INN). The five most represented therapeutic areas include nervous (32.0%), immune (24.0%), musculoskeletal (18.0%), genitourinary (14.8%), and respiratory (13.2%) systems. Signs and symptoms are the most frequently measured PROs while HRQOL represents 20.7%. The database can be searched by INN, commercial name, marketing authorization holder, indication, PRO, approval date, and agency. Further information displayed include description of clinical studies supporting the label, the product_s pharmacological action, and data source. Conclusion: The PROLabels database is a useful on-line tool to find which products have obtained a PROLabel when reviewed by the EMEA or the FDA. The database will be updated on a weekly basis and expanded to include other data sources such as Canada and mutual recognition/decentralized procedure in Europe. 98 EXAMINING THE PERFORMANCE OF DISCRIMINATION AND RESPONSIVENESS OF A CONDITION-SPECIFIC PREFERENCEBASED INSTRUMENT: THE ASTHMA QUALITY OF LIFE UTILITY INDEX (AQL-5D) Yang Y, Brazier JE, and Tsuchiya A Health Economics and Decision Making Section, Sheffield University, Sheffield, United Kingdom Purpose: There has been potential demand for the conditionspecific preference-based measures for calculating Quality Adjusted Life Years (QALYs), given the generic measures such as EQ-5D, HUI and SF-6D may lack sensibility to capture changes in quality of life. Based on the Asthma Quality of Life Questionnaire (AQLQ), an asthma specific preference-based utility index AQL-5D has been derived and a valuation survey has been carried out. The resulting utility scoring algorithm makes it possible to calculate QALYs using existing and future AQLQ data sets. The purpose of this presentation is to examine the psychometric properties of the AQL-5D in terms of discrimination and responsiveness. Methods: The AQL-5D is applied to an asthma dataset which covers a wide range of asthma patients, and contains the AQLQ and EQ-5D. Descriptive statistics for the AQLQ scores, AQL-5D and EQ- 5D indices are computed. The relationship between AQLQ scores with AQL-5D indices, and AQL-5D indices with EQ-5D indices were tested using Pearson_s product moment correlation coefficient and bivariate linear regression. In order to examine the discriminative ability of the AQLQ, EQ-5D and AQL-5D to detect difference in independent indicators of health, patients were divided into groups according to a generic health index and an asthma specific index. Effect sizes by patient groups are calculated for each instrument and compared. A similar analysis is undertaken for changes over time. Results and Conclusions: Analysis is ongoing. 99 EVIDENCE BASED GUIDELINES, TIME BASED HEALTH OUTCOMES AND THE MATTHEW EFFECT Essink-Bot ML,1 Kruijshaar ME,1 Barendregt JJ,1,2 and Bonneux LGA3 1Department of Public Health, Erasmus MC, Rotterdam, the Netherlands; 2University of Queensland, Brisbane, Australia; 3[PE: city?] Purpose: To show that evidence based practice guidelines hide value choices, depending on the measure of health benefit, and explore how experts in guideline development deal with the resulting dilemmas. ABSTRACTS Methods: Epidemiological modelling to simulate the effects of spending a fixed budget to either avoid a maximum number of deaths (risk based) or to buy the maximum number of life years (time based); interactive Internet survey among 123 medical experts involved in cardiovascular disease (CVD) guideline development. Results: Maximising the number of deaths prevented leads to allocation of a CVD-preventive drug to subjects with the highest risk, that is, elderly; smokers; obese persons; subjects of low socioeconomic status (SES). Maximising the number of life years gained favours the young, non-smokers, subjects of normal weight, high SES, or: _the one who has will be given more_ (Matthew 25:29). The experts found the choice dilemmas difficult, contributing to a rather low response rate (31%). A majority of the respondents favoured maximisation of life years. They did not reject an upper age limit (70-75 years), and preferred to partially exclude the additional risk of smoking and obesity (_blaming the victim_). However, they would favour some investment in narrowing the SES gap in life expectancy. Conclusions: Evidence-based allocation of expensive therapies can lead to differential distributions, depending on the choice between events prevented or time gained as health outcomes. The choice for either one introduces moral choices because of the variety of origins of increased disease risk. 100 IS THE GERMAN INSTITUTE FOR QUALITY AND EFFICIENCY IN HEALTH CARE (IQWIG) PREPARED FOR COMPETITIVE DECISION MAKING? Mueller E, Rosery H, Walzer S, and Bergemann R Analytica International, Loerrach, Germany Purpose: To review the approach and results of the recently founded German IQWiG (Institut für Qualität und Wirtschaftlichkeit im Gesundheitswesen) in health technology assessment and reimbursement recommendations. Methods: The published guidelines of IQWiG and available appraisals are reviewed with respect to guidelines and reports from NICE, ECHTA and Cochrane checklists. The methodology recommendations and performance instructions are systematically compared and evaluated. The decision process of IQWiG statements and its binding character is assessed and ranked for transparency. Results: Published guidelines and checklists give valuable recommendations for performing adequate health technology assessments. NICE offers substantial guidance for manufacturers with respect to contents, quality and time-frame of data needed for an assessment. The decision process is transparent and opportunities for comments or appeals on decisions are given. IQWiG has not adapted these methodologies but defined own guidelines and decision approaches. Up to now neither the guidelines (_methodology paper_) nor the first published reports seem to be adequate to provide sufficient support for assessments and decision making in the German health care system. Conclusions: The IQWiG guidelines lack of concise definitions and performance instructions. Though not legally binding, the IQWiG recommendations may have an important impact on decision making by the G-BA (Gemeinsamer Bundesausschuss). Therefore it is necessary to improve the guidance for respective assessments as well as the transparency of evaluations. 101 MULTICRITERIA DECISION MODEL IN SELECTING A HEALTH INSTITUTION Oddershede AM,1 Carrasco RA,2 and Barham E1 1Universidad de Santiago de Chile, Santiago, Chile; 2University of Newcastle, Newcastle, United Kingdom Purpose: To provide with a decision making model to examine patient criteria and its impact in selecting a health institution. Using Analytic Hierarchic Process (AHP) different criteria as technology, cost and other are analysed and evaluated. Methods: The challenge health institutions face is to improve quality of service. Measuring quality in health care system involves competing goals: satisfying citizen necessities and proportionate resources for health professionals_ aspirations. Since service does not often meet patient expectations, a case study is pursued in a Chilean region applying AHP approach. A comparison process based on decisive factors and judgements from an expert panel is carried out to derive a criteria priority matrix. Results: The priorisation results indicated that a criterion with most effect in the selection is opportune service. Physicians_ skills and its availability obtained the second priority, followed by cost and technology attributes. Location, comfort, billing were considered less important. Overall result indicated preference for private hospital (41.4%). Sensitivity analysis showed that as criteria relative weights vary, public institution is selected. Conclusions: The multicriteria analysis revealed attributes and its ranking from patient perspective. To obtain appropriate medical intervention immediately following a health requirement has the greatest impact over quality of service. Cost attained a lower priority, although expert group included patients from different economic sectors. Technology provision, appropriate information availability and communication technology system is strongly desired by the patient. The AHP methodology gives support to hospital management decision makers for resource distribution when taking actions for service quality improvement. 102 ATTRIBUTE WEIGHTS IN A MULTIDIMENSIONAL OBSTETRICAL DECISION Bijlenga DB, Birnie E, and Bonsel GJ Academic Medical Centre, Amsterdam, the Netherlands Purpose: Patient preferences are substantial in medical decision making. Present study focuses on the choice between induction and waiting in term risk pregnancies. Modelling is complex because of multiple actors, domains and time frames. Objective is to obtain relative weights of pertinent attributes for subsequent discrete choice experiments. Methods: Based on a list of 40 high-risk pregnancy aspects, 6 attributes with their associate levels were identified. These were incorporated to 26 vignettes; 8 antepartum, 5 intrapartum, 4 maternal and 9 child outcomes. Thirteen laypersons ordered the vignettes from good to bad, and valued them on a 100-points Visual Analogue Scale (VAS) ranging from _worst imaginable situation_ to _best imaginable situation._ Results: Antepartum maternal discomfort (Mean 61.8, CI 46.8- 76.8), secondary caesarean section (Mean 49.1; CI 33.9-64.3) and maternal (Mean 41.5, CI 28.5-54.2) and child (Mean 39.0, CI 31.1- 46.9) moderate outcomes are in the middle segment. Severe child outcome (Mean 16.3, CI 9.0-23.7) is in the lowest segment. Mild but long-lasting complications are preferred over severe but shortlasting complications both antepartum (CI 10.6-26.6, p .000) and postpartum (CI 16.6-28.8, p .000). All attributes have a significant effect on the ratings (p .025). Conclusions: The attributes have a hierarchy in the decisionmaking problem, and all attributes influence the decision. The results will guide selection of attributes for subsequent discrete choice experiments. 103 RISK PERCEPTION OF PREGNANT WOMEN BEING OFFERED PRENATAL SCREENING Timmermans DRM,1 Kleinveld JH,1 Van den Berg M,1 Van Vugt JMG,2 Ten Kate L,3 and Van der Wal G1 ABSTRACTS ABSTRACTS E29 E30 _ MEDICAL DECISION MAKING/JUL_AUG 2007 1Department of Public and Occupational Health, Amsterdam, the Netherlands; 2Department of Obstetrics and Gynecology, Amsterdam, the Netherlands; 3Department of Clinical Genetics, Amsterdam, the Netherlands Purpose: Ideally, the decision for or against prenatal genetic screening should be based on the active decision-making of pregnant women who should be fully informed in particular about the risks involved. Methods: In a large randomized controlled trial (N = 2785) with two groups being offered prenatal screening and a control group, we measured pregnant women_s perceived risk of giving birth to a child with Down_s syndrome. This was measured with a numeric risk scale and recalculated into three categories: (1) a risk between 1 out of 2 and 1 out of 200; (2) a risk between 1 out of 200 and 1 out of 1000; (3) and a risk smaller than 1 out of 1000. Results: Participants generally perceived their risk as rather low: about 77% of the women thought they had a risk below 1 out of 1000 before information was offered while this was about 43% after test information was given as compared to 72% of women in the control group (a significant difference _;2 (4) = 162.8, p .001). Before the test offer, about 50% of women correctly classified their risk. After the offer but before the test was done, significantly more women classified their risk correctly compared with the control group (62.1% versus 54.5%; _;2(2) = 9.90, p .01). Conclusion: Although there is an increase in accurate risk perception, it is worrisome that still a large number of women had an inadequate perception of their risk of giving birth to a child with Down syndrome. 104 REAL TIME DECISION SUPPORT: INTRODUCING A PRELIMINARY DECISION SUPPORT SYSTEM FOR ANESTHETISTS Pott CM1 and Ballast A2 1University of Groningen, Groningen, the Netherlands; 2University Medical Center Groningen (UMCG), Groningen, the Netherlands Purpose: To enhance patient safety and to improve working conditions of anesthetists, we developed a decision support system (DSS) to support monitoring and diagnosing tasks of anesthetists in real time. The main goal of the DSS is to improve the situation awareness of anesthetists whenever a critical period appears, that is, their _big picture_ of the state of the patient. Methods: Based on the results of an international survey we conducted in 2002, we modified an existing cognitive process model adding urgent conditions. Together with a set of production rules for diagnoses derived from a textbook of anesthesia, we used our model to develop a perioperative DSS. Results: Patient data measured by commercial monitoring devices are forwarded to production rules which calculate the probability of potential diagnoses. Up to six diagnoses are displayed on the interface of the DSS. The diagnoses are grouped for cardiovascular, respiratory and anesthetic depth. For urgent (acute life threatening) states, 3 alarm icons are introduced: heart, lungs and eye. An indication of level of urgency and the trend is added. In case a patient problem cannot be categorized, our DSS displays the probabilities of differential diagnoses and their histories as a diagnostic help. Conclusions: We focused on drawing the attention of the anesthetist to relevant information, rather than replacing existing information. Errors caused by human failure like overlooking information or cognitive tunneling might be prevented by our DSS. Currently, the effects of using the DSS on situation awareness are tested in a anesthesia simulator. 105 IT-BASED EXPERIENCE SUPPORT FOR CLINICIANS Dahl FA and Gulbrandsen P Helse Øst Health Services Research Unit, Akershus University Hospital, Oslo, Norway Purpose: We want to use information about clinical decisions made under uncertainty available for the teams of clinicians responsible for the decisions, as feedback and incitement to improvement. We study the effect of this effort in terms of improved decisions in the team, converging decisions between doctors within the team, and possibly patient outcomes. We will also study the learning process itself, using qualitative methods. Methods: We approach clinical teams and let them decide which clinical decisions they want to study, and which information to collect. For each decision, we use IT to collect the information related to the decision, and patient outcome measured after some predefined time period. Data will be collected also when the actual decision is not to provide active treatment. Team and individual results will be presented and discussed within the team. Variance between doctors will be used as a pedagogic tool. Results: Five different clinical groups in our hospital have responded positively to the initiative: orthopaedic surgeons (choice of treatment for medial hip fracture), endocrinologists (whether or not to prescribe insulin when controlling diabetes in pregnancy), pediatricians (whether or not to use chest X-ray in bronchiolitis), gynaecologists (choice of operation method for vaginal prolapse), and cardiologists (handling of acute chest pain). We now start the process of data collection. By the time of the conference, we expect to have some preliminary results. Conclusions: The design of this quality improvement initiative has received positive response among clinicians, and they have easily found relevant decisions to study. 106 A METHOD FOR ANALYSING CLINICAL DECISION MAKING THAT CAN BE USED TO IDENTIFY TACIT AND CONTEXTUAL INFLUENCES Buckingham CD1 and Adams AE2 1Aston University, Birmingham, United Kingdom; 2University of Warwick, Coventry, United Kingdom Purpose: To present a method for analysing clinical decision making (CDM), that exposes contextual influences on patient consultations. Methods: We conceptualise CDM as three linked classification tasks: generating differential diagnoses; assessing potential outcomes of untreated diagnoses; and determining interventions. Each task involves doctors identifying relevant patient cues that should influence classification, and integrating them to determine their combined influences on potential diagnostic, outcome, and intervention classes to which patients may belong. The classification process is deconstructed into constituent elements common to all three tasks, such as the psychological representations of cues, directions of reasoning, and the associations of cue patterns with potential classes. A coding scheme has been developed to identify these elements, which permits analysis of psychological mechanisms influencing each CDM stage. Results: Coding reveals doctors_ dynamic reasoning processes, which we have captured by asking them to view videotapes of simulated patient consultations, and then provide retrospective debriefings. These debriefings approximate _think aloud_ data but are more suited for tasks that simultaneously require listening to consultations. When applied to transcriptions, coding highlights the cues used by doctors, and how they are processed to produce classification decisions. Thus coding can identify sociological and organisational factors influencing psychological processes, indicating potential sources of bias and inequalities in health care. This has been achieved successfully in studies showing, for example, how age and gender influence health care in ways not easily justified medically. ABSTRACTS Conclusions: The developed classification model and its associated coding framework provides an important method for analysing CDM during consultations. 107 HOW DOES THE MULTIDISCIPLINARY TEAM MEETING (MDM) AFFECT DECISION-MAKING IN LUNG CANCER? Roques TW, Cremin M, Noonan-Shearer K, Martin WMC, Harnett AN, and Barrett A Norfolk and Norwich University Hospital, Norwich, United Kingdom Purpose: To study how an established weekly lung cancer MDM affects decisions about oncological therapy. Methods: All non-surgical patients presented at the MDM over 4 months were studied at 3 time points. The presenting physician recommended treatment prior to the MDM (T1). An observer recorded the MDM discussion and recommended outcome (T2). These were compared with the actual decision later taken by the patient in consultation with the doctor (T3). Results: 55 patients were presented. The MDM discussion lasted a median of 282 seconds (range 100-462 seconds). A mean of 5 professionals contributed to each discussion (range 3-8). At T1, 21% of recommendations were for curative treatment and 79% palliative. The MDM discussion changed the treatment intent in 13%. The final treatment intent differed from the MDM recommendation in 8%. The exact treatment given (eg, chemotherapy, radiotherapy, combination etc.) was predicted by only 34% of presenting physicians and by 57% of MDT discussions. The commonest reasons for differences between T2 and T3 were that the patient_s general condition had deteriorated so they were not well enough for treatment or that the symptoms that needed palliating at T3 were different from those at T2. Conclusions: The treatment accepted by the patient is often different from that recommended by the MDM or referring physician. While the MDM may have educational, team-building and organisational benefits, its role in influencing and improving decisionmaking merits further research. 108 HOW DO MULTIDISCIPLINARY CANCER MEETINGS (MDM) AFFECT DECISIONS IN BREAST CANCER MANAGEMENT? Roques TW, Noonan-Shearer KD, Cremin MS, Harnett AN, Martin WMC, and Barrett A Norfolk and Norwich University Hospital, Norwich, United Kingdom Purpose: To study how a weekly breast cancer MDM affects clinical decisions about adjuvant therapy. Methods: Prospective study of patients with stage 1-3 breast cancer discussed at the MDM over 4 months at 3 time points. The presenting clinician predicted whether the patient would receive adjuvant chemotherapy, hormone therapy and radiotherapy before the MDM discussion (T1). An observer recorded the MDM discussion and recommended outcome (T2). These were compared with the actual decision later taken by the patient in consultation with the doctor (T3). Results: 72 patients were studied. For chemotherapy, the treatment decision differed between T1 and T2 in 2% of patients and between T2 and T3 in 9%. For hormone therapy, the corresponding figures were 12% and 14% and for radiotherapy 8% and 13%. The overall treatment plan agreed by doctor and patient was recommended by only 72% of MDM discussions and 76% of presenting doctors. The treatment plan was the same at all 3 time points in 69%. The median length of MDM discussion was 88 seconds. The breast care nurse had met all patients prior to MDM but was only involved in the discussion in 6%. The mean cost of personnel for each meeting was £774. Conclusions: This breast cancer MDM does not reliably inform adjuvant treatment decisions. The MDM discussion is usually short and based on pathological variables rather than patient preferences. Whilst these meetings may have many educational and organisational benefits, their role in the decision-making process warrants closer scrutiny. 109 THE DECISION TO BREAK CONFIDENTIALITY IN NURSING Dalgleish LI1 and Wilson K2 1Department of Nursing Midwifery, University of Stirling, Stirling, United Kingdom; 2University of Queensland, Brisbane, Australia Purpose: Confidentiality is an important medical ethical principle. Nurses are confronted with ethical dilemmas about confidentiality daily. However, little empirical work on this decision has been conducted. Methods: The decision to break confidentiality was studied by asking 71 nursing students to make such decisions for five scenarios. The scenarios were chosen to have different types of confidentiality dilemmas. The moral (or ethical) intensity of each scenario was measured as well as ratings of its six components: magnitude of the consequences, social consensus, probability of effect, temporal immediacy, proximity and concentration of effect. Individual difference variables such as ethical position and empathy were also measured. Results: Results showed that the components of moral intensity predictive of ethical intensity and decision to break confidentiality varied across scenarios. For some scenarios, empathy and idealism predicted the decision to break confidentiality. Conclusions: Future research using judgment analysis to measure the relative importance of situation factors and signal detection theory to measure thresholds for breaking confidentiality is suggested. 110 EVALUATING AN APPROACH TO SETTING PRIORITIES FOR PATIENT SAFETY INTERVENTIONS Lamont T National Patient Safety Agency, London, United Kingdom Methods: The National Patient Safety Agency (NPSA) was launched in 2001 with a broad remit. In 2004, the NPSA piloted an approach to setting priorities for new work, including the scoring of topics against criteria by an expert panel. This paper describes the results of an evaluation of the pilot in spring 2005, using views of stakeholders and documentary analysis. Results: The evaluation involved limited surveys and meetings of key stakeholders; review of submissions received; analysis of panel scoring and process of decision-making. The process was revised to address some key weaknesses identified in the evaluation, including: the following _ gaps in submissions and panel membership, _ problems in availability and use of supporting evidence, _ lack of clarity in use of criteria and _ timing of process and need for agreed fast-tracking process. Conclusions: The NPSA approach draws on other experiences of prioritising health care and research (eg, health technology assessment, research payback, local resource allocation) and developing theory to support decision-making (eg, information analysis. Some methods such as multi-attribute risk ranking have not yet been widely used in health settings. There is a rich and developing body of theory around priority setting, but some is difficult to apply in practice. In addition, there is limited evidence in many key areas of patient safety. The revised approach to priority-setting by the NPSA attempts to combine a more structured approach to decision-making with the values of key stakeholders, recognising the political nature of its work to improve patient safety. ABSTRACTS ABSTRACTS E31 </meta-value>
</custom-meta>
</custom-meta-wrap>
</article-meta>
</front>
</article>
</istex:document>
</istex:metadataXml>
<mods version="3.6">
<titleInfo lang="en">
<title>2006 EUROPEAN SOCIETY FOR MEDICAL DECISION MAKING MEETING</title>
</titleInfo>
<titleInfo type="alternative" lang="en" contentType="CDATA">
<title>2006 EUROPEAN SOCIETY FOR MEDICAL DECISION MAKING MEETING</title>
</titleInfo>
<typeOfResource>text</typeOfResource>
<genre type="other" displayLabel="other" authority="ISTEX" authorityURI="https://content-type.data.istex.fr" valueURI="https://content-type.data.istex.fr/ark:/67375/XTP-7474895G-0">other</genre>
<originInfo>
<publisher>Sage Publications</publisher>
<place>
<placeTerm type="text">Sage CA: Los Angeles, CA</placeTerm>
</place>
<dateIssued encoding="w3cdtf">2007-07</dateIssued>
<copyrightDate encoding="w3cdtf">2007</copyrightDate>
</originInfo>
<language>
<languageTerm type="code" authority="iso639-2b">eng</languageTerm>
<languageTerm type="code" authority="rfc3066">en</languageTerm>
</language>
<relatedItem type="host">
<titleInfo>
<title>Medical Decision Making</title>
</titleInfo>
<genre type="journal" authority="ISTEX" authorityURI="https://publication-type.data.istex.fr" valueURI="https://publication-type.data.istex.fr/ark:/67375/JMC-0GLKJH51-B">journal</genre>
<identifier type="ISSN">0272-989X</identifier>
<identifier type="eISSN">1552-681X</identifier>
<identifier type="PublisherID">MDM</identifier>
<identifier type="PublisherID-hwp">spmdm</identifier>
<part>
<date>2007</date>
<detail type="volume">
<caption>vol.</caption>
<number>27</number>
</detail>
<detail type="issue">
<caption>no.</caption>
<number>4</number>
</detail>
<extent unit="pages">
<start>e-3</start>
<end>e-4</end>
</extent>
</part>
</relatedItem>
<identifier type="istex">525CB0AB84383F0EC9BA6467590F85311F2844B5</identifier>
<identifier type="ark">ark:/67375/M70-DPPD2M40-0</identifier>
<identifier type="DOI">10.1177/0272989X080270040104</identifier>
<identifier type="ArticleID">10.1177_0272989X080270040104</identifier>
<recordInfo>
<recordContentSource authority="ISTEX" authorityURI="https://loaded-corpus.data.istex.fr" valueURI="https://loaded-corpus.data.istex.fr/ark:/67375/XBH-0J1N7DQT-B">sage</recordContentSource>
</recordInfo>
</mods>
<json:item>
<extension>json</extension>
<original>false</original>
<mimetype>application/json</mimetype>
<uri>https://api.istex.fr/ark:/67375/M70-DPPD2M40-0/record.json</uri>
</json:item>
</metadata>
<author></author>
<serie></serie>
</istex>
</record>

Pour manipuler ce document sous Unix (Dilib)

EXPLOR_STEP=$WICRI_ROOT/Sante/explor/PandemieGrippaleV1/Data/Istex/Corpus
HfdSelect -h $EXPLOR_STEP/biblio.hfd -nk 002538 | SxmlIndent | more

Ou

HfdSelect -h $EXPLOR_AREA/Data/Istex/Corpus/biblio.hfd -nk 002538 | SxmlIndent | more

Pour mettre un lien sur cette page dans le réseau Wicri

{{Explor lien
   |wiki=    Sante
   |area=    PandemieGrippaleV1
   |flux=    Istex
   |étape=   Corpus
   |type=    RBID
   |clé=     ISTEX:525CB0AB84383F0EC9BA6467590F85311F2844B5
   |texte=   2006 EUROPEAN SOCIETY FOR MEDICAL DECISION MAKING MEETING
}}

Wicri

This area was generated with Dilib version V0.6.34.
Data generation: Wed Jun 10 11:04:28 2020. Site generation: Sun Mar 28 09:10:28 2021