Design of a cohort study TIME direction of inquiry people exposed disease population without the no disease disease not exposed disease no disease “at risk”
Design of a case-control study TIME direction of inquiry Start with: Exposed cases (people with disease) Not exposed Population Exposed controls (people without disease) Not exposed
Questions to ask when an association is reported in the literature (eg estrogen and CHD) ExplanationFinding Association Bias in selectionYesNo or measurement ChanceLikelyUnlikely Confounding Yes No Cause
Association and cause ExplanationFinding Association Bias in selectionYesNo or measurement ChanceLikelyUnlikely Confounding Yes No Cause Case report? Case series?
Association and cause ExplanationFinding Association Bias in selectionYesNo or measurement ChanceLikelyUnlikely Confounding Yes No Cause P value
Association and cause ExplanationFinding Association Bias in selectionYesNo or measurement ChanceLikelyUnlikely Confounding Yes No Cause
Criteria for causation 1. Is there evidence from true experimentation in humans? 2. Is the association strong? 3. Is the association consistent from study to study? 4. Is the temporal association correct? 5. Is there a dose-response gradient? 6. Does the association make biological sense? 7. Is the association specific? (Adapted from Bradford Hill)
Randomization 1. Guarantees equal probability of receiving control/experimental treatment to all participants (removes investigator bias) 2. Protects against imbalances in known and unknown confounders 3. Provides basis for statistical analysis
Hierarchy of study methods for clinical decision-making Systematic reviews
Is observational evidence equivalent to experimental evidence ? Benson NEJM 2000;342:1877 Concato NEJM 2000;342:1887 In some cases - YES In others- NO !
Clinical trials are selective! -Select group gets in - chance decides who gets treatment; treatment effect decides who does better Observational evidence is also selective: self-selection of exposure: this may decide who does better
How much of Medicine is Evidence-Based ? Matzen P. Ugeskr laeger 2003;165: General Internal Medicine - 50% Psychiatry- 65% Others (surgery, general practice, dermatology) - less Lai Br. J Ophthal. 2003;4:385-90: 42.9% of patient interventions were based on evidence from RCT, meta- analysis or systematic reviews (23% on no evidence)
Major Disadvantage of RCTs: Selectivity
Clinical trials Should not be performed unless there is a realistic chance of providing a valid/reliable answer to a well- defined medical question
But … SHEP study Of 447,921 (100%) identified 31,960 (11.6%) met initial criteria 4,736 (1.03%) randomized
Men, gen’l pop Wom., gen’l pop SHEP pop % of Pop DM CVD CHF MI Angina
Coke tastes better…….! Than what?
Coke tastes better…….!
CAVEATS in using RCT evidence to guide patient management 1. Tendency to extend application of new treatments to patient groups other that those for whom data exist 2. Extrapolation of data to agents of the same class but untested for specific indication
Don’t drown in the evidence Use predigested sources ACP J Club, Bandolier, POEMS, clinical evidence, Cochrane Reviews, Clinical guidelines clearinghouse