Potential differences in apparently similar clinical trials:
(1) differences in the inclusion and exclusion criteria (and how they are applied)
(2) differences in baseline states of patients
(3) differences in treatment and control interventions (doses, timing, brand, compliance)
(4) differences in management (additional pharmacologic therapy, patient care setting, responses to intermediate outcomes)
(5) differences in how outcomes are measured (follow-up times, use of cause-specific mortality, etc.)
(6) differences in analysis and differences in handling withdrawals, drop-outs, cross-overs, or other problems
(7) differences in the quality of design and execution
(8) differences in individual estimates of treatment effect (due to bias, imprecision or other causes)
where:
• Patients may differ in other ways in addition to the baseline state, such as development of comorbid conditions that do not exclude.
• One way to implement this is to score each item on a Likert scale (no, minimal, moderate and marked differences).
• Even if you are successful in getting all apples, there can be considerable variation between kinds of apples.
