|Clinical Study Design||Relative Cost||Design and Data Collection||Patient Population||Potential for Bias||Advantages and Disadvantages|
|Observational studies (including registry studies)||$||Can be retrospective or prospective in design; data quality is variable||Typically unselected population (e.g., Medicare)||Without randomization, comparative effectiveness studies cannot be performed||Large population; often many unmeasured variables or unexplained factors|
|Traditional RCTs||$$$$-$$$$$||Prospective design; data collection occurs at specialized study centers||Highly-selected patient population at study centers; may lead to results that are not generalizable||Randomization eliminates confounding bias||Current gold standard for comparative-effectiveness studies|
|Registry-based RCTs||$$-$$$||Prospective design; data collection often occurs at diverse clinical sites||Typically designed to study a specific patient population (e.g., those undergoing PCI)||Randomization eliminates confounding bias||Large number of outcomes; harnesses power of already-established clinical registry|
|Large, pragmatic clinical trials||$$-$$$$||Prospective design; data is collected ubiquitously as part of clinical care||Depending on electronic infrastructure, can be broad or selective; can incorporate enrichment criteria||Randomization eliminates confounding bias||Simple design; large number of outcomes; requires infrastructure that can facilitate easy and quick enrollment|
PCI = percutaneous coronary intervention; RCT = randomized controlled trial.