Use of instrumental variable in prescription drug research with observational data: a systematic review.
OBJECTIVE To assess the effect of promoting a bundle of fixed doses of a generic statin and angiotensin-converting enzyme inhibitor/angiotensin receptor blocker (ACEI/ARB), delivered with minimal outpatient visits, laboratory testing, and dosage titration, to people with diabetes, coronary artery disease (CAD), or both in a large integrated healthcare system. STUDY DESIGN Three-year observational study of 170,024 Kaiser Permanente members with diabetes, CAD, or both. METHODS Using instrumental variable analysis, we assessed the impact of promoting the cardioprotective bundle on hospitalization rates for stroke and myocardial infarction (MI). RESULTS In 2004 and 2005, 47,268 of 170,024 individuals received "low exposure" (medication possession on 1 to 365 days). Their risk of hospitalization for MI or stroke in 2006 was lowered by 15 events per 1000 person-years (95% confidence interval [CI] = 1, 30), preventing events in 726 people. Furthermore, 21,292 of 170,024 individuals received "high exposure" (medication possession on 366 to 730 days). Their risk of hospitalization for MI or stroke was reduced by 26 events per 1000 person-years (95% CI = 17, 34), preventing events in 545 people. CONCLUSION A simplified method for bundling fixed doses of a generic statin and an ACEI/ARB was successfully implemented in a large, diverse population in an integrated healthcare delivery system, reducing the risk of hospitalization for MI and stroke.