Method: Primary care practices (N=38) in three organizations of the HMO Research Network's Center for Education and Research on Therapeutics (CERTs) were randomized to receive group academic detailing with feedback (intervention) or usual care. All practice patients dispensed three medications (angiotensin-converting enzyme (ACE) inhibitors/angiotensin receptor blockers (ARBs), diuretics, and digoxin) were categorized as initial or continuing medication users. Generalized estimating equation approaches to logistic regression were used to assess adjusted differences in monitoring rates in the 12-month period following the intervention.

Results: Among the 142 primary care physicians invited to participate, 120 attended a first session, 103 attended a follow-up session and 90 attended both. Laboratory testing among patients with an initial dispensing for a diuretic improved with the intervention (OR = 1.24, 95% CI 1.08 – 1.43). An improvement was also detected among patients with an initial ACE/ARB dispensing, but it was not uniform across organization. While there was no effect for two smaller organizations (OR = 0.93, 95% CI 0.74 – 1.18; OR = 0.88, 95% CI 0.51 – 1.53), there was a significant effect for the largest site (OR = 1.36, 95% CI 1.18 – 1.58). No significant difference was found by study arm among patients with an initial digoxin dispensing. Among continuing users, the intervention significantly improved laboratory testing among patients dispensed an ACE/ARB (OR = 1.39, 95% CI 1.11 - 1.75) or a diuretic (OR = 1.28, 95% CI 1.02 - 1.60). However, no significant difference was found for continuing digoxin users. Adjusted differences in testing rates between study arms were modest, ranging from 4.96% among patients with an initial diuretic dispensing to 2.54% among continuing users of diuretics.

Conclusion: While informatic tools likely hold promise in reducing medication errors among outpatients, the availability of such systems is far from commonplace in the outpatient setting. Thus, educational approaches that are central to “low tech” solutions remain needed, especially in the short run. Although our findings were mixed in terms of ability to impact routine laboratory monitoring by academic detailing, it seems important to understand the effectiveness of more “low tech” solutions to improving medication safety within outpatient care.