Incidence- Versus Prevalence-Based Measures of Inappropriate Prescribing in the Veterans Health Administration.

OBJECTIVES

To describe variations in potentially inappropriate prescribing (PIP) and characterize the extent to which switching to an incidence-based indicator would affect health system quality rankings.

DESIGN

Observational study.

SETTING

Veterans Health Administration in 2011.

PARTICIPANTS

Older adults receiving outpatient primary care.

MEASUREMENTS

PIP was defined according to the National Committee for Quality Assurance High-Risk Medications in the Elderly list. Ranks were separately assigned for prevalent and incident PIP at the regional, network, and healthcare system levels.

RESULTS

National PIP prevalence was 12.3% (167,766/1,360,251), and incidence was 5.8% (78,604/1,360,251). PIP prevalence ranged from 3.5% to 33.1% across healthcare systems (interquartile range (IQR) = 9.2-15.5%). PIP incidence ranged from 1.2% to 14.9% (IQR = 4.1-7.2%). Rank order in PIP prevalence and incidence was correlated (Spearman correlation; ρ = 0.934, P < .001), although substantial changes in ranks were seen for some healthcare systems, with seven of 139 (5.0%) systems shifting more than 30 rank positions and 21 (15.1%) systems shifting 16 to 30 positions.

CONCLUSION

Prevalence- and incidence-based indicators of prescribing quality were strongly correlated. Transitioning to incidence-based indicators would not produce an initial disruption in quality rankings for most healthcare systems and might yield more-salient measures for tracking healthcare quality.