Building a better quality measure: are some patients with 'poor quality' actually getting good care?

BACKGROUND: National performance measures monitor the proportion of diabetic patients with low-density lipoprotein (LDL) levels >/=130 mg/dL, but such simple intermediate outcomes measure poor control, not necessarily poor care. "Tightly linked" quality measures define good quality either by a good intermediate outcome (LDL <130 mg/dL) or by evidence of appropriate responses to poor control (eg, starting or optimizing medications for high LDL or not doing so in the face of contraindications). OBJECTIVES: We examined hyperlipidemia therapy for patients with diabetes to determine the relative accuracy of quality assessment using simple intermediate outcome versus tightly linked quality measures. RESEARCH DESIGN: Retrospective longitudinal cohort. SUBJECTS: A total of 1154 diabetic patients with an LDL test done between October 1, 1998, and March 31, 1999, in 2 large VA facilities. MEASURES: LDL levels, medication treatment, and explanations for poor quality. RESULTS: Although 27% (307 of 1154) of patients had an LDL >/=130 mg/dL using the simple intermediate outcome measure, only 13% (148 of 1154) were classified as having substandard quality using the tightly linked measure. Among the 159 reclassified to adequate quality, 117 had lipid-lowering medication started or increased within 6 months of an LDL >/=130 mg/dL, 8 were already on high-dose medication, 12 had a repeat LDL <130 mg/dL, and 22 had contraindications to treatment. CONCLUSION: Simple intermediate outcome measures can be an inaccurate reflection of true quality of care, and many patients classified as having substandard quality by "poor control" might actually be receiving good quality of care.