Pharmacist recommendations to improve the quality of diabetes care: a randomized controlled trial

Abstract

Background: Diabetes mellitus continues to result in substantial morbidity and mortality despite receiving much attention from health care providers. Automated clinician reminder systems have been developed to improve adherence to diabetes care guidelines, but these reminder systems do not always provide actionable information and may be unable to detect relevant, subjective patient information that affects clinical decision making. Face-to-face visits with pharmacists, who have knowledge of care guidelines and medication management strategies, may assist in improving diabetes care. It is unknown if the combination of pharmacist chart review and clinician reminders could improve diabetes care without requiring face-to-face visits.

Objective: To assess the effects of a comprehensive, pharmacistdelivered, primary care, physician-focused intervention in a large hospital based primary care practice to improve the quality of care for patients with diabetes including rates of semiannual hemoglobin A1c testing and other biomarker and process measures.

Methods: This was a prospective, randomized, controlled study conducted in a hospital-based, primary care practice, composed of 37 faculty primary care physicians (PCPs) and 95 internal medicine residents. The initial sample included 346 patients with diabetes and 72 PCPs caring for them. PCPs were randomized to receive either a personalized letter from a practicing pharmacist containing treatment recommendations for patients with upcoming primary care visits (intervention, n = 33) or to usual care without the letters (control, n = 39). The letter included patient-specific recommendations regarding overdue testing as well as drug therapy to achieve diabetes-related treatment targets. The intervention included addition of the letter to the electronic medical record (EMR) and presentation of the letter to the PCP at the time of the index primary care visit that occurred between November 2003 and August 2004. Follow-up chart review was performed after the primary care visit to determine changes in 5 process and 3 biomarker outcome measures of diabetes care within 30 days of the index visit. The primary study outcome was a process measure, change in rates of semiannual A1c testing from baseline to 30-day follow-up. Baseline differences were tested for statistical significance using Pearson chisquare. The statistical significance of the intervention's effect was tested using logistic regression models predicting achievement of each study outcome, with randomization status (intervention vs. control) as the predictor variable of interest, controlling for baseline performance for each measure.

Results: 171 patients were in the 4 medical clinic suites with 33 PCPs who received the intervention, and 175 patients were in the 4 suites with 39 PCPs in usual care. 30-day outcomes were analyzed for 301 patients (87.0%) who attended their scheduled index primary care visit. Of these 301 patients, 44.5% were black, 65.8% were female, and the mean age was 63 years. At baseline, there were no significant differences between the intervention group (n = 150) and the usual care (control) group (n = 151) in the 3 biomarker measures (proportion with A1c less than 7%, proportion with low-density lipoprotein cholesterol [LDL-C] less than 100 milligrams per deciliter [mg per dL], or blood pressure less than 130/80 millimeters mercury [mm Hg]). There were no significant baseline differences in 4 of the 5 process measures; however, the rate of annual LDL-C testing was significantly higher for the intervention than for the control group at baseline (86.0% vs. 74.8%, respectively, P = 0.015). In logistic regression analysis, rates of semiannual A1c testing were not significantly different between the intervention and control groups, increasing from baseline to follow-up by 16% in the intervention group and 9% in the control group (P = 0.146). The proportion of patients with A1c less than 7% at follow-up was 43.3% in the intervention group versus 37.7% in the control group (intervention effect P = 0.099). The only statistically significant difference between the 2 groups in the 8 outcome measures was a higher proportion with an annual eye exam at follow-up in the intervention group (60.0%) versus the usual care group (50.3%, intervention effect P = 0.017).

Conclusions: Pharmacist-generated recommendations delivered by letter to PCPs in an academic medical practice were not associated with statistically significant improvements in most quality measures for diabetes care assessed at 30 days following the intervention. Further research is needed with more patients and a longer follow-up time to determine how best to improve the quality of care of patients with diabetes using focused recommendations for therapy changes and reminder notices to clinicians.