Sensitivity of patient outcomes to pharmacist interventions. Part I: systematic review and meta-analysis in diabetes management.

Background: Pharmacists participate in managing diabetes therapy. Despite many reviews, few have quantified the impact of pharmacists' interventions.

Objectives: To identify outcomes sensitive to pharmacists' interventions and quantify their impact through critical literature review.

Methods: All original research describing the impact of pharmacists' interventions in the management of diabetic pharmacotherapy was sought in International Pharmaceutical Abstracts, MEDLINE, Embase, Cochrane Register, and Cumulative Index to Nursing & Allied Health Literature from inception through 2006. Two independent reviewers identified articles, compared results, and settled differences through consensus. The Downs-Black scale was used to assess quality. Data included intervention type, patient numbers, demographics, study characteristics, instruments used, data compared, and outcomes reported. A random-effects meta-analysis combined amenable results.

Results: Of 302 articles identified, 108 involved pharmacists' interventions; 36 addressed diabetes (14 medical clinics, 11 community pharmacies, 7 ambulatory care clinics, 4 hospital wards, 1 physician's office, 1 prison, and 3 in both medical clinics and community pharmacies; 1 did not describe its practice site). Research designs included randomized (n = 18) and nonrandomized (n = 9) controlled trials, pre- and postobservational cohorts (n = 2), retrospective cohort study (n = 1), chart reviews (n = 5), and database study (n = 1). Diabetes education (69%) and medication management (61%) were the most frequently used interventions. Mean +/- SD quality was 62 +/- 11% (fair). Fifty-one (69%) study results were sensitive. Meta-analysis of data from 2247 patients in 16 studies found a significant reduction in hemoglobin A1C (A1C) levels in the pharmacists' intervention group (1.00 +/- 0.28%; p < 0.001) but not in controls (0.28 +/- 0.29%; p = 0.335). Pharmacists' interventions further reduced A1C values 0.62 +/- 0.29% (p = 0.03) over controls.

Conclusions: A1C is sensitive to pharmacists' interventions. Several potentially sensitive outcomes were identified, but too few studies were available for quantitative summaries. More research is needed.