Article type
Year
Abstract
Background: Financial Contribution from the Health Transition Fund, Health Canada Evaluations of educational interventions to improve physician performance are often poorly designed and measure questionable outcomes. Objectives: To conduct an educational trial using rigorous scientific principles, including randomization, control conditions, blinding, clinically relevant and statistically sensitive outcomes, and a priori specification of hypotheses and analyses.
Methods: The setting was in British Columbia (BC) urban and small-town family practices. We studied physicians (n=202) in 28 small groups already enrolled in a case-based learning program, who volunteered for the trial. A factorial design with randomization of matched groups of physicians to four blocks was used. The blocks EP, EN, CP, and CN, were defined as follows. Evidence-based case-based self-instructional modules (E) of 6-8 pages, versus unrelated control modules of the group's choice (C). One-page prescribing portraits (P) with histograms comparing individual and group prescribing rates, each with a succinct evidence-based message, versus no portraits (N). E and P covered 4 clinical conditions: hypertension, type-2 diabetes, otitis media, and congestive heart failure. Outcome measures included: 1) Changes in prescribing of target drugs to new patients (a more sensitive outcome than all prescriptions) measured using the PharmaNet database, which captures all prescriptions in BC. 2) Qualitative and semi-quantitative measures of physician attitudes.
Results: Randomization balanced baseline characteristics in the 4 blocks: average group size (6.4 to 8.4 physicians), percent female (39% to 45%), percent solo practices (16% to 19%), mean years in practice (14 to 16), and mean patient visits per week (113 to 121). Analysis of hypertension showed that both interventions increased thiazide (1.25 [1.03,1.46]) and decreased calcium channel blocker (0.50 [0.36,0.72]) prescribing to new patients.
Conclusions: Proper conduct of educational trials is necessary and feasible to advance knowledge of what influences prescribing. Complicated interventions can have significant impacts on sensitive measures of physician prescribing.
Methods: The setting was in British Columbia (BC) urban and small-town family practices. We studied physicians (n=202) in 28 small groups already enrolled in a case-based learning program, who volunteered for the trial. A factorial design with randomization of matched groups of physicians to four blocks was used. The blocks EP, EN, CP, and CN, were defined as follows. Evidence-based case-based self-instructional modules (E) of 6-8 pages, versus unrelated control modules of the group's choice (C). One-page prescribing portraits (P) with histograms comparing individual and group prescribing rates, each with a succinct evidence-based message, versus no portraits (N). E and P covered 4 clinical conditions: hypertension, type-2 diabetes, otitis media, and congestive heart failure. Outcome measures included: 1) Changes in prescribing of target drugs to new patients (a more sensitive outcome than all prescriptions) measured using the PharmaNet database, which captures all prescriptions in BC. 2) Qualitative and semi-quantitative measures of physician attitudes.
Results: Randomization balanced baseline characteristics in the 4 blocks: average group size (6.4 to 8.4 physicians), percent female (39% to 45%), percent solo practices (16% to 19%), mean years in practice (14 to 16), and mean patient visits per week (113 to 121). Analysis of hypertension showed that both interventions increased thiazide (1.25 [1.03,1.46]) and decreased calcium channel blocker (0.50 [0.36,0.72]) prescribing to new patients.
Conclusions: Proper conduct of educational trials is necessary and feasible to advance knowledge of what influences prescribing. Complicated interventions can have significant impacts on sensitive measures of physician prescribing.