Abstract:

Objective
Our objective was to assess the effectiveness of e-learning interventions among physicians for clinical practice and patient outcomes.

Search strategy
We conducted a systematic review of randomized controlled trials (RCTs). The search strategy was created and assessed by three independent research librarians. Two investigators conducted the search independently using the following computerized databases: MEDLINE, EMBASE, and ERIC, from root to December 2008.

Selection criteria
The studies selected were RCTs investigating practicing physicians’ clinical behavior (Kirkpatrick level III) or patient outcome (level IV) after the physicians received an e-learning intervention with comparison to traditional learning or no intervention. Studies were considered if the e-learning intervention was online, on-time, structured, and interactive.

Data collection and analysis
Two authors with expertise in health research methodology and health education independently conducted citation identification, study selection, quality assessment, and data abstraction using pre-piloted forms. Agreement was assessed and when consensus was not reached, a third author resolved disagreements. We calculated standardized mean difference (SMD) for continuous data and relative risk (RR) for dichotomous data using a random-effects model.

Main results
Of the nine selected RCTs that fulfilled our eligibility criteria, seven were published, one was unpublished and received from an author, and one was a protocol of a trial in progress. Our study revealed moderate quality of evidence that e-learning compared to traditional learning improves physicians’ quality of care by increasing the proportion of patients receiving appropriate treatment or screening; the size of the effect was small (pooled relative risk estimate from four trials: 0.96, 95% confidence interval 0.95 to 0.98). Other outcomes consistently favored the e-learning group over traditional learning. The six trials comparing e-learning interventions with no intervention were generally of low methodological quality and had mixed results with three RCTs showing no significant difference and three demonstrating a significant difference in favor of the e-learning intervention. No RCTs assessed the effectiveness of e-learning for patient outcomes.

Authors' conclusions
This review has shown moderate quality of evidence from RCTs that e-learning is more beneficial than (or at least as beneficial as) traditional learning in improving physician patient care behavior. The research comparing e-learning to no intervention is inconclusive due to methodological weaknesses and clinical heterogeneity across studies.