Objective: To determine if a preceptor and individualized feedback improves the performance of physicians in searching MEDLINE in clinical settings.

Design: Randomized controlled trial with 2 to 10 months follow-up.

Setting: A 300-bed teaching hospital.

Participants: All 392 physicians and physicians-in-training from 6 major clinical departments were invited to participate if they made patient-care decisions during the study period; 79.4% agreed.

Interventions: All participants were given 2 hours of basic training, then randomized to a control group (no further intervention) or an intervention group in which each person chose a clinical preceptor experienced in MEDLINE searching and received individualized feedback from a study librarian on each of their 1st 10 searches.

Main outcome measures: The number and proportion of relevant and irrelevant references retrieved for 1st, 4th, and 8th searches of participants were compared with independent librarian searches on the same topics.

Results: Intervention group members did not search more often than controls (5.9 searches per month versus 4.7, respectively; P = 0.26) and there were no significant differences in the quality of searches. Rather, search performance for both groups improved, with the average number of relevant references retrieved per search increasing from 4.5 to 7.4 (P < 0.01). The librarian retrieved more relevant citations than participants for the 1st search (P = 0.001) but not for the 4th (P = 0.60) or 8th (P = 0.76) searches.

Conclusions: A program of assigning preceptors and providing feedback on individual searches did not enhance the quantity or quality of searches. Soon after a basic introduction to searching, however, clinicians in both groups improved their search performance.