A comparison of the referral rates of trainees and trainers in an academic teaching practice

Abstract

Objective: To compare the referral rates of trainee and trainer family physicians. Setting: The practice of the Department of Family Medicine at the Medical University of Southern Africa, Pretoria. Design: Analysis of 43 028 problem encounters selected from one in-service practice database. Participants: Thirteen junior registrars, seven senior registrars in a Master's programme and seven senior physicians. Main measure: Referral rates compared by the Generalized Linear Mixed Model to allow for case mix and variation between the three study groups. Results: Adjusted referral rates per thousand problem encounters were 97.7 for junior registrars (95% CI 79.4 - 120.7), 77.1 for senior registrars (95%CI 59.3 - 99.5) and 73.7 for senior physicians (95% CI 54.4 - 99.2). Differences between the groups were not statistically significant (Wald chi-square = 3.90; df = 2; P = 0.195). There was insufficient evidence to show that the large amount of variation in the referral rates of doctors within study groups was different between the three groups. Conclusions: Using a performance-oriented database and an advanced method for adjusting for case mix makes a difference to referral rates. There was no significant difference between the mean referral rates of trainees and trainers. There was a large amount of variation within all three groups. Together, these findings support the thesis that factors other than clinical diagnosis in the behaviour of doctors or their interaction with patients are determinants of the referral decision. This points to the value of peer reviewing of referral rates for both trainees and trainers during vocational training, as well as in group practices.