Self-reported versus actual test ordering behavior among primary care clinicians.

To determine the accuracy of clinician self-reports, 38 clinicians were surveyed in a university-based health maintenance organization (HMO) practice on their thyroid function testing (TFT) patterns one year after the distribution of an educational intervention (the intervention). The mean of the absolute difference between self-reported and actual test ordering rates was 12%; the difference was greater than 20% for only 3 of 33 clinicians who ordered TFT for at least five patients during the study period. A strong association was found between self-reported change and the actual mean change in TFT postintervention ordering rates (p less than 0.0001). Results suggest that clinicians, at least in certain settings, can accurately estimate their utilization patterns and the effect of practice guidelines.