Comparison of housestaff's estimates of their workday activities with results of a random work-sampling study.

BACKGROUND: Accurately quantifying housestaff's workday activities is acquiring increasing importance as resources become constrained and programs become more accountable for medical education. The authors compared a traditional method of time analysis based on housestaff's estimates of how they spent their workdays with the results of a formal time-analysis study based on random work sampling.
METHOD: All housestaff (18 interns and 18 residents) rotating on a general medicine service at Duke University Medical Center between December 1991 and March 1992 participated in the study. Twenty-six of the housestaff first provided estimates of how they spent their workdays, and then all 36 wore random reminder beepers and recorded what they were doing (activity) and with whom (contact) at each beep.
RESULTS: The housestaff overestimated the amounts of time spent in patient evaluation (e.g., the mean estimated proportion of time spent performing histories and physical examinations was 29%, whereas the mean actual proportion was 17%) and in educational activities (e.g., the mean estimated proportion of reading time was 8.4%, whereas the mean actual proportion was 2.7%). The housestaff underestimated the amount of supervision by attending physicians: the mean estimated proportion was 7.7%, whereas the mean actual proportion was 16.9%.
CONCLUSION: The Housestaff's estimates of workday times differed from the observed times measured by random work sampling. These inaccuracies were manifest in several important areas, such as patient evaluation, educational activities, and attending physicians' supervision. These results suggest that program directors who seek to describe housestaff's work activities or wish to determine the effects of administrative interventions should use random work sampling as the measure.