Time-of-day changes in physician clinical decision making: A retrospective study

Background Time of day has been associated with variations in certain clinical practices such as cancer screening rates. In this study, we assessed how more general process measures of physician activity, particularly rates of diagnostic test ordering and diagnostic assessments, might be affected by time of day. Methods We conducted a retrospective chart review of 3,342 appointments by 20 attending physicians at five outpatient clinics, matching appointments by physician and comparing the average diagnostic tests ordered and average diagnoses assessed per appointment in the first hour of the day versus the last hour of the day. Statistical analyses used sign tests, two-sample t-tests, Wilcoxon tests, Kruskal Wallis tests, and multivariate linear regression. Results Examining physicians individually, four and six physicians, respectively, had statistically significant first- versus last-hour differences in the number of diagnostic tests ordered and number of diagnoses assessed per patient visit (p <= 0.04). As a group, 16 of 20 physicians ordered more tests on average in the first versus last hour (p = 0.012 for equal chance to order more in each time period). Substantial intra-clinic heterogeneity was found in both outcomes for four of five clinics (p < 0.01). Conclusions There is some statistical evidence on an individual and group level to support the presence of time-of-day effects on the number of diagnostic tests ordered per patient visit. These findings suggest that time of day may be a factor influencing fundamental physician behavior and processes. Notably, many physicians exhibited significant variation in the primary outcomes compared to same-specialty peers. Additional work is necessary to clarify temporal and inter-physician variation in the outcomes of interest.