Introduction: The effect of changes in doctors' rosters is rarely subjected to scientific evaluation. We describe how a natural experiment (NE) study design can be used to evaluate if a managerial decision about doctors' rosters has an effect on patient flow in an emergency department (ED). We hypothesized that an extra doctor each morning from 6 a.m. (i.e., a modified "casino shift") might improve the productivity of a hospital's ED. Methods: This was an NE observational study using data on patient flow in the ED of Zealand University Hospital, Denmark, between April 1, 2016, and April 1, 2017. We compared days on which the 6 a.m. Emergency physician called in sick (case days) with data from the same weekday a week later where staffing was as scheduled (control days). Results: Patient caseload did not did differ significantly on days with and without the extra doctor from 6 a.m. (measured by number of admissions, triage scores and mean patient age). Door-to-doctor time was 70 minutes (mean, standard deviation [SD], 49) on days without the extra doctor and 56 minutes (mean, SD 41) on days with the early-morning doctor present (p > 0.05). ED length of stay was 250 minutes (mean, SD 119) on days without the extra doctor and 209 minutes (mean, SD 109) on days with the early-morning doctor present (p > 0.05). Conclusion: In our setting, an extra doctor in the ED from 6 a.m. Did not change patient flow. These results suggest that the workflow in the ED should be viewed as a connected supply chain. The study also demonstrates that a natural experiment study design can be used to evaluate ED managerial decisions.