Petersen and associates [1] conclude that cross-coverage increases the risk for adverse events; therefore, efforts to improve housestaff working hours, which require increased cross-coverage, will lead to more adverse events. The problem with this reasoning is that their study does not adequately prove either that cross-coverage increases errors or that reform of resident work hours will worsen the error rate.

To prove that errors increase with improved resident work hours, the authors should have compared adverse events in patients on a medical service in a hospital without cross-coverage with patients in a hospital with cross-coverage and better-rested physicians. Their control group consisted instead of patients cared for by relatively well-rested house officers. Therefore, no conclusions can be drawn about the relation between improved work hours and the error rate.

Further, the data do not convincingly show a causal relation between cross-coverage and increased adverse outcomes. If one considers the gross rate of adverse events rather than a comparison of cases with controls, only 26% of all adverse events occurred during cross-coverage hours, even though cross-coverage accounted for 30% of the time during which patients received care. Even these numbers are suspect, given that many adverse events are the result of multiple, cumulative decisions. This makes it impossible to accurately determine when the decision leading to the adverse event was made.

Petersen and colleagues are correct when they state that resident work hours reform is "a laudable attempt to make the training process more humane." Their study does not provide any proof that policymakers should limit reform efforts because of increased error rates during cross-coverage.