Tuesday Oct 14, 2014
Latest Study of Resident Work Restrictions Doesn't Settle Debate
A recent study published in Health Affairs(content.healthaffairs.org) reported on data collected from 2003 to 2009 that showed no evidence of a decline in mortality for patients under the care of physicians who trained with the 80-hour workweek restrictions promulgated by the Accreditation Council for Graduate Medical Education (ACGME). On the surface, this may indicate that there is little or no difference between the training received before and after the duty hour restrictions were imposed, but the limitations of data collection and the nature of the study make this conclusion an impossible one to proclaim with any certainty.
This particular study's findings notwithstanding, the issues surrounding duty-hour restrictions carry a high emotional charge. Opinions are numerous, but most folks settle into one of two camps, either for or against the restrictions. Proponents point to the dangers of sleep deprivation and its effects on concentration and focus. Opponents point to the missed opportunities for learning and the problems inherent in frequent patient handoffs. The answer, as in most cases, probably lies somewhere firmly in the middle, but collecting empirical data on these issues is difficult, at best.
At heart, all physicians are scientists. We formulate and test hypotheses -- collecting data, refining the question, and using the results to guide further questions and actions. That said, much like the general public, doctors often misinterpret or overestimate the ability of statistical analysis to "prove" anything. Studies like this one are great at giving us a place to begin, but they are often used to draw broader conclusions than is statistically possible, making the reporting of these data difficult and fraught with misinterpretation. To that end, I'll offer my thoughts on the interpretation that this article espouses.
First, the data came only from Florida. Although the data should be somewhat representative of the country, the diversity seen in training approaches and locations nationwide, even with ACGME standards in place, is far broader than data from a single state can encompass. This limits the generalizability of the data to any group outside that state. For example, there may have been novel programs instituted in Florida to accommodate the new duty-hour restrictions, giving residents trained in Florida a completely different experience than those trained in Colorado or Maine or Nevada.
This possibility is not as far-fetched as it sounds. When the ACGME's 2003 duty-hour restrictions grew even tighter in 2011, my residency program in South Carolina modified schedules to include mandatory nap time, which in effect allowed interns to have overnight call without a "night float" system. Across the country, the night float has become a standard program to compensate for the 16-hour contiguous work restriction on interns, but the modified system we had was far different than any other solution I've seen since. Novel programs such as this were not accounted for directly in this study.
Second, the study has so many possible confounders that being able to accurately conclude anything other than similar mortality before and after the changes is highly improbable. The study authors themselves admit that there were many innovations in medicine during that time, not to mention the increasingly ubiquitous use of health information technology, from electronic health records to Epocrates and Up-to-Date. There is no statistically sound model that can accurately take those use case scenarios into account.
Simply put, any mortality rate change attributable to duty-hour restrictions may have been masked by other changes that could have independently shifted mortality up or down. Isolating the change in mortality due solely to the duty-hour restrictions requires far deeper analysis of the other factors influencing resident performance and patient mortality than were analyzed in this study.
Third, although patient death is an easily measured outcome, it may not be the best indicator of change. Quality of care, patient morbidity, readmission rates, and errors in patient handoffs and communication could all be added to outcomes measures to better understand the impact of duty-hour restrictions, but most of these markers are difficult to measure, link to a specific physician, and parse accurately in a retrospective fashion. However, they may provide a better picture of the overall impact of the duty-hour restrictions. For example, if patient handoffs necessitated by the duty-hour restrictions created more medical errors resulting in morbidity but not mortality, that needs to be depicted as part of the overall impact.
There's little doubt that the debate about duty-hour restrictions and their impact on physicians and patients will continue for now. Coupled with the emotional biases tied to our educational strategies, the idea that "we do it this way because that's the way it's always been done" leads to a highly charged issue.
The good news is that ACGME has acknowledged the limitations of available research on this issue, and more data are coming -- eventually. ACGME announced in March that multicenter trials designed to investigate the effect of resident duty-hour restrictions are underway. Analyses of those studies, which are focused on surgery and internal medicine residencies, are expected to be completed in 2016 and 2019, respectively.
Hopefully, those controlled trials will shed more light on what effect, if any, the restrictions have, moving us past conjecture and assumption to a place of better empirical understanding.
Gerry Tolbert, M.D., is a board-certified family physician who practices in northern Kentucky. A lifelong technophile, his interests include the intersection of medicine and technology. You can follow him on Twitter @DrTolbert.
Posted at 09:37PM Oct 14, 2014 by Gerry Tolbert, M.D.