Browne et al. used a very interesting approach to begin to understand possible unintended consequences of the regulations of the Accreditation Council for Graduate Medical Education (ACGME) with regard to resident duty hours—they made use of the Nationwide Inpatient Sample (NIS), the largest database on all-payer hospital inpatient care in the United States (including data for patients with Medicare, Medicaid, and private insurance as well as data for patients without insurance). Previous studies have made use of data from patients with Medicare only, patients in U.S. Veterans Affairs facilities, or patients in individual hospitals. The NIS database is obviously more inclusive of all hospitalized patients and most likely better reflects any changes in inpatient care due to resident duty-hour reform.
The authors have also approached the issue of unintended consequences of duty-hour restrictions on patient care with regard to a specific common orthopaedic condition—hip fracture. This is the only paper that I am aware of that focuses on orthopaedic surgery—other papers have focused on medical or general surgical services. Also, importantly, the authors have wisely selected a common and costly condition to study—hip fractures. Such a study has implications for and commonalities with all orthopaedic residency programs.
In their unique approach, the authors also compared teaching hospitals with nonteaching hospitals, which is made possible when using the NIS database. They compared teaching hospitals with themselves before and after duty hours became regulated (July 2003), nonteaching hospitals with themselves before and after July 2003, and teaching hospitals with nonteaching hospitals before and after July 2003. There are three concerns that I have with this method: (1) They included rural hospitals with nonteaching hospitals. Are there rural hospitals that have residents rotating on orthopaedic services? If so, these hospitals should have been excluded or added to the teaching hospital category. (2) After data acquisition, 64% (31,002) of the patients were from nonteaching hospitals and only 36% (17,428) were from teaching hospitals. It would have been better if the two patient populations were of similar size. (3) The authors admit that they did not know if the teaching hospitals had an existing orthopaedic residency. Because their findings have important implications, I think the authors should be absolutely accurate regarding whether or not any of the hospitals, teaching or nonteaching, had an active orthopaedic residency program.
Preventing fatigue in resident physicians who have enormous responsibility while caring for even the very sickest patients appears to be the correct path. While many educators support this concept, other educators do not, stating that such policies fragment care and create shift work for physicians-in-training. This results in these young physicians continuing this practice into their careers in both medicine and surgery and, overall, it weakens their postgraduate training by limiting the number, variety, and even total duration of surgical procedures that surgical residents can experience. Interestingly, most orthopaedic surgeons and surgeons in general fall into one of these two camps—mostly the latter. Not many surgeons fall "in-between." It is also interesting that the "rules" only apply to residents-in-training. The medical profession in the United States has not called for a reduction or limitation of work hours for practicing physicians and/or surgeons as has occurred in Western Europe and Great Britain. To the public, it must be confusing to see this dichotomy—that is, residents who are given ten hours off duty while attending surgeons are allowed to work as long as they want, no matter how tired.
There have been earlier attempts to determine any untoward effects of limiting residents' work hours. These attempts have mainly involved the reporting of mortality rates before and after the ACGME ruling. There were no consistent significant findings for either medical or surgical patients. The authors of this paper also did not report any significant changes in patient mortality in teaching or nonteaching hospitals following the implementation of the ACGME duty-hours regulations. They did, however, report a significant increase in morbidity (pneumonia [p = 0.04], hematoma [p = 0.03], need for transfusion [p < 0.01], and renal complications [p = 0.03]) in patients with hip fractures in teaching hospitals after July 2003. Significant increases in inflation-adjusted cost (p < 0.01) and length of stay (p < 0.02) were also reported in teaching hospitals after July 2003. Interestingly, however, using the Deyo index1,2, the authors found that there was an overall increase in the morbidity of these patients in both teaching and nonteaching hospitals after July 2003. Only by making use of regression and bivariate analyses were they able to demonstrate "significant differences over time between teaching and nonteaching facilities" with these above-listed morbidities, costs, and lengths of stay. Otherwise it appears that morbidity increased in patients in both types of facilities between the two study periods—an increase in the Deyo Index by 33% in nonteaching hospitals and by 48% in teaching hospitals. Why the more recently treated patients were sicker is not known. Teaching hospitals had a larger percentage of male patients with hip fractures. Were they sicker than the female patients in nonteaching hospitals? Teaching hospitals also had a larger percentage of black and "other" minority (not Hispanic) patients than nonteaching hospitals had. Were these groups sicker? Furthermore, teaching hospitals had a larger percentage of patients with Medicaid than nonteaching hospitals had. Were these patients sicker or did they have more untreated medical problems? The geographic distribution of patients in teaching and nonteaching hospitals was dissimilar. Could there be an unrecognized effect related to morbidity in these two groups?
There are additional weaknesses of the study. The selection of the years 2004 and 2005 was relatively soon after ACGME reform. Many programs were most likely not in conformity, some programs were actively adjusting to the changes, and many residents, for different reasons, were not accurately reporting their duty hours. The authors do not mention patients who sustained polytrauma; patients with polytrauma have a higher rate of complications than do patients with an isolated hip fracture. The authors also do not take into account the role of case managers, the use of physician extenders such as physician assistants and nurse practitioners, the use of hospitalists, the role of fellows, any effects of an increased number of residents in a program, and, importantly, the change in rotations for residents, such as the night float system.
In summary, this paper is an important first—a study of possible unintended consequences (i.e., increased patient morbidity) in orthopaedic surgery (i.e., hip fractures) as a result of the ACGME resident duty hours regulations, which became effective in July 2003. It is the first of many expected studies that will be performed to determine if limiting resident duty hours is detrimental or beneficial to patient care. But this study raises more questions than it answers. I do not believe that it provides the evidence to reverse, modify, or stop duty-hour limitations; however, it will be the start of a healthy debate about the issue. Larger and even more detailed studies are needed to provide the best evidence regarding the impact on patient care that may result from the limitation of work hours for residents in training. If and when there is ample evidence that there is a need to prevent medical errors that result in increased patient morbidity, increased mortality, added costs, and longer hospital stays due to fatigue in all physicians—not just residents—it will be essential for our profession to correct the problem.
*The author did not receive any outside funding or grants in support of his research for or preparation of this work. Neither he nor a member of his immediate family received payments or other benefits or a commitment or agreement to provide such benefits from a commercial entity.
1. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol. 1992;45:613-9.
2. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chron Dis. 1987;40:373-83.