Should U.S. hospitals go 24/7?
Article Outline
It is hard to escape the notion that, in a service economy, services should be available continuously. People can surf the Web, trade stocks, go to restaurants, and buy a carton of milk 24-hours a day; but, as currently structured, our nation's hospitals are not fully operational on weekends and holidays. During those off-days, it is assumed that emergency care is unaffected, but routine care is certainly restricted. Fewer patients are admitted on weekends, and fewer procedures are performed. As a result, both clinical and nonclinical staffing is reduced to contain costs. There has been concern, however, that the low numbers of essential and ancillary staff may affect care negatively, especially in light of research to suggest that lower staffing across hospitals is related to a higher frequency of adverse events and other poor outcomes 1, 2, 3. The leap of faith is that a similar effect on quality will occur within hospitals when staffing is reduced on weekends.
In this issue of the Journal, Cram and colleagues found that hospitals in California in 1998 had higher mortality rates on weekends. This research adds to a limited but growing body of evidence on the relationship between staffing and the safety of patients (4). Bell and Redelmeier were among the first to show mortality effects among patients admitted on weekends for a selective group of conditions in Canada. More recently, also in California, Gould et al found fewer births but higher neonatal mortality on weekends (5).
The major limitation to any nonexperimental study of differences in quality and outcomes is the suspicion that unmeasured aspects of case mix may confound the analysis. Hospital workers know that patients who present on weekends, even to the emergency room, are often triaged to weekdays, unless they necessarily require a weekend admission (6). Indeed, the ratio of admissions occurring on weekends to those occurring on weekdays is routinely well below the expected 40% ratio (2 days vs. 5 days) if admissions were proportional throughout the week. Cram et al deserve credit for applying restrictive cohort criteria to isolate a study group that is most likely to be homogeneous. By focusing on unscheduled admissions through the emergency department, the authors achieved a ratio of weekend to weekday admissions of 39%, thus reducing the expected weekend bias (and the effect). They also estimated regression coefficients for the risk of mortality associated with various comorbid conditions and used them to adjust for differences in severity. They found a small increase (about 3%) in mortality during weekends. Following the approach of other researchers, this study also examined results for individual conditions. However, only three of 50 high mortality diagnoses displayed significant weekend effects, despite a sample size in excess of 1.5 million patients. The authors did not provide weekend to weekday ratios for individual diagnoses, so it is not known whether an admission bias might exist for these individual conditions.
Although the overall result was modest, one finding that sets this study apart is that a significant weekend effect existed among major teaching hospitals, but not among minor or nonteaching hospitals (4, Table 4). The authors attribute the differential effect to “a confluence of factors that ultimately result in reduced quality of care”. Certainly another possible explanation is a difference in unmeasured severity. Controlling for cross-hospital severity is even more important than controlling for within-hospital severity. One study found patient characteristics to be 315 times more important than hospital characteristics in predicting mortality (7), and patients at teaching hospitals tend to be more acutely ill than patients at nonteaching hospitals. Given the modest weekend effect found in this study, as well as the study's reliance on administrative data, even small errors in the measurement of the case mix could result in a spurious conclusion. Another possible contributing factor that may account for the increased weekend effect in teaching hospitals is the July phenomenon. Inefficiencies and errors are considered by some to be disproportionately present during July and August in teaching hospitals compared with nonteaching hospitals due to the presence of new residents, who begin their training during those months 8, 9, 10. It would be illuminating to see whether the teaching-weekend effect is also greater in July and August than during the rest of the year.
Even with more detailed clinical data and different analyses, it may never be possible to provide a definitive answer about the weekend effect (6). Not only do unmeasured case mix differences loom large, but it is difficult to measure staffing patterns precisely, much less to determine the effects of training and experience on quality, or to assess the interactions that take place between staff and patients or between clinical and nonclinical staff. The authors remark, “All healthcare workers should take extra vigilance when caring for patients outside of traditional working hours.” (4). Further research would be particularly useful here. For example, would more nursing staff or more highly trained nurses on weekends eliminate the weekend effect? Would more physicians help? If other researchers confirm weekend effects for particular diagnoses, would increasing staff for just the units treating those patients be sufficient?
A related and important policy question is whether, in addition to addressing concerns surrounding safety, hospitals should begin to reengineer themselves toward a “Seven-Day Operation”. Hospitals are extremely expensive entities with high fixed costs. To use these resources efficiently, perhaps we should not “waste” weekend days by failing to offer a full range of services. One study found that cardiac patients who were not quite ready to be discharged on Friday usually stayed over the weekend and through Tuesday (the extra day was assumed to be needed for a post-weekend workup) (11). If emergency department physicians work on weekends and overnights, can other specialists do the same? Given such unintended cost consequences of weekend slowdowns, more research is needed to determine whether 7-day operations, if managed efficiently, would be more or less expensive than the current method. If future research provides evidence that patients are at higher risk of injury or death on weekends, then changing to a 7-day operation might save both lives and money.
Providing optimal care regardless of the day of week is an important goal for hospitals. The question all providers should ask themselves is “Would you send your mother to this hospital on a weekend?” If the answer is no or maybe, we should start changing how things are done.
References
- . Hospital nurse staffing and patient mortality, nurse burnout, and job dissatisfaction. JAMA. 2002;288:1987–1993
- . Nurse-staffing levels and the quality of care in hospitals. [comment]. New Engl J Med. 2002;346:1715–1722
- . Physician staffing patterns and clinical outcomes in critically ill patients (a systematic review). JAMA. 2002;288:2151–2162
- . Effects of weekend admission and hospital teaching status on in-hospital mortality. Am J Med. 2004;117:151–157
- . Neonatal mortality in weekend vs weekday births. JAMA. 2003;289:2958–2962
- . Why do hospital death rates vary?. N Engl J Med. 2001;345:692–694
- Silber JH, Rosenbaum PR. A spurious correlation between hospital mortality and complication rates: the importance of severity adjustment. Med Care. Oct 1997;35(Suppl):OS77-OS92
- . The relationship of house staff experience to the cost and quality of inpatient care. JAMA. 1990;263:953–957
- . The killing season–fact or fiction?. BMJ. 1994;309:1690
- . Indirect costs for medical education. Is there a July phenomenon?. Arch Intern Med. 1989;149:765–768
- . Restricted weekend service inappropriately delays discharge after acute myocardial infarction. Heart. 2002;87:216–219
PII: S0002-9343(04)00254-2
doi:10.1016/j.amjmed.2004.04.005
© 2004 Elsevier Inc. All rights reserved.
