Interns' Work Hours
http://www.100md.com
《新英格兰医药杂志》
To the Editor: Landrigan et al. (Oct. 28 issue)1 show that reducing interns' work hours in the intensive care unit (ICU) decreased serious medical errors, probably prompting widespread reevaluation of traditional ICU on-call schedules. No one would argue that limiting work hours is not a vitally important topic. However, having experienced the intervention schedule firsthand as intern subjects in the study, we have serious concerns about the authors' conclusions.
Residents in the study worked traditional every-three-day on-call schedules and routinely worked extra hours to cover for the interns. Worried residents and attending physicians, aware that the interns on the intervention schedule were poorly informed, took a more active role in patient care, making the majority of decisions and more closely supervising the interns' actions. This hypervigilance may have strongly biased the study toward a positive result. The authors address this issue by citing the similar numbers of orders and procedures in the two groups, but these tasks are traditionally performed exclusively by interns and are not directly reflective of decision making. One effect not quantified in the study was the interns' learning, which we felt was compromised by the intervention schedule.
Nathan A. Pennell, M.D., Ph.D.
Joyce F. Liu, M.D.
Michael J. Mazzini, M.D.
Brigham and Women's Hospital
Boston, MA 02115
References
Landrigan CP, Rothschild JM, Cronin JW, et al. Effect of reducing interns' work hours on serious medical errors in intensive care units. N Engl J Med 2004;351:1838-1848.
To the Editor: The study by Landrigan et al. demonstrates that, as compared with traditional work hours among ICU interns, reduced hours are associated with a similar number of adverse events but a decreased number of serious medical errors. As the results of these studies are applied toward practical solutions to the 80-hour workweek mandated by the Accreditation Council for Graduate Medical Education (ACGME), it is important to remember that the results were achieved by means of a 33 percent increase in intern staffing on the ICU. Without the hiring of new interns, this means that other members of the intern class either lost free or educational time or faced an increased burden of responsibility on other wards — changes that may translate into a worse quality of life, more medical errors, or more adverse events for the intern class, taken as a whole. Future studies might examine the quality of life per 1000 intern-days, just as Landrigan et al. have examined errors per 1000 patient-days.
Ian G. Harnik, M.D.
Mount Sinai Hospital
New York, NY 10029
ian.harnik@mssm.edu
To the Editor: The studies by Lockley et al. (Oct. 28 issue)1 and Landrigan et al. demonstrate that reducing interns' work hours in the ICU reduced their attentional failures and errors. However, interns' hours were reduced in the ICU by assigning a fourth intern to the rotation. This may have facilitated the benefits that the authors document, since without added staffing, there would have been fewer, albeit better rested, interns providing care. More interns on the ICU rotation meant that there were fewer interns — and conceivably, more errors — elsewhere in the training program.
The fundamental barrier to generalizability is economic. Highly motivated house officers have long been a source of inexpensive hospital labor. Increasing the number of trainees by 33 percent or supplementing them with other professionals would be costly. Although the public should be spared medical errors caused by sleep deprivation, the work-hour mandates of the ACGME do not solve the problem. Their effect on house-staff sleep is an important advance, but their effect on patient safety may be much less than that suggested by these two studies.
Henry E. Fessler, M.D.
Johns Hopkins Medical Institutions
Baltimore, MD 21287
hfessler@jhmi.edu
References
Lockley SW, Cronin JW, Evans EE, et al. Effect of reducing interns' weekly work hours on sleep and attentional failures. N Engl J Med 2004;351:1829-1837.
To the Editor: Landrigan and colleagues used binomial distributions to compare medical error rates during two different work schedules and concluded that the intervention schedule led to fewer medical errors than the traditional schedule (100.1 vs. 136.0 errors per 1000 patient-days, P<0.001). Using this method, the authors assumed that each error (event) was independent, but this assumption does not account for clustering — an important concern, since most interns made multiple mistakes. It is conceivable, for example, that a single intern on the traditional schedule had a bad month, committing most of the errors attributed to the traditional schedule. With this error-prone intern omitted from the analysis, the schedules might have identical error rates. Although this extreme scenario is unlikely, perhaps the authors can reanalyze their data and present a P value that accounts for clustering. Alternatively, since the interns rotated on each schedule, a within-subject paired t-test (with each intern as a separate subject) could be used, as in the companion article by Lockley et al. on attentional failures.
Daniel J. Brotman, M.D.
Cleveland Clinic Foundation
Cleveland, OH 44195
brotmad@ccf.org
To the Editor: In his accompanying editorial on work-hour reductions and interns' errors, Drazen (Oct. 28 issue)1 reports his experience as an attending physician in both groups of the study and states that the interns "often knew very little about the patients who had been admitted the night before." We applaud his concluding suggestions but feel they merit strengthening. The challenge facing medicine practiced by teams of physicians lies in the effective transferral of care from one practitioner to another. As a chief medical resident and a program director, we implemented systems meant to improve work-hour compliance and maintain or improve the quality of care and education. We submit that care transitions are underappreciated as a source of error, understudied as such, and assumed to be a skill learned during residency training. It is not sufficient for senior physicians to model ideal behavior; instead, we must devote resources to developing systems that specifically address these deficiencies. The Society of Hospital Medicine is implementing such a tool,2 but these efforts are in their infancy, at best. Innovative electronic sign-out systems may succeed in this regard.3
Jamie P. Dwyer, M.D.
Vanderbilt University Medical Center
Nashville, TN 37232
Marc D. Cohen, M.D.
Mayo Clinic
Jacksonville, FL 32224
References
Drazen JM. Awake and informed. N Engl J Med 2004;351:1884-1884.
Society of Hospital Medicine quality improvement tools. (Accessed January 28, 2005, at http://www.hospitalmedicine.org/presentation/Improvement.asp.)
Lasslo R, Balsbaugh T, Malyj W. Web-based patient signout system -- evolution of a quality improvement tool. Medinfo 2004;2004:1707-1707.
To the Editor: Dr. Drazen's reflections on errors in the ICU are informative and motivating, including his candid conclusion that a rate of "1 serious, preventable adverse event . . . every three to four weeks . . . is unacceptably high." I agree but would note that the observed rate of 38 such events per 1000 patient-days, reported by Landrigan et al., means that 1 serious, preventable adverse event occurs every two to three days in Dr. Drazen's 10-bed ICU (1000 patient-days divided by 10 patients per day, divided by 38 adverse events equals 2.6 days). Thus, permit me to amplify 10-fold Dr. Drazen's exhortation that we "be awake and informed."
Arthur T. Evans, M.D., M.P.H.
Cook County Hospital
Chicago, IL 60612
aevans@cchil.org
The authors reply: We agree with Drs. Dwyer and Cohen that technological advances may improve care transitions,1 although we would hypothesize that the integration of new technology with teamwork training, a redesign of rounding structures to support shorter work hours, and a cultural shift toward shared patient responsibility is required to decrease sign-out errors substantially. Although we appreciate the concerns regarding redistribution of personnel expressed by Dr. Fessler and Dr. Harnik, the extreme working conditions and error rates in ICUs2 may justify short-term reallocations. Additional hiring will ultimately be needed to prevent errors due to excessive work hours, but since preventable adverse drug events have been shown to cost a 700-bed hospital $2.8 million per year,3 additional hiring will probably prove cost-effective. For perspective, two days of patient billings from a 10-bed ICU exceed the annual salary of the additional intern in our intervention.
We welcome the firsthand impressions of Dr. Pennell and colleagues. Most errors detected in the study, however, were errors not in decision making but, rather, in the execution of previously determined plans (e.g., entering medication orders or performing complete examinations). "Hypervigilance" on the part of residents and attending physicians, although not undesirable, is unlikely to prevent such errors and cannot explain the increase in attentional failures that paralleled the higher error rates during the traditional schedule. Although we did not study resident education, we disagree that decreased time in the hospital necessarily undermines education. Interns on the traditional schedule recurrently experienced acute total sleep deprivation, which significantly impairs learning.4
On a technical note, Dr. Brotman correctly states that the analyses of interns' errors did not account for within-subject effects. This was deliberate. It was necessary to conduct unpaired analyses in order to maintain consistency with analyses of errors in the units as a whole, for which within-subject comparisons were not possible. The extreme scenario postulated by Dr. Brotman did not occur; errors were widely distributed among the interns.
Finally, we wholeheartedly agree with Dr. Fessler that "the work-hour mandates of the ACGME do not solve the problem." First, these standards apply only to trainees; there are no limits for practicing physicians in the United States. Second, our traditional schedule, which met the current ACGME standards of 320 hours every four weeks and 30 consecutive hours every other shift, led to significant decrements in interns' alertness and patient safety. Even our intervention schedule often resulted in 16 to 20 consecutive hours of work, a duration exceeding federal standards in other safety-sensitive industries and associated with a profoundly increased risk of accidents.5 Consequently, we would recommend that future reforms focus primarily on reducing the duration, not the frequency, of extended work shifts. As long as 24-hours shifts remain the norm, it is unlikely that scheduling modifications will lead to substantial gains in patient safety.
Chistopher P. Landrigan, M.D., M.P.H.
Steven W. Lockley, Ph.D.
Charles A. Czeisler, Ph.D., M.D.
Brigham and Women's Hospital
Boston, MA 02115
clandrigan@rics.bwh.edu
for the Harvard Work Hours, Health, and Safety Group
References
Petersen LA, Orav EJ, Teich JM, O'Neil AC, Brennan TA. Using a computerized sign-out program to improve continuity of inpatient care and prevent adverse events. Jt Comm J Qual Improv 1998;24:77-87.
Cullen DJ, Sweitzer BJ, Bates DW, Burdick E, Edmondson A, Leape LL. Preventable adverse drug events in hospitalized patients: a comparative study of intensive care and general care units. Crit Care Med 1997;25:1289-1297.
Bates DW, Spell N, Cullen DJ, et al. The costs of adverse drug events in hospitalized patients. JAMA 1997;277:307-311.
Stickgold R, James L, Hobson JA. Visual discrimination learning requires sleep after training. Nat Neurosci 2000;3:1237-1238.
Department of Transportation, Federal Motor Carrier Safety Administration. Hours of service of drivers; driver rest and sleep for safe operations; proposed rule. Fed Regist 2000;65:25541-25611.
Residents in the study worked traditional every-three-day on-call schedules and routinely worked extra hours to cover for the interns. Worried residents and attending physicians, aware that the interns on the intervention schedule were poorly informed, took a more active role in patient care, making the majority of decisions and more closely supervising the interns' actions. This hypervigilance may have strongly biased the study toward a positive result. The authors address this issue by citing the similar numbers of orders and procedures in the two groups, but these tasks are traditionally performed exclusively by interns and are not directly reflective of decision making. One effect not quantified in the study was the interns' learning, which we felt was compromised by the intervention schedule.
Nathan A. Pennell, M.D., Ph.D.
Joyce F. Liu, M.D.
Michael J. Mazzini, M.D.
Brigham and Women's Hospital
Boston, MA 02115
References
Landrigan CP, Rothschild JM, Cronin JW, et al. Effect of reducing interns' work hours on serious medical errors in intensive care units. N Engl J Med 2004;351:1838-1848.
To the Editor: The study by Landrigan et al. demonstrates that, as compared with traditional work hours among ICU interns, reduced hours are associated with a similar number of adverse events but a decreased number of serious medical errors. As the results of these studies are applied toward practical solutions to the 80-hour workweek mandated by the Accreditation Council for Graduate Medical Education (ACGME), it is important to remember that the results were achieved by means of a 33 percent increase in intern staffing on the ICU. Without the hiring of new interns, this means that other members of the intern class either lost free or educational time or faced an increased burden of responsibility on other wards — changes that may translate into a worse quality of life, more medical errors, or more adverse events for the intern class, taken as a whole. Future studies might examine the quality of life per 1000 intern-days, just as Landrigan et al. have examined errors per 1000 patient-days.
Ian G. Harnik, M.D.
Mount Sinai Hospital
New York, NY 10029
ian.harnik@mssm.edu
To the Editor: The studies by Lockley et al. (Oct. 28 issue)1 and Landrigan et al. demonstrate that reducing interns' work hours in the ICU reduced their attentional failures and errors. However, interns' hours were reduced in the ICU by assigning a fourth intern to the rotation. This may have facilitated the benefits that the authors document, since without added staffing, there would have been fewer, albeit better rested, interns providing care. More interns on the ICU rotation meant that there were fewer interns — and conceivably, more errors — elsewhere in the training program.
The fundamental barrier to generalizability is economic. Highly motivated house officers have long been a source of inexpensive hospital labor. Increasing the number of trainees by 33 percent or supplementing them with other professionals would be costly. Although the public should be spared medical errors caused by sleep deprivation, the work-hour mandates of the ACGME do not solve the problem. Their effect on house-staff sleep is an important advance, but their effect on patient safety may be much less than that suggested by these two studies.
Henry E. Fessler, M.D.
Johns Hopkins Medical Institutions
Baltimore, MD 21287
hfessler@jhmi.edu
References
Lockley SW, Cronin JW, Evans EE, et al. Effect of reducing interns' weekly work hours on sleep and attentional failures. N Engl J Med 2004;351:1829-1837.
To the Editor: Landrigan and colleagues used binomial distributions to compare medical error rates during two different work schedules and concluded that the intervention schedule led to fewer medical errors than the traditional schedule (100.1 vs. 136.0 errors per 1000 patient-days, P<0.001). Using this method, the authors assumed that each error (event) was independent, but this assumption does not account for clustering — an important concern, since most interns made multiple mistakes. It is conceivable, for example, that a single intern on the traditional schedule had a bad month, committing most of the errors attributed to the traditional schedule. With this error-prone intern omitted from the analysis, the schedules might have identical error rates. Although this extreme scenario is unlikely, perhaps the authors can reanalyze their data and present a P value that accounts for clustering. Alternatively, since the interns rotated on each schedule, a within-subject paired t-test (with each intern as a separate subject) could be used, as in the companion article by Lockley et al. on attentional failures.
Daniel J. Brotman, M.D.
Cleveland Clinic Foundation
Cleveland, OH 44195
brotmad@ccf.org
To the Editor: In his accompanying editorial on work-hour reductions and interns' errors, Drazen (Oct. 28 issue)1 reports his experience as an attending physician in both groups of the study and states that the interns "often knew very little about the patients who had been admitted the night before." We applaud his concluding suggestions but feel they merit strengthening. The challenge facing medicine practiced by teams of physicians lies in the effective transferral of care from one practitioner to another. As a chief medical resident and a program director, we implemented systems meant to improve work-hour compliance and maintain or improve the quality of care and education. We submit that care transitions are underappreciated as a source of error, understudied as such, and assumed to be a skill learned during residency training. It is not sufficient for senior physicians to model ideal behavior; instead, we must devote resources to developing systems that specifically address these deficiencies. The Society of Hospital Medicine is implementing such a tool,2 but these efforts are in their infancy, at best. Innovative electronic sign-out systems may succeed in this regard.3
Jamie P. Dwyer, M.D.
Vanderbilt University Medical Center
Nashville, TN 37232
Marc D. Cohen, M.D.
Mayo Clinic
Jacksonville, FL 32224
References
Drazen JM. Awake and informed. N Engl J Med 2004;351:1884-1884.
Society of Hospital Medicine quality improvement tools. (Accessed January 28, 2005, at http://www.hospitalmedicine.org/presentation/Improvement.asp.)
Lasslo R, Balsbaugh T, Malyj W. Web-based patient signout system -- evolution of a quality improvement tool. Medinfo 2004;2004:1707-1707.
To the Editor: Dr. Drazen's reflections on errors in the ICU are informative and motivating, including his candid conclusion that a rate of "1 serious, preventable adverse event . . . every three to four weeks . . . is unacceptably high." I agree but would note that the observed rate of 38 such events per 1000 patient-days, reported by Landrigan et al., means that 1 serious, preventable adverse event occurs every two to three days in Dr. Drazen's 10-bed ICU (1000 patient-days divided by 10 patients per day, divided by 38 adverse events equals 2.6 days). Thus, permit me to amplify 10-fold Dr. Drazen's exhortation that we "be awake and informed."
Arthur T. Evans, M.D., M.P.H.
Cook County Hospital
Chicago, IL 60612
aevans@cchil.org
The authors reply: We agree with Drs. Dwyer and Cohen that technological advances may improve care transitions,1 although we would hypothesize that the integration of new technology with teamwork training, a redesign of rounding structures to support shorter work hours, and a cultural shift toward shared patient responsibility is required to decrease sign-out errors substantially. Although we appreciate the concerns regarding redistribution of personnel expressed by Dr. Fessler and Dr. Harnik, the extreme working conditions and error rates in ICUs2 may justify short-term reallocations. Additional hiring will ultimately be needed to prevent errors due to excessive work hours, but since preventable adverse drug events have been shown to cost a 700-bed hospital $2.8 million per year,3 additional hiring will probably prove cost-effective. For perspective, two days of patient billings from a 10-bed ICU exceed the annual salary of the additional intern in our intervention.
We welcome the firsthand impressions of Dr. Pennell and colleagues. Most errors detected in the study, however, were errors not in decision making but, rather, in the execution of previously determined plans (e.g., entering medication orders or performing complete examinations). "Hypervigilance" on the part of residents and attending physicians, although not undesirable, is unlikely to prevent such errors and cannot explain the increase in attentional failures that paralleled the higher error rates during the traditional schedule. Although we did not study resident education, we disagree that decreased time in the hospital necessarily undermines education. Interns on the traditional schedule recurrently experienced acute total sleep deprivation, which significantly impairs learning.4
On a technical note, Dr. Brotman correctly states that the analyses of interns' errors did not account for within-subject effects. This was deliberate. It was necessary to conduct unpaired analyses in order to maintain consistency with analyses of errors in the units as a whole, for which within-subject comparisons were not possible. The extreme scenario postulated by Dr. Brotman did not occur; errors were widely distributed among the interns.
Finally, we wholeheartedly agree with Dr. Fessler that "the work-hour mandates of the ACGME do not solve the problem." First, these standards apply only to trainees; there are no limits for practicing physicians in the United States. Second, our traditional schedule, which met the current ACGME standards of 320 hours every four weeks and 30 consecutive hours every other shift, led to significant decrements in interns' alertness and patient safety. Even our intervention schedule often resulted in 16 to 20 consecutive hours of work, a duration exceeding federal standards in other safety-sensitive industries and associated with a profoundly increased risk of accidents.5 Consequently, we would recommend that future reforms focus primarily on reducing the duration, not the frequency, of extended work shifts. As long as 24-hours shifts remain the norm, it is unlikely that scheduling modifications will lead to substantial gains in patient safety.
Chistopher P. Landrigan, M.D., M.P.H.
Steven W. Lockley, Ph.D.
Charles A. Czeisler, Ph.D., M.D.
Brigham and Women's Hospital
Boston, MA 02115
clandrigan@rics.bwh.edu
for the Harvard Work Hours, Health, and Safety Group
References
Petersen LA, Orav EJ, Teich JM, O'Neil AC, Brennan TA. Using a computerized sign-out program to improve continuity of inpatient care and prevent adverse events. Jt Comm J Qual Improv 1998;24:77-87.
Cullen DJ, Sweitzer BJ, Bates DW, Burdick E, Edmondson A, Leape LL. Preventable adverse drug events in hospitalized patients: a comparative study of intensive care and general care units. Crit Care Med 1997;25:1289-1297.
Bates DW, Spell N, Cullen DJ, et al. The costs of adverse drug events in hospitalized patients. JAMA 1997;277:307-311.
Stickgold R, James L, Hobson JA. Visual discrimination learning requires sleep after training. Nat Neurosci 2000;3:1237-1238.
Department of Transportation, Federal Motor Carrier Safety Administration. Hours of service of drivers; driver rest and sleep for safe operations; proposed rule. Fed Regist 2000;65:25541-25611.