The effect of duty hour regulations on outcomes of neurological surgery in training hospitals in the United States: duty hour regulations and patient outcomes

Clinical article

Free access

Object

The effects of sleep deprivation on performance have been well documented and have led to changes in duty hour regulation. New York State implemented stricter duty hours in 1989 after sleep deprivation among residents was thought to have contributed to a patient's death. The goal of this study was to determine if increased regulation of resident duty hours results in measurable changes in patient outcomes.

Methods

Using the Nationwide Inpatient Sample (NIS), patients undergoing neurosurgical procedures at hospitals with neurosurgery training programs were identified and screened for in-hospital complications, in-hospital procedures, discharge disposition, and in-hospital mortality. Comparisons in the above outcomes were made between New York hospitals and non–New York hospitals before and after the Accreditation Council for Graduate Medical Education (ACGME) regulations were put into effect in 2003.

Results

Analysis of discharge disposition demonstrated that 81.9% of patients in the New York group 2000–2002 were discharged to home compared with 84.1% in the non–New York group 2000–2002 (p = 0.6, adjusted multivariate analysis). In-hospital mortality did not significantly differ (p = 0.7). After the regulations were implemented, there was a nonsignificant decrease in patients discharged to home in the non–New York group: 84.1% of patients in the 2000–2002 group compared with 81.5% in the 2004–2006 group (p = 0.6). In-hospital mortality did not significantly change (p = 0.9). In New York there was no significant change in patient outcomes with the implementation of the regulations; 81.9% of patients in the 2000–2002 group were discharged to home compared with 78.0% in the 2004–2006 group (p = 0.3). In-hospital mortality did not significantly change (p = 0.4). After the regulations were in place, analysis of discharge disposition demonstrated that 81.5% of patients in the non–New York group 2004–2006 were discharged to home compared with 78.0% in the New York group 2004–2006 (p = 0.01). In-hospital mortality was not significantly different (p = 0.3).

Conclusions

Regulation of resident duty hours has not resulted in significant changes in outcomes among neurosurgical patients.

Abbreviations used in this paper:ACGME = Accreditation Committee on Graduate Medical Education; AHRQ = Agency for Healthcare Research and Quality; APR-DRG = All Patient Refined Diagnosis Related Group; HCUP = Healthcare Cost and Utilization Project; ICH = intracerebral hemorrhage; LOS = length of stay; MI = myocardial infarction; NIS = Nationwide Inpatient Sample; SEER = Surveillance, Epidemiology, and End Results; UTI = urinary tract infection.

Abstract

Object

The effects of sleep deprivation on performance have been well documented and have led to changes in duty hour regulation. New York State implemented stricter duty hours in 1989 after sleep deprivation among residents was thought to have contributed to a patient's death. The goal of this study was to determine if increased regulation of resident duty hours results in measurable changes in patient outcomes.

Methods

Using the Nationwide Inpatient Sample (NIS), patients undergoing neurosurgical procedures at hospitals with neurosurgery training programs were identified and screened for in-hospital complications, in-hospital procedures, discharge disposition, and in-hospital mortality. Comparisons in the above outcomes were made between New York hospitals and non–New York hospitals before and after the Accreditation Council for Graduate Medical Education (ACGME) regulations were put into effect in 2003.

Results

Analysis of discharge disposition demonstrated that 81.9% of patients in the New York group 2000–2002 were discharged to home compared with 84.1% in the non–New York group 2000–2002 (p = 0.6, adjusted multivariate analysis). In-hospital mortality did not significantly differ (p = 0.7). After the regulations were implemented, there was a nonsignificant decrease in patients discharged to home in the non–New York group: 84.1% of patients in the 2000–2002 group compared with 81.5% in the 2004–2006 group (p = 0.6). In-hospital mortality did not significantly change (p = 0.9). In New York there was no significant change in patient outcomes with the implementation of the regulations; 81.9% of patients in the 2000–2002 group were discharged to home compared with 78.0% in the 2004–2006 group (p = 0.3). In-hospital mortality did not significantly change (p = 0.4). After the regulations were in place, analysis of discharge disposition demonstrated that 81.5% of patients in the non–New York group 2004–2006 were discharged to home compared with 78.0% in the New York group 2004–2006 (p = 0.01). In-hospital mortality was not significantly different (p = 0.3).

Conclusions

Regulation of resident duty hours has not resulted in significant changes in outcomes among neurosurgical patients.

Adoption of resident duty hour regulations by the Accreditation Committee on Graduate Medical Education (ACGME) culminated after decades of rising concern regarding the effect of strenuous work schedules on the performance of residents in training.1 Despite over 9 years of regulation, there remain relatively scarce data regarding the impact of resident duty hour regulations on clinical outcomes. Several research publications and professional advocacy groups had favored the adoption of the recent ACGME resident duty hour regulations, citing concerns such as resident quality of life,16 patient safety,16 and fatigue-related decline in clinical performance.22,24,26,29,34 A recent review of the sleep deprivation literature concluded that physicians' sleep schedules are important components of patient care, especially when fatigued physicians manage challenging clinical presentations.34 Caruso et al. identified 48 studies looking at the impact of resident duty hours on patient safety and found 27 studies showing a positive effect of limiting duty hours, 17 studies showing no clear effect, and 4 studies showing a negative effect.5 The authors noted that positive effects were more likely to be found in smaller studies with a median number of 11,402 patients while the median number of subjects in the studies reporting mixed effects was more than 4 million patients.

A number of studies have addressed how changes in resident work schedules/hours have impacted the broader health care community in areas of patient safety and out-comes; however, the results remain inconclusive.14,27,33 To better serve patients as well as provide excellent education, it becomes increasingly important to specifically address measurable clinical outcomes related to resident physicians' duty hours. We sought to determine if regulation of neurosurgery resident duty hours has resulted in significant changes in readily available measures of clinical outcome such as in-hospital mortality, discharge disposition, in-hospital complications, or in-hospital procedures.

We chose to study this question using a large administrative database (the Nationwide Inpatient Sample [NIS], http://www.hcup-us.ahrq.gov/nisoverview.jsp). The institution of duty hour regulations (405 Bell Regulations) by New York State on July 1, 1989,36 and nationwide by the ACGME on July 1, 2003,1 created a “natural experiment” in which the intervention (implementation of duty hours regulation) was applied at different times and in different places. This presented the opportunity to compare outcomes with and without the effect of duty hour regulation.

Large administrative databases have previously been used to demonstrate significant changes in important clinical outcomes in a number of circumstances. Using the NIS, Giles and colleagues demonstrated statistically and clinically significant reductions in mortality from abdominal aortic aneurysm repair after the introduction of endovascular aneurysm repair.13 Chihara and colleagues used vital statistics databases from Japan and the Surveillance, Epidemiology, and End Results (SEER) database of the National Cancer Institute of the US to examine the effect of mortality from chronic myelogenous leukemia and were able to document clinically and statistically significant reductions in mortality from chronic myelogenous leukemia by comparing patients treated before and after the introduction of imatinib into clinical practice.6 Johnson and O'Neill used the SEER database to demonstrate a statistically significant increase in survival of glioblastoma patients after the introduction of adjuvant temozolomide therapy while controlling for other factors known to affect survival in these patients.18 These examples established that if an intervention has important clinical effects in actual practice, those effects can be identified in large administrative databases despite the fact that there are long-term trends leading to generalized improvements in outcomes over time.

Methods

Study Objectives, Intervention, and Hypothesis

The study objective was to determine if resident duty hour regulation resulted in significant changes in measures of clinical outcome. Duty hour regulations were first instituted in New York State in 1989 and then nationally by the ACGME in 2003. The null hypothesis is that the implementation of duty hour regulations has had no measurable impact on important clinical outcomes in patients treated in neurosurgical training programs.

Study Design

To test this hypothesis retrospectively with an administrative database (the NIS) requires that certain potential biases be addressed. The most obvious of these biases is the effect of passage of time on clinical outcomes irrespective of individual interventions.11,25 The structure of this study specifically addresses this issue by asking the following 4 questions.

  1. Did clinical outcomes differ between regulated New York State training hospitals and unregulated training hospitals in other states before the ACGME regulations were implemented (New York 2000–2002 vs non–New York 2000–2002)? This is the first natural experiment. By the year 2000, New York State training programs had been operating under duty hour regulations for 10 years. The intervention (duty hour regulation) was mature. Outcomes in New York State hospitals could be directly and contemporaneously compared, thereby controlling for outcome bias introduced by the passage of time.

  2. Did clinical outcomes change in the non–New York training hospitals after regulations were implemented (non–New York 2000–2002 vs non–New York 2004–2006)? This is the second natural experiment. However, passage of time bias could have potentially affected the clinical outcome. To examine these effects, we analyzed the third experiment.

  3. Did clinical outcomes differ significantly between New York training hospitals before the ACGME regulations were in place compared with after (New York 2000–2002 vs New York 2004–2006)? If there is a general trend to different outcomes between the study periods (2000–2002 and 2004–2006) unrelated to duty hour regulation, it should be evident in this comparison.

  4. The New York State hospital regulations had been in effect for 14 years before the study period 2004–2006, while for non–New York hospitals these were the first 2 years of regulation. To study the maturation effect of duty hour regulation, we asked the question, “Did clinical outcomes differ after the regulations were implemented in non–New York State training hospitals compared with New York State training hospitals (non–New York 2004–2006 vs New York 2004–2006)?” Figure 1 shows a schematic representation of the study design.

Fig. 1.
Fig. 1.

Schematic representation of study design. Comparison 1: 2000–2002 New York group with 2000–2002 non–New York group. (Did clinical outcomes differ between regulated New York State training hospitals and unregulated training hospitals in other states before the ACGME regulations were implemented?) Comparison 2: 2000–2002 non–New York group with 2004–2006 non–New York group. (Did clinical outcomes change in the non–New York training hospitals after regulations were implemented?) Comparison 3: 2000–2002 New York group with 2004–2006 New York group. (Did clinical outcomes differ significantly between New York training hospitals before the ACGME regulations were in place compared with after?) Comparison 4: 2004–2006 New York group with 2004–2006 non–New York group. (Did clinical outcomes differ after the regulations were implemented in non–New York State training hospitals compared with New York State training hospitals?)

Subjects

We chose to study a single specialty, neurological surgery, for several reasons. Using a single specialty reduces the variation introduced by including multiple organ systems, diseases, and methods of treatment. The nature of neurosurgical diseases would allow us to identify important objective outcomes, such as death and severe disability. Duty hours in most neurosurgical training programs are widely thought to be well above the proposed limits, magnifying the effect of imposed reductions. The number of residents in neurosurgical training programs is small (typically 1 or 2 in each year of training), and therefore restricted duty hours may be more likely to have a measurable impact than such restrictions would have in larger programs.

Data Collection

The data were obtained from the NIS, which is part of the Healthcare Cost and Utilization Project (HCUP) sponsored by the Agency for Healthcare Research and Quality (AHRQ). The NIS is the largest all-payer inpatient care database in the US and is designed to identify, track, and analyze national trends in health care utilization, access, charges, quality, and outcomes. Since 1988 the NIS has collected data of approximately 5–8 million hospital stays annually, representing an approximately 20% stratified sample of US hospitals. Detailed information on the design of the NIS is available at http://www.hcup-us.ahrq.gov. We used the NIS data from 2000 to 2002 and from 2004 to 2006. To allow for a period of implementation of the rules before measuring outcomes, 2003 was taken as a neutral year because the duty hour regulations were implemented in July 2003 in non–New York teaching hospitals.

Patient Identification

From the years 2000–2002 and 2004–2006, procedural code fields were screened for the following ICD-9-CM codes to identify patients 18 years or older who had undergone neurosurgical procedures: (01) incision and excision of skull, brain, and cerebral meninges; (02) other operations on skull, brain, and cerebral meninges; (03) operations on spinal cord and spinal canal structures; (04) operations on cranial and peripheral nerves; and (05) operations on sympathetic nerves or ganglia. From this cohort, we selected only patients who had been treated at teaching hospitals. The NIS defines a teaching hospital as a hospital with an American Medical Association–approved residency program and either membership in the Council of Teaching Hospitals or a ratio of 0.25 or higher for full-time equivalent interns and residents to beds. Out of this hospital cohort, we then identified the hospitals affiliated with neurosurgical residency programs from the provided hospital ID codes. The control group was identified as patients treated at New York neurosurgical training hospitals from 2000 to 2002 when duty hour regulations were in effect in the state of New York. The comparison groups were identified as patients treated at non–New York neurosurgical training hospitals from 2000 to 2002 (unregulated group) and 2004 to 2006 (regulated group) and New York neurosurgical training hospitals from 2004 to 2006 (doubly regulated group).

Clinical Variables

From the NIS data set we identified information regarding patients' age, sex, race/ethnicity, and comorbidities, such as hypertension, diabetes mellitus, dyslipidemia, atrial fibrillation, congestive heart failure, chronic lung disease, coagulopathy, renal failure, and alcohol and tobacco use. In addition, we used the 3M Health Information Systems All Patient Refined Diagnosis Related Group (3M APR-DRG) mortality risk algorithm as a surrogate marker of disease severity. The APR-DRG is a joint development of 3M Health Information Systems and the National Association of Children's Hospitals and Related Institutions (http://www.ahrq.gov/professionals/quality-patient-safety/quality-resources/tools/mortality/Hughessumm.pdf). The 4 severity-of-illness subclasses and the 4 subclasses based on risk of mortality are numbered sequentially from 1 to 4 indicating, respectively, minor, moderate, major, and extreme severity of illness or risk of mortality. This 4-point ordinal system is based on a patient's age and primary and secondary diagnoses, with additional adjustments made for in-hospital procedures. This algorithm has been demonstrated to be a reliable and valid risk-of-mortality adjustment system and has been used previously for adjustment of disease severity classification.21,30

Primary and Secondary Outcomes

Primary outcome measures included discharge disposition (home, long-term care facility, or other) and in-hospital mortality. Secondary outcomes included length of stay (LOS), hospitalization charges, complications, and in-hospital procedures. We used ICD-9-CM secondary diagnosis codes to identify complications such as pneumonia (486, 481, 482.8, and 482.3), deep venous thrombosis (451.1, 451.2, 451.81, 451.9, 453.1, 453.2, 453.8, and 453.9), urinary tract infection ([UTI]; 599.0 and 590.9), sepsis (995.91, 996.64, 038, 995.92, and 999.3), pulmonary embolism (415.1), and myocardial infarction ([MI]; 410, 411, 412, 413, and 414). Procedure-related complications, including postoperative neurological complications such as infarction or intracerebral hemorrhage ([ICH]; 997.00–997.09) and repeat craniotomy (0123) were identified. The in-hospital procedures were identified based on ICD-9-CM procedure codes, including administration of thrombolytics (99.10), performance of gastrostomy (43.11–43.19), performance of mechanical ventilation (96.70–96.72), performance of tracheostomy (31.1–31.29), and transfusion of packed red blood cells (99.04).

Statistical Methods

We used SAS software (version 9.3, SAS Institute Inc.) to convert raw counts generated from the NIS database into weighted counts that represent national estimates. All analyses accounted for the complex sampling design and sample discharge weights of the NIS, following HCUP-NIS recommendations (HCUP Methods Series Report No. 2003–2). These methods are recommended for national estimates and have been described in previous publications based on the NIS.20,28 We used the Wald chi-square test in SAS version 9.3 for categorical data and analysis of variance for continuous data to detect any significant differences in variables among patients. We also used a Bonferroni correction p value (< 0.05) for 2 × 2 tables. Individual variables were tested in 2 × n table format. For the chi-square test, a p value ≤ 0.05 was considered significant. Logistic regression analysis was performed to determine the association between intervention and odds of discharge to nursing home/extended care facility. We adjusted for age and sex in the initial models and subsequently adjusted for the APR-DRG severity index and other potential confounders identified from the univariate analysis.

Results

Comparison 1 (New York 2000–2002 vs non–New York 2000–2002)

Clinical outcomes did not differ between regulated New York State training hospitals and unregulated training hospitals in other states before the ACGME regulations were implemented (Tables 1 and 2).

TABLE 1:

Comparison 1 (2000–2002 New York group and 2000–2002 non–New York group): Did clinical outcomes differ between regulated New York State training hospitals and unregulated training hospitals in other states before the ACGME regulations were implemented?*

Parameter2000–2002 NY2000–2002 Non-NYp Value
estimated no. of patients94,651476,339
mean age in yrs40.6 ± 5634 ± 560.02
women55,541 (58.6)251,859 (52.9)0.4
race/ethnicity0.1
 white60,723 (66.1)217,074 (66)
 black9971 (10.8)37,215 (11.3)
 Hispanic4239 (4.6)51,602 (15.7)
 other16,975 (18.5)22,944 (7)
comorbid conditions
 hypertension4342 (17)24,405 (17)0.9
 diabetes mellitus1697 (6.6)7923 (5.5)0.3
 dyslipidemia3735 (3.9)10,995 (2.3)0.01
 atrial fibrillation2530 (2.7)10,321 (2.2)0.1
 congestive heart failure490.7 (1.9)2760 (1.9)0.9
 chronic lung disease1468 (5.7)10,424 (7.3)0.05
 coagulopathy464 (1.8)3478 (2.4)0.2
 renal failure254.7 (1)2109 (1.5)0.4
 alcohol512.8 (2)2993 (2.1)0.9
 smoking3832 (4)22,188 (4.7)0.8
in-hospital complications
 pneumonia2168 (2.3)14,537 (3.1)0.4
 deep venous thrombosis1271 (1.3)4714 (1)0.09
 UTI4800 (5.1)25,368 (5.3)0.8
 sepsis78.4 (0.1)674.4 (0.1)0.3
 pulmonary embolism269.7 (0.3)1517 (0.3)0.7
 MI586.4 (0.6)2713 (0.6)0.8
 postop stroke1652 (1.7)8239 (1.7)0.9
 ICH2602 (2.7)12,379 (2.6)0.8
 repeat craniotomy165.3 (0.2)1770 (0.4)0.09
in-hospital procedure
 thrombolytics87.9 (0.1)345.7 (0.1)0.5
 gastrostomy1729 (1.8)8707 (1.8)0.9
 mechanical ventilation3037 (3.2)21,586 (4.5)0.3
 tracheostomy274.9 (0.3)723.3 (0.2)0.4
 transfusion5724 (6)22,448 (4.7)0.3
APR-DRG severity0.08
 no class specified4.2 (0)88 (0.1)
 minor loss of function12,189 (47.6)55,491 (38.7)
 moderate to major loss of function11,858 (46.4)75,851 (52.8)
 extreme loss of function1531 (6)12,036 (8.4)
mean LOS in days8.3 ± 318.1 ± 30<0.0001
mean hospital charges$29,012 ± 113,068$43,762 ± 165,672<0.0001
discharge disposition0.1
 home77,530 (81.9)399,782 (84.1)
 long-term care facility14,003 (14.8)60,357 (12.7)
 other380.9 (0.4)1036 (0.2)
in-hospital mortality2737 (2.9)14,463 (3)0.8

Values are number of patients (%) unless specified otherwise. Mean values are ± SD.

TABLE 2:

Multivariate regression analysis of primary outcome measures for all groups

OutcomeUnadjustedAdjusted for Age, Sex, & Risk Factors*
OR (95% CI)p ValueOR (95% CI)p Value
Comparison 1
 home0.8 (0.5–1.6)0.11.1 (0.4–2.8)0.6
 long-term facility1.2 (0.7–2.1)0.10.9 (0.4–2.0)0.7
 in-hospital mortality0.9 (0.5–1.8)0.80.8 (0.3–2.1)0.7
Comparison 2
 home0.8 (0.6–1.1)0.10.9 (0.7–1.2)0.6
 long-term facility1.2 (1.1–1.5)0.011.1 (0.9–1.3)0.3
 in-hospital mortality1.0 (0.8–1.3)0.81.0 (0.8–1.2)0.9
Comparison 3
 home0.8 (0.5–1.3)0.50.6 (0.3–1.4)0.3
 long-term facility1.3 (0.8–2.0)0.51.5 (0.8–2.9)0.2
 in-hospital mortality1.2 (0.7–2.0)0.51.3 (0.6–2.5)0.4
Comparison 4
 home0.8 (0.6–1.0)0.20.7 (0.6–0.9)0.01
 long-term facility1.2 (0.9–1.5)0.21.3 (1.1–1.5)0.006
 in-hospital mortality1.1 (0.8–1.4)0.51.1 (0.9–1.5)0.3

Risk factors include APR-DRG severity score, any variables that were significant between their respective groups out of the comorbidities, and in-hospital complications.

After applying the inclusion criteria to the NIS data set, 12 New York hospitals and 71 non–New York hospitals were identified for the years 2000–2002. From 2000 to 2002, an estimated 94,651 patients underwent neurosurgical procedures at the New York hospitals and 476,339 at the non–New York hospitals (Table 1). Patients in the 2000–2002 non–New York group were younger than those in the 2000–2002 New York group (34 vs 40.6 years, respectively, p = 0.02). There was no difference in the sex and racial distribution between the 2 groups. The primary outcome analysis of discharge disposition demonstrated that 81.9% of patients in the New York group were discharged to home compared with 84.1% in the non–New York group (p = 0.1). The rate of in-hospital mortality was 2.9% in the New York group and 3.0% in the non–New York group (p = 0.8). Looking at in-hospital complications, there were no significant differences between the 2 groups in the frequency of pneumonia (p = 0.4), deep venous thrombosis (p = 0.09), UTI (p = 0.8), sepsis (p = 0.3), pulmonary embolism (p = 0.7), MI (0.8), postoperative stroke (p = 0.9), ICH (p = 0.8), or repeat craniotomies (p = 0.09). The frequency of in-hospital procedures was not significantly different between the 2 groups: use of thrombolytics (p = 0.5), gastrostomy (p = 0.9), mechanical ventilation (p = 0.3), tracheostomy (p = 0.4), and transfusion (p = 0.3). The difference in length of stay and hospital charges was statistically significant between the New York and non–New York groups (8.3 vs 8.1 days, respectively, p < 0.0001; and $29,012 vs $43,762, respectively, p < 0.0001).

We did not find any significant difference between the 2 groups in terms of frequency of comorbid conditions such as hypertension, diabetes mellitus, atrial fibrillation, congestive heart failure, coagulopathy, renal failure, alcohol use, or smoking. The frequencies of dyslipidemia (3.9% in the New York group and 2.3% in the non–New York group, p = 0.01) and chronic lung disease (5.7% in the New York group vs 7.3% in the non–New York group, p = 0.05) were significantly different between the 2 groups. The APR-DRG severity index was not different between the 2 groups. The New York group had 47.6% of patients with minor loss of function, 46.4% with moderate to major loss of function, and 6.0% with extreme loss of function versus the non–New York group in which 38.7% had minor loss of function, 52.8% had moderate to major loss of function, and 8.4% had extreme loss of function (p = 0.08). After adjusting for age, sex, medical comorbidities, and APR-DRG disease severity scale score, the odds for discharge to home, discharge to long-term care facility, and in-hospital mortality were 1.1 (95% CI 0.4–2.8, p = 0.6), 0.9 (95% CI 0.4–2.0, p = 0.7), and 0.8 (95% CI 0.3–2.1, p = 0.7), respectively, for the 2000–2002 New York group compared with the 2000–2002 non–New York group (Table 2).

Comparison 2 (non–New York 2000–2002 vs non–New York 2004–2006)

Clinical outcomes did not change in the non–New York training hospitals after regulations were implemented (Tables 2 and 3).

TABLE 3:

Comparison 2 (2000–2002 non–New York group with 2004–2006 non–New York group): Did clinical outcomes change in the non–New York training hospitals after regulations were implemented?*

Parameter2000–2002 Non-NY2004–2006 Non-NYp Value
estimated no. of patients476,339511,778
mean age in yrs34 ± 5637.1 ± 58<0.0001
women251,859 (52.9)259,446 (51.2)0.6
race/ethnicity0.4
 white217,074 (66)270,343 (72.3)
 black37,215 (11.3)36,494 (9.8)
 Hispanic51,602 (15.7)42,964 (11.5)
 other22,944 (7)24,034 (6.4)
comorbid conditions
 hypertension24,405 (17)117,845 (23)0.02
 diabetes mellitus7923 (5.5)35,170 (6.9)0.1
 dyslipidemia10,995 (2.3)20,462 (4)0.001
 atrial fibrillation10,321 (2.2)15,473 (3)0.01
 congestive heart failure2760 (1.9)11,166 (2.2)0.4
 chronic lung disease10,424 (7.3)42,230 (8.3)0.1
 coagulopathy3478 (2.4)14,738 (2.9)0.1
 renal failure2109 (1.5)9642 (1.9)0.1
 alcohol2993 (2.1)12,452 (2.4)0.3
 smoking22,188 (4.7)32,405 (6.3)0.01
in-hospital complications
 pneumonia14,537 (3.1)16,250 (3.2)0.7
 deep venous thrombosis4714 (1)5142 (1)0.9
 UTI25,368 (5.3)31,090 (6.1)0.1
 sepsis674.4 (0.1)10,291 (2)<0.0001
 pulmonary embolism1517 (0.3)2961 (0.6)0.0005
 MI2713 (0.6)3499 (0.7)0.1
 postop stroke8239 (1.7)9014 (1.8)0.8
 ICH12,379 (2.6)15,890 (3.1)0.1
 repeat craniotomy1770 (0.4)1846 (0.4)0.8
in-hospital procedure
 thrombolytics345.7 (0.07)756.1 (0.1)0.01
 gastrostomy8707 (1.8)10,687 (2.1)0.3
 mechanical ventilation21,586 (4.5)25,135 (4.9)0.4
 tracheostomy723.3 (0.2)1399 (0.3)0.1
 transfusion22,448 (4.7)29,218 (5.7)0.4
APR-DRG severity0.6
 no class specified88 (0.1)399.9 (0.1)
 minor loss of function55,491 (38.7)185,723 (36.3)
 moderate to major loss of function75,851 (52.8)275,175 (53.8)
 extreme loss of function12,036 (8.4)50,479 (9.8)
mean LOS in days8.1 ± 308.2 ± 30<0.0001
mean hospital charges$43,762 ± 165,672$54,725 ± 195,067<0.0001
discharge disposition0.01
 home399,782 (84.1)417,240 (81.5)
 long-term care facility60,357 (12.7)77,036 (15.1)
 other1036 (0.2)1404 (0.3)
in-hospital mortality14,463 (3)15,992 (3.1)0.8

Values are number of patients (%) unless specified otherwise. Mean values are ± SD.

After applying the inclusion criteria to the NIS data set, there were 69 non–New York hospitals identified for the years 2004–2006. From 2004 to 2006, an estimated 511,778 patients underwent neurosurgical procedures at the non–New York hospitals (Table 3). Patients in the 2000–2002 non–New York group were younger than those in the 2004–2006 non–New York group (34 years vs 37.1 years, respectively, p < 0.0001). There was no difference in the sex and racial distribution between the 2 groups. The primary outcome analysis of discharge disposition demonstrated that 84.1% of patients in the 2000–2002 non–New York group were discharged to home compared with 81.5% in the 2004–2006 non–New York group (p = 0.01). The rate of in-hospital mortality was 3.0% in the 2000–2002 non–New York group and 3.1% in the 2004–2006 non–New York group (p = 0.8). Looking at in-hospital complications, there was a significant increase in the number of patients suffering from sepsis (0.1% vs 2.0%, p < 0.0001) and pulmonary embolism (0.3% vs 0.6%, p = 0.0005). There was no significant difference between the 2 groups in the frequency of pneumonia (p = 0.7), deep venous thrombosis (p = 0.9), UTI (p = 0.1), MI (p = 0.1), postoperative stroke (p = 0.8), ICH (p = 0.1), or repeat craniotomies (p = 0.8). There was a significant increase in the number of patients undergoing thrombolytic treatment: 0.07% in the 2000–2002 non–New York group and 0.1% in the 2004–2006 non–New York group (p = 0.01). The frequency of in-hospital procedures, including use of gastrostomy (p = 0.3), mechanical ventilation (p = 0.4), tracheostomy (p = 0.1), and transfusion (p = 0.4), was not significantly different between the 2 groups. The differences in LOS and hospital charges were statistically significant between the 2000–2002 non–New York and 2004–2006 non–New York group (8.1 vs 8.2 days, respectively, p < 0.0001; and $43,762 vs $54,725, respectively, p < 0.0001).

There were some significant differences between the 2 groups in terms of comorbid conditions; there were increases in patients with hypertension (17% vs 23%, p = 0.02), dyslipidemia (2.3% s 4%, p = 0.001), and atrial fibrillation (2.2% vs 3%, p = 0.01), and in the number of patients who were smokers (4.7% vs 6.3%, p = 0.01) in the 2004–2006 non–New York group compared with the 2000–2002 non–New York group. We did not find any significant difference between the 2 groups in terms of other conditions such as diabetes mellitus, congestive heart failure, chronic lung disease, coagulopathy, renal failure, or alcohol use. Based on APR-DRG severity, there was no difference in disease severity between the 2 groups. The 2000–2002 non–New York group had 38.7% of patients with minor loss of function, 52.8% with moderate to major loss of function, and 8.4% with extreme loss of function versus the 2004–2006 non–New York group, in which 36.3% had minor loss of function, 53.8% had moderate to major loss of function, and 9.8% had extreme loss of function (p = 0.6). After adjusting for age, sex, medical comorbidities, and APR-DRG disease severity scale score, the odds for discharge to home, long-term care facility, and in-hospital mortality were 0.9 (95% CI 0.7–1.2, p = 0.6), 1.1 (95% CI 0.9–1.3, p = 0.3), and 1.0 (95% CI 0.8–1.2, p = 0.9), respectively, for the 2000–2002 non–New York group compared with the 2004–2006 non–New York group (Table 2).

Comparison 3 (New York 2000–2002 vs New York 2004–2006)

Clinical outcomes did not differ significantly between New York training hospitals before the ACGME regulations were in place compared with after.

After applying the inclusion criteria to the NIS data set, there were 17 New York hospitals identified for the years 2004–2006. From 2004 to 2006, an estimated 117,323 patients underwent neurosurgical procedures at the New York hospitals (Table 4). Patients in the 2004–2006 New York group were younger than those in the 2000–2002 New York group (37.1 vs 40.6 years, respectively, p < 0.0001). There was no difference in the sex and racial distribution between the 2 groups. The primary outcome analysis of discharge disposition demonstrated that 81.9% of patients in the 2000–2002 New York group were discharged to home compared with 78.0% in the 2004–2006 New York group (p = 0.5). The rate of inhospital mortality was 2.9% in the 2000–2002 New York group and 3.4% in the 2004–2006 New York group (p = 0.5). Looking at in-hospital complications, there was a significant increase in the patients suffering from sepsis: 0.1% in the 2000–2002 New York group and 2.5% in the 2004–2006 New York group (p = 0.009). Otherwise, there was no significant difference between the 2 groups in the frequency of pneumonia (p = 0.1), deep venous thrombosis (p = 0.5), UTI (p = 0.3), pulmonary embolism (p = 0.1), MI (p = 0.9), postoperative stroke (p = 0.3), ICH (p = 0.2), or repeat craniotomies (p = 0.07). There was a significant increase in the number of patients who were placed on mechanical ventilation (3.2% in the 2000–2002 New York group vs 6.4% in the 2004–2006 New York group, p = 0.05) and the rate of transfusions over time (6.0% in the 2000–2002 New York group vs 10.5% in the 2004–2006 New York group, p = 0.05). The frequency of other inhospital procedures, including use of thrombolytics (p = 0.07), gastrostomy (p = 0.1), and tracheostomy (p = 0.1), was not significantly different between the 2 groups. The difference in LOS and hospital charges was statistically significant between the 2000–2002 New York and 2004–2006 New York groups (8.3 vs 9.6 days, respectively, p < 0.0001; and $29,012 vs $56,073, respectively, p < 0.0001).

TABLE 4:

Comparison 3 (2000–2002 New York group and 2004–2006 New York group): Did clinical outcomes differ significantly between New York training hospitals before the ACGME regulations were in place compared with after?*

Parameter2000–2002 NY2004–2006 NYp Value
estimated no. of patients94,651117,323
mean age in yrs40.6 ± 56.237.1 ± 60<0.0001
women55,541 (58.6)61,140 (52.1)0.3
race/ethnicity0.5
 white60,723 (66.1)72,710 (64.5)
 black9971 (10.8)17,204 (15.3)
 Hispanic4239 (4.6)12,179 (10.8)
 other16,975 (18.5)10,618 (9.4)
comorbid conditions
 hypertension4342 (17)29,795 (25.4)0.07
 diabetes mellitus1697 (6.6)9531 (8.1)0.2
 dyslipidemia3735 (3.9)7379 (6.3)0.1
 atrial fibrillation2530 (2.7)3740 (3.2)0.3
 congestive heart failure490.7 (1.9)3220 (2.7)0.2
 chronic lung disease1468 (5.7)11,777 (10)0.03
 coagulopathy464 (1.8)3170 (2.7)0.1
 renal failure254.7 (1)2472 (2.1)0.1
 alcohol512.8 (2)2315 (2)0.9
 smoking3832 (4)7378 (6.3)0.3
in-hospital complications
 pneumonia2168 (2.3)4168 (3.6)0.1
 deep venous thrombosis1271 (1.3)1389 (1.2)0.5
 UTI4800 (5.1)7451 (6.4)0.3
 sepsis78.4 (0.1)2875 (2.5)0.009
 pulmonary embolism269.7 (0.3)620 (0.5)0.1
 MI586.4 (0.6)738 (0.6)0.9
 postop stroke1652 (1.7)2523 (2.2)0.3
 ICH2602 (2.7)4189 (3.6)0.2
 repeat craniotomy165.3 (0.2)524 (0.4)0.07
in-hospital procedure
 thrombolytics87.9 (0.1)261 (0.2)0.07
 gastrostomy1729 (1.8)3483 (3)0.1
 mechanical ventilation3037 (3.2)7470 (6.4)0.05
 tracheostomy274.9 (0.3)769 (0.7)0.1
 transfusion5724 (6)12,368 (10.5)0.05
APR-DRG severity0.6
 no class specified4.2 (0)71 (0.1)
 minor loss of function12,189 (47.6)42,275 (36)
 moderate to major loss of function11,858 (46.4)63,313 (46.4)
 extreme loss of function1531 (6)11,664 (9.9)
mean LOS in days8.3 ± 31.19.6 ± 35.5<0.0001
mean hospital charges$29,012 ± 113,068$56,073 ± 193,666<0.0001
discharge disposition0.5
 home77,530 (81.9)91,566 (78)
 long-term care facility14,003 (14.8)21,330 (18.3)
 other380.9 (0.4)424 (0.4)
in-hospital mortality2737 (2.9)4004 (3.4)0.5

Values are number of patients (%) unless specified otherwise. Mean values are ± SD.

We did not find any significant difference between the 2 groups in terms of the frequency of comorbid conditions such as hypertension, diabetes mellitus, dyslipidemia, atrial fibrillation, congestive heart failure, coagulopathy, renal failure, alcohol use, or smoking. The frequency of chronic lung disease (5.7% in the 2000–2002 New York group vs 10.0% in the 2004–2006 New York group, p = 0.03) was significantly different between the 2 groups. Based on APR-DRG severity, there was no difference in disease severity between the 2 groups. The 2000–2002 New York group had 47.6% of patients with minor loss of function, 46.4% with moderate to major loss of function, and 6.0% with extreme loss of function versus the 2004–2006 New York group where 36.0% had minor loss of function, 46.4% had moderate to major loss of function, and 9.9% had extreme loss of function (p = 0.6). After adjusting for age, sex, medical comorbidities, and APRDRG disease severity scale score, the odds for discharge to home, long-term care facility, and in-hospital mortality were 0.6 (95% CI 0.3–1.4, p = 0.3), 1.5 (95% CI 0.8–2.9, p = 0.2), and 1.3 (95% CI 0.6–2.5, p = 0.4), respectively, for the 2000–2002 New York group compared with the 2004–2006 New York group (Table 2).

Comparison 4 (non–New York 2004–2006 vs New York 2004–2006)

Clinical outcomes did differ after the regulations were implemented in non–New York State training hospitals compared with New York State training hospitals. There were more discharges to long-term care facilities and fewer discharges to home from the New York hospitals, which had been regulated for a longer period of time (Tables 2 and 5).

TABLE 5:

Comparison 4 (2004–2006 New York group and 2004–2006 non–New York group): Did clinical outcomes differ after the regulations were implemented in non–New York State training hospitals compared with New York State training hospitals?*

Parameter2004–2006 NY2004–2006 Non-NYp Value
estimated no. of patients117,323511,672
mean age in yrs37.1 ± 6037.1 ± 58.5<0.0001
women61,140 (52.1)259,446 (51.2)0.7
race/ethnicity0.008
 white72,710 (64.5)270,343 (72.3)
 black17,204 (15.3)36,494 (9.8)
 Hispanic12,179 (10.8)42,964 (11.5)
 other10,618 (9.4)24,034 (6.4)
comorbid conditions
 hypertension29,795 (25.4)117,845 (23)0.3
 diabetes mellitus9531 (8.1)35,170 (6.9)0.1
 dyslipidemia7379 (6.3)20,462 (4)0.03
 atrial fibrillation3740 (3.2)15,473 (3)0.6
 congestive heart failure3220 (2.7)11,166 (2.2)0.2
 chronic lung disease11,777 (10)42,230 (8.3)0.1
 coagulopathy3170 (2.7)14,738 (2.9)0.4
 renal failure2472 (2.1)9642 (1.9)0.5
 alcohol2315 (2)12,452 (2.4)0.2
 smoking7378 (6.3)32,405 (6.3)0.9
in-hospital complications
 pneumonia4168 (3.6)16,250 (3.2)0.3
 deep venous thrombosis1389 (1.2)5142 (1)0.4
 UTI7451 (6.4)31,090 (6.1)0.6
 sepsis2875 (2.5)10,291 (2)0.3
 pulmonary embolism620 (0.5)2961 (0.6)0.5
 MI738 (0.6)3499 (0.7)0.6
 postop stroke2523 (2.2)9014 (1.2)0.2
 ICH4189 (3.6)15,890 (3.1)0.6
 repeat craniotomy524 (0.4)1846 (0.4)0.4
in-hospital procedure
 thrombolytics261 (0.2)756.1 (0.1)0.2
 gastrostomy3483 (3)10,687 (2.1)0.08
 mechanical ventilation7470 (6.4)25,135 (4.9)0.1
 tracheostomy769 (0.7)1399 (0.3)0.1
 transfusion12,368 (10.5)29,218 (5.7)0.02
APR-DRG severity0.9
 no class specified71 (0.1)400 (0.1)
 minor loss of function42,275 (36)185,723 (36.3)
 moderate to major loss of function63,313 (46.4)275,175 (53.8)
 extreme loss of function11,664 (9.9)50,479 (9.9)
mean LOS in days9.6 ± 35.58.2 ± 30.3<0.0001
mean hospital charges$56,073 ± 193,666$54,725 ± 195,0670.1
discharge disposition0.2
 home91,566 (78)417,240 (81.5)
 long-term care facility21,330 (18.3)77,036 (15.1)
 other424 (0.4)1404 (0.3)
in-hospital mortality4004 (3.4)15,992 (3.1)0.5

Values are number of patients (%) unless specified otherwise. Mean values are ± SD.

Although the mean ages of patients in the two groups were the same, patients in the 2004–2006 non–New York group were younger than the 2004–2006 New York group based on the standard deviation and distribution (37.1 years, p < 0.0001). There was no difference in the sex distribution between the 2 groups. The racial distribution between the 2 groups was significantly different: more non–New York patients were white than were New York patients (72.3% vs 64.5%, p = 0.008). The primary outcome analysis of discharge disposition demonstrated that 81.5% of patients in the non–New York group were discharged to home compared with 78.0% in the New York group (p = 0.2). The rate of in-hospital mortality was 3.4% in the New York group and 3.1% in the non–New York group (p = 0.5). Looking at in-hospital complications, there was no significant difference between the 2 groups in the frequency of pneumonia (p = 0.3), deep venous thrombosis (p = 0.4), UTI (p = 0.6), sepsis (p = 0.3), pulmonary embolism (p = 0.5), MI (0.6), postoperative stroke (p = 0.2), ICH (p = 0.6), or repeat craniotomies (p = 0.4). The rate of blood transfusions was significantly higher in the New York group: 10.5% vs 5.7% (p = 0.02). Otherwise, the frequency of in-hospital procedures including use of thrombolytics (p = 0.2), gastrostomy (p = 0.08), mechanical ventilation (p = 0.1), and tracheostomy (p = 0.1) was not significantly different between the 2 groups. The difference in LOS was statistically significant between the New York and non–New York groups (9.6 vs 8.2 days, respectively, p < 0.0001); however, the amount of hospital charges was not significantly different ($56,073 vs $54,725, respectively, p = 0.1) (Table 5).

We did not find any significant difference between the 2 groups in terms of frequency of comorbid conditions such as hypertension, diabetes mellitus, atrial fibrillation, congestive heart failure, chronic lung disease, coagulopathy, renal failure, alcohol use, or smoking. The frequency of dyslipidemia (6.3% in the New York group and 4.0% in the non–New York group, p = 0.03), however, was significantly different between the 2 groups. Based on APRDRG severity, there was no difference in disease severity between the 2 groups. The New York group had 36.0% of patients with minor loss of function, 46.4% with moderate to major loss of function, and 9.9% with extreme loss of function versus the non–New York group where 36.3% had minor loss of function, 53.8% had moderate to major loss of function, and 9.9% had extreme loss of function (p = 0.9). After adjusting for age, sex, medical comorbidities, and APR-DRG disease severity scale score, the odds for discharge to home, long-term care facility, and in-hospital mortality were 0.7 (95% CI 0.6–0.9, p = 0.01), 1.3 (95% CI 1.1–1.5, p = 0.006), and 1.1 (95% CI 0.9–1.5, p = 0.3), respectively, for the 2004–2006 non–New York group compared with the 2004–2006 New York group (Table 2).

Summary

Returning to our 4 questions, there were no significant differences in patient outcomes in the New York group under New York regulations compared with the non–New York group prior to the 2003 implementation of the ACGME stricter regulations (New York vs non–New York 2000–2002, Comparison 1). There was an increase in discharge to long-term care facility without a change in in-hospital mortality after the stricter duty hours were implemented among the non–New York groups (non–New York 2000–2002 vs non–New York 2004–2006, Comparison 2); however, the significance of this decrease disappeared with multivariate analysis (Table 2). The change in regulation in New York to adopt the ACGME regulations did not significantly affect patient outcomes in terms of discharge to long-term care facility or in-hospital mortality (New York 2000–2002 vs New York 2004–2006, Comparison 3). There was a significant decrease in patients discharged to home and an increase in patients discharged to long-term care facilities, without a significant difference in in-hospital mortality, in hospitals in which the regulations had been in place longer (New York 2004–2006 vs non–New York 2004–2006, Comparison 4) (Figs. 2 and 3).

Fig. 2.
Fig. 2.

Schematic representation of primary outcome measures. Comparison 1: 2000–2002 New York group with 2000–2002 non–New York group. (Did clinical outcomes differ between regulated New York State training hospitals and unregulated training hospitals in other states before the ACGME regulations were implemented? Answer: no.) Comparison 2: 2000–2002 non–New York group with 2004–2006 non–New York group. (Did clinical outcomes change in the non–New York training hospitals after regulations were implemented? Answer: no.) Comparison 3: 2000–2002 New York group with 2004–2006 New York group. (Did clinical outcomes differ significantly between New York training hospitals before the ACGME regulations were in place compared with after? Answer: no.) Comparison 4: 2004–2006 New York group with 2004–2006 non–New York group. (Did clinical outcomes differ after the regulations were implemented in non–New York State training hospitals compared with New York State training hospitals? Answer: yes. There were fewer discharges to home [*p = 0.01] and more to long-term care facilities [*p = 0.006] from the New York hospitals after national implementation of duty hours regulations.)

Fig. 3.
Fig. 3.

Graph showing discharge disposition and in-hospital mortality by New York and non–New York groups. *p = 0.01 and p = 0.006 in the comparison of discharge disposition between New York and non–New York groups, 2004–2006.

Discussion

Effect of Fatigue on Performance

Empirically validated investigations aimed at understanding the relationship between sleep deprivation and cognitive performance have demonstrated adverse effects of loss of sleep on neuropsychological measures of cognitive functioning. Many of the studies suggest that subjects who are sleep deprived tend to do about 1.4 SDs worse than controls.34 A meta-analysis of the effects of sleep deprivation26 demonstrated decreased performance on standard tests of cognitive functioning (for example, logical reasoning tasks, mental addition, and Torrence tasks) and altered mood states following sleep deprivation. Similar findings reported in Nature demonstrated that individuals who are awake for 24 hours perform similarly to individuals with a 0.1% blood alcohol level.8 The effect of a lack of sleep on surgical dexterity among surgical residents has also been documented.10,15,23,31 The effects of sleep deprivation seem to differ depending on the specialty, as a recent article in Neurosurgery illustrates.12 Given different surgical tasks, neurosurgery residents did not show a statistically significant difference in the pre- and postcall states. However, in a similar study there was a significant decline in surgical skills among general surgery residents. A study among obstetric and gynecological residents found that sleep deprivation caused a decline in fine motor coordination skills.2 According to these examples, the effects of sleep deprivation are not uniform and seem to vary depending on specialty.

Effect of Policy Change (ACGME Duty Hour Regulations) on Clinical Outcome

We found little evidence that resident duty hour regulation is associated with a significant improvement in major health care outcomes when applied to the training of residents in a high-intensity surgical subspecialty traditionally associated with long duty hours and resident fatigue. An analysis of our primary outcome measures revealed no significant difference in mortality or discharge to a long-term care facility among the hospitals in New York, which had been operating under work hour regulations for 10 years and those in other states that were operating without such regulations. Implementation of the ACGME duty hour regulations in the non–New York group led to no observed difference in the in-hospital mortality or discharge to longterm care facilities (1.1 [95% CI 0.9–1.3], p = 0.3). Similarly, after ACGME regulations were implemented in New York there was no significant difference either in hospital mortality (1.3 [95% CI 0.6–2.5], p = 0.4) or discharge to long-term care facilities (1.5 [95% CI 0.8–2.9], p = 0.2). In hospitals in which the regulations had been in place longer, there was a significant decrease in patients discharged to home (0.7 [95% CI 0.6–0.9], p = 0.01) and an increase in patients discharged to long-term care facilities (1.3 [95% CI 1.1–1.5], p = 0.006) (New York 2 004–2006 vs non–New York 2004–2006) without a significant difference in in-hospital mortality (1.1 [95% CI 0.9–1.5], p = 0.3). The increase in frequency of long-term care facility transfer may indicate changing trends in health care over the past decade. This may be the result of improved awareness or improved patient care; however, the exact nature or effect is difficult to assess from this data set, especially considering that the frequency of mortality has not changed.

The secondary outcome measures (in-hospital complications and procedures) were either unchanged or worse in the years immediately following the implementation of duty hour regulations in the non–New York group. Using the 2000–2002 New York group as the gold standard, in-hospital complications and procedures in the non–New York group prior to the implementation of duty hour regulations (2000–2002) were comparable. After the implementation of duty hour regulations in non–New York hospitals, in-hospital complications and procedures were not significantly different. When we compared non–New York groups before and after the duty hour regulation implementation, we found that there was a significant increase in the rate of postoperative sepsis and pulmonary embolism. Our findings are similar to those of other studies of surgical specialties looking at patient outcomes before and after the duty hour regulations were in place.3,4,9,19,33 In a single institution's experience with neurosurgical patients, Dumont et al.9 found that overall morbidity increased in patients treated after the duty hour regulations were in place but that mortality remained the same. In a single institution study of obstetric and gynecological residents, there was minimal improvement in some patient outcomes such as postpartum hemorrhage and neonatal resuscitations; however, there was no significant difference in rates of cesarean delivery for nonreassuring fetal status, failed induction, labor abnormality, or repeat cesarean delivery.3 Using the NIS data, Poulose et al.27 found that after the implementation of work hour regulations in New York State the incidence of accidental puncture or laceration as well as postoperative pulmonary embolus or deep venous thrombosis increased. There were significant increases in the in-hospital mortality and postoperative surgical complication rates of surgical patients in Switzerland after duty hour regulations were imposed.4,19 Volpp et al. looked at both medical and surgical Medicare patients from the years 2000 to 2005 and found no significant increases or decreases in mortality before or after the regulation reform.33 Gottlieb et al.14 reported shorter hospital stays, fewer laboratory tests, and fewer medication errors after the implementation of regulations in the internal medicine department of a large universityaffiliated Veterans Affairs Medical Center. The effect of these improvements on measurable patient outcomes was not reported in this study.

Effect of Duty Hour Regulation on Residency Training in Neurosurgery

Our findings are similar to those of other recent studies specifically addressing the impact of the duty hour regulations on neurosurgical programs and residents. An initial survey done by Cohen-Gadol et al. reported that 61% (95% CI 53%–69%) of the residents and 79% (95% CI 63%–89%) of the neurosurgery program directors noted that the duty hour guidelines have had a negative effect on their training programs.7 Looking at several objective measures of the quality of neurosurgery programs and resident performance, Jagannathan et al.17 found that American Board of Neurological Surgery written examination scores for residents taking the examination for self-assessment decreased from 310 in 2002 to 259 in 2006 (16% decrease, p < 0.05). In addition, although there was an increase in the number of resident registrations to the AANS meetings, the number of abstracts presented by residents decreased from 345 in 2002 to 318 in 2007 (7% decrease, p < 0.05).

Criticisms of Study Design

The study will be criticized for using simple and relatively crude measures of health outcomes and relying on a large administrative database. However, the concerns that led to the imposition of duty hour regulations were based on mortality and major morbidity resulting from errors made by tired trainees, and health care policy decision makers often rely heavily on evidence obtained from such large administrative databases. We measured in-hospital complications, in-hospital mortality, and discharge disposition. We also included measures of less serious outcomes that could be associated with decline in performance but not affect outcome. Length of stay can be affected by inefficiencies in care as well as complications. Unanticipated return to the operating room and readmission to the hospital do not necessarily result in poor outcomes but may result from errors in care. No important differences in outcome were identified for any of these measures.

It would have been desirable to analyze the difference in outcome in New York State hospitals before and after the 1989 implementation of the New York State duty hour regulations. Unfortunately, the NIS database includes data only from 1988 onward, and therefore there are not sufficient data prior to the implementation of the New York State regulations to allow for a reasonably statistically powerful comparison.

This is a retrospective review of a large database, and as a result several inherent limitations would have affected our results.35 Patient selection was not randomized; therefore, selection bias may have affected our results. Regional and cultural differences in patient care, physician practices, and socioeconomic factors between different parts of the country may have an impact. In addition, there may be a selection bias based on coding practices because many different hospitals code injuries and/or procedures differently. The nature of the database means that we are not comparing exactly the same hospitals in the period 2000–2002 and 2004–2006. We included all available data from NIS to obtain a homogeneous mixture of the patient population in various parts of the country to minimize that potential bias. Moreover, the NIS database does not provide details on severity of neurological deficits, results of diagnostic studies, or use of procedures. Even though the APR-DRG disease severity scale scores were not different between these groups, we were unable to use conventional grades of neurosurgical disease to assess severity of disease in this analysis. The non–New York patients in the post–regulation era had slightly higher frequencies of hypertension, dyslipidemia, and smoking, which may have impacted our results. We recognize the need for further detailed studies to explore the impact of duty hour regulations not only on the quality of patient care but also on the quality of training of future generations of physicians in all subspecialties.

Accepting the results of this analysis requires accepting the following premises: 1) To justify a major and costly change in regulatory policy in duty hours for residents in training, the outcome improvements achieved through regulation should be large enough to be measurable in a large discharge database study. 2) The data in the NIS are reliable and representative of the US population treated in hospitals. 3) The quality and effectiveness of treatment of neurosurgical disease does not differ in significant ways between New York and non–New York hospitals or between ACGME-accredited neurosurgical training programs.

Future Directions and Considerations

There remain a lack of data and consensus regarding the impact of resident duty hour regulations on many different aspects of health care and patient outcomes, patient satisfaction, cost, and resident training. Tan et al.32 suggested that the annual labor cost of implementing the duty hour regulations was estimated at $1.6 billion for all programs; it remains to be seen how hospitals have made up this cost and how it has affected the broader health care system. As the ACGME moves toward further modifying the duty hours, concerns of negative impact on training programs continue to grow. It may result in longer residencies, an increase in the number of required residents to fulfill the ever-growing clinical tasks, less operative exposure especially for specialties such as neurosurgery, and, as a result, a negative impact on overall training of the next generation of neurosurgeons.

Conclusions

Our study demonstrates that regulation of resident duty hours has not resulted in statistically significant improvements in major outcomes among neurosurgical patients. Interestingly, the regulations appear to be associated with an increase in the frequency of postoperative complications and discharge to long-term care facilities in the years immediately following their implementation in the neurosurgical training programs in the US. If the effects of fatigue are such that they do not produce noticeable deterioration in outcomes, the intensity of the discussion and concern about resident duty hours should lessen. Given the large cost of implementing these regulations and the potential unintended negative impact on resident training, we question the wisdom of continuing to promote these regulations in the absence of data that clearly demonstrate a clinically important positive effect on patient welfare.

Acknowledgment

The authors thank Dr. Rachel Nygaard, Department of General Surgery at Hennepin County Medical Center, for editorial comments.

Disclosure

Dr. Stephen Haines was program director of the Department of Neurosurgery at the University of Minnesota during the years of this study. The authors otherwise report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper.

Author contributions to the study and manuscript preparation include the following. Conception and design: Haines. Acquisition of data: Norby, Siddiq. Analysis and interpretation of data: all authors. Drafting the article: Norby, Siddiq, Haines. Critically revising the article: all authors. Reviewed submitted version of manuscript: all authors. Approved the final version of the manuscript on behalf of all authors: Norby. Statistical analysis: Siddiq, Adil.

References

  • 1

    Accreditation Council for Graduate Medical Education: Report of the ACGME Work Group on resident duty hours. Premier Inc. (https://www.premierinc.com/safety/safety-share/08-02_downloads/03_wkgreport_602.pdf) [Accessed April 8 2014]

  • 2

    Ayalon RDFriedman F Jr: The effect of sleep deprivation on fine motor coordination in obstetrics and gynecology residents. Am J Obstet Gynecol 199:576.e1576.e52008

  • 3

    Bailit JLBlanchard MH: The effect of house staff working hours on the quality of obstetric and gynecologic care. Obstet Gynecol 103:6136162004

  • 4

    Businger APLaffer UKaderli R: Resident work hour restrictions do not improve patient safety in surgery: a critical appraisal based on 7 years of experience in Switzerland. Patient Saf Surg 6:172012

  • 5

    Caruso JVeloski JGrasberger MBoex JPaskin DKairys JAppendix C: Systematic review of the literature on the impact of variation in residents' duty hour schedules on patient safety. Philibert IAmis SJ: The ACGME 2011 Duty Hour Standards: Enhancing Quality of Care Supervision and Resident Professional Development ChicagoACGME2011

  • 6

    Chihara DIto HMatsuda TKatanoda KShibata ASaika K: Decreasing trend in mortality of chronic myelogenous leukemia patients after introduction of imatinib in Japan and the U.S. Oncologist 17:154715502012

  • 7

    Cohen-Gadol AAPiepgras DGKrishnamurthy SFessler RD: Resident duty hours reform: results of a national survey of the program directors and residents in neurosurgery training programs. Neurosurgery 56:3984032005

  • 8

    Dawson DReid K: Fatigue, alcohol and performance impairment. Nature 388:2351997. (Letter)

  • 9

    Dumont TMRughani AIPenar PLHorgan MATranmer BIJewell RP: Increased rate of complications on a neurological surgery service after implementation of the Accreditation Council for Graduate Medical Education work-hour restriction. Clinical article. J Neurosurg 116:4834862012

  • 10

    Eastridge BJHamilton ECO'Keefe GERege RVValentine RJJones DJ: Effect of sleep deprivation on the performance of simulated laparoscopic surgical skill. Am J Surg 186:1691742003

  • 11

    Fisher AAForeit JR: Chapter 7: Intervention Study Designs. Designing HIV/AIDS Intervention Studies: An Operations Research Handbook Washington, DCPopulation Council2002. 4562

  • 12

    Ganju AKahol KLee PSimonian NQuinn SJFerrara JJ: The effect of call on neurosurgery residents' skills: implications for policy regarding resident call periods. Clinical article. J Neurosurg 116:4784822012

  • 13

    Giles KAPomposelli FHamdan AWyers MJhaveri ASchermerhorn ML: Decrease in total aneurysm-related deaths in the era of endovascular aneurysm repair. J Vasc Surg 49:5435512009

  • 14

    Gottlieb DJParenti CMPeterson CALofgren RP: Effect of a change in house staff work schedule on resource utilization and patient care. Arch Intern Med 151:206520701991

  • 15

    Grantcharov TPBardram LFunch-Jensen PRosenberg J: Laparoscopic performance after one night on call in a surgical department: prospective study. BMJ 323:122212232001

  • 16

    Gurjala ALurie PHaroona LRising JPBell BStrohl KP: Petition requesting medical residents work hour limits. Public Citizen (http://www.citizen.org/hrg1570) [Accessed April 8 2014]

  • 17

    Jagannathan JVates GEPouratian NSheehan JPPatrie JGrady MS: Impact of the Accreditation Council for Graduate Medical Education work-hour regulations on neurosurgical resident education and productivity. Special topic. J Neurosurg 110:8208272009

  • 18

    Johnson DRO'Neill BP: Glioblastoma survival in the United States before and during the temozolomide era. J Neurooncol 107:3593642012

  • 19

    Kaderli RBusinger AOesch AStefenelli ULaffer U: Morbidity in surgery: impact of the 50-hour work-week limitation in Switzerland. Swiss Med Wkly 142:w135062012

  • 20

    Khatri RChaudhry SAVazquez GRodriguez GJHassan AESuri MF: Age differential between outcomes of carotid angioplasty and stent placement and carotid endarterectomy in general practice. J Vasc Surg 55:72782012

  • 21

    Khatri RTariq NVazquez GSuri MFEzzeddine MAQureshi AI: Outcomes after nontraumatic subarachnoid hemorrhage at hospitals offering angioplasty for cerebral vasospasm: a national level analysis in the United States. Neurocrit Care 15:34412011

  • 22

    Koslowsky MBabkoff H: Meta-analysis of the relationship between total sleep deprivation and performance. Chronobiol Int 9:1321361992

  • 23

    Leff DRAggarwal RRana MNakhjavani BPurkayastha SKhullar V: Laparoscopic skills suffer on the first shift of sequential night shifts: program directors beware and residents prepare. Ann Surg 247:5305392008

  • 24

    Leung LBecker CE: Sleep deprivation and house staff performance. Update 1984–1991. J Occup Med 34:115311601992

  • 25

    Paradis C: Bias in surgical research. Ann Surg 248:1801882008

  • 26

    Pilcher JJHuffcutt AI: Effects of sleep deprivation on performance: a meta-analysis. Sleep 19:3183261996

  • 27

    Poulose BKRay WAArbogast PGNeedleman JBuerhaus PIGriffin MR: Resident work hour limits and patient safety. Ann Surg 241:8478602005

  • 28

    Qureshi AIChaudhry SAHassan AEZacharatos HRodriguez GJSuri MF: Thrombolytic treatment of patients with acute ischemic stroke related to underlying arterial dissection in the United States. Arch Neurol 68:153615422011

  • 29

    Samkoff JSJacques CH: A review of studies concerning effects of sleep deprivation and fatigue on residents' performance. Acad Med 66:6876931991

  • 30

    Shukla RFisher RFisher R: Testing of 3M's APR-DRG risk adjustment for hospital mortality outcomes. Abstr Acad Health Serv Res Health Policy Meet 19:112002

  • 31

    Taffinder NJMcManus ICGul YRussell RCDarzi A: Effect of sleep deprivation on surgeons' dexterity on laparoscopy simulator. Lancet 352:11911998. (Letter)

  • 32

    Tan PHogle NJWidmann WD: Limiting PGY 1 residents to 16 hours of duty: review and report of a workshop. J Surg Educ 69:3553592012

  • 33

    Volpp KGRosen AKRosenbaum PRRomano PSEven-Shoshan OWang Y: Mortality among hospitalized Medicare beneficiaries in the first 2 years following ACGME resident duty hour reform. JAMA 298:9759832007

  • 34

    Weinger MBAncoli-Israel S: Sleep deprivation and clinical performance. JAMA 287:9559572002

  • 35

    Woodworth GFBaird CJGarces-Ambrossi GTonascia JTamargo RJ: Inaccuracy of the administrative database: comparative analysis of two databases for the diagnosis and treatment of intracranial aneurysms. Neurosurgery 65:2512572009

  • 36

    10 N.Y. Comp. Codes R. & Regs. §405.4. (http://www.health.ny.gov/professionals/doctors/graduate_medical_education/other_related_information/405_4.htm) [Accessed April 14 2014]

If the inline PDF is not rendering correctly, you can download the PDF file here.

Article Information

Address correspondence to: Kiersten Norby, M.D., Department of Surgery, Hennepin County Medical Center, 701 Park Ave. S., Minneapolis, MN 55415. email: kierstennorby@gmail.com.

Please include this information when citing this paper: published online June 3, 2014; DOI: 10.3171/2014.4.JNS131191.

© AANS, except where prohibited by US copyright law.

Headings

Figures

  • View in gallery

    Schematic representation of study design. Comparison 1: 2000–2002 New York group with 2000–2002 non–New York group. (Did clinical outcomes differ between regulated New York State training hospitals and unregulated training hospitals in other states before the ACGME regulations were implemented?) Comparison 2: 2000–2002 non–New York group with 2004–2006 non–New York group. (Did clinical outcomes change in the non–New York training hospitals after regulations were implemented?) Comparison 3: 2000–2002 New York group with 2004–2006 New York group. (Did clinical outcomes differ significantly between New York training hospitals before the ACGME regulations were in place compared with after?) Comparison 4: 2004–2006 New York group with 2004–2006 non–New York group. (Did clinical outcomes differ after the regulations were implemented in non–New York State training hospitals compared with New York State training hospitals?)

  • View in gallery

    Schematic representation of primary outcome measures. Comparison 1: 2000–2002 New York group with 2000–2002 non–New York group. (Did clinical outcomes differ between regulated New York State training hospitals and unregulated training hospitals in other states before the ACGME regulations were implemented? Answer: no.) Comparison 2: 2000–2002 non–New York group with 2004–2006 non–New York group. (Did clinical outcomes change in the non–New York training hospitals after regulations were implemented? Answer: no.) Comparison 3: 2000–2002 New York group with 2004–2006 New York group. (Did clinical outcomes differ significantly between New York training hospitals before the ACGME regulations were in place compared with after? Answer: no.) Comparison 4: 2004–2006 New York group with 2004–2006 non–New York group. (Did clinical outcomes differ after the regulations were implemented in non–New York State training hospitals compared with New York State training hospitals? Answer: yes. There were fewer discharges to home [*p = 0.01] and more to long-term care facilities [*p = 0.006] from the New York hospitals after national implementation of duty hours regulations.)

  • View in gallery

    Graph showing discharge disposition and in-hospital mortality by New York and non–New York groups. *p = 0.01 and p = 0.006 in the comparison of discharge disposition between New York and non–New York groups, 2004–2006.

References

1

Accreditation Council for Graduate Medical Education: Report of the ACGME Work Group on resident duty hours. Premier Inc. (https://www.premierinc.com/safety/safety-share/08-02_downloads/03_wkgreport_602.pdf) [Accessed April 8 2014]

2

Ayalon RDFriedman F Jr: The effect of sleep deprivation on fine motor coordination in obstetrics and gynecology residents. Am J Obstet Gynecol 199:576.e1576.e52008

3

Bailit JLBlanchard MH: The effect of house staff working hours on the quality of obstetric and gynecologic care. Obstet Gynecol 103:6136162004

4

Businger APLaffer UKaderli R: Resident work hour restrictions do not improve patient safety in surgery: a critical appraisal based on 7 years of experience in Switzerland. Patient Saf Surg 6:172012

5

Caruso JVeloski JGrasberger MBoex JPaskin DKairys JAppendix C: Systematic review of the literature on the impact of variation in residents' duty hour schedules on patient safety. Philibert IAmis SJ: The ACGME 2011 Duty Hour Standards: Enhancing Quality of Care Supervision and Resident Professional Development ChicagoACGME2011

6

Chihara DIto HMatsuda TKatanoda KShibata ASaika K: Decreasing trend in mortality of chronic myelogenous leukemia patients after introduction of imatinib in Japan and the U.S. Oncologist 17:154715502012

7

Cohen-Gadol AAPiepgras DGKrishnamurthy SFessler RD: Resident duty hours reform: results of a national survey of the program directors and residents in neurosurgery training programs. Neurosurgery 56:3984032005

8

Dawson DReid K: Fatigue, alcohol and performance impairment. Nature 388:2351997. (Letter)

9

Dumont TMRughani AIPenar PLHorgan MATranmer BIJewell RP: Increased rate of complications on a neurological surgery service after implementation of the Accreditation Council for Graduate Medical Education work-hour restriction. Clinical article. J Neurosurg 116:4834862012

10

Eastridge BJHamilton ECO'Keefe GERege RVValentine RJJones DJ: Effect of sleep deprivation on the performance of simulated laparoscopic surgical skill. Am J Surg 186:1691742003

11

Fisher AAForeit JR: Chapter 7: Intervention Study Designs. Designing HIV/AIDS Intervention Studies: An Operations Research Handbook Washington, DCPopulation Council2002. 4562

12

Ganju AKahol KLee PSimonian NQuinn SJFerrara JJ: The effect of call on neurosurgery residents' skills: implications for policy regarding resident call periods. Clinical article. J Neurosurg 116:4784822012

13

Giles KAPomposelli FHamdan AWyers MJhaveri ASchermerhorn ML: Decrease in total aneurysm-related deaths in the era of endovascular aneurysm repair. J Vasc Surg 49:5435512009

14

Gottlieb DJParenti CMPeterson CALofgren RP: Effect of a change in house staff work schedule on resource utilization and patient care. Arch Intern Med 151:206520701991

15

Grantcharov TPBardram LFunch-Jensen PRosenberg J: Laparoscopic performance after one night on call in a surgical department: prospective study. BMJ 323:122212232001

16

Gurjala ALurie PHaroona LRising JPBell BStrohl KP: Petition requesting medical residents work hour limits. Public Citizen (http://www.citizen.org/hrg1570) [Accessed April 8 2014]

17

Jagannathan JVates GEPouratian NSheehan JPPatrie JGrady MS: Impact of the Accreditation Council for Graduate Medical Education work-hour regulations on neurosurgical resident education and productivity. Special topic. J Neurosurg 110:8208272009

18

Johnson DRO'Neill BP: Glioblastoma survival in the United States before and during the temozolomide era. J Neurooncol 107:3593642012

19

Kaderli RBusinger AOesch AStefenelli ULaffer U: Morbidity in surgery: impact of the 50-hour work-week limitation in Switzerland. Swiss Med Wkly 142:w135062012

20

Khatri RChaudhry SAVazquez GRodriguez GJHassan AESuri MF: Age differential between outcomes of carotid angioplasty and stent placement and carotid endarterectomy in general practice. J Vasc Surg 55:72782012

21

Khatri RTariq NVazquez GSuri MFEzzeddine MAQureshi AI: Outcomes after nontraumatic subarachnoid hemorrhage at hospitals offering angioplasty for cerebral vasospasm: a national level analysis in the United States. Neurocrit Care 15:34412011

22

Koslowsky MBabkoff H: Meta-analysis of the relationship between total sleep deprivation and performance. Chronobiol Int 9:1321361992

23

Leff DRAggarwal RRana MNakhjavani BPurkayastha SKhullar V: Laparoscopic skills suffer on the first shift of sequential night shifts: program directors beware and residents prepare. Ann Surg 247:5305392008

24

Leung LBecker CE: Sleep deprivation and house staff performance. Update 1984–1991. J Occup Med 34:115311601992

25

Paradis C: Bias in surgical research. Ann Surg 248:1801882008

26

Pilcher JJHuffcutt AI: Effects of sleep deprivation on performance: a meta-analysis. Sleep 19:3183261996

27

Poulose BKRay WAArbogast PGNeedleman JBuerhaus PIGriffin MR: Resident work hour limits and patient safety. Ann Surg 241:8478602005

28

Qureshi AIChaudhry SAHassan AEZacharatos HRodriguez GJSuri MF: Thrombolytic treatment of patients with acute ischemic stroke related to underlying arterial dissection in the United States. Arch Neurol 68:153615422011

29

Samkoff JSJacques CH: A review of studies concerning effects of sleep deprivation and fatigue on residents' performance. Acad Med 66:6876931991

30

Shukla RFisher RFisher R: Testing of 3M's APR-DRG risk adjustment for hospital mortality outcomes. Abstr Acad Health Serv Res Health Policy Meet 19:112002

31

Taffinder NJMcManus ICGul YRussell RCDarzi A: Effect of sleep deprivation on surgeons' dexterity on laparoscopy simulator. Lancet 352:11911998. (Letter)

32

Tan PHogle NJWidmann WD: Limiting PGY 1 residents to 16 hours of duty: review and report of a workshop. J Surg Educ 69:3553592012

33

Volpp KGRosen AKRosenbaum PRRomano PSEven-Shoshan OWang Y: Mortality among hospitalized Medicare beneficiaries in the first 2 years following ACGME resident duty hour reform. JAMA 298:9759832007

34

Weinger MBAncoli-Israel S: Sleep deprivation and clinical performance. JAMA 287:9559572002

35

Woodworth GFBaird CJGarces-Ambrossi GTonascia JTamargo RJ: Inaccuracy of the administrative database: comparative analysis of two databases for the diagnosis and treatment of intracranial aneurysms. Neurosurgery 65:2512572009

36

10 N.Y. Comp. Codes R. & Regs. §405.4. (http://www.health.ny.gov/professionals/doctors/graduate_medical_education/other_related_information/405_4.htm) [Accessed April 14 2014]

TrendMD

Metrics

Metrics

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 123 123 51
PDF Downloads 91 91 36
EPUB Downloads 0 0 0

PubMed

Google Scholar