After Libby Zion died tragically in a teaching hospital, much attention from the medical community, public and government, was directed at resident training and the number of hours worked consecutively. This marked a new era heralded by close scrutiny of the time residents spent in the hospital. Two dramatic reductions were mandated nationally by the Accreditation Council for Graduate Medical Education (ACGME) in 2003 and 2011. It was hypothesized that better rested residents would be more efficient in delivering patient care, such that, markers of utilization would decrease at teaching hospitals. Because non-teaching hospitals were not affected by reforms, we did not expect to see any changes across the same time periods of transitions, thereby making them an ideal control group.
Methods:
To assembly a nationally representative data set, we used the Agency for Healthcare Research and Quality’s (AHRQ’s) Healthcare Cost and Utilization Project-Nationwide Inpatient Sample (HCUP-NIS) database files between 2001-2005 and 2009-2013. All patients admitted to hospitals participating in the NIS from July 1, 2001 to June 30, 2005 and from July 1, 2009 to June 30, 2013 comprised our patient sample. This encompassed 2 years of hospitalizations before and after each of the ACGME reforms (July 1, 2003 and July 1, 2011).
After several exclusions, our final patient population was 117,060,957 from the first time period and 117,087,577 from the second time period, hospitalized in either teaching or non-teaching hospitals. We examined in-hospital mortality rates, length of hospital stay (LOS), total hospital charges and hospitalization costs. Patient- and hospital- level characteristics were compared between pre-reform and post-reform subgroups within non-teaching and teaching hospitals using the Pearson χ2 test for categorical variables and univariate linear regression (1-way ANOVA) for continuous variables.
Results:
As compared to pre-reform, in-hospital mortality decreased following the 2003 ACGME reform in both teaching [AOR=0.93, 95% CI (0.90-0.97)] and non-teaching [AOR=0.92, 95% CI (0.90-0.94)] hospitals. When segmented regression analysis was used, no statistically significant changes in adjusted hospital mortality were present in the first month following the 2003 and 2011 ACGME reforms in either teaching [0.67 (-1.36 to 2.69) & -0.35 (-1.80 to 1.09)] or non-teaching hospitals [-0.27 (-2.50 to 1.95) & -0.46 (-1.91 to 1.00)].
Following the 2003 ACGME reform, hospital LOS was similar to that of pre-reform in teaching [5.02 vs. 4.96 days] and nonteaching [4.45 vs. 4.46 days] hospitals. However, hospital LOS was significantly shorter post-reform in teaching hospitals [4.87 vs. 4.80 days] but not in non-teaching hospitals [4.24 vs. 4.20 days] following the 2011 ACGME reform. Hospital charges and hospitalization costs were higher following the 2003 ACGME reform in both teaching and non-teaching hospitals when compared to pre-reform. Using segmented regression analysis, no statistically significant change in hospital charges and hospitalizations costs were noted after the first month of the 2003 and 2011 ACGME reforms in both teaching and nonteaching hospitals.
Conclusions:
Our study found that the implementation of the ACGME duty hour reforms was not associated with any significant changes in hospital mortality, length of stay, or costs. We hope these findings might inform the direction and decisions of the upcoming 2017 ACGME Task Force proposed changes.