Points represent the differences from baseline (academic year 2001-2002) in adjusted estimates of Likert scores generated from mixed-effects multivariate linear regression models using month and study year indicator variables (reference group: academic year 2001-2002). For “time for teaching,” the greatest decline in satisfaction occurred in 2003-2004, corresponding to Accreditation Council for Graduate Medical Education duty-hour restrictions. Changes after implementation of duty-hour restrictions represent statistically significant decreases. Similar patterns were observed for hours per week of didactic teaching, intern involvement with decisions, and sufficient autonomy for residents. In contrast, the decline in satisfaction with “professional growth” was incremental over the period with no relationship with implementation of duty-hour restrictions (year 3 [2003-2004]). This pattern was observed with items relating to professional development. Error bars indicate 95% confidence interval for estimate.
Arora V, Meltzer D. Effect of ACGME Duty Hours on Attending Physician Teaching and Satisfaction. Arch Intern Med. 2008;168(11):1226-1231. doi:10.1001/archinte.168.11.1226
Copyright 2008 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2008
The implementation of restricted duty hours by the Accreditation Council for Graduate Medical Education (ACGME) have raised concerns regarding the potential negative effects on resident education and patient care.1 Equally concerning are the potential effects on teaching faculty. Surgical studies report the “trickle-up” effect, or, increased faculty workload due to absent residents.2,3 In a recent survey after the implementation of ACGME duty hours, faculty retrospectively reported decreased satisfaction with teaching and diminished career satisfaction.4 This study assesses the effect of ACGME duty-hour restrictions on inpatient attending physician teaching and satisfaction using data collected for 5 years, before and after implementation at a single institution.
From July 2001 to July 2006, inpatient attending physicians at the University of Chicago, Chicago, Illinois, were surveyed after their inpatient general medicine (GENS) rotation. The institutional review board of the University of Chicago approved this study. From July 2001 to July 2003, residents were not restricted by work hours. Each GENS team consisted of 1 attending physician, 1 resident, and 2 interns and took calls every fifth night, with 2 days of short-call shifts. In July 2003, to comply with ACGME restrictions, 8 teams with the same composition were introduced. Two teams took calls each night in a fourth-night rotation.5 New admissions alternated between teams until midnight. After admitting 10 patients or midnight (whichever came first), a night-float resident admitted new patients.6 These admissions were “held over” for the next day's call team. During the study period, 1 resident team each month was led by a faculty hospitalist, who was an attending physician 3 months yearly. Resident hours were consistent with ACGME restrictions.7
Attending physicians were asked to complete a 32-item end-of-rotation survey, adapted from prior studies.8,9 Attending physicians were asked to estimate the following: total, bedside, and didactic teaching (hours per week); feedback to housestaff (times per month); conferences missed due to rounds (times per week); and percentage of patients seen on day of admission. Attending physicians also rated their overall satisfaction and their satisfaction with items related to teaching, relationships, patient care, and professional development using a 5-point Likert scale ranging from 5 (very satisfied) to 1 (not at all satisfied). Lastly, space was provided for additional comments.
Because attending physicians were surveyed repeatedly, mixed-effects regression models controlling for attending physician were used to determine the effect of duty hours on teaching and satisfaction. Models included month indicators to account for seasonal variation. To investigate temporal trends, secondary analysis using study year indicators (and no duty-hour effect) was performed. All statistical tests were performed using Intercooled Stata 9.0 statistical software (StataCorp, College Station, Texas), with statistical significance defined as P < .05. Open-ended comments were reviewed for mention of duty hours.
Data are reported as before vs after implementation mean value; difference (95% confidence interval of the difference) and P value from mixed effects regression model.
Of the 465 end-of-rotation surveys, 314 (68%) were received from attending physicians. Of the 113 individual attending physicians on GENS rotation during the study, 92 completed at least 1 survey before and after duty hours, for a total of 300 surveys (65%) for multivariate analyses. After implementation of duty-hour restrictions, attending physicians reported fewer hours of didactic teaching (3.92 hours vs 3.13 hours; −0.79 hours [−1.28 to −0.30 hours]) (P = .01), more times per week when residents missed conferences because of rounds (1.22 vs 1.69; 0.46 [0.13 to 0.80]) (P = .007), and a lower percentage of patients seen on the day of admission (40.01% vs 29.26%; −10.75% [−15.72% to −5.77%]) (P = .001). According to the Likert scale scores, attending physicians were less satisfied after implementation of duty-hour restrictions (4.26 vs 4.00; −0.26 [−0.47 to −0.05]) (P = .02), specifically for the following items: time for teaching (3.40 vs 2.92; −0.48 [−0.73 to −0.24]) (P = .001), professional growth and development (4.06 vs 3.72; −0.34 [−0.53 to −0.15]) (P = .001), educational stimulation of the work (4.41 vs 4.15; −0.26 [−0.43 to −0.09]) (P = .002), ability to determine length of stay (3.68 vs 3.27; −0.41 [−0.63 to −0.20]) (P = .001), and influence on hospital policy (2.65 vs 2.42; −0.23 [−0.40 to −0.07]) (P = .006). After implementation of duty-hour restrictions, attending physicians were also less likely to report that interns were truly involved in clinical decisions (4.50 vs 4.27; −0.24 [−0.37 to −0.10]) (P = .001) or that residents had sufficient autonomy (4.57 vs 4.40; −0.16 [−0.30 to −0.02]) (P = .02).
Using temporal trend models, we confirmed that changes related to housestaff teaching (time for teaching) were the greatest in the year immediately after implementation of duty-hour restrictions and were not likely to be due to temporal trend alone (Figure). For items relating to professional development, an incremental pattern of declining satisfaction even before implementation of duty-hour restrictions emerged, suggesting the influence of secular forces.
Attending physician comments confirmed these findings. One attending physician stated “success in dealing with reduced housestaff hours but at the cost of teaching.” Another attending physician expressed “the 80-hour work week . . . limits interns and residents' ability to provide continuity of care and be at the front line in decision making . . . increasingly dependent on attendings to make management decisions.” Lastly, another attending physician described their changing role, stating “ . . . too often I found myself ‘resi-terning’ . . . writing orders, following up on laboratories, calling consults. With new work hour rules, on multiple occasions, I met other GENS attendings also still in the hospital at 8 PM.”
After ACGME duty hours, attending physicians reported diminished satisfaction with the inpatient rotation. Changes related to housestaff teaching appeared to correspond with the implementation of duty-hour restrictions. However, decreased satisfaction with professional development appeared to be the result of secular trend, with no discernible relationship to duty-hour restrictions. These findings highlight the importance of secular trend when interpreting retrospective evaluations of duty hours. Studies that do not account for secular trend should be interpreted with caution.
The decline in satisfaction with teaching may be due to structural changes in the residency to achieve duty-hour compliance. For example, the institution of a night-float system results in more patients “held over” for an on-call team, with lower percentages of patients seen on the “initial day of admission.” These patients may not be ideal teaching cases, since housestaff did not admit the patient. Alternatives that balance duty hours with the opportunity for housestaff to actively participate in admitting patients they will follow should be explored.
Because inpatient attending physicians serve as career role models for medical students, understanding declines in attending physician satisfaction with professional development is important.8 One possibility is that over time, clinical work interferes with increasing pressures to produce scholarly work. Clinical work may be less educationally stimulating over time due to repeated hospitalization of the same patients or patients with the same disease (eg, sickle cell disease) or increased time spent addressing nonmedical issues (eg, nursing home placement). Future work can test these hypotheses.
There are several limitations of this study. This single-institution study limits generalizability. While the effects of secular trend or accumulating physician experience—both overall and institutional—cannot be excluded, additional analyses suggest that changes relating to housestaff teaching were consistent with the implementation of duty-hour restrictions. Preconceived attitudes toward duty hours may have biased responses. The survey was initially designed to test the effect of hospitalists and may have missed other effects of duty hours. While more satisfied attending physicians may have responded, this would lower the likelihood of detecting any changes. Because mixed-effects models may underestimate standard errors, we also used fixed-effects models, which yielded similar results that remained significant except for overall satisfaction (Likert scale score change, 0.23 [range, −0.47 to 0.09]) (P = .06).
In conclusion, after implementation of ACGME duty hours, inpatient attending physicians reported diminished satisfaction with teaching and professional development. Although changes in teaching corresponds with duty hours, decreased satisfaction with professional development appears to be due to secular trend.
Correspondence: Dr Arora, 5841 S Maryland Ave, MC 2007, AMB B217, Chicago, IL 60637 (firstname.lastname@example.org).
Author Contributions:Study concept and design: Arora and Meltzer. Acquisition of data: Arora and Meltzer. Analysis and interpretation of data: Arora and Meltzer. Drafting of the manuscript: Arora. Critical revision of the manuscript for important intellectual content: Arora and Meltzer. Statistical analysis: Arora and Meltzer. Obtained funding: Arora and Meltzer. Administrative, technical, and material support: Arora and Meltzer. Study supervision:Meltzer.
Financial Disclosure: None reported.
Funding/Support: This study was supported by the Agency for Healthcare Research and Quality grant R01 10597-01A2 A Multicenter Trial of Academic Hospitalists.
Additional Contributions: Veronica Tirado, BA, and Katie Chiu, BA, provided excellent research assistance; Kimberly Alvarez, BA, helped prepare the manuscript; and Juned Siddique, DrPH, provided analytic insights. We are grateful to our attending physician colleagues at the University of Chicago for their enthusiastic and continued completion of surveys.