Abstract
This cross-sectional study uses data from a voluntary survey that was administered during the American College of Physicians Internal Medicine In-Training Examination to assess whether residents in programs with nocturnists perceived nighttime supervision differently than those in programs without nocturnists.
Overnight supervision of internal medicine residents has traditionally been indirect1 and provided by a physician outside the hospital. During the past 15 years, the complexity and volume of overnight inpatient medical services have increased.2,3 This, coupled with data that increased resident supervision is associated with improved patient safety,4 has led many teaching hospitals to employ hospitalists in-house overnight (nocturnists).1,5
Little is known about the national landscape of resident overnight supervision, including how commonly internal medicine residency programs use nocturnists and how they affect residents’ supervision. We tested whether residents in programs with nocturnists perceived nighttime supervision differently than those in programs without nocturnists.
Methods
The American College of Physicians Internal Medicine In-Training Examination (IM-ITE) includes a voluntary survey. The 2017 survey asked residents, “How often do you receive adequate supervision to ensure patient safety when you are working overnight?” using a 5-point Likert scale (responses of “always,” “most of the time,” “sometimes,” “rarely,” or “never”). Presence or absence of nocturnists was determined by a yes or no response to “Does your internal medicine residency program have an attending physician who stays in the hospital overnight and is responsible for directly supervising internal medicine residents providing care on medicine wards?”
This analysis was confined to residents in US programs. Because patient safety is a minimum standard of supervision, we used Pearson χ2 or adjusted Wald χ2 statistics to compare the “always” response for receiving adequate supervision with all other responses in respondents with and without nocturnists, with an α level of P ≤ .05. We also performed group-based hypothesis testing among all responses using the Pearson χ2 statistic. Because senior residents may supervise their juniors, we conducted analyses overall and stratified by postgraduate year using the adjusted Wald χ2 statistic and the Kendall τb correlation coefficient interacting presence or absence of nocturnist and postgraduate year with all responses and an α level of P ≤ .01. The data analysis was conducted in Stata SE version 14.2 (StataCorp).
This study was declared exempt by the George Washington University institutional review board.
Results
Of 24 620 internal medicine residents in US programs who completed the IM-ITE in 2017, 23 553 participated in the survey. After excluding 2809 residents who did not allow their data to be used for research purposes, responses from 20 744 internal medicine residents (84% response rate) were analyzed (Table 1). Fifty-six percent were male and 86.1% were residents in categorical programs, with no differences between respondents with nocturnists and without nocturnists.
Table 1. Characteristics of Respondentsa.
| Characteristic | No. (%) | P valueb | ||
|---|---|---|---|---|
| Overall (N = 20 744) |
With nocturnists (n = 10 902) |
Without nocturnists (n = 9842) |
||
| Sex | ||||
| Male | 11 742 (56.6)c | 6288 (57.7) | 5454 (55.4) | .82 |
| Female | 9002 (43.4) | 4614 (42.3) | 4388 (44.2) | |
| Postgraduate year | ||||
| 1 | 7017 (33.8) | 4242 (38.9) | 2775 (28.2) | .06 |
| 2 | 7222 (34.8) | 3612 (33.1) | 3610 (36.7) | .44 |
| 3 | 6505 (31.4) | 3048 (28.0) | 3457 (35.1) | .09 |
| Type of program | ||||
| Categorical | 17 868 (86.1) | 9459 (86.8) | 8409 (85.4) | .89 |
| Primary care | 1410 (6.8) | 692 (6.4) | 718 (7.3) | .92d |
| Medicine-pediatrics | 957 (4.6) | 440 (4.0) | 517 (5.3) | .97d |
| Preliminary internship | 311 (1.5) | 200 (1.8) | 111 (1.1) | .34d |
| Other (not specified) | 198 (1.0) | 111 (1.0) | 87 (0.9) | .97d |
| Medical school | ||||
| US medical graduate | 12 870 (62.0)c | 6467 (59.3) | 6403 (65.1) | .22 |
| Non-US medical graduate | 7874 (38.0) | 4435 (40.7) | 3439 (34.9) | |
Respondents not affiliated with US residency programs and those who opted out of allowing access to their data for analysis were excluded. Data from 20 744 respondents were analyzed.
Calculated using the Pearson χ21 bivariate test and compares the group with nocturnists vs the group without nocturnists (the α level was .05).
Nationally, 58% of internal medicine residents are male and 49.4% are graduates of US medical schools.6
Calculated using the adjusted Wald χ2 (bivariate test) and compares the group with nocturnists vs the group without nocturnists (the α level was .05).
Overall, 51.4% of respondents reported that their supervision at night was “always” adequate to ensure patient safety, 31.9% responded “most of the time,” 10.6% responded “sometimes,” 4.4% responded “rarely,” and 1.7% responded “never” (Table 2). Fifty-three percent of residents reported having a nocturnist. Significantly more residents who reported having a nocturnist reported “always” receiving adequate supervision to ensure patient safety overnight compared with residents who reported not having a nocturnist (61.3% vs 40.5%, respectively; P < .001). The results for residents having a nocturnist vs not having a nocturnist were statistically significant when “always” vs all other responses were compared and when all response categories were compared. Results were similar by postgraduate year (Table 2).
Table 2. Perception of Adequacy of Supervision to Ensure Patient Safety Overnight.
| Total | How often do you receive adequate supervision to ensure patient safety when you are working overnight? | |||||
|---|---|---|---|---|---|---|
| Alwaysa | Most of the time | Sometimes | Rarely | Never | ||
| All respondentsa | 20 744 | 10 662 (51.4) | 6614 (31.9) | 2201 (10.6) | 920 (4.4) | 347 (1.7) |
| With nocturnist | 10 902 | 6678 (61.3) | 3337 (30.6) | 735 (6.7) | 127 (1.2) | 25 (0.2) |
| Without nocturnist | 9842 | 3984 (40.5) | 3277 (33.3) | 1466 (14.9) | 793 (8.1) | 322 (3.3) |
| PGY-3b | ||||||
| With nocturnist | 3048 | 1911 (62.7) | 894 (29.3) | 199 (6.5) | 35 (1.2) | 9 (0.3) |
| Without nocturnist | 3457 | 1431 (41.4) | 1125 (32.5) | 471 (13.6) | 311 (9.0) | 119 (3.4) |
| PGY-2 | ||||||
| With nocturnist | 3612 | 2152 (59.6) | 1151 (31.9) | 253 (7.0) | 49 (1.4) | 7 (0.2) |
| Without nocturnist | 3610 | 1375 (38.1) | 1236 (34.2) | 590 (16.3) | 292 (8.1) | 117 (3.2) |
| PGY-1 | ||||||
| With nocturnist | 4242 | 2615 (61.7) | 1292 (30.5) | 283 (6.7) | 43 (1.0) | 9 (0.2) |
| Without nocturnist | 2775 | 1178 (42.5) | 916 (33.0) | 405 (14.6) | 190 (6.9) | 86 (3.1) |
The Pearson χ21 or adjusted Wald χ21 bivariate test statistics were used to compare the group with nocturnists vs the group without nocturnists for a dichotomous comparison between the “always” response and the 4 other response options combined. The α level was .05; P < .001 for all group-based comparisons.
To ensure no confounding of the results based on postgraduate year (PGY) or the presence or absence of a nocturnist, all response options were tested for independence using the adjusted Wald χ2 test stratified by PGY level (the α level was .05; P < .001) and the Kendall τb correlation coefficient, −0.0353 (the α level was .01; P < .001).
Discussion
Fifty-one percent of internal medicine residents reported that their overnight supervision was always adequate to ensure patient safety. When nocturnists were present, more residents (61%) reported always receiving adequate supervision than when they were not (40%). However, a substantial percentage did not report always receiving adequate supervision despite having a nocturnist, possibly reflecting variability in nocturnist job descriptions across institutions. Some nocturnists provide care independently while being available for indirect supervision, whereas others primarily provide direct supervision.1,5
Strengths of this study include its large size, high response rate, and capture of a substantial majority of US internal medicine residents. Limitations include measurement of perception rather than objective data and cross-sectional survey design. Program characteristics (eg, university-based or community-based) were not collected and confounding due to program-level variability is possible. These data do not account for other factors that may affect perceived supervision, including safety culture and other personnel.
Hospitals should assess adequacy of their nighttime supervision of residents to ensure patient safety. Best practices for roles of nocturnists merit further study.
Section Editor: Jody W. Zylke, MD, Deputy Editor.
References
- 1.Farnan JM, Burger A, Boonyasai RT, et al. ; SGIM Housestaff Oversight Subcommittee . Survey of overnight academic hospitalist supervision of trainees. J Hosp Med. 2012;7(7):521-523. doi: 10.1002/jhm.1961 [DOI] [PubMed] [Google Scholar]
- 2.Hundert SA, White AA, Reilly DF. Number of general medicine hospital admissions performed by internal medicine residents before and after the 2011 duty-hour regulations. South Med J. 2015;108(8):476-481. [DOI] [PubMed] [Google Scholar]
- 3.Clark AV, LoPresti CM, Smith TI. Trends in inpatient admission comorbidity and electronic health data: implications for resident workload intensity. J Hosp Med. 2018;13(8):570-572. doi: 10.12788/jhm.2954 [DOI] [PubMed] [Google Scholar]
- 4.Blum AB, Shea S, Czeisler CA, Landrigan CP, Leape L. Implementing the 2009 Institute of Medicine recommendations on resident physician work hours, supervision, and safety. Nat Sci Sleep. 2011;3:47-85. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Trowbridge RL, Almeder L, Jacquet M, Fairfield KM. The effect of overnight in-house attending coverage on perceptions of care and education on a general medical service. J Grad Med Educ. 2010;2(1):53-56. doi: 10.4300/JGME-D-09-00056.1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Brotherton SE, Etzel SI. Graduate medical education, 2017-2018. JAMA. 2018;320(10):1051-1070. doi: 10.1001/jama.2018.10650 [DOI] [PubMed] [Google Scholar]
