Adjusting to duty hour reforms: residents' perception of the safety climate in interdisciplinary night-float rotations.

Background
New scheduling models were needed to adjust to residents' duty hour reforms while maintaining safe patient care. In interdisciplinary night-float rotations, four to six residents from most residency programs collaborated for after-hours cross-coverage of most adult hospitalised patients as part of a Faculty-led rotation. Residents worked sixteen 12-hour night shifts over a month.


Methods
We measured residents' perception of the patient safety climate during implementation of night-float rotations in five tertiary hospitals. We surveyed 267 residents who had completed the rotation in 2015-2016 with an online version of the Safety Attitudes Questionnaire. First year residents came from most residency programs, second- and third-year residents came from internal medicine.


Results
One-hundred-and-thirty residents completed the questionnaire. Scores did not differ across hospitals and residents' years of training for all six safety-related climate factors: teamwork climate, job satisfaction, perceptions of management, safety climate, working conditions, and stress recognition.


Conclusion
Simultaneous implementation in five hospitals of a Faculty-led interdisciplinary night-float rotation for most junior residents proved to be logistically feasible and showed similar and reassuring patient safety climate scores.


Introduction
Duty hour reforms for medical residents were driven by the expectation that new on-call schedules would provide better conditions for learning and safer patient care. 1 However, systematic reviews of the last decade of duty hour reforms did not find convincing evidence of improved patient safety. [1][2][3] Some studies in acute care and surgery settings even raised concerns regarding an increased morbidity and mortality after the implementation of new scheduling systems. [4][5][6][7] A 16-hour workday restriction has been in effect in the Province of Québec (Canada) since 2012 and night calls are limited to twelve hours. [8][9][10] Replacing 24hour on-call periods, new scheduling models were needed to address both patient safety and educational standards. [10][11][12] At our institution, due to mandatory pre-and post-call absence, most of these scheduling models led to fifty percent workday attendance at clinical duties and a catastrophic discontinuity in clinical and educational activities.
To overcome this challenge, we implemented an innovative model of interdisciplinary night-float rotation. 13 For four consecutive weeks, we grouped three to six residents from most residency programs to work as teams for after-hours consultations and cross-coverage of hospitalised patients. 13 We used the term "float" because residents shared the workload of multiple services and "rotation" because it was scheduled in the residency curriculum and assessed like other mandatory clinical rotation.
Some raised legitimate concerns regarding the working climate resulting from this unusual combination of residents from different backgrounds working in a new environment with supervisors from all disciplines and providing coverage for almost all hospitalised patients. A positive working climate ensuring residents' wellbeing and patient safety being at the origin of duty hour reforms, residents' perception in that regard was a priority of the quality assurance process. 10 The Safety Attitudes Questionnaire (SAQ) was benchmarked in multiple care units, including general inpatient and acute care settings, and studied in association with patient outcomes. [14][15][16][17][18] Its goal is to elicit a snapshot of the safety culture through surveys of frontline worker perceptions. 15 Within the first fourteen months of implementation of our night-float rotations and using the SAQ we measured residents' attitude regarding patient safety in five hospitals.

Setting
We implemented night-float rotations simultaneously in five tertiary hospitals of Université Laval in Québec City, Canada (1500 hospitalised patients). In each site, night-float teams gathered two to four first-year residents and one or two second-or third-year internal medicine residents. Residents worked together to ensure cross-coverage of adult patient in-house calls, including new consultations at the emergency room (all sites), on most medical and surgical hospitalisation units (all sites) and in some intensive care units (sites 2 and 4). Residents worked under the immediate supervision of senior residents and on-call physicians depending on the speciality involved in each case. Residents were assigned to 16 shifts over four consecutive weeks. They worked Monday to Thursday from 8 p.m. to 8 a.m. and were exempted from daytime activities. University personnel managed residents' distribution in collaboration with residency programs.
Night-float rotations were mandatory for first-year residents from most residency programs. Instructional methods comprised video recordings, online resources, reflexive activities and logbooks. For each period, in each site, a clinical supervisor recruited among all participating specialities was responsible for residents' follow-up and assessment. Group meetings with supervisors took place weekly. Assessment criteria focused on collaboration, professionalism and recognition of medical and surgical emergencies.

Population
We surveyed by email, with two reminders, all 267 residents who had completed the night-float rotation from August 2015 to November 2016. First year residents came from most residency programs, second-and third-year residents from internal medicine.

Measurement of patient safety climate
With permission from the authors, we translated the short form of the SAQ in Canadian-French (SAQ-FR), confirmed by four bilingual educators. 19 The questionnaire was translated back to English electronically and all sentences maintained their semantic properties. Completed online, all questions began with "In your last night-float rotation […]." Thirty-one items of the 36-item SAQ elicit residents' attitude toward six safety-related climate factors (five filler items). 15 Answers were given on a Likert scale of 1 to 5: 1=disagree strongly, 2=disagree slightly, 3=neutral, 4=agree slightly, 5=agree strongly. In accordance with Sexton et al. 15 mean scores for each of the six factors were reported on a 100-point scale using this formula: (mean score of the items of a factor -1)*25.

Patient safety climate analysis
We calculated differences between residents' years of training and between sites with multivariate analysis of variance (MANOVA) to consider simultaneously the six dimensions of the SAQ. 15,18 As the sample is fairly small and groups unequal, Pillai's trace was preferred for its robustness. 20 We used Post-hoc Bonferroni tests to investigate significant differences among groups. We performed reliability analyses accordingly to classical test theory, hence we conducted item analysis to obtain Cronbach's alpha as internal consistency coefficients and corrected item-total correlations as discrimination indices. We used IBM SPSS statistics for Windows Version 20 (IBM Corp., Armonk, NY, USA).
The Research Ethics Committee of Laval University, applying rule 2.5 for quality improvement of educational projects, waived ethical approval. All participants were provided a consent form before completing the questionnaires, which remained anonymous.

Results
As presented in Table 1, 130 residents voluntarily participated in the study (48% response rate) and none were excluded. All surveys were completed without missing data and hence included for analysis. Descriptive statistics for the whole sample, across hospitals, and across residents' years of training are summarized in Table 2.
Scores on the SAQ-FR do not seem to vary across training levels or sites. Mahalanobis distance revealed 21 cases being multivariate outliers, which were excluded from analysis. MANOVA across residents' years of training revealed no significant differences between the first-year and the secondand third-year groups (p = .112). As for mean differences between different sites, MANOVA revealed no significant differences between the firstyear and second-and third-year groups (p = .076). When both considered in the same analysis, the effect of training level*site was not statistically significant either (p = .105), nor were the effects of training level (p = .202) and site (p = .069). However, the sample remained small; hence, statistical power was limited to observe significant differences. When performing individual ANOVAs which require smaller samples but are more risky because type I error is multiplied with each combined ANOVA, 20 differences between groups were significant for Perception of management between sites (F[4,129] = 2.446, p = .050), but no Bonferroni test was significant. The ANOVAs on training levels were also non-significant (p = .201 to .878). SAQ-FR showed variable internal consistency levels as presented in Table 3, with  e114 Cronbach's α from of .54 for working conditions to .85 for job satisfaction, with .86 for the whole scale.

Discussion
Faculty-led simultaneous implementation in five hospitals of a mandatory interdisciplinary night-float rotation for most junior residents was logistically feasible and showed similar and reassuring patient safety climate scores. Provided for illustrative purposes, Figure 1 contains mean SAQ scores of intensive care units where health professionals are also working at night in stressful environments. 15 Patient safety climate scores were similar across all five sites for all safety-related climate factors. Those results resonate with survey studies where residents in a night-float system felt an improvement in patient care, reporting that "better care was provided by a rested physician in spite of being less familiar with the patient." 21,22 Awaiting confirmation from a qualitative study underway, we believe certain characteristics of our night-float rotation had positive effects on the working climate. For example, the unique interdisciplinary nature of this rotation may be reflected in teamwork climate scores. Also, clear rules were given to residents and hospital personnel regarding call dispatch among the team to regulate the workload. Influencing the perception of management, each team could report to and had scheduled meetings with a supervisor who provided feedback and guidance. Although supervisors were not constantly on-site, all new consultations were discussed immediately with the on-call physician or senior resident. Also, second-and third-year internal medicine residents contributed to role modelling, feedback and bedside teaching, known as key elements for the educational value of night-float assignments by residents and faculty.

Limitations
The average response rate of SAQ in other studies is 67%. 15 Subjected to sampling bias, our response rate of 48% should be interpreted as residents' opinion regarding the safety climate. Response rates over 60% in each night-float teams would be needed for definitive cultural assessment. 14 Night-float teams were constantly moving from one physical work unit to the other. Therefore, the safety climate was assessed among the night-float teams of residents, who share common management, clinical and operational objectives. We did not include responses from the large number of health professionals in contact with residents during their rotation. Based on previous surveys, residents' self-assessment may overestimate the quality of care and differ from what would be observed by nurses and supervisors during night floats. 24 Incompletely tested in the SAQ, nightfloat rotations, although they offer more stability, can be associated with the sleep and alertness disturbances associated with night work.