Comparison of Anesthesia Beginning at On-hours Versus Off-hours on Outcomes in Patients Undergoing Hip Surgery: A Retrospective, Propensity Score–matched Cohort Study

Objective: To explore the impact of the begin time of anesthesia on in-hospital mortality and early prognosis of patients undergoing hip surgery. Methods: All patients who had hip surgery between January 1, 2015, and December 31, 2020, were evaluated in this retrospective cohort study. The primary outcome was in-hospital mortality. Secondary outcomes were: (1) postoperative hospital length of stay (postoperative LOS) and total LOS; (2) ICU admission; (3) the ratio of postoperative complications, including renal dysfunction, anemia, hypotension, deep vein thrombosis (DVT), arrhythmia, coronary artery disease (CAD) or heart failure, pulmonary infection, electrolyte disturbance, hyoxemia and delirium; (4) intraoperative outcomes, including blood loss, urine volume, concentrated red blood cells (CRBC), fresh frozen plasma (FFP), equilibrium liquid, colloidal solution, intraoperative sufentanil, and using vasoactive drugs. A propensity score–matched analysis was used to adjust for confounders to make baseline characteristics more similar within the cohort. Results: We identied 1,843 patients, of whom 1,727 had anesthesia begun at on-hours and 116 had anesthesia begun at off-hours. Before propensity score matching (PSM), in-hospital mortality (risk ratio, 19.85; 95% CI, 4.39-89.78; P <0.001), postoperative LOS (11 days [8-16] vs. 9 days [7-14]; P =0.001) and total LOS (16 days [12-25.3] vs. 14 days [12-19]; P =0.020), the proportion of ICU admission (risk ratio, 4.70; 95% CI, 1.84-12.01; P <0.001), hypotension (risk ratio, 5.96; 95% CI, 1.84-19.29; P =0.004), pulmonary infection (risk ratio, 4.74; 95% CI, 1.98-11.33; P <0.001) and hyoxemia (risk ratio, 5.32; 95% CI, 1.88-15.03; P <0.001) was higher in the off-hours group. Intraoperative CRBC (0 U [0-2] vs. 0 U [0-0]; P <0.001), FFP (0 mL [0-37.5] vs. 0 mL [0-0]; P <0.001) and the intraoperative dosage of sufentanil (24.5 vs. 19.3 μg; P =0.003) was higher in the off-hours group. After PSM, 110 patients in the on-hours group were matched to similar patients in the off-hours group. Intraoperative CRBC (0 U [0-2] vs. 0 U [0-0]; P =0.040) and FFP (0 mL [0-0] vs. 0 mL [0-0]; P =0.015) was higher in the off-hours group. And postoperative renal dysfunction (risk ratio, 1.67; 95% CI, 0.86-3.21; P =0.050) and hyoxemia (4.5% vs. 0%; P =0.060) had a tendency to be different. judgment requires further


Introduction
It is shown that working hours per week is an independent variable associated with burnout, with 59% of Chinese anesthetists working more than 50 hours per week, which would in turn lead to worse quality of patient care (Li et al., 2018). It is also reported that there is an exponential increase in occupational accidents beyond the 9th hour at work (Hanecke et al., 1998). Moreover, anesthetists are shown to report frequently working under time pressure and with frequent overtime (Lederer et al., 2018), especially during the COVID-19 epidemic (Jain et al., 2020, Lee et al., 2020). Since off-hours anesthesia becomes more and more common, it is imperative to nd the correlation between off-hours anesthesia and the prognosis of patients to guide clinical practice.
Many studies (Almallouhi et  In general, regardless of the surgery type, it's currently known that night-time surgery is associated with intraoperative adverse events and postoperative pulmonary complications (Cortegiani et al., 2019), and surgeries starting after 4:00 PM is associated with increased perioperative mortality (Whitlock et al., 2015). Based on these facts, here, we aimed at exploring the "off-hours effect" independently, without considering the "weekend effect". And we hypothesized that off-hours anesthesia is associated with adverse perioperative outcomes in patients undergoing hip surgery.

Study Design
This study has been reviewed and approved by the Hospital Institutional Review Board of Tongji Hospital, Huazhong University of Science and Technology, Wuhan, China (TJ-IRB20210305) and has been registered at ClinicalTrials.gov (NCT04957121). The need for informed consent has been waived for this study. We set out this retrospective cohort study using the Mediston Anesthesia Information System of our operating room, thus ensuring that the patient's perioperative information can be completely and accurately recorded.

Study Cohort
We identi ed all patients who underwent hip surgery in our hospital from January 1, 2015, to December 31, 2020. The surgical methods mainly included hemiarthroplasty (HA) and THA, while also containing other surgery types conducting on the hip joint. Patients who were under 18 years old or had incomplete data were excluded from this study. The principal exposure under investigation was anesthesia beginning at on-hours versus off-hours. Hip surgeries with the start time of anesthesia between 8:00 and 17:59 were coded as "on-hours", and those with anesthesia beginning between 18:00 and 7:59 were coded as "offhours". Besides, taking the long duration of hip surgery into account, if the anesthesia began before 18:00 but ended after 20:00, it was also de ned as "off-hours". We had our grouping pattern modi ed, based primarily on the grouping in previous studies (Cortegiani et
The Nottingham Hip Fracture Score (hereinafter referred to as NHFS) has been developed and validated for speci cally predicting 30-day mortality after hip fracture surgery (Maxwell et al., 2008, Moppett et al., 2012. The NHFS was assessed for individuals to keep the differences of the baseline characteristics from affecting the interpretation of outcomes. Brie y, NHFS consists of seven items: (1) 3 marks for age between 66 to 85, 4 marks for age ≥ 86; (2) 1 mark for the male gender; (3) 1 mark for the number of comorbidities ≥ 2; (4) 1 mark for mini-mental test score ≤ 6; (5) 1 mark for admission hemoglobin ≤ 100 g/L; (6) 1 mark for living in an institution; (7) 1 mark for presence of malignant disease.

Statistical Analysis
Descriptive statistics were provided of all the baseline characteristics and study outcomes. Continuous variables conforming to normal distribution were described as means (with standard deviation), and an independent-samples t test was used for comparison. Continuous variables of non-normal distribution were reported as medians (with interquartile range), and Mann-Whitney U test was used for comparison. Categorical variables were presented as counts and percentages, and chi-square test or Fisher exact test was used for comparison.

Propensity Score Matching
Propensity score matching (PSM) was used to adjust for baseline differences in our retrospective cohort. The variables (gender, age, anesthesia method, surgical method, baseline anemia, hypertension and electrolyte disturbance) which differed signi cantly across baseline characteristics were included in the propensity score calculation as covariates. And all the other baseline variables were included as additional covariates. Nearest-neighbor matching (1:1) with caliper width (0.2) without replacement was used to form a matched sample, thus proposed causal exposure baseline variables were balanced between the on-hours group and the off-hours group. Covariate balance was assessed in the matched sample by checking absolute standardized difference together with P value between two groups. We considered a covariate to be well balanced if the standardized difference was less than or close to 10%.
A P value less than 0.05 was considered to indicate statistical signi cance. All statistical tests were two tailed. Statistical analyses were performed using SPSS (Version 22; IBM, Armonk, New York). PSM was conducted using the SPSS PS Matching command.

Study population
After exclusions for 162 cases whose age < 18 yrs old and 113 missing data, there were 1843 cases nally included in study analyses, with 1727 cases in the on-hours group and 116 cases in the off-hours group (Fig. 1). The off-hours group had higher proportion of male patients, accounting for 63.8%, compared to 49.1% in the on-hours group (P = 0.002). And the mean age of the off-hours group is younger than that of the on-hours group (50.2 vs. 56.6 yrs; P < 0.001). The proportion of baseline anemia (22.4% vs. 10.5%; P < 0.001) and electrolyte disturbance (36.2% vs. 20.7%; P < 0.001) in the off-hours group was higher than that in the on-hours group, while the proportion of baseline hypertension (16.4% vs. 24.4%; P = 0.049) in the off-hours group was lower than that in the on-hours group. Furthermore, compared to the on-hours group, patients in the off-hours group were more likely to have other types of hip surgery (P < 0.001), and more likely to have their hip surgeries conducted under GA (P < 0.001) ( Table 1).   Table 2). Anesthesia beginning at off-hours was also associated with higher intraoperative dosage of sufentanil (24.5 vs. 19.3 µg; P = 0.003) ( Table 2). As to the comparison of the postoperative outcomes, the off-hours group had higher ratio of (1) [7][8][9][10][11][12][13][14]; P = 0.001) and total LOS (16 days [12-25.3] vs. 14 days [12][13][14][15][16][17][18][19]; P = 0.020) ( Table 3). Propensity Score-matched Cohort As depicted in Fig. 1, there were 220 cases in the propensity score-matched cohort, with each group containing 110 cases. We considered all the covariates well balanced according to the absolute standardized difference and P value (Table 4).   Table 5). There was no statistical signi cance of postoperative variables between the two groups after PSM, but the off-hours group tended to have a higher ratio of postoperative renal dysfunction (27.3% vs. 16.4%; RR, 1.67; 95% CI, 0.86-3.21; P = 0.050) and hyoxemia (4.5% vs. 0%; P = 0.060) ( Table 6).

Discussion
In this retrospective study, we found that off-hours anesthesia was associated with higher intraoperative CRBC and FFP in both the unmatched cohort and the propensity score-matched cohort. The intraoperative dosage of sufentanil was higher in the off-hours group before PSM, although not in the propensity score-matched analysis. Before PSM, postoperative LOS and total LOS was longer in the offhours group, while there was no statistical difference after PSM. Some postoperative adverse outcomes, including in-hospital mortality, ICU admission, hypotension, pulmonary infection, and hyoxemia, also happened more often in the off-hours group in the unmatched analysis, although not after PSM.
The anesthesia method, which have been added into propensity score matching as a covariate, is a main embodiment of the anesthetist's decision and judgment. It is worth noticing that in the off-hours group, al., 2020) over GA for the prognosis of hip surgery patients, this phenomenon might be an important factor for worse postoperative outcomes of patients in the off-hours group before PSM. It is also explicable a phenomenon that off-hours working might affect the decision and judgment of the anesthetist. It is found that, at night, fatigue may cause worse task performance and mood of anesthetists (Cao et al., 2008). Also, after night-shift sleep deprivation, anesthetists present longer reaction time and greater reliance on avoidance as a coping strategy, which have a potential negative impact on patients (Saadat et al., 2017). Besides, from another perspective, during off-hours, the spare resources of anesthesia were insu cient and the time of preoperative preparation was limited, as a consequence, GA became to be the most applicable choice for the anesthetist. Therefore, we reasoned that anesthesia starting at off-hours may negatively affect the decision and judgment of the anesthetist during hip surgery, thus leading to the occurrence of the perioperative adverse outcomes.
It is consistent that the intraoperative CRBC and FFP were higher in the off-hours group before and after PSM. Ren et al. had similar nding that patients undergoing liver transplantation in the night group had more requirement of intraoperative RBC suspension (Ren et al., 2019), while their nding that the intraoperative blood loss was also more in the night group cannot be reproduced in our analysis. It remained to be further studied why the intraoperative CRBC and FFP was higher in the off-hours group even if the intraoperative blood loss or the urine volume had no difference in the propensity scorematched cohort. Even worse, although the intraoperative CRBC and FFP were higher in the off-hours group, the ratio of postoperative renal dysfunction and hyoxemia tended to be higher in the off-hours group after PSM. These ndings might re ect that, during off-hours anesthesia, (1) the patient's intraoperative status might have been "mis-assessed" by the anesthetist, and (2) the clinical measures taken by the anesthetist might not be "su cient" to improve the patient's uid loss or hypoxia. These ndings further support that the decision and judgment of anesthetist could be greatly impaired during off-hours anesthesia.
In general, anesthesia starting at off-hours may negatively affect anesthesia method choosing and intraoperative anesthesia management. Instead of discussing the direct in uence of off-hours anesthesia on the prognosis of patient undergoing hip surgery, we are supposed to pay more attention to the adverse effects of off-hours anesthesia on the anesthetist, which may not only indirectly lead to poor prognosis of the hip surgery patient, but also affect the physical and mental health of the anesthetist.

Strengths and Limitations
As a retrospective study, although we have used PSM to minimize the baseline differences between two groups, some bias still could not be avoided. For example, due to information bias or recall bias, the incidence of some postoperative complications documented in our study was lower than in some previous prospective studies, which may bring errors to our results. In addition, due to the small sample size in this study, the proportion of some postoperative outcomes in the off-hours group may increase when the sample size is enlarged, and accordingly, the differences between two groups may become statistical.
Furthermore, we adopted a relatively arbitrary either-or grouping (Cortegiani et

Conclusion
Patients undergoing hip surgery with off-hours anesthesia is associated with more intraoperative CRBC and FFP, and also possibly associated with more often occurrence of postoperative renal dysfunction and hyoxemia. The negative effects that off-hours working has on anesthetist's mood and task performance may partially explain these adverse outcomes.

Declarations
Ethics approval and consent to participate This study has been reviewed and approved by the Hospital Institutional Review Board of Tongji Hospital, Huazhong University of Science and Technology, Wuhan, China (TJ-IRB20210305). The need for informed consent has been waived for this study.

Consent for publication
Not applicable.
Availability of data and materials The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests
All authors report no competing interests. Flowchart of the study population.