In this study, the physician was the unit of analysis. We hypothesized that the number of hours physicians worked per week would decrease, and that their satisfaction would increase, as a result of closing the night clinic.

A difference-in-difference model (DID) was used to examine whether closing the night clinic was associated with the number of hours worked per week and the satisfaction of physicians. The DID is widely applied to evaluate the net effect of policies on the targets of an intervention. Its basic approach involves measuring the change in a certain indicator experienced by both an intervention and a control group, by comparing data gathered before and after the implementation of a policy. The effect of the policy is assessed by comparing the changes in the intervention group with those in the control group. Finally, the net effect is calculated through linear regression analysis performed while controlling for confounders. In this study, the intervention group was drawn from a tertiary teaching hospital located in Beijing that closed its night clinic in 2017. The control group consisted of physicians from a tertiary teaching hospital in the same city that still operated a night clinic. The DID model can be expressed as follows:

Y =β₀+β₁group +β₂year +β₃groupxyear +β_iX_i+ε.

In this study, Y is the number of hours physicians worked per week plus the scores for work–family balance and job satisfaction. We set the value of the dummy variable as 1 for physicians from the hospital that closed its night clinic (intervention group), and as 0 for the control group. β₁ thus captures all time-independent differences between the intervention and control groups. Year was also treated as a dummy variable, with a value equal to 1 for the period of time following the closure of the night clinic (2017), and a value of 0 for the period of time before its closure (2016). β₂ reflects differences in the variables of interest before versus after the night clinic closed. The coefficient β₃, which reflects the interaction between the intervention and time in the regression analysis, represents the DID estimate of the net effect of closing the night clinic. X_i refers to the control variables, including gender, age, years in practice, academic rank, educational background, marital status, number of children, management position, specialty, and so on; this, X reflectsthe unobservable factors that may affect the outcome variables.

Baseline and final investigations were conducted in December 2016 and December 2017, respectively. The China–Japan Friendship Hospital, which closed its night clinic at the beginning of 2017, was treated as the intervention group. Peking Union Medical College Hospital was randomly selected from among the hospitals that still operated a night clinic during this period to serve as the control group. Both hospitals were tertiary teaching hospitals located in Beijing. Our sample reflected the actual proportion of physicians holding each of three academic ranks (residents, attendings, and chiefs). This method yielded 200 and 300 physicians from each sampled hospital in 2016 and 2017, respectively. The final sample in 2016 included 150 physicians from the intervention hospital and 163 from the control hospital; the 2017 sample included 256 and 260 physicians from the intervention and control hospitals, respectively.

The survey protocol was reviewed and approved by the Research Ethics Committee of the Chinese Academy of Medical Science Peking Union Medical College. An informed consent form, including an explanation of the purpose of the study, was included on the first page of the survey. After accepting the terms of participation, the physicians completed the online questionnaire.

The selected physicians were invited to complete online self-administered questionnaires that solicited information on their personal characteristics (e.g., age, gender, marital status, educational background, and number of children), practice characteristics (e.g., specialty, hours worked per week, years in practice, management position, and academic rank), and satisfaction. The following items were used to measure physicians’ level of satisfaction: ‘I maintain a good balance between work and family’ and ‘I am satisfied with my current work’. Respondents rated each of these items on the following five-point Likert scale: ‘strongly agree’, ‘agree’, ‘neither agree nor disagree’, ‘disagree’, or ‘strongly disagree’.

Descriptive statistics were used to compare workload and satisfaction indicators between hospitals with and without a night clinic. Continuous variables are presented as means ± standard deviation (SD), and differences between intervention and control groups were compared with Student’s t-tests. Categorical variables are presented as percentages, and chi-square tests were used to analyze differences between the two groups. The DID model was used to calculate the effect of closing the night clinic on the variables of interest. After controlling for confounders, including age, gender, marital status, and number of children, we used general linear regression to estimate the net effect of closing the night clinic. All analyses were performed using the SAS software packages (ver. 9.4; SAS Institute Inc., Cary, NC, USA). All tests for statistical significance were two-sided, and the significance level was set at α=0.05.

As shown in Table 1, the baseline survey was completed by 150 and 163 physicians from the intervention and control groups, respectively. Except for monthly salary, there were no significant between-group differences in gender (p=0.098), age (p=0.515), years in practice (p=0.508), or educational background (p=0.580). In total, 38.0% of physicians in the intervention group reported that their salary was less than 5000 CNY, which was higher than the proportion in the control group (8.0%, P<0.001).

The post-survey included 256 physicians in the intervention group and 260 in the control group. The results were similar to those of the baseline survey, in that there were no significant between-group differences in gender (p=0.786), age (p=0.310), years in practice (p=0.955), or educational background (p=0.955). The physicians in the intervention group again reported lower salaries than those in the control group (P<0.001). The proportion of those with a monthly salary more than 10 thousand CNY was 6.7% in the intervention group, whereas it was 79.2% in the control group (Table 1).

Table 2 compares the hours worked per week and scores for work–life balance and job satisfaction between physicians in the intervention group and those in the control group, based on all data gathered over the 2-year study period. According to the unadjusted analysis (i.e., not controlling for monthly salary, gender, age, marital status, or number of children), the DID estimate was negative for the workload indicator and positive for the satisfaction indicators. The hours worked per week decreased by 1.16 hours after the night clinic was closed. The scores for ‘I maintain a good balance between work and family’ and ‘I am satisfied with my current work’ increased by 0.5 and 0.62, respectively, from the baseline to the final measurement.

Table 3 shows the net effect of the policy of night clinic closure based on the DID analysis. The effect of this policy on reducing the number of hours worked per week was statistically insignificant (P=0.822), whereas the policy significantly improved physicians’ work–life balance and job satisfaction (p=0.019 and p<0.001, respectively).

Additionally, compared with specialists in internal medicine, those in obstetrics (p<0.001), and emergency medicine (p=0.022), reported working more hours per week. Respondents with more children were less likely to maintain a good balance between work and family (p=0.017), and single physicians (i.e., unmarried, divorced, or widowed) had lower satisfaction scores compared with married physicians (p=0.039). Salary had no significant effect on physicians’ satisfaction (p=0.083).

This work was supported by the Beijing Health and Family Planning Commission. We appreciate the support of the Human Resources Departments of the China–Japan Friendship Hospital and the Peking Union Medical College Hospital.