Effect of public hospital managers’ risk and gain perception on their attitude towards physician dual practice: a cross-national study in 31 provinces of China

Background This study aims to explore the effect of public hospital managers’ risk and gain perception on their attitude towards physician dual practice (PDP). Methods A cross-sectional study enrolling 1513 managers from public hospitals in the East, Middle and West of China was conducted. Generalized linear mixed models (GLMM) were used to identify the determinants of managers’ support for PDP. Results The rate of managers’ support for allowing PDP or implementing PDP with restriction, was 94.3% (95% CI: 0.93, 0.95). The mean score of managers’ risk perception was 67.7 ± 14.46, and the mean score of managers’ gain perception was 24.0 ± 5.56. After controlling for individual and institutional characteristics, the GLMM presented the score for risk perception increased 1 score and the rate of managers’ support for PDP decreased by 5% (OR = 0.95, 95% CI: 0.93, 0.97); while the score for gain perception increased 1 score and the rate of managers’ support increased by 18% (OR = 1.18, 95% CI: 1.12, 1.24). Conclusions Our data demonstrate that the majority of Chinese public hospital managers are in favor of allowing or implementing PDP with restrictions. Although gain perception is comparatively weaker than risk perception, a stronger influence in determining public hospital managers’ support for PDP is demonstrated.


Background
In contrast to population health care needs, health care service provision is in shortage, which is quite common in both developed and developing countries [1,2]. Hence, physician dual practice (PDP), a phenomenon that a physician can perform clinical practice in both the public and private sectors [3][4][5], is believed to provide greater access to health care services for more patients [6][7][8]. According to common opinions from 15 of the world's top medical experts and health policy makers, as well as from evaluations from the World Health Organization (WHO), PDP ranks the second among global health care human resources research concerns [9]. Through PDP, physicians can increase their income and professional satisfaction [10]. PDP may also reduce the financial burden on governments to retain high quality medical service in the public sector [7,11]. However, debates on the disadvantages of PDP never cease, and negative effects of physicians engaging in public sectors and private sectors by far exceed the positive [12,13]. Considering limits of the resources, physicians engaging in PDP may shirk their public sector duties due to competition for time, and reduce the quantity or quality of their services in public sector due to conflict of interest [14]. Furthermore, PDP also gives opportunity to hidden outflow of public sector resources, especially highly skilled physicians, rich and low complicated patients [15]. Therefore, further exploration and investigation are needed as effect might change in different contexts [5].
In 2009, China initiated new health-care reform with aims of achieving accessible and affordable health care service with full coverage for the entire population [16]. As a result, "Public Hospital Reform" and "Encouraging Private Hospital" became more important than before. Acting as a bridge between the public and private sectors, PDP draws great attention from both the government and general public for the equity, efficiency and quality of health care provision [17]. The policies and regulations on PDP underwent favorable changes in the last decade (Fig. 1). However, PDP did not enter a booming stage as expected, and substantial there are hot debates on how PDP impacts public hospitals' performance has ensued [10]. According to a survey of 35,500 Chinese physicians, 50% of practitioners did not participate in the PDP due to the reluctant attitudes public hospitals towards PDP [18]. This finding is supported by a previous study, which identified that 71.6% of the Chinese physicians reported that the resistance to PDP originated from the administration of the public hospitals [19]. In China, public hospitals take most of the responsibilities for their physicians professional development, and invest a lot during physicians career life. In turn, physicians make public hospitals trustworthy and the priority for patients. However, the intrinsic characteristic of PDP is mobility, which threatens the interdependence between public hospitals and their physicians. Moreover, PDP greatly challenges the continuity and constancy of public hospitals' personnel management, performance appraisal, compensation system, social security payment, medical risk insurance etc., thus affects the supervision capacities of public hospitals and increases their administration costs [20]. Hence, it is important to investigate the public hospital managers' attitude towards PDP, as well as the determinants.
This study which analyzed survey data collected in the East, Middle and West of China, aims to investigate the determinants influencing public hospital managers' support for PDP by using risk and gain perspectives that are based on valence-instrumentality-expectancy (VIE) theory. VIE theory is widely applied in psychological and behavioral studies to explain decision process and job performance, with a three-way interaction between (a) the importance attached to certain outcomes (Valence), (b) the extent to which performance is believed to result in these outcomes (Instrumentality), and (c) the extent to which extended effort is believed to result in effective performance (Expectancy) [21,22]. In other words, an individual's decision is given by the sum of positive and negative values (Valences), individual attaches to outcomes and individual performance-outcome probability (Instrumentality) multiplied by individual effort-performance probability (Expectancy). VIE theory argues that role perceptions of positive and negative values are important determinants of job performance. This study hypothesizes that a managers' risk perception (a negative value), as well as their gain perception (positive values), could influence their decision in supporting PDP in a public hospital setting.

Data source
From January to October in 2019, we conducted a hospital-based survey with coverage of 31 provinces in the East, Middle and West of China. Managers of public hospitals who signed the informed consent were enrolled. This study used a convenience sample, at a series of meetings, conferences, and training sessions organized by the National Health Commission for public hospital managers, including: China Modern Hospital Management Seminars, National Tertiary Public Hospital Performance Evaluation Training Courses, Conference of Experience Exchange on China's Medical Reform, Conference of Experience Exchange on Hospital Discipline Development in China and so on. Presidents, vice-presidents in charge of clinical and medical affairs, and directors of medical affairs departments in China public hospitals were included in this study. In total, 1513 questionnaires were completed.

Development of the instrument
The risk and gain perception of managers was assessed based on a questionnaire (Supplementary material 1). The questionnaire developed as follows: 1) the two variables of risk and gain perception, derived from consumer purchase decision research, were modified according to the specific scenarios of PDP; 2) two rounds of expert consultation were conducted to make preliminary amendments to the initial scale. The expert group was composed of the managers from the Medical Administration Bureau and public hospitals, professors in the public health sectors, and doctors with a professional title. Items on the scale of risk and gain perception were discussed and measured carefully by experts before the questionnaire was finally approved; 3) a presurvey was conducted with a sample of 252 managers from public hospitals in Wenzhou city and Hangzhou city; 4) homogeneity test and the change of Cronbach's α coefficient were both applied for item selection. The homogeneity test was used to calculate the correlation value (Corrected Item-Total Correlation, CITC) between each item's score and the total score of the scale as a criterion for deleting the item [23]. The CITC value of each item was higher than 0.5, indicating that all the items of manager's risk and gain perception variables were highly homogeneous. Finally, Cronbach's α was used as a test for reliability [24], and confirmatory factor analysis was used as a validity test of the questionnaire [25].

Study variables and sample size
Managers were asked to evaluate their risk perception by answering 20 questions (Section B in Supplementary material 1) and their gain perception by answering 8 questions (Section C in Supplementary material 1) using a 5-step Likert scale. The answers were reported as: completely disagree (score 1), less agree (score 2), neutral (score 3), more agree (score 4) or completely agree (score 5). The outcome variable was managers' support for PDP (the first question in Section E). This category variable was divided between completely prohibited, and permitted (including allowing with restrictions and completely allowing). In this study, binary variables (0 = completely prohibited; 1 = permitted) were used for the convenience of explaining a managers' support for PDP. According to the results of a previous study, the rate of support for managers for PDP was 44.5% [26]. Therefore, the sample size was calculated as 673 with a permissible error of δ = 3%, a type I error of α = 0.05 and an expected 20% non-response rate. We recruited a total of 1513 managers from the East, Middle and West of China in the public hospitals which met the requirement for statistical analysis.

Quality control
In order to ensure the results of the survey were accurate and reliable, we passed the results to experts with statistics expertise and experience of field research. Prior to the formal survey, research assistants were trained in accordance with a unified training plan and teaching materials. The training content included: the main purpose of the research, content, related terminology of this study and duration of this investigation; instructions for questionnaire filling and guidance for each item; explanations that may be necessary at the scene. This protocol ensured that all research assistants were familiar with the purpose and techniques required for survey, the meaning of the indicators and the content of the questionnaire before data collection. Each research assistant underwent a rigorous simulated training session before formal data collection. Regular monitoring of data collection was conducted throughout the data collection process.

Statistical analysis
In this study, all records have been checked for any missing data or outliers. The missing data in managers' support for PDP (outcome variable) and outliers in all selected variables were excluded prior to the data analysis. The scores of risk perception and gain perception were described with means and standard deviations of categorical variables including: sex, age, education, major, length of service, level of managers' position, institutional category, institutional level, institutional territory and location, which were ultimately summarized as percentages. The rates of managers' support for PDP were compared with Chi-squared test to identify the determinants. Generalized linear mixed models (GLMM) including fixed and random effects were used in this study to identify the determinants of managers' support for PDP when controlling for other confounding factors. The scores of risk and gain perception were specified as fixed effects and managers' work place as a random effect; managers' sex, age, education, major, length of service, level of managers' position, institutional category, institutional level, institutional territory and location were included as covariates (the detailed description was showed in Table 1). There are three models to explore the determinants: Three GLMMs were established from odds ratios (OR) together with 95% confidence intervals (CIs) for each determinant regarding managers' support. Model 1 included the demographic determinants of managers' sex, age, and education. Model 2 included demographic determinants from model 1 plus managers' educational major, length of service, and level of managers' position. Model 3 included determinants from models 1 and 2 in addition to work environment determinants of managers' institutional category, institutional level, and institutional territory and location. All analyses were performed

Reliability and validity of questionnaire
In this study, the Cronbach α coefficient of managers' risk and gain perception was more than 0.7, which met the internal consistency requirements of the instruments. As a confirmatory factor analysis demonstrated, from the perspective of the standardized factor load of each measurement item corresponding to the risk and gain perception, both of confirmatory factor analysis were higher than 0.5, with P value less than 0.001; CR was greater than 0.72; and AVE was greater than 0.46. Therefore, the risk and gain perception could achieve effective convergence aggregation, and the scale had good convergence validity.

Demographics of managers in the public hospital
The demographics of the sample of managers are listed in Table 2 In contrast, the mean score of managers' gain perception of 24.0 ± 5.56 was correlated with managers' age, length of service and institutional level (P < 0.05). The score of managers' gain perception was highest in managers aged less than 40 years old (24.5 ± 4.96). Managers with length of service of 6-10 years held higher gain perception scores than those with less than 5 years or more than 11 years of service. It is noteworthy that managers in tertiary hospitals had a higher risk perception (68.6 ± 14.24, P < 0.001), but lower gain perception (23.8 ± 5.51, P = 0.003) than managers from public hospitals. (Table 3).

Distribution of demographics of managers among the rate of support for PDP
The rate of managers' support for allowing PDP and implementing PDP with restrictions was 94.3% (95% CI: 0.93, Determinants of managers' support for PDP Table 5 shows the multivariate analysis of the determinants of managers' support for PDP by a generalized linear mixed model. According to the model 1, the total score of risk perception, gain perception, and age had a significant effect on the rate of managers' support for PDP. By controlling demographic characteristics from model 1, when the risk perception increased 1 score, the rate of managers' support for PDP decreased 6% (OR = 0.94, 95% CI: 0.92, 0.96). By controlling work characteristics and working environment characteristics, the results in model 2 and model 3 remained statistically significant (P < 0.001). Based on the results from model 1, when the gain perception increased 1 score, the rate of managers' support for PDP increased by 17% (OR = 1.17, 95% CI: 1.12, 1.23). The results in model 3 remained statistically significant (P < 0.001) after controlling for work and working environment characteristics. Therefore, we can identify that risk perception is negatively correlated with managers' support for PDP, but gain perception is positively correlated with managers' support for PDP. The age of managers was significantly correlated with their support for PDP in model 1. The rate of managers' support decreased 52% for those aged 41-50 years old (OR = 0.48, 95% CI: 0.24, 0.96). By controlling demographic characteristics and work characteristics in model 2, the rate of managers' support decrease to 59% for those aged 41-50 years old (OR = 0.41, 95% CI: 0.17, 0.97). However, when controlling for the demographic characteristics of work and working environment characteristics, there was no significant association between managers' age and their support for PDP.

Discussion
In this study, we found that the rate of managers' support for PDP in China is 94.3%, including support with restrictions as well as complete support, which is consistent with the goals of policy introduced after 2009 [27]. According to the previous report, PDP had direct implications for access, equity, price and quality of the whole nation's health care service, and thus benefits the whole coverage of universal health [28]. This outcome could possibly explain why most managers in the public hospitals support PDP, as they have realized the advantages of PDP for improving health care service access to those under-served areas in China. Judging by the results of our study, we are not surprise to find that the managers in the East of China (38.39%) held the highest rate for favoring PDP, followed by those from the West of China (32.63%). While managers from the Middle of China showed the lowest rate for favoring PDP (28.98%). These results may be explained by the extensive and distinguished Electronic Business (E-Business) and Information Technology (IT) industry development in the East, which connects all of China and brings easier access of PDP for both physicians and patients. According to China Internet Hospital Development Research Report 2020, the domestic health care market based on E-Business and IT has experienced an exponential growth. The scale of internet hospitals exceeded 49 billion RMB in 2018, 67.95 billion RMB in 2019, and was expected to exceed 100 billion RMB in 2020, with average daily visits of internet hospitals 2000-3000 [29].
According to our findings, risk perception and gain perception were significantly correlated with managers' support for PDP. That is, managers' support for PDP would increase when gain perception was being enhanced or risk perception was being controlled. Risk perception has been well identified as an intuitive and emotional structure closely related to human perception and behavior [30][31][32]. Obviously, it is important to reduce risk perception of managers. According to white book on medical practitioners in China 2018, involving 44,600 hospitals and 146, 200 physicians, the average weekly working hours in tertiary hospitals and secondary hospitals are 51.05 h and 51.13 h respectively, merely 19.2% physicians perceive themselves in good status during self-assessment, and psychological exhaustion of physicians is significantly greater than that of workers from other industries [33]. Constrained by limited time and energy, PDP may present as absenteeism, tardiness, inefficiency, and lack of motivation among public hospitals. So managers worry about physicians engaged in dual practice compromising service delivery. For Chinese policy-makers, promoting PDP through internet may greatly extend the outreach of PDP for physicians on one hand, and strength the capacities of monitoring and supervision for managers on the other. We could expect wide application of IT in health care industry, such as 5G, internet+, artificial intelligence and telemedicine, could pave the way for the popularity of PDP [34], and reduce the risk of physicians skimping of time and efforts in the public hospitals.
More importantly, this study also found that gain perception held more effects on the managers' support for PDP than risk perception, which may have some implications for the health-care authority in China. Previously, policy makers in China had long been neglecting the public hospital managers' gain perception towards PDP [35]. It could be largely explained by the negative attitudes of public hospital managers [36]. Many countries have prioritised gain perception [14]. For example, in the U.K., dual practitioners within the National Health Service (NHS) are required to pay 9% of their incomes as compensation to NHS. While in Germany, dual practitioners pay according to what is prescribed in their contracts with public hospitals. On the other hand, many countries also developed effective measures to reduce risk perceptions of public hospitals, such as in France and Ireland, where private practice is encouraged within public hospitals during off-duty time for easier and better monitoring. Again in U.K., physicians of the NHS are required to work at least 4 days per week within public hospitals. While in Australia, public hospitals employ lawyers to lower the negative effects on normal running of the hospital once medical disputes concerning PDP occurred [14]. Therefore, it is effective ways for Chinese policymakers to establish reasonable benefit distribution mechanism, cost sharing mechanism, economic loss identification and compensation mechanism among the public and private sectors, so as to achieve the public hospital managers' sense of gain perception.

Limitations
Firstly, the determinants of managers' support for PDP in this cross-sectional study were constrained by the pre-set questions in the surveys. There may have been some potential unobserved confounding factors were not controlled in the generalised linear mixed model. Secondly, the investigation and survey were conducted for managers of public hospitals. Other PDP stakeholders, such as managers of private hospitals and health care authorities were excluded from the study. Thirdly, the relationship between physicians' support for PDP and that of hospital managers was unclear. More evidence based on quantitative studies and randomized controlled trials may throw light on the unknowns of the mutual interference of stakeholders' attitudes and behaviors towards PDP. This study used a convenience sampling method which may have resulted in a sample that is not representative of the larger population of public hospital managers in China. In order to yield more evidence based findings, randomized studies in this field are recommended.

Conclusions
This study found that the majority of public hospital managers in China were not against PDP, rather managers' support rate for PDP, including support with restrictions, was high. In other words, although public hospital managers in China disagreed with PDP according to various public reports, by implementing PDP with proper conditions left much flexibility and feasibility for the government to achieve optimizing the productivity and capacity of medical resources.