Physician shortages are an emerging international concern [1–4]. More critically, there is no immediate solution to this problem. The workforce dynamic model, proposed by Sklar, suggests three phase of physician workforce dynamic and their various pressure points to which interventions can target [4]. Specifically, the attrition and outflow of physician threatens the functioning of health care systems. High physician turnover can lead to an inadequate supply of health services, thereby compromising people’s access to health care and endangering the quality of patient care and safety [5–7]. In addition, the adverse consequences of physician turnover may include financial loss, low patient satisfaction, and poor organizational practices [8].

Rapid changes in health care environments and increasing health care costs have led to great pressures on physicians. Hospital physicians may be particularly susceptible to these factors because of the nature and environment of their work. Compared with general practitioners, hospital physicians typically treat more patients with critical conditions, and they have a higher workload and lower job control. Furthermore, previous studies have indicated that psychosocial job stressors, high levels of distress, frequent sleep problems, heavy workloads, job dissatisfaction, and poor workplace relationships are plausible factors associated with high physician turnover [9–11]. Many of these work-related stress, distress, and dissatisfaction may be resulted partly from hospital physicians’ long work hours [12, 13].

The situation may be worse in many newly developed countries. Several studies have indicated that the average work hours among physicians in Taiwan are greater than those of physicians in other developed countries [14, 15]. Although some studies are available in Asia or in Taiwan, these studies suffer from the methodological issues such as low response rates, poor generalizability, and a lack of detailed analyses on the relationship between work hours and turnover intention. Because physicians play a crucial role in ensuring a well-functioning health care system, many countries (e.g., France, Canada, and the United States) have attempted to restrict the maximum work hours of resident physicians [16]. However, in newly developed countries, understanding of these problems remains limited and few efforts have been made to regulate the work hours of hospital physicians, especially in attending physicians.

From a management perspective, a possible and intuitive approach to employee retention is to increase income. The National Health Insurance program in Taiwan is famous for low expenditure, high medical service usage, high coverage rate, and acceptable medical quality [17]. Low expenditure with high usage rate, combined with ageing problem and increased cost due to technology advancement, undoubtedly put great pressure on hospital’s financial balance [18]. As a result, health professionals’ income is always the target of controlling cost, including physicians. Recently in Taiwan, malpractice lawsuit, excessive workload, and unfair income are three most frequently mentioned causes of the physician shortage in hospitals. Whether elevating the pay satisfaction of physician is a useful approach to decrease the impact of long work hours on turnover rate is a critical question for hospital managers and health policy makers.

This is the first article to directly analyze the effect of work hours on turnover intention, and also put pay satisfaction into consideration to estimate the possible moderating effect. There are three interesting findings. First, the average work hours among hospital physicians who are above 35 years old in Taiwan was 59.8 (19.9) hours per week, which was considerably higher than that of physicians in United States (49.6 h per week for nonresident physicians) [24]. More importantly, 214 (8.6%) of the surveyed physicians worked more than 88 h per week. Second, a clear independent relationship was observed between work hours and turnover intention. These findings not only raise serious concerns regarding the health and wellbeing of physicians but also indicate that overtime work may increase physicians’ intentions to leave their current practice; thus, adding unnecessary administrative cost and more critically, endangering the continuity of patient care. The physician shortage problem in the hospital sector may threaten the function of the health care system. Unlike other factors associated with physicians’ turnover intentions (e.g., work stress, job satisfaction, etc.), work hours is a tangible and actionable factor. By building a reasonable work hour regulation, we can improve the wellbeing of physicians and diminish the possibility of leaving current practice setting.

Third, we know that even though financial incentive plays an important role in retaining physicians, it still cannot effectively moderate the adverse effect of long work hours on turnover intention. Similar results have been proposed by previous studies. One cross-sectional study in Ghana revealed that the dimensions of motivation and job satisfaction significantly associated with turnover intention included career development, workload, management, organizational commitment, and burnout, but not remuneration [25]. Another article conducted in England to retain the general practitioner workforce showed that reduced intensity and volume of workload were more important than incentive payment [26]. A review article to survey incentives for retaining health workers also documented financial incentives alone would exert limited effort. Working conditions, supervision and management, and education and training opportunities were also important and needed to be addressed together [27]. As a result, hospital managers and government should not overlook the influential role of work hours. Money cannot solve everything, and there are still root causes which drive the turnover intention of physician, such as long work hours. Taking action on improving the overload of work can assist elevating the wellbeing of physicians, and resulting in mitigating turnover intention, which assure the well function of health system.

However, there were some limitations in this study. First, because this was a cross-sectional study, we were unable to ascertain the causal relationship between work hours and turnover intention, which could be addressed by future longitudinal studies. Second, because of data limitations, some relevant information, such as the physicians’ rank (intern, resident, fellow, or attending physician) or specialty were unavailable, thus preventing detailed analyses. However, we have tried to use the variables of age and clinical settings to offer more information. Third, because we were not assured the exact number of physicians who attained the questionnaire, it’s impossible to calculate the specific physicians’ response rate in this study. Forth, work hours were self-reported and based on recall, which may have resulted in the figures being over-reported or under-reported because of perceived differences related to work. If the reporting is non-differential, the effects of work hours may be stronger than we observed. Besides, some points require further discussion. Although we used weekly work hours as our primary independent variable, the concept of work hours should be refined to elucidate the turnover problem further. For example, the number of work hours is not necessarily representative of work intensity [28]. Other variables (e.g., job control and work schedule flexibility) may also affect the psychological health of physicians [29]. The stress feeling at work, the interruptions and other variables regarding work situation was incomplete, thus, hinder further understanding about the true root of work hour issue. Moreover, the intentions to retire, take a career break, or reduce clinical hours of work are important questions about physician workforce which were lack in the survey. Finally, although we sought to include as many as possible confounding variables in the analyses according to literature review, there may still be other factors that need to be explored further. For future studies, post-questionnaire focus groups or individual one-to-one interviews would be useful to get a more in-depth understanding of the factors responsible for employment unhappiness.

In order to implement the findings of this study, we may need to rethink of the medical system. If some physicians quit form the job due to long work hours, then those who stay in position will encounter more work hours, and thus lead to stronger intention of leaving their hospital. From the results of this article, we should break up the pernicious circle between long work hours and turnover intention. However, if cutting number of hours each physician does during a week is the proposed solution, then more physicians will presumably be needed to fulfill the same workload as now. Some plausible methods should be attempted to solve the possible workforce shortage problem. For instance, task shifting to replacement medical staff, such as enhancing the role and increasing the number of nurse practitioners or physician assistants may be suitable. Developing more effective information systems may help physicians to work more efficiently during their work hour. Strengthening the ability of primary care system and spreading the ideas of integrated medicine will also reduce the workload of hospital specialty physicians. Based on above method or other innovative solutions, we can reduce physician’s long work hours without necessarily increasing the total number of physicians.

In this study, we conducted a large-scale survey of hospital physicians. The sample comprised 2423 physicians working at 100 hospitals in Taiwan, assuring good generalizability of the study. The medical centers and regional hospitals where physician shortages were the most serious in Taiwan were included in the sample. The number of total hospital physicians working in Taiwan in 2011 was 24,552, and this study included 9.9% of them to conduct analyses [30]. The sample size of 2423 physicians ensured high statistical power, not only in determining the significant effects of work hours on turnover intentions but also in examining whether pay satisfaction moderates this association. The findings show that overtime work is prevalent among hospital physicians in Taiwan. Both the Taiwanese government and hospitals must take action to address the emerging problem of physician high turnover rate. Regularly assessing physicians’ work conditions (e.g., work hours), and limiting excess work hours may be suitable policy tools. Furthermore, hospitals should not consider relying solely on financial incentives to solve the problem. There was a clear need to face the overload work hours of hospital physicians. Although it may be difficult and challenging, this study encouraged tackling work hour problem, which would lead to the possibility of solving high turnover intention among hospital physicians in Taiwan.