One main finding in this study was that the increase in job satisfaction over a period of five years only occurred among those who changed job position over this period, even when we controlled for changes in work hours and work-home interface stress over the same period. As expected, the physicians who changed position from being senior house officers experienced an increase in job satisfaction. Interestingly, also physicians other than senior house officers experienced an increase in job satisfaction after changing job position.
This study is in line with previous studies that show a higher level of job satisfaction among general practitioners than among hospital doctors in Norway [6–8]. Norwegian health-care services were reformed in recent years with more cost- and consumer-based alternatives for hospital doctors provided by the Norwegian hospital reform of 2002 [25, 26] and the establishment of a patient-list general practice outside of hospitals in 2001 . A study in 2010 showed that these reforms had little impact on physician job satisfaction, and the article concluded that it seemed likely that physician job satisfaction was based more on internal values than external changes . Whether a physician is employed or self-employed has been shown to be of importance to the physicians' satisfaction in the U.S. Employed physicians were more satisfied with leisure and family time  and with their profession and access to resources . In Norway, most general practitioners and private practice specialists are remunerated by a combination of fee for service (60%) and capitation (40%), whereas most doctors working in hospitals are employed by the state owned health trusts, which complicates direct comparisons between e.g. the U.S. and Norway.
Job insecurity among senior house officers may be relevant in explaining their lower satisfaction at T1, and they may have experienced greater satisfaction when they changed to more secure and permanent jobs (e.g. as chiefs specialists). This may also be related to natural course of career over the years, such as higher perceived competence and autonomy, increased wages etc. at T2 .
All other physicians who change job position also experience an increase in job satisfaction, even when we have controlled for work-home stress and work hours. The gravitational hypothesis from 1972 [31, 32] predicts that individuals will gravitate toward jobs that best fit their ability level. This may be a good explanation for the higher satisfaction experienced by physicians who changed jobs in our study. The physicians who changed job position were less satisfied in their job at T1 compared with their colleagues, but they were equally satisfied at T2 after they moved to a new job. A longitudinal study from 1996 supports the gravitational hypothesis . Other studies have shown that physicians whose work conditions fitted their preferred work situations and work hours experienced less burnout than their colleagues [34, 35]. Previous reports suggest there is a connection between change in job position and job satisfaction [36, 37], but to our knowledge the present study is the first to show this in a representative sample of physicians.
Our findings may have important implications for understanding physicians' career choices. First, those who experience low satisfaction with their work might find that a change in position induces greater well-being at work. Second, this principle may be valid beyond the natural career shift experienced by physicians in training, because any change in position appeared to entail higher levels of work satisfaction. Third, explicit career counselling and advice should be a focus of medical associations and those who care for a doctor's well-being. Advice and counselling should involve helping the physicians finding jobs where they can experience most possible fit early in their career. This is a challenging task in an ongoing changing profession with both intrinsic and extrinsic stressors , but nevertheless very important. Understanding the physicians' background, personality characteristics, the teaching program they have followed in medical school, their role models and other factors influencing their career path so far might help the advisers lead the physicians in the right direction [39, 40]. Also more exposure to variety of different specialities might help physicians find the position that best fits them at an earlier point in time . If a doctor is not satisfied in his/her job, career advice should also involve suggestion regarding potential change in job position, which might lead to an increased job satisfaction. The importance of career counselling and advice has been emphasized by other authors [29, 30, 42–44], but the empirical foundation for giving such advice is strengthened by the present study.
A Dutch study of the general population found no difference in work-family interface between those who stayed in the same job and those who changed jobs . We found an independent predictor effect of reduction in work-home interface stress on job satisfaction in our cohort of physicians. This suggests that a reduction in the wear and tear of life-work stress and the difficulties of balancing a doctor's work and social life over the years has an independent effect on doctors' work satisfaction, including those who change position over the years. This may be related to particularly demanding and disruptive working conditions in different job positions. The unique demands and responsibilities of patient work can take a heavy toll on a doctor's social life. Providing assistance with balancing work and private life should be a part of the career counselling and the responsibility of the employer, but these pressures may also be alleviated by support from colleagues and marital partner .
We found that a reduction in the number of work hours seemed to be of no importance to the increase in job satisfaction. Previous studies have shown that working fewer hours does not improve a physician's career satisfaction . In Norway, work hours are highly regulated, and this may be a reason for the lack of effect on satisfaction . A study of oncologists from the USA showed an effect on well-being when weekly working hours were above 60, which is a threshold far exceeding the 41 to 43 hours that our Norwegian doctors reported . Furthermore, a recent comparative study of Norwegian and German doctors linked the higher job satisfaction found among Norwegian doctors to a higher satisfaction with working hours compared with German doctors who worked longer hours .
Somewhat unexpectedly we found no differences between the genders with respect to the effects on job satisfaction of a change in position or a reduction in work-home stress. Both male and female doctors experienced a better working life when they changed position and they experienced the same effect of a reduction in work-home stress, which was possibly due to their children being older in this phase of their life. Our group previously found that female doctors at T1 worked 3.3 hours less per week than their male counterparts , and one might expect a larger reduction in work-home stress among women. Nevertheless, in the present study we controlled for the number of working hours, and the effect of this reduction was the same for men. We believe that this may be due to the gender equality of Scandinavian countries. Gender roles are not highly differentiated, and the sharing of childcare and house chores is common, at least in the highly educated population like medical doctors. In contrast, a strong impact of gender was found with respect to career choices and work-home balance in young Swiss medical doctors [45, 49].
We lack solid knowledge about the effects of job satisfaction among doctors on their work and patient care. There is increasing evidence that considerable dissatisfaction, as measured by depressive symptoms and burnout, in doctors will negatively influence their work [50–52]. It is not well known whether an increase in work satisfaction among doctors will induce better patient care, and this is a relationship that is not sufficiently explored in other occupational groups . Future prospective studies should investigate this issue.
Strengths and limitations
To our knowledge there have been no other cohort studies among physicians addressing this issue. A major strength was the longitudinal design and a relatively good response rate compared with many other studies of physicians. The weaknesses are that the study relied on self-reported measures of symptoms, and we do not know whether the increase in job satisfaction after a change in job position had any impact on a doctor's work with their patients. We did not control for the number of on-call working hours, and this might be another limitation. However, we would not expect this to be so important among relatively established doctors, who usually engage less in stressful on-call work.