A large number of old and new problems enhance nowadays the normal burden on physicians' general conditions. Physicians and other health workers are distributed in a very uneven manner: countries with the lowest relative need have the highest number of health workers. Health care is anyway the world's largest single industry employing in excess of 59 million staff (Gottret and Schieber, 2006; WHO, 2006). In addition, even developed countries are experiencing an important shrinking of the domestic resources attributable to the welfare services. Moreover, health systems are often vulnerable during times of crisis, such as the last one, which is still ongoing. Many health systems are functioning on limited budgets and are currently understaffed. The current need for sustainable health services is increasingly dealing with issues such as the need to reduce bureaucracy, increasing cost-effectiveness, improving efficiency, decentralizing services and allocating resources to address the needs of the population (Johnson and Stoskopf, 2010). In this context, the government of the country where this study was carried out attempted during the last decade to transfer management techniques and economic/organizational models from the private sector as a basis to reorganize and “reform” public services, which are often portrayed as monolithic, inflexible, inefficient and unable to innovate. It also recently proposed to reduce significantly physicians' autonomy in their clinical practice.

Additional problems, exacerbating physicians general conditions, are related to the exponential growth of malpractice costs, related both to liability insurance and defensive medicine costs. Premium's costs for physicians are growing relating also to the increasing number of patients' complaints. Today medicine is much more complex than 50 years ago, it often requires teams instead of a single professional and very expensive drugs and technologies, moreover the exponential growth of medical knowledge is often coupled with additional relational competences which have to be reached within fields such as inter-professional team work, physicians/patient relationship, quality of care and safety (Dyrbye and Shanafelt, 2016). Patients are nowadays much more informed and therefore willing to protect and defend their right to health.

Since the economic downturn of 2008 the trend rates of complaints increased across medical specialities (Kynes et al., 2013) and across services (consumer complaints increased also during the same period from 7 to 10%). Different ergonomic and environmental conditions such as insurances' costs, health-care institutions in general, technical aspects of care and access to information, may be correlated to a number of complaint risk factors for physicians. Despite this, interesting findings are pointing at analyzing complaints by their type, including thereby physicians' variable of individual difference. Recent studies, in fact, (Kynes et al., 2013; Zengin et al., 2014; Schaad et al., 2015) agree on pointing out that the main reason which motivates patient to consult a complaints center, or to take legal action in order to restore their safety, are converging toward “the quality of interpersonal relationship with health care professionals” (Schaad et al., 2015) and communication (Kynes et al., 2013).

Despite the large number of studies on physicians' job satisfaction (Lichtenstein, 1984; Cooper et al., 1989; Konrad et al., 1999; Williams et al., 1999; Lavanchy et al., 2004; Bell et al., 2006; Etchegaray et al., 2010; Mcintyre and Mcintyre, 2010; Hann et al., 2011; Mohr and Burgess, 2011; Jönsson, 2012; Bouwkamp-Memmer et al., 2013) and well-being (Firth-Cozens, 2001; Arnetz et al., 2002; Wallace and Lemaire, 2007; Janisse, 2008; Fang et al., 2011; Hills et al., 2012; Rockey, 2012; Alsuwaida et al., 2013; Bell, 2013; Aalto et al., 2014; Heponiemi et al., 2014; Simon and Durand-Bush, 2014; Min et al., 2015; Scheepers et al., 2015; Tucker et al., 2015), there is apparently little interest in physicians' dissatisfaction and its roots.

During the early'80 physicians dissatisfaction gained social and academic attention. The first study was carried out in 1984 (Lichtenstein). Even though this first attempt allowed researcher to develop an instrument for physicians' satisfaction and dissatisfaction which was found to have both convergent and discriminant validity it is totally out of date. Previous studies (Nathanson and Becker, 1973; Lichtenstein, 1984) don't help us to understand the multifaceted and complex nature of physicians dissatisfaction because researchers used either a single overall question or multiple question derived from literature, that simply were summed and reported as a single general level of physicians dissatisfaction. Various job satisfaction scales, including single-item, and multi-item scales have been employed during the last decades. The 10-item mono-factorial scale (Hills et al., 2012) used to assess job satisfaction in Europe, UK, North America and Australasia does not allow an in-depth, multifaceted approach to physician's dissatisfaction, thus preventing the differentiation between intrinsic and extrinsic factors.

The first systematic longitudinal survey on physicians' dissatisfaction was carried out in the US in the period from 1996 to 2001 (Landon et al., 2003). Findings demonstrate that physicians' dissatisfaction ranges did not change dramatically. Major driving forces of dissatisfaction are those related to threat of physicians' autonomy, to their ability to manage their day-today patient interactions and to their ability to provide high-quality care.

A recent study on dissatisfaction included a comparative assessment of results on Indian physicians' dissatisfaction with those of developed countries (Sharma et al., 2014). Dissatisfaction magnitude and its causes are similar, working environment and rates of pay were identified as major satisfaction/dissatisfaction factors. Work-overload and deprivation of autonomy associated to management outcome/ efficacy and efficiency evaluations are also reported as important dissatisfaction's factors. A study in Europe (Janus et al., 2008) is congruent with those findings. According to the AA. non-monetary factors are important determinant of physicians dissatisfaction, even more than monetary incentives. Seven principal factors were identified: decision making and social recognition, continuous education and job security and the increasing amount of administrative tasks.

The above-mentioned studies were carried out using measures mainly based on previous studies and on literature; their approach is therefore quantitative and focused on confirming previous results and already tested frameworks of reference. Because of the global need for enhancing integration of health systems all over the world (Lister and Labonté, 2009) and the widespread lack of economic resources, health systems constantly undergo a large number of changes. In addition research so far, largely failed to analyze the potential impact of different service systems within and between countries on physicians' well-being/dissatisfaction and performance (Linzer et al., 2002; Siegrist et al., 2010) We assume therefore there is an urgent need of accurate data on physicians' dissatisfaction addressing a 360° view of their perception relating to individual, environmental, and social causes.

According to the 2015 Medscape Physician Compensation Report physicians' dissatisfaction is increasing. Results of several studies (Reschovsky et al., 1998; Barnard and Tong, 2000; Landon et al., 2006) show that dissatisfied physicians are 2–3 times more likely to leave medicine than satisfied physicians. AA. demonstrate that dissatisfied physicians are also more likely to retire and to take that step at an early age. Women physician during their childbearing years and high-paying specialties, such as surgery, compared with lower-paid specialties, such as internal medicine, are more likely to reduce their clinical practice or retire. These finding account for a lack of relationship of this decision with income. Previous research demonstrate that unhappy physicians provide lower quality care (Linn et al., 1985; Grembowsky et al., 2005) and are unable to cope with increased physical demands and stress related to surgical specialties. Physicians' job satisfaction affects positively patients' adherence to treatment and different actions in managing their disease (Williams and Skinner, 2003) in fact dissatisfied physician have a riskier prescribing profile, less satisfied patients, less adherent patients and thus their global performance drops to a reduced quality of care (Wallace et al., 2009). In addition physicians' job satisfaction is related to patients' satisfaction rates (Haas et al., 2000; Williams and Skinner, 2003; Stelfox et al., 2005). Comparative analysis (Siegrist et al., 2010) of physicians' dissatisfaction between countries, ruled by different health-care provision systems, highlights that the highest levels of work stress is reported by physicians in Germany and the lowest level by physicians in the UK, with US physicians reporting intermediate levels. Differences are largely due to the reward component. Moreover, lower levels of administrative and clinical autonomy are associated with higher work stress in all countries. A wide range of factors may potentially raise physicians' risks of work stress (Dyrbye and Shanafelt, 2016) and dissatisfaction: frequent organizational changes, worldwide lasting crisis and subsequent lack of resources, the health system management's need for stringent outcome evaluation, increasing frequency of patients complaints, the strengthening of parameters to assess services' quality (Lister and Labonté, 2009), the growing amount of evidence based strict protocols implementation and the increasing lack of professional autonomy. Long-term exposure to stressful experiences at work and lasting job dissatisfaction can often result in burnout. Scholars (Dewa et al., 2014) show that both burnout dimensions, that is emotional exhaustion and depersonalization, are significantly associated with more use of sick leave days and with the intention to leave a current position. Moreover, physicians who have high scores for both EE and DP dimensions of burnout incur significantly greater odds of having self-perceived “insufficient” work ability (Ruitenburg et al., 2012).

The PhyDis scale consisted of 21 items. Participants were asked to indicate to which extent each item indicates a perceived subjective source of dissatisfaction. Each item was answered on a 6-point Likert scale, ranging from 1 (not at all) to 6 (extremely). On line questionnaire reached 3070 physicians of the province. The study was presented as research on physicians' dissatisfaction.

The survey was administered to all physicians of the Verona province in Northern Italy. Three thousand and seventy physicians were contacted by e-mail. A total of 1251 questionnaires were completed, with a response rate of 40.75%. We eliminated 398 questionnaires because they had more than 20% of missing responses in the PhyDis Scale and in the Global Job Satisfaction Scale (311 questionnaires), and/or they had all missing answers in the Global Job Satisfaction Scale (88 questionnaires). The final sample consisted of 853 physicians. The gender distribution was 510 males (59.8%) and 323 females (37.9%); 20 participants have not indicated the gender (2.3%). The mean age was 51.38 years (SD = 11.71; range = 24–83; 120 missing data, 14.07%), and the mean length of service was 17.94 years (SD = 12.20; range = 0–55; 101 missing data, 11.84%).

The five-factor model with 21 items showed an acceptable fit: [χ(179)2 = 789.36, p ≅ 0.00; CFI = 0.99; RMSEA = 0.03; SRMR = 0.05]. Although, the χ² was significant, all the other indices met the respective criterion. However, the inspection of loadings revealed that one item of the factor Responsibility toward patients (“Patients' suffering and pain”) showed a low loading (0.30, completely standardized coefficient). After dropping it, the fit indices for the model were: [χ(160)2 = 683.79, p ≅ 0.00; CFI = 0.99; RMSEA = 0.03; SRMR = 0.05]. The loadings were all significant (see Table 1). Correlations between the five factors (Table 2) were moderate to large (range = 0.31–0.82). The chi-square difference test showed that the correlations >0.80 were significantly different from 1 (i.e., the latent construct were distinct factors): all [χd(4)2 ≥ 58.37, ps ≤ 0.00]. The one-factor model showed an acceptable fit: [χ(170)2 = 1318.90, p ≅ 0.00; CFI = 0.96; RMSEA = 0.05; SRMR = 0.07], although worse than the five-factor model. AIC values confirmed that the five-factor solution (AIC = 369.30) was preferable compared to the one-factor model (AIC = 639.88).

The reliability of each dimension was examined by computing Cronbach's alphas for the five dimensions. Cronbach's alphas were 0.75 for Personal Costs, 0.69 for Bureaucracy, 0.68 for Physician's responsibility toward hopelessly ill patients, 0.67 for Decline in Public Image and Role Uncertainty, and 0.63 for Relationship with colleagues and staff.

The means, standard deviations, and correlations of study variables are presented in Table 2. As shown in the table, the Role Uncertainty was rated as the most dissatisfying factor, Bureaucracy immediately following. Relationship with colleagues and staff was the aspect that generated less dissatisfaction.

The correlations reveal that the five sources of dissatisfaction were positively related. Global Job Satisfaction was negatively correlated with each of the five identified sources of dissatisfaction. These results support that the two sets of measures are discriminated from each other.

Age and Length of Service were positively correlated with Global Job Satisfaction and with the dissatisfaction due to the Bureaucracy. Moreover, Age and Length of Service were negatively correlated with dissatisfaction about Personal Costs and Role Uncertainty.

There were few sex differences (Table 3). Women reported higher levels of dissatisfaction than men, with specific reference to decline in public image and role uncertainty (p = 0.001), physician's responsibility toward patients (p < 0.001); and relationship with colleagues and staff (p = 0.002); and on average, women were less satisfied (p = 0.041). There weren't differences regarding Personal Costs, and Bureaucracy.

We divided participants into two groups on the basis of their age, using the median split method. We split the sample in two groups relating to the average of age (from 24 to 53 years and from 55 to 83 years).

Younger subjects and senior ones rated differently relating to personal costs (p < 0.001); Decline in Public Image and Role Uncertainty (p = 0.007); Bureaucracy (p = 0.039); Relationship with colleagues and staff (p = 0.044); and Global Job Satisfaction (p < 0.001). The younger population perceived lower levels of job satisfaction and higher levels of dissatisfaction; but they suffered less than senior to the bureaucratization. There weren't differences regarding physician's Responsibility toward patients.

One hundred and twenty-two physicians did not complete this question. 41.6% of the subjects classified themselves as secure, 13.3% as anxious, and 45.1% as avoidant.

In line with up to date literature, in our sample the attachment assessment allowed us to identified differences between secure and insecure subjects when confronted with relational issues: insecure subjects feel more uncomfortable than secure ones when charged with the responsibility of severely ill patients (p = 0.006), and when in trouble because of relational problems with colleagues and staff (p = 0.013). Moreover, insecure subjects feel less satisfied than secure physicians (p = 0.037). No differences emerged relating to Personal costs, Bureaucracy, and Role Uncertainty.

To examine invariance of factor structure between women and men subgroups, and between senior and young subgroups, and between secure and insecure subgroups, the multi-sample procedure was applied. In four models, we tested: (a) the configural invariance (i.e., the same number of factors across the groups; Baseline model); (b) the invariance of factor loadings (λ_x; Model 1); (c) the invariance of error variances (θ_δ; Model 2); (d) the invariance of correlations and variances (Φ; Model 3; see Bobbio and Manganelli, 2011). The chi-square difference test was used to compare nested models.

Fit indices and chi-square differences are summarized in Tables 4–6.

As can be seen, the invariance of factor loadings and of error variances was not confirmed for the women-men comparison. We then constrained each factor loading, one at time, to be equal across the two groups (with the exception of the first factor loading of each factor, which was fixed to one for model identification). Each model was compared to the baseline model using the chi-square difference test. Results showed significant differences across the two groups on the loadings items Item 3 (0.66 and 0.74, ps < 0.001 for men and women, respectively), Item 4 (0.61 and 0.72, ps < 0.001 for men and women, respectively), Item 11 (0.50 and 0.45, ps < 0.001 for men and women, respectively), Item 20 (0.35 and 0.56, ps < 0.001 for men and women, respectively; see Table 1). All [χΔ(1)2 ≥ 3.71, ps ≤ 0.054].

The same procedure was applied to error variances; the model including the equality constraint was compared to the model testing the invariance of factor loadings (Model 1); significant differences were found between men and women on the error variance of items Item16 (0.84 and 0.81, for men and women, respectively, ps < 0.001), Item 15 (0.77 and 0.66, for men and women, respectively, ps < 0.001), Item 13 (0.64 and 0.63, for men and women, respectively, ps < 0.001), Item 19 (0.79 and 0.65, for men and women, respectively, ps < 0.001), all χΔ(1)2 ≥ 7.12, ps ≤ 0.01.

Invariance of factor loadings, error variances and correlations was confirmed for the senior-young comparison (see Table 5), and for the secure-insecure comparison (see Table 6).

Multiple regression analyses were used to explore the effects of physicians' dissatisfaction on Global Job Satisfaction.

There was a constant effect of age, ranging from β = 0.12, SE = 0.01, p = 0.046, to β = 0.17, SE = 0.01, p = 0.004).

The regression analysis [F_{(6, 732)} = 19.63, p < 0.001, R² = 0.14], showed that high levels of dissatisfaction relating to personal cost (β = −0.09, SE = 0.03, p = 0.030), bureaucracy (β = −0.19, SE = 0.05, p < 0.001), and problems with the staff (β = −0.14, SE = 0.05, p = 0.001), predict low job satisfaction.

The regression analysis showed some sex differences. Whereas males' low job satisfaction is predicted by increasing bureaucratic procedures [F_{(6, 441)} = 13.20, p < 0.001, R² = 0.15; β = −0.27, SE = 0.06, p < 0.001]; relational problems with colleagues and staff (β = −0.24, SE = 0.08, p = 0.002), predict females' low rates of job satisfaction [F_{(6, 271)} = 6.77, p < 0.001, R² = 0.13].

The variable which predicts low job satisfaction in insecure attached physicians is bureaucracy [F_{(6, 361)} = 8.39, p < 0.001, R² = 0.12; β = −0.19, SE = 0.08, p = 0.006]; whereas secure attached physicians' low job satisfaction is foremost due to difficulties and relational problems with colleagues/staff (β = −0.17, SE = 0.08, p = 0.027), and bureaucracy [F_{(6, 263)} = 8.61, p < 0.001, R² = 0.17; β = −0.19, SE = 0.08, p = 0.006]. Secure attached physicians are more sensitive to emotional and relational issues and therefore more prone to the risk of feeling uncomfortable when confronted with uncooperative staff's and colleagues' attitudes.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.