Business Law paper

Suicidal tendency, physical health problems and addictive behaviours among general practitioners: their relationship with burnout Florent Lheureux , Didier Truchot and Xavier Borteyrou

Laboratoire de Psychologie (EA3188), Université de Franche-Comté, UFR SLHS, Besançon, France

ABSTRACT The aim of this article is to analyse further the association of burnout with (poor) physical health, addictive behaviours and suicidal tendency among general practitioners (GPs). Four hypotheses were studied: (H1): burnout (i.e. emotional exhaustion, EE, and depersonalization, DP) will be positively associated with suicidal tendency; (H2): will be negatively related to physical health (i.e. large number of physical symptoms and long-lasting impairment); and (H3): positively linked to addictive behaviours (i.e. addiction to alcohol and psychotropic medication) of GPs. Based on the “spiral of losses” depicted by the conservation of resources theory, we also considered whether physical health mediates the relationships of burnout/suicidal tendency and burnout/addictive behaviours (H4). 1890 French GPs completed a questionnaire administered by phone. Information was collected on burnout, three physical health indicators (BMI, number of physical symptoms and lasting physical health problems), four health behaviours (consumption of tobacco, alcohol, anxiolytics and antidepressants) and suicidal tendency (ideation, plan and attempt). Concerning EE, the results supported the hypotheses, except for alcohol consumption. However, the findings showed that DP was associated with more positive outcomes when controlling for exhaustion. The difference in findings for EE and DP are discussed, together with the need for GPs to develop strategies for resilience.

ARTICLE HISTORY Received 18 May 2014 Accepted 7 July 2015

KEYWORDS General practitioners; burnout; physical health; suicidal tendency; alcohol; psychotropic medication; conservation of resources theory; work-related stress

Introduction

The aim of this article is to analyse further the association of burnout with (poor) physical health, addictive behaviours and suicidal tendency among general practitioners (GPs). The prevalence of these factors is generally higher for GPs than for the general population or comparable populations and they have frequently been studied. However, their relation- ships have rarely been analysed in the same study and even more rarely studied among GPs. Thus, they need to be more fully examined, especially the role of burnout.

The prevalence of burnout among GPs has been extensively studied. The term “burnout” was introduced in the 1970s by Freudenberger (1974) and Maslach (1976) to refer to an occupational stress outcome that occurs among professionals confronted

© 2016 Informa UK Limited, trading as Taylor & Francis Group

CONTACT Florent Lheureux [email protected]

WORK & STRESS, 2016 VOL. 30, NO. 2, 173–192 http://dx.doi.org/10.1080/02678373.2016.1171806

with demanding and emotionally charged relationships with clients or patients. Burnout is generally defined as a psychological syndrome consisting of three dimensions: emotional exhaustion, depersonalization (or cynicism) and reduced personal accomplishment (Maslach & Jackson, 1981). Hence, burnout is now considered a serious and pervasive work problem not only for professionals, but also for their clients/patients as well as for organizations. For instance, burnout is known to reduce the quality of care (e.g. Williams, Manwell, Konrad, & Linzer, 2007). Several empirical studies suggest that GPs are a pro- fessional group particularly prone to burnout (e.g. Grassi & Magnani, 2000). Although the predictors of burnout among GPs have been well documented (see Lee, Seo, Hladkyj, Lovell, & Schwartzmann, 2013), the consequences of burnout on their health remain little studied and knowledge on this subject is still fragmentary.

Understanding the links between burnout and GPs’ health: insights from the conservation of resources theory

Drawing on the Conservation of Resources theory (COR, Hobfoll, 1989, 2001) this study aimed at investigating four general hypotheses regarding the links between burnout and GPs’ health. These are: burnout will be positively associated with their suicidal tendency (H1); will be positively associated with their addictive behaviours (H2); will be negatively linked to their physical health (H3) and will mediate the relationships between burnout/ suicidal tendency and burnout/addictive behaviours (H4).

The COR theory posits that individuals are motivated to obtain, retain and protect their “resources”. Resources have been defined as “objects, personal characteristics, conditions and energies that are valued by the individual or that serve as a means for attainment of these objects, personal characteristics, conditions, or energies” (Hobfoll, 1989, p. 516). Such resources are necessary for individuals to “create a world that will provide them plea- sure and success” (p. 516). Consequently, stress arises when these resources are threatened, lost or unsuccessfully invested to gain another resource.

In line with the COR theory, burnout has been defined as “an affective state character- ized by one’s feelings of being depleted of one’s physical, emotional and cognitive energies” and “follows prolonged exposure to stress” (Shirom & Melamed, 2005, p. 603). These emotional, physical and cognitive energetic resources are used by people to fulfil their pro- fessional duties and to cope with situations that potentially threaten what they value (Hobfoll & Shirom, 1993). Accordingly, burnout results from chronic exposure to the threat of valued resources (objects, conditions or personal characteristics), which necessi- tates investing all available energetic resources during a (too) long period without succeed- ing in protecting or recovering the threatened resources. Because resource conservation or recovery requires investing other resources (Principle 2 of COR theory), if individuals do not have a strong resource pool, they are less likely to succeed in resource conservation or recovery (Principle 2 corollary 1). Thus, the individual could be engaged in a “spiral of losses” (corollary 2), given that losing one kind of resource increases the likelihood of the subsequent loss of other resources if work demands remain at a too high level. This “spiral of losses” hypothesis has received growing empirical support (e.g. Armon, Shirom, Shapira, & Melamed, 2008; De Cuyper, Mäkikangas, Kinnunen, Mauno, & De Witte, 2012).

174 F. LHEUREUX ET AL.

GPs’ suicidal tendency and addictive behaviours as “outcomes” of burnout

Relying on the spiral of losses hypothesis we can hypothesize that the depletion of ener- getic resources (e.g. emotional and physical) precedes and favours both suicidal tendency and addictive behaviours. The continuous threat to valued resources (e.g. material resources, personal characteristics or conditions) coupled with insufficient and decreasing energetic resources (necessary to implement active/problem-solving coping strategies) is likely to increase feelings of learned helplessness (McMullen & Krantz, 1988) as well as depressive symptoms (Hobfoll & Shirom, 2001), thus making suicidal ideation more prevalent and suicidal plan and attempt more likely. Moreover, the highly aversive nature of this situation, as well as the inability to implement active/problem-solving coping strategies, logically suggest that the depletion of energetic resources (e.g. emotional exhaustion) favours the consumption of alcohol and psychotropic medication, considered problem-avoidance and (bad) mood-regulatory coping behaviours (Carver, Scheier, & Weintraub, 1989). Furthermore, this phenomenon is probably strengthened by the fact that resource investment is intrinsically taxing (Schönpflug, 1985), making people with limited resources reluctant to invest them and more prone to “defensive” coping.

Available supporting empirical evidence. Both suicidal tendency (ideation, plan and attempt) and addictive behaviours have been identified as especially prevalent among GPs or other medical specialties.

The risk of suicide among physicians is significantly higher than for the general popu- lation and other professionals (e.g. Gold, Sen, & Schwenk, 2013; Schernhammer & Colditz, 2004). Similarly, suicidal ideations have been identified as relatively frequent among GPs and medical students (e.g. Dyrbye et al., 2008; Hem, Grenvold, Aasland, & Ekeberg, 2000) and as proximal predictors of suicidal planning and attempts (e.g. Dennis et al., 2009; Kessler, Borges, & Walters, 1999). Several empirical papers have observed a relationship between burnout and suicidal ideation among GPs or other medical specialties (e.g. Cathé- bras, Begon, Laporte, Bois, & Truchot, 2004; Van der Heijden, Dillingh, Bakker, & Prins, 2008).

Abuse of psychoactive substances among physicians has been extensively studied for many years (see Baldisseri, 2007; O’Connor & Spickard, 1997). Especially, the con- sumption of alcohol and psychotropic medication (e.g. anxiolytics and antidepressants) have been found to be more prevalent in this population (e.g. Cathébras et al., 2004; Hughes et al., 1992; Sebo, Bouvier Gallacchi, Goehring, Künzi, & Bovier, 2007). Several studies observed that burnout was associated with alcohol consumption and the use of psychotropic medication in the physician population (which usually includes GPs: Cathébras et al., 2004; Juntunen et al., 1988; Soler, Yaman, & Esteva, 2007). Similar observations were made in the general population or in other professions (e. g. Ahola, Toppinen-Tanner, Huuhtanen, Koskinen, & Väänänen, 2009; Chen & Cunradi, 2008; Leiter et al., 2013). However, some studies found no relationship between burnout and alcohol consumption (Blanchard et al., 2010; Kuerer et al., 2007).

Therefore, both theoretical reasoning and empirical evidence lead to the formulation of the following two hypotheses: Hypothesis 1: burnout of GPs will be positively associated with their suicidal tendency (i.e. suicidal ideation, plan and attempt) (H1). Hypothesis 2: burnout of GPs will be positively linked to their addictive behaviours. More exactly,

WORK & STRESS 175

it is assumed that burned-out GPs will consume more alcohol (H2a), anxiolytics (H2b) and antidepressants (H2c) than non-burned-out GPs.

GPs’ burnout, decrement of physical health and outcomes

At this point, one question has not been addressed by this theoretical reasoning: why and how are burnout and the physical health of GPs interrelated? According to Hobfoll’s approach (1989, 2001), physical functioning and abilities as well as the body’s responsive- ness to environmental demands can be viewed as resources, and can be lost like emotional resources. Given that emotional exhaustion has been identified as the primary syndrome of the burnout process (Taris, Le Blanc, Schaufeli, & Schreurs, 2005), especially among health-care providers (Maslach, 1976; Maslach & Jackson, 1981), and constitutes its “core” component (Lee et al., 2013), emotional resources (such as the capacity to express positive feelings towards the recipients and to respond empathically to their emotional needs) are likely to be depleted first, thus increasing the likelihood of the sub- sequent loss of physical resources, which is manifested through somatic symptoms and limitation of abilities. And so, this successive loss of resources (emotional exhaustion fol- lowed by physical impairment) probably favours addictive behaviours and suicidal ten- dency in GPs.

Available supporting empirical evidence. Few studies concern the prevalence of physical health problems among physicians (including GPs) in comparison to the general popu- lation and they report conflicting results (Stavem, Hofoss, Aasland, & Loge, 2001; Töyry et al., 2000; Tyssen, 2007).

In the general population as well as in other health-care workers, the links between burnout and physical health have been fairly well documented (see Schaufeli & Enzmann, 1998; Shirom & Melamed, 2005). For instance, burnout was found to be associ- ated mainly with cardiovascular disorders, sleep disturbances, dizziness, tachycardia, diar- rhoea, loss of appetite, nausea, musculoskeletal disorders, diseases of the circulatory system or disabilities (e.g. Ahola et al., 2009; Armon et al., 2008; Kim, Ji, & Kao, 2011; Kuerer et al., 2007; Melamed, Shirom, Toker, Berliner, & Shapira, 2006; Toppinen- Tanner, Ahola, Koskinen, & Väänänen, 2009). Three studies directly demonstrated a link between burnout and the physical health of GPs (Lee, Lovell, & Brotheridge, 2010; Vela-Bueno et al., 2008).

Complementarily, numerous studies have shown a link between physical health pro- blems and suicidal tendency in the general population (e.g. Chan, Liu, Chau, & Chang, 2011; Dennis et al., 2009; Webb et al., 2012). To our knowledge, no study concerning the relationship between physical health problems of GPs and their suicidal tendency has been conducted to date.

In line with this rationale and the empirical evidence reviewed above, two more hypoth- eses can be added: Hypothesis 3: burnout of GPs will be negatively linked to their physical health. In particular, burnout is expected to be associated with a large number of physical problems/symptoms (H3a) (e.g. sleep disorders, gastric problems, dizziness, tachycardia, colitis, etc.) and with the appearance of lasting physical health problems (limitation of abil- ities, impairment) (H3b). Hypothesis 4: physical impairment will partially mediate the burnout-outcomes relationships. More precisely, the relationships between burnout and suicidal tendency (H4a), as well as with the consumption of alcohol (H4b), anxiolytics

176 F. LHEUREUX ET AL.

(H4c) and antidepressants (H4d), will be both direct and indirect (via the decrement of physical functioning and abilities).

Figure 1 summarizes all the hypotheses under study.

Method

Participants and procedure

1890 French GPs participated in the survey. 74% were men and ages ranged from 30 to 72 years (M = 50.6, SD = 7.6). On average, GPs worked 11.1 hours a day (SD = 2.3). They were part of a sample of GPs willing to participate in research on working conditions, randomly constituted from a nationally representative database of GPs in France by five URMLs (French regional associations of private practitioners) who collaborated in the study. Being a GP was the only inclusion criterion and the response rate was 94.5%. Each par- ticipant was paid the equivalent of two consultations for his/her participation. Each eli- gible GP first received a pre-notification letter describing the survey’s purpose and inviting them to participate. Telephone appointments were scheduled to administer the questionnaire. Interviewers received specific instructions on the optimal strategies to collect data in an efficient manner from GPs.

Measures

Burnout Emotional exhaustion (EE) and depersonalization (DP) were assessed with items derived from the French version of the Maslach Burnout Inventory-Human Services Survey (MBI- HSS, Maslach & Jackson, 1981), which is the most used measure of burnout. The French translation of the MBI-HSS is widely used in French-speaking countries. This tool has been validated in a sample of 383 Quebec health-workers by Dion and Tessier (1994) (i.e. factorial validity, internal consistency, long-range stability, convergent validity and

Figure 1. Summary of hypotheses investigated.

WORK & STRESS 177

hypothetico-deductive validity). The nine items of the EE scale refer to feelings of being exhausted by one’s work (e.g. “I feel fatigued when I get up in the morning and have to face another day on the job”). The five items of the DP scale assess a detached and imper- sonal response towards the recipients (e.g. “I don’t really care what happens to some reci- pients”). Ratings were given on a 7-point Likert-type scale ranging from 0 (never) to 6 (daily). The internal consistency of each sub-scale was satisfactory (EE α = .81; DP α = .64). The personal accomplishment (PA) sub-scale was not used here, given that PA is considered a distinct construct (e.g. an individual characteristic) and not a symptom of job burnout (Cordes & Dougherty, 1993; Schaufeli & Taris, 2005).

Physical health indexes The three indexes used to assess the physical health of GPs were the Body Mass Index (BMI), a somatic symptomatology index and a lasting physical health problem index. BMI corresponds to the weight in kilograms divided by the height in meters squared and rounded to 1 decimal place.

The somatic symptomatology index derives from the number of repeated physical symp- toms present. This was measured with eight questions referring to (1) sleep disorders, (2) eating problems (loss of appetite, anorexia or bulimia), (3) gastric problems (heartburn, gastro-oesophageal reflux), (4) heart palpitations, tachycardia, (5) feeling unwell, dizzi- ness, vertigo or glare, (6) breathlessness, breathing difficulties, (7) colitis, chronic intestinal pains, constipation and (8) other physical symptom, with a binary response format. The repeated presence of the symptom was coded 1 and its absence was coded 0. The total score can vary between 0 and 8 and refers to the number of physical symptoms that repeat- edly affect the GP (index of somatic symptomatology).

The lasting physical health problems index reflects the presence, the severity and the fre- quency of physical impairments and functional limitations during the last six months. This was assessed with three items. First, one question measured the presence of a chronic physical health problem during the last six months, with a binary response format (yes or no). If the answer was “yes” then two supplementary questions assessed the severity and frequency of the induced disabilities experienced, with the same response choice. The total score corresponds to the number of “yes” responses and can vary between 0 (no long-lasting health problem) and 3 (presence of a long-lasting physical health problem with severe and frequent disabilities).

Indexes of addictive behaviours Alcohol consumption was assessed with three items derived from the Alcohol Use Dis- orders Identification Test (AUDIT-C, Bush, Kivlahan, McDonell, Fihn, & Bradley, 1998). The first question measured consumption frequency during the last year, from 0 (never) to 4 (4 or more times a week). The second question only concerned consumers and assessed the number of standard drinks containing alcohol on a typical day during the last year from 0 (1–2 drinks) to 4 (10 or more). The third question assessed the fre- quency during the last year of high consumption on one occasion (i.e. six drinks or more), from 0 (never) to 4 (daily or almost daily). The alcohol consumption score corre- sponds to the sum of the three ratings and can vary between 0 and 12.

The consumption of anxiolytic medication was assessed with two questions. The first question concerned the use of anti-anxiety medication with a binary response format

178 F. LHEUREUX ET AL.

0 (no) or 1 (yes). When the answer was “yes”, a second question measured the frequency of consumption, from 1 (rarely) to 4 (daily). The total score can vary between 0 and 4.

Antidepressant consumption was assessed with the same two questions as for anxiolytic consumption. The total score can also vary between 0 and 4.

Tobacco consumption was assessed with two questions. The first one referred to current tobacco smoking status and was coded by 0 (abstinent), 1 (occasional smoker) and 2 (daily smoker). Smokers were asked a second question concerning the number of tobacco products they smoked in a day and was coded by 1 (less than 10), 2 (between 11 and 20), 3 (between 21 and 30) or 4 (more than 30). The tobacco consumption score corresponds to the product of the two ratings (frequency*quantity). The total score can vary between 0 and 8.

Suicidal tendency index Suicidal tendency was investigated with six questions. Presence/absence of suicidal idea- tion, suicide plan and suicide attempt were each measured with a binary response format 0 (no) and 1 (yes). Two periods were successively investigated: the whole life and the last 12 months. The total score corresponds to the sum of the six answers and can vary between 0 and 6.

Note that the internal consistency of these measures was not reported because they were “indexes”, which must be differentiated from “scales”. Indexes are used to assess for- mative constructs (i.e. that are formed through the accumulation of factors that theoreti- cally contribute to the same encompassing variable but are not necessarily correlated, such as for quality of life, life stress, etc., see Edwards & Bagozzi, 2000). In contrast, scales include reflective indicators of the same latent construct, which are theoretically expected to correlate strongly, making the use of internal consistency coefficients relevant (see Strei- ner, 2003).

Data analyses

First, descriptive statistics and correlations between all variables were analysed. Then, six hierarchical linear regression analyses were carried out. The four hypothesized dependent variables (suicidal tendency/consumption of alcohol/or anxiolytics/or antidepressants) were first regressed on gender, age, the average number of hours worked per day, BMI and tobacco consumption taken as control variables (Step 1). Then (Step 2), burnout indi- cators (EE, DP) were added as predictors in order to estimate their incremental predictive value (ΔR2). Last, the hypothesized mediators (lasting health problems and the number of physical symptoms) were included in Step 3. During two supplementary analyses, the two hypothesized mediators were also regressed first on control variables (Step 1), while EE and DP were subsequently included (Step 2). Given that the distributions were positively skewed with a high share of 0 score, we also performed several logistic regression analyses (72.6% of 0 score for suicidal tendency, 64.8% for lasting health problems, 94% for anti- depressant consumption, 80.3% for anxiolytics consumption, 82.6% for tobacco consump- tion, 56.9% of 0 or 1 score for the number of physical symptoms and 57% of 0, 1 or 2 score for alcohol consumption). Although the transformation of ordinal variables into binary data has a number of disadvantages (e.g. loss of information and subjectivity in the deter- mination of the cut-off value), logistic regression analysis could be viewed as more suited

WORK & STRESS 179

to the observed distributions. As the six hierarchical logistic regression analyses that were performed gave very similar results we do not report them here (they are available from the authors upon request).

Finally and complementarily, for each burnout indicator (EE and DP), mediation ana- lyses were applied using the PROCESS macro for SPSS (Hayes, 2013) in order to test the hypothesized mediations more thoroughly (with gender, age, the average number of hours worked per day, BMI, tobacco consumption and the other burnout indicator as control variables). Hayes’ approach is an integration and extension of known mediation/moder- ation analyses, which can estimate the statistical significance of indirect effects using a bootstrap procedure. Multiple samples were computed from the original sample by random replacements of values in order to test the robustness of regression coefficients by estimating a confidence interval (CI) for each indirect (i.e. mediated) effect (the 0 value must not be comprised in the CI). The original sample was resampled 2000 times and the bias-corrected percentile method was used to create 95% CI.

Note that age and gender were included in all analyses because they were regularly identified in past research as predictors of burnout, physical health, addictive behaviours and suicidal tendency (Ahola et al., 2009; Peisah, Latif, Wilhelm, & Williams, 2009; Pur- vanova & Muros, 2010; Schernhammer & Colditz, 2004).

Results

Descriptive statistics and correlations between variables

Table 1 shows the descriptive statistics and intercorrelations of the variables included in this study. Overall, GPs were quite exhausted (M = 27.45, SD = 10.08; theoretical range from 0 to 54) and slightly depersonalized their recipients (M = 11.15, SD = 11.40; theoreti- cal range from 0 to 30). These scores were somewhat higher than those observed in other French samples of GPs (Cathébras et al., 2004; Truchot, 2003, 2009). The number of phys- ical symptoms (.44), anxiolytic consumption (.22), suicidal tendency (.17), antidepressant consumption (.13), lasting physical health problems (.13) and BMI (.08) were significantly correlated with EE. On the other hand, EE was not associated with alcohol or tobacco con- sumption. Correlations with DP were lower.

DP was significantly associated with BMI (.09) and the number of physical symptoms (.09). Lasting health problems, number of physical symptoms, suicidal tendency and con- sumption of anxiolytics and antidepressants were also moderately interrelated, whereas correlations of BMI, alcohol consumption and tobacco consumption with other health indexes were quite weak. Age was positively correlated with BMI (.18), lasting health pro- blems (.22) and alcohol consumption (.23). Being a man (coded 1) was associated with a higher level of DP (.11), a higher BMI (.28) and a higher consumption of alcohol (.26), whereas being a woman (coded –1) was associated with more physical symptoms (−.11) and more intake of antidepressants (−.07).

Complementary analyses: hierarchical linear regression and mediation analyses

Table 2 presents the results of the six multiple hierarchical linear regression analyses. Except for alcohol consumption, including EE and DP always increased the explained

180 F. LHEUREUX ET AL.

Table 1. Descriptive statistics (means and standard deviations) and zero-order correlation matrix of variables under study. Variables M SD 1 2 3 4 5 6 7 8 9 10 11

1. Age 50.6 7.1 2. Gender: % men (coded men = 1, women = −1) 74% .27* 3. Emotional exhaustion 27.45 10.08 −.06 −0.02 4. Depersonalization 11.15 11.40 .00 .11* .37* 5. Body Mass Index 24.25 3.09 .18* .28** .08* .09* 6. Lasting health problems 0.51 0.78 .22* .04 .13* .04 .13* 7. No. of physical symptoms 1.58 1.55 −.01 −.11* .44* .09* .10* .29* 8. Alcohol consumption 2.45 1.82 .23* .26* −.02 .05 .12* .09* .01 9. Tobacco consumption 0.33 1.16 .04 .03 .03 −.01 −.03 .01 .06* .10* 10. Anxiolytic consumption 0.42 0.94 .03 −.05 .22* .01 −.02 .14* .25* .00 .03 11. Antidepressant consumption 0.18 0.84 −.01 −.07* .13* .00 .03 .16* .19* .02 .07* .29* 12. Suicidal tendency 0.36 0.69 .01 −.02 .17* .03 −.01 .14* .26* .05 .08* .19* .26* *p < .01.

W O R K & S T R E S S

1 8 1

Table 2. Results of the six hierarchical linear regression analyses (beta coefficients and 95% confidence intervals in brackets). Alcohol consumption Anxiolytic consumption Antidepressant consumption

Step 1 Step 2 Step 3 Step 1 Step 2 Step 3 Step 1 Step 2 Step 3 β [CI] β [CI] β [CI] β [CI] β [CI] β [CI] β [CI] β [CI] β [CI]

Age .18** [.12/.24]

.18** [.12/.24]

.16** [.10/.22]

.07* [.01/.12]

.08** [.03/.14]

.07* [.01/.12]

.01 [−.04/.07]

.02 [−.03/.08]

−.01 [−.06/.05]

Gendera .19** [.13/.25]

.19** [.13/.25]

.19** [.14/.25]

−.08* [−.13/−.02]

−.04 [−.10/.01]

−.02 [−.08/.03]

−.10** [−.16/−.04]

−.09** [−.14/−.03]

−.07* [−.12/−.01]

Average hours/day −.06* [−.11/−.00]

−.06 [−.11/.00]

−.05 [−.11/.01]

.05 [−.01/.10]

−.02 [−.07/.04]

−.01 [−.06/.05]

.03 [−.03/.09]

−.00 [−.06/.05]

.01 [−.05/.06]

Body Mass Index .05 [−.01/.10]

.05 [−.01/.10]

.03 [−.02/.09]

−.02 [−.08/.03]

−.04 [−.09/.02]

−.06* [−.12/.−01]

.06* [.01/.12]

.06* [.00/.11]

.03 [−.03/.08]

Tobacco consumpt. .10* [.05/.16]

.10** [.05/.16]

.10** [.05/.16]

.01 [−.04/.07]

−.01 [−.06/.05]

−01 [−.07/.04]

.09** [.04/.15]

.08** [.03/.13]

.07** [.02/.13]

Emotional exhaust. −.01 [−.07/.05]

−.04 [−.11/.02]

.30** [.24/.36]

.21** [.15/.27]

.17** [.11/.23]

.09** [.02/.15]

Depersonalization .03 [−.03/.09]

.04 [−.02/.10]

−.12** [−.18/−.07]

−.10** [−.16/−.04]

−.07* [−.12/−.01]

−.05 [−.10/.01]

Lasting health prob. .07* [.01/.13]

.05 [−.01/.10]

.11** [.05/.17]

No. phys. symptoms .04 [−.03/.10]

.18** [.12/.24]

.13** [.07/.19]

R2 .11 .11 .12 .01 .08 .11 .02 .04 .07 ΔR2 .11** .00 .01* .01* .07** .03** .02** .02** .03** aMen = 1, Women = −1. *p < .05. **p < .01.

1 8 2

F.LH E U R E U X E T A L.

Table 2 (continued). Results of the six hierarchical linear regression analyses (beta coefficients and 95% confidence intervals in brackets). Suicidal tendency Number of physical symptoms Lasting health problems

Step 1 Step 2 Step 3 Step 1 Step 2 Step 1 Step 2 β [CI] β [CI] β [CI] β [CI] β [CI] β [CI] β [CI]

Age .03 [−.02/.09]

.05 [−.01/.10]

.02 [−.04/.07]

−.01 [−.07/.04]

.02 [−.03/.07]

.23** [.18/.29]

.24** [.18/.29]

Gendera −.03 [−.07/.03]

−.01 [−.07/.05]

.02 [−.04/.07]

−.14** [−.20/−.08]

−.10** [−.15/−.05]

−.06* [−.12/−.00]

−.05 [−.10/.01]

Average no. hours/day .03 [−.03/.08]

−.02 [−.07/.04]

−.01 [−.06/.05]

.05 [−.00/.11]

−.04 [−.09/.01]

−.02 [−.07/.03]

−.05 [−.10/.00]

Body Mass Index −.01 [−.07/.04]

−.03 [−.08/.03]

−.06* [−.11/−.00]

.14** [.08/.20]

.11** [.06/.16]

.13** [.07/.18]

.12** [.06/.17]

Tobacco consumption .05 [−.00/.11]

.04 [−.01/.09]

.03 [−.02/.08]

.08** [.02/.13]

.05 [−.00/.09]

−.00 [−.06/.05]

−.04 [−.06/.04]

Emotional exhaustion .23** [.17/.29]

.13** [.06/.19]

.49** [.44/.54]

.15** [.09/.20]

Depersonalization −.06* [−.12/−.00]

−.04 [−.09/.02]

−.12** [−.17/−.07]

−.04 [−.10/.02]

Lasting health problems .10** [.05/.16]

No. of physical symptoms .18** [.11/.24]

R2 .01 .05 .09 .03 .23 .07 .09 ΔR2 .01 .04** .04** .03** .20** .07** .02** aMen = 1, Women = −1. *p < .05. **p < .01.

W O R K & S T R E S S

1 8 3

variance (Step 2, ΔR2 from .02 to .07, p < .01). EE was significantly and positively associ- ated with anxiolytic (β = .30, p < .01) and antidepressant (β = .17, p < .01) consumption, as well as with suicidal tendency (β = .23, p < .01), the number of physical symptoms (β = .49, p < .01) and lasting health problems (β = .15, p < .01). However, the association with alcohol consumption was non-significant (β = −.01). After the inclusion of the two hypothesized mediators at Step 3 (i.e. lasting health problems and the number of physical symptoms), EE remained significantly linked to anxiolytic consumption (β = .21, p < .01, Δβ = .09), antidepressant consumption (β = .09, p < .01, Δβ = .08) and suicidal tendency (β = .13, p < .01, Δβ = .10), although each beta coefficient decreased (see Δβ).

The results obtained concerning DP were different. Overall, the standardized beta weights were lower (mean |β| = .07 at Step 2) than for EE (mean |β| = .23 at Step 2). More- over, while the zero-order correlations with the dependent and mediating variables were non-significant (except with the number of physical symptoms, r = .09, p < .01, see Table 1), DP was negatively associated with four of these variables at Step 2 (partial correlations): anxiolytic consumption (β = −.12, p < .01), antidepressant consumption (β = −.07, p < .05), suicidal tendency (β = −.06, p < .05) and the number of physical symptoms (β = −.12, p < .01). Furthermore, DP was no longer significantly linked to antidepressant consumption and suicidal tendency once the hypothesized mediators were included at Step 3 (β = −.05 and β = −.04, respectively). Like for EE, DP was not significantly linked to alcohol consumption (β = .03).

Lasting health problems and the number of physical symptoms were both significantly and positively associated with antidepressant consumption (β = .11 and β = .13, respect- ively, p < .01) and with suicidal tendency (β = .10 and β = .18, respectively, p < .01). Alcohol consumption was only linked to lasting health problems (β = .07, p < .05) and anxiolytic consumption was only associated with the number of physical symptoms (β = .18, p < .01). Their inclusion at Step 3 systematically increased the explained variance (ΔR2 from .01, p < .05 to .04, p < .01).

Hayes’ (2013) mediation analyses corroborated the idea that EE had an indirect effect via the number of physical symptoms on suicidal tendency (.05; CI: .03/.08; Sobel’s z = 5.32, p < .001), anxiolytic consumption (.07; CI: .05/.11; z = 5.46, p < .001) and antidepress- ant consumption (.05; CI: .02/.08; z = 3.97, p < .001), but not on alcohol consumption (.03; CI: −.03/.08; z = 1.13, p = .26). Lasting health problems also partially mediated the effect of EE on suicidal tendency (.01; CI: .00/.02; z = 2.82, p < .01), antidepressant consumption (.01; CI: .00/.02; z = 2.93, p < .001) and alcohol consumption (.02; CI: .00/.04; z = 2.08, p < .04), but not on anxiolytic consumption (.01; CI: −.00/.02; z = 1.47, p = .14). Direct effects of EE on suicidal tendency (.08; CI: .04/.12; p < .001), anxiolytic consumption (.18; CI: .13/.24; p < .001) and antidepressant consumption (.06; CI: .02/.11; p < .01) were significant, whereas alcohol consumption was not directly affected by EE (−.07; CI: −.18/.04, p = .21).

Concerning DP, its negative association with suicidal tendency was significantly mediated by the number of physical symptoms (−.01; CI: −.03/−.01; z = −3.38, p < .001). Similar results were observed for anxiolytic consumption (−.02; CI: −.04/ −.01; z = −3.40, p < .001) and antidepressant consumption (−.01; CI: −.02/−.00; z = −2.93, p < .001), but not for alcohol consumption (−.01; CI: −.03/.01; –z = 1.07, p = .29). Lasting health problems significantly mediated none of the indirect effects of

184 F. LHEUREUX ET AL.

DP. The sole significant direct effect of DP was observed on anxiolytic consumption (−.10; CI: −.15/−.04; p < .001).

Discussion

Main findings

The results concerning the links between burnout dimensions and suicidal tendency (H1) showed that EE was associated with suicidal tendency among our sample of GPs, thus con- firming past studies (e.g. Cathébras et al., 2004; Van der Heijden et al., 2008). As EE can be considered the core dimension of burnout (Lee et al., 2013), its correlation with suicidal tendency somewhat supports our first hypothesis. Moreover, its correlation remained sig- nificant in multiple regression analysis. However, the non-significance of the zero-order correlation of DP suggests that it is only the exhaustion component of burnout that favours suicidal tendency, not its psychological withdrawal component, which seemed to have reverse indirect effects (see below). Hence, our first hypothesis was partially corroborated.

The links between burnout and use of alcohol were weak. The finding that GPs suffer- ing from burnout have a higher alcohol consumption (e.g. Cathébras et al., 2004; Juntunen et al., 1988) was not confirmed here. However, the fact that anxiolytic and antidepressant consumptions were related to EE was consistent with previous research (Blanchard et al., 2010; Soler et al., 2007). Although there is some support for our second hypothesis, regard- ing anxiolytic (H2b) and antidepressant (H2c) consumptions, the links between burnout and alcohol consumption (H2a) were not conclusive. It is possible that a third variable moderates the burnout/alcohol consumption relationship. Alcohol consumption may be viewed as a strategy adopted by workers in order to cope with demanding situations (Carver et al., 1989). It is a protective strategy that facilitates mental disengagement and, by doing so, temporarily reduces exposure to stressors, while impeding adoption of a more functional problem-solving strategy. Opting for one coping strategy or another largely depends on the primary appraisal of environmental demands and the sec- ondary appraisal of available resources and possibilities of action (Lazarus & Folkman, 1987). A deeper analysis of appraisals made by GPs facing excessive demands would prob- ably help to identify the existence of a moderating variable at that level.

The results concerning the links between burnout dimensions and physical health indi- cators (H3) underlined the preponderant role of EE. The highest observed correlation (.44) was between EE and the number of physical symptoms, thus confirming previous studies that highlighted the physical consequences of burnout among GPs (e.g. Lee et al., 2010; Vela-Bueno et al., 2008) and supporting our hypothesis H2a. The association between burnout and lasting health problems (H2b) was only partially supported because the magnitude of the correlation was lower for EE than it was with the number of physical symptoms and did not reach statistical significance for DP. Overall, our results also pointed out that associations between DP and physical health indicators or behaviours were lower.

Hierarchical multiple linear regression analyses and mediation analyses corroborated the idea that EE has direct effects on suicidal tendency and on the consumption of psycho- tropic drugs, as well as indirect/mediated effects via the number of physical symptoms and

WORK & STRESS 185

lasting health problems. The number of physical symptoms appeared to be the main mediator, as illustrated by regression coefficients in mediation analyses. Indirect effects on suicidal tendency and on the consumption of psychotropic medication were also observed for DP. However, contrary to our fourth hypothesis, lasting health problems did not mediate these effects. Furthermore, these effects were negative, suggesting that DP somewhat preserves GPs’ health. Although they were unexpected, these results can be explained. In fact, they are consistent with the conceptualization of DP as a coping strategy adopted by workers in order to reduce relational strain (see Taris et al., 2005) and, accordingly, they suggest that DP may be a rather protective strategy, at least in the short term.

On the whole these results corroborated most of the hypotheses under study. Generally speaking, they thus corroborated the COR theory. More precisely, the fact that EE was positively associated with a poor physical health, suicidal tendency and addiction to psy- chotropic medication is consistent with the “spiral of losses” hypothesis: principle 2/cor- ollary 1 of COR theory. These results indicate that burned-out GPs were less capable of conserving their remaining resources (e.g. physical functioning and abilities, optimism) than their colleagues because of a stronger depletion of their emotional energetic resources. Furthermore, the fact that the decrement of physical health was identified as mediating the association of EE with suicidal tendency and substance abuse is consistent with COR theory.

Gender and age differences

The higher consumption of alcohol by men compared to women in our sample was similar to previous findings in the general population (e.g. Wilsnack, Vogeltanz, Wilsnack, & Harris, 2000). The results of a higher consumption of anxiolytics and antidepressants by women were also in agreement with various studies carried out in the general popu- lation (e.g. Van der Heyden et al., 2009).

The fact that suicidal tendency did not differ significantly according to gender among GPs was not consistent with the literature (e.g. Gold et al., 2013; Schernhammer & Colditz, 2004). However, as noted by Schernhammer and Colditz (2004), most physicians are men. Thus, the lower number of women included in our representative sample and the fact that suicide and suicidal ideation are, fortunately, low-occurrence phenomena strongly limited statistical inferences.

Gender differences regarding the number of physical symptoms might be interpreted as reflecting the tendency of women to somatize more than men (Wool & Barsky, 1994), although the observed relationship in our study was weak and the gender/somatization associations in the literature depend on the presence of moderators or confounders (Creed & Barsky, 2004).

The fact that male physicians tended to score higher on DP was consistent with Schau- feli and Enzmann (1998) and Prins and collaborators (2007). However, these studies also reported significant differences concerning EE, with higher scores for women. In our sample, gender and EE score was not significantly linked. As noted by Pretty, McCarthy, and Catano (1992), the results in the literature concerning the effects of gender on burnout dimensions are mixed. In their study, they observed that job level moderates the relation- ship between these two variables: men were more emotionally exhausted in managerial

186 F. LHEUREUX ET AL.

positions and women were more exhausted in non-managerial positions. Given that such a distinction between managerial and non-managerial positions does not really concern GPs, the absence of difference between females and males in our sample appears to be quite logical. Furthermore, the gender effect in burnout is lower in European countries (see meta-analysis of Purvanova & Muros, 2010).

Concerning age, there was no difference between younger and older GPs with respect to EE and DP. The links between burnout and age among physicians are inconsistent. Some studies reported an absence of correlation (e.g. Kirwan & Armstrong, 1995) whereas other empirical studies showed that doctors with many years of experience had significantly lower scores on MBI sub-scales of DP and EE (e.g. Peisah et al., 2009). These conflicting findings regarding the burnout and age relationship could be explained by confounders like experience, position or status (Maslach, Schaufeli, & Leiter, 2001); thus age has no specific effect per se.

Limitations of the study and avenues for future research

Because this study is correlational, inferences about causal relationships between variables remain limited. More precisely, as illustrated by Taris and Kompier (2006), the results of cross-sectional mediation analyses should be approached with caution. They found that longitudinal models are better suited for examining mediation than cross-sectional models. In the latter, the temporal indeterminacy makes it impossible to estimate reliably the causal order of variables under study because of the one-phase data collection. In the present study, the application of Hayes’ procedure (2013) corroborated the idea that the number of physical symptoms that repeatedly affected GP health (e.g. sleep disorders, gastric problems, tachycardia, etc.) mediated the association of burnout (more precisely EE) with their suicidal tendency, antidepressant consumption and anxiolytic consump- tion. However, it could be argued that it was chronic physical impairment that favoured EE and, finally, suicidal tendency and addiction to psychotropic medication. Accordingly, additional analyses of EE mediating the effects of the number of physical symptoms on suicidal tendency, anxiolytic consumption and antidepressant consumption gave similar results (not shown here). The cross-sectional design used here does not enable this ques- tion to be investigated further. As recommended by Taris and Kompier (2006), a multi- phase longitudinal study, with at least three waves of data collection, is required to confirm our results.

Another limitation has to be mentioned: the use of several indexes that were developed specifically for the study somewhat reduced the comparability of our research with pre- vious studies. Thus, in addition to using a longitudinal design, future studies will need to replicate these results using measures that have already been used in previous research.

Finally, although the reported results are statistically significant, they were of relatively weak magnitude (R2 between .07 and .23 for the full models). This indicates that other predictors have to be included in future research, such as a family history of substance abuse or mental illness, or certain personality traits (e.g. mood/anxiety disorders) (Baldis- seri, 2007; Kessler et al., 1999). Taking into account the differences between GPs who are isolated and lack support (e.g. in rural areas) as opposed to GPs working in group practices who have more support from colleagues (e.g. in urban areas) would be also of great rel- evance and could strengthen the explanatory power of our model, given that the social

WORK & STRESS 187

support of colleagues has been identified as a protecting resource against burnout (e.g. Prins et al., 2007) Thus, the consideration of these probable missing covariates would allow us to test the hypotheses concerning the influence of burnout on physical health, addictive behaviours and suicidal tendency in GPs in a more robust manner.

Practical implications

Burnout is known to decrease the quality of care provided by physicians (e.g. Williams et al., 2007). Furthermore, as illustrated by our results, it could have long-term deleterious effects on their health, which would in turn decrease the overall efficacy of the health-care system given the central role of GPs in the early prevention, diagnosis and treatment of diseases (Baldisseri, 2007). Thus, apart from the fact that helping them to preserve their health is intrinsically a desirable purpose, preventing burnout of GPs has benefits for both the patients and the entire society. Because occupational antecedents of EE are gen- erally difficult to modify, recent research has focused on the development of resilience strategies. For example, these strategies can consist of leisure-time activities, self- demarcation, limitation of working hours, continuous professional development, accep- tance of professional and personal boundaries, personal reflexivity and focusing on the positive aspects of work (Zwack & Schweitzer, 2013). Attempts to develop such resilience strategies among health-care workers include mindfulness-based stress-reduction pro- grammes (Shapiro, Astin, Bishop, & Cordova, 2005) and cognitive behavioural approaches (Gardiner, Lovell, & Williamson, 2004). In the light of our results, information on the burnout process and its consequences coupled with stress-reduction programmes could help GPs to recognize the spiral of losses and develop resilience strategies.

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

Florent Lheureux http://orcid.org/0000-0002-1740-0356

References

Ahola, K., Toppinen-Tanner, S., Huuhtanen, P., Koskinen, A., & Väänänen, A. (2009). Occupational burnout and chronic work disability: An eight-year cohort study on pensioning among Finnish forest industry workers. Journal of Affective Disorders, 115(1–2), 150–159. doi:10.1016/j.jad.2008.09.021

Armon, G., Shirom, A., Shapira, I., & Melamed, S. (2008). On the nature of burnout-insomnia relationships: A prospective study of employed adults. Journal of Psychosomatic Research, 65 (1), 5–12. doi:10.1016/j.jpsychores.2008.01.012

Baldisseri, M. R. (2007). Impaired healthcare professional. Critical Care Medicine, 35(2 Suppl.), 106–116. doi:10.1097/01.CCM.0000252918.87746.96

Blanchard, P., Truchot, D., Albiges-Sauvin, L., Dewas, S., Pointreau, Y., Rodrigues, M., … Kantor, G. (2010). Prevalence and causes of burnout amongst oncology residents: A comprehensive nationwide cross-sectional study. European Journal of Cancer, 46(15), 2708–2715. doi:10.1016/ j.ejca.2010.05.014

188 F. LHEUREUX ET AL.

Bush, K., Kivlahan, D. R., McDonell, M. B., Fihn, S. D., & Bradley, K. A. (1998). The AUDIT Alcohol Consumption Questions (AUDIT-C). An effective brief screening test for problem drinking. Archives of Internal Medicine, 158(16), 1789–1795. doi:10.1001/archinte.158.16.1789

Carver, C. S., Scheier, M. F., & Weintraub, J. K. (1989). Assessing coping strategies: A theoretically based approach. Journal of Personality and Social Psychology, 56(2), 267–283. doi:10.1037/0022- 3514.56.2.267

Cathébras, P., Begon, A., Laporte, S., Bois, C., & Truchot, D. (2004). Epuisement professionnel chez les médecins généralistes. Presse Médicale, 33, 1569–1574. doi:10.1016/s0755-4982(04)98994-4

Chan, H., Liu, C., Chau, Y., & Chang, C.-M. (2011). Prevalence and association of suicide ideation among Taiwanese elderly – A population-based cross-sectional study. Chang Gung Medical Journal, 34, 197–204.

Chen, M.-J., & Cunradi, C. (2008). Job stress, burnout and substance use among urban transit operators: The potential mediating role of coping behaviour. Work & Stress, 22(4), 327–340. doi:10.1080/02678370802573992

Cordes, C. L., & Dougherty, T. W. (1993). A review and an integration of research on job burnout. Academy of Management Review, 18(4), 621–656.

Creed, F., & Barsky, A. (2004). A systematic review of the epidemiology of somatisation disorder and hypochondriasis. Journal of Psychosomatic Research, 56(4), 391–408. doi:10.1016/S0022- 3999(03)00622-6

De Cuyper, N., Mäkikangas, A., Kinnunen, U., Mauno, S., & De Witte, H. (2012). Cross-lagged associations between perceived external employability, job insecurity, and exhaustion: Testing gain and loss spirals according to the Conservation of Resources Theory. Journal of Organizational Behavior, 33(6), 770–788. doi:10.1002/job.1800

Dennis, M., Baillon, S., Brugha, T., Lindesay, J., Stewart, R., & Meltzer, H. (2009). The influence of limitation in activity of daily living and physical health on suicidal ideation: Results from a popu- lation survey of Great Britain. Social Psychiatry and Psychiatric Epidemiology, 44, 608–613. doi:10.1007/s00127-008-0474-2

Dion, G., & Tessier, R. (1994). Validation de la traduction de l’Inventaire d’épuisement profession- nel de Maslach et Jackson [Validation of a French translation of the Maslach Burnout Inventory (MBI)]. Canadian Journal of Behavioural Science/Revue Canadienne Des Sciences Du Comportement, 26, 210–227. doi:10.1037/0008-400x.26.2.210

Dyrbye, L. N., Thomas, M. R., Massie, F. S., Power, D. V, Eacker, A., Harper, W., … Shanafelt, T. D. (2008). Burnout and suicidal ideation among U.S. medical students. Annals of Internal Medicine, 149, 334–341. doi:149/5/334

Edwards, J. R., & Bagozzi, R. P. (2000). On the nature and direction of relationships between con- structs and measures. Psychological Methods, 5(2), 155–174. doi:10.1037/1082-989X.5.2.155

Freudenberger, H. J. (1974). Staff burn-out. Journal of Social Issues, 30, 159–165. doi:10.1111/j. 1540-4560.1974.tb00706.x

Gardiner, M., Lovell, G., & Williamson, P. (2004). Physician you can heal yourself! Cognitive behavioural training reduces stress in GPs. Family Practice, 21, 545–551. doi:10.1093/fampra/ cmh511

Gold, K. J., Sen, A., & Schwenk, T. L. (2013). Details on suicide among US physicians: Data from the national violent death reporting system. General Hospital Psychiatry, 35(1), 45–49. doi:10.1016/j. genhosppsych.2012.08.005

Grassi, L., & Magnani, K. (2000). Psychiatric morbidity and burnout in the medical profession: An Italian study of general practitioners and hospital physicians. Psychotherapy and Psychosomatics, 69(6), 329–334. doi:10.1159/000012416

Hayes, A. F. (2013). Introduction to mediation, moderation, and conditional process analysis. A regression-based approach. New York, NY: The Guilford Press.

Hem, E., Grenvold, N. T., Aasland, O. G., & Ekeberg, O. (2000). The prevalence of suicidal ideation and suicidal attempts among Norwegian physicians. Results from a cross-sectional survey of a nationwide sample. European Psychiatry, 15(3), 183–189. doi:10.1016/S0924-9338(00)00227-3

Hobfoll, S. E. (1989). Conservation of resources. A new attempt at conceptualizing stress. American Psychologist, 44(3), 513–524. doi:10.1037/0003-066X.44.3.513

WORK & STRESS 189

Hobfoll, S. E., & Shirom, A. (1993). Stress and burnout in work organizations. In R. T. Golembiewski (Ed.), Handbook of organization behavior (pp. 41–61). New York, NY: Marcel Dekker.

Hobfoll, S. E. (2001). The influence of culture, community, and the nested-self in the stress process: Advancing conservation of resources theory. Applied Psychology, 50(3), 337–421. doi:10.1111/ 1464-0597.00062

Hobfoll, S. E., & Shirom, A. (2001). Conservation of resources theory: Applications to stress and management in the workplace. In R. T. Golembiewski (Ed.), Handbook of organizational behav- ior (pp. 57–80). New York, NY: Marcel Dekker.

Hughes, P. H., Brandenburg, N., Baldwin, D. C. J., Storr, C. L., Williams, K. M., Anthony, J. C., & Sheehan, D. V. (1992). Prevalence of substance use among US physicians. JAMA: The Journal of the American Medical Association, 267(17), 2333–2339. doi:10.1001/jama.1992.03480170059029

Juntunen, J., Asp, S., Olkinuora, M., Aärimaa, M., Strid, L., & Kauttu, K. (1988). Doctors’ drinking habits and consumption of alcohol. British Medical Journal, 297(6654), 951–954. doi:10.1136/ bmj.297.6654.951

Kessler, R. C., Borges, G., & Walters, E. E. (1999). Prevalence of and risk factors for lifetime suicide attempts in the National Comorbidity Survey. Archives of General Psychiatry, 56(7), 617–626. doi:10.1001/archpsyc.56.7.617

Kim, H., Ji, J., & Kao, D. (2011). Burnout and physical health among social workers: A three-year longitudinal study. Social Work, 56(3), 258–268. doi:10.1093/sw/56.3.258

Kirwan, M., & Armstrong, D. (1995). Investigation of burnout in a sample of British general prac- titioners. The British Journal of General Practice : The Journal of the Royal College of General Practitioners, 45, 259–260.

Kuerer, H. M., Eberlein, T. J., Pollock, R. E., Huschka, M., Baile, W. F., Morrow, M., … Shanafelt, T. D. (2007). Career satisfaction, practice patterns and burnout among surgical oncologists: Report on the quality of life of members of the Society of Surgical Oncology. Annals of Surgical Oncology, 14(11), 3043–3053. doi:10.1245/s10434-007-9579-1

Lazarus, R. S., & Folkman, S. (1987). Transactional theory and research on emotions and coping. European Journal of Personality, 1(3), 141–169. doi:10.1002/per.2410010304

Lee, R. T., Lovell, B. L., & Brotheridge, C. M. (2010). Tenderness and steadiness: Relating job and interpersonal demands and resources with burnout and physical symptoms of stress in Canadian physicians. Journal of Applied Social Psychology, 40(9), 2319–2342. doi:10.1111/j.1559-1816. 2010.00658.x

Lee, R. T., Seo, B., Hladkyj, S., Lovell, B. L., & Schwartzmann, L. (2013). Correlates of physician burnout across regions and specialties: A meta-analysis. Human Resources for Health, 11(1), 48. doi:10.1186/1478-4491-11-48

Leiter, M. P., Hakanen, J. J., Ahola, K., Toppinen-Tanner, S., Koskinen, A., & Väänänen, A. (2013). Organizational predictors and health consequences of changes in burnout: A 12-year cohort study. Journal of Organizational Behavior, 34, 959–973. doi:10.1002/job.1830

Maslach, C. (1976). Burned out. Human Behavior, 9(5), 16–22. Maslach, C., & Jackson, S. E. (1981). The measurement of experienced burnout. Journal of

Organizational Behavior, 2(2), 99–113. doi:10.1002/job.4030020205 Maslach, C., Schaufeli, W. B., & Leiter, M. P. (2001). Job burnout. Annual Review of Psychology, 52,

397–422. doi:10.1146/annurev.psych.52.1.397 McMullen, M. B., & Krantz, M. (1988). Burnout in day care workers: The effects of learned help-

lessness and self-esteem. Child & Youth Care Quarterly, 17(4), 275–280. doi:10.1007/ BF01083908

Melamed, S., Shirom, A., Toker, S., Berliner, S., & Shapira, I. (2006). Burnout and risk of cardiovas- cular disease: Evidence, possible causal paths, and promising research directions. Psychological Bulletin, 132(3), 327–353. doi:10.1037/0033-2909.132.3.327

O’Connor, P. G., & Spickard, A. (1997). Physician impairment by substance abuse. Medical Clinics of North America, 81(4), 1037–1052. doi:10.1016/S0025-7125(05)70562-9

Peisah, C., Latif, E., Wilhelm, K., & Williams, B. (2009). Secrets to psychological success: Why older doctors might have lower psychological distress and burnout than younger doctors. Aging & Mental Health, 13(2), 300–307. doi:10.1080/13607860802459831

190 F. LHEUREUX ET AL.

Pretty, G. M. H., McCarthy, M. E., & Catano, V. M. (1992). Psychological environments and burnout: Gender considerations within the corporation. Journal of Organizational Behavior, 13(7), 701–711. doi:10.1002/job.4030130706

Prins, J. T., Hoekstra-Weebers, J. E. H. M., Gazendam-Donofrio, S. M., Van De Wiel, H. B. M., Sprangers, F., Jaspers, F. C. A., & Van Der Heijden, F. M. M. A. (2007). The role of social support in burnout among Dutch medical residents. Psychology, Health & Medicine, 12(1), 1–6. doi:10.1080/13548500600782214

Purvanova, R. K., & Muros, J. P. (2010). Gender differences in burnout: A meta-analysis. Journal of Vocational Behavior, 77(2), 168–185. doi:10.1016/j.jvb.2010.04.006

Schaufeli, W. B., & Taris, T. W. (2005). The conceptualization and measurement of burnout: Common ground and worlds apart. Work & Stress, 19(3), 256–262. doi:10.1080/ 02678370500385913

Schaufeli, W., & Enzmann, D. (1998). The burnout companion to study and practice: A critical analysis. London: Taylor & Francis.

Schernhammer, E. S., & Colditz, G. A. (2004). Suicide rates among physicians: A quantitative and gender assessment (meta-analysis). The American Journal of Psychiatry, 161(12), 2295–2302. doi:10.1176/appi.ajp.161.12.2295

Schönpflug, W. (1985). Goal-directed behavior as a source of stress: Psychological origins and con- sequences of inefficiency. In M. Frese & J. Sabini (Eds.), The concept of action in psychology (pp. 172–188). Hillsdale, NJ: Lawrence Erlbaum.

Sebo, P., Bouvier Gallacchi, M., Goehring, C., Künzi, B., & Bovier, P. A. (2007). Use of tobacco and alcohol by Swiss primary care physicians: A cross-sectional survey. BMC Public Health, 7, 5. doi:10.1186/1471-2458-7-5

Shapiro, S. L., Astin, J. A., Bishop, S. R., & Cordova, M. (2005). Mindfulness-based stress reduction for health care professionals: Results from a randomized trial. International Journal of Stress Management, 12(2), 164–176. doi:10.1037/1072-5245.12.2.164

Shirom, A., & Melamed, S. (2005). Does burnout affect physical health? A review of the evidence. In A.-S. G. Antoniou & C. L. Cooper (Eds.), Research companion to organizational health psychol- ogy (pp. 599–622). Northampton, MA: Edward Edgar.

Soler, J. K., Yaman, H., & Esteva, M. (2007). Burnout in European general practice and family medi- cine. Social Behavior and Personality, 35(8), 1149–1150. doi:10.2224/sbp.2007.35.8.1149

Stavem, K., Hofoss, D., Aasland, O. G., & Loge, J. H. (2001). The self-perceived health status of Norwegian physicians compared with a reference population and foreign physicians. Scandinavian Journal of Public Health, 29(3), 194–199. doi:10.1177/14034948010290030701

Streiner, D. L. (2003). Being inconsistent about consistency: When coefficient alpha does and doesn’t matter. Journal of Personality Assessment, 80(3), 217–222. doi:10.1207/S15327752JPA8003_01

Taris, T. W., & Kompier, M. A. J. (2006). Games researchers play – Extreme-groups analysis and mediation analysis in longitudinal occupational health research. Scandinavian Journal of Work, Environment & Health, 32(6), 463–472.

Taris, T. W., Le Blanc, P. M., Schaufeli, W. B., & Schreurs, P. J. G. (2005). Are there causal relation- ships between the dimensions of the Maslach burnout inventory? A review and two longitudinal tests. Work & Stress, 19(3), 238–255. doi:10.1080/02678370500270453

Toppinen-Tanner, S., Ahola, K., Koskinen, A., & Väänänen, A. (2009). Burnout predicts hospital- ization for mental and cardiovascular disorders: 10-year prospective results from industrial sector. Stress and Health, 25(4), 287–296. doi:10.1002/smi.1282

Töyry, S., Räsänen, K., Kujala, S., Aärimaa, M., Juntunen, J., Kalimo, R., … Husman, K. (2000). Self- reported health, illness, and self-care among Finnish physicians: A national survey. Archives of Family Medicine, 9(10), 1079–1085. doi:10.1001/archfami.9.10.1079

Truchot, D. (2003). Le burnout des médecins libéraux de Champagne Ardenne: Rapport de recherche (p. 43). Reims, France. Retrieved from http://ireps-ors-paysdelaloire.centredoc.fr/ opac/index.php?lvl=notice_display&id=19164

Truchot, D. (2009). Le burn-out des médecins généralistes: influence de l’iniquité perçue et de l’or- ientation communautaire. Annales Médico-Psychologiques, 167(6), 422–428. doi:10.1016/j.amp. 2009.03.018

WORK & STRESS 191

Tyssen, R. (2007). Health problems and the use of health services among physicians: A review article with particular emphasis on Norwegian studies. Industrial Health, 45(5), 599–610. doi:10.2486/indhealth.45.599

Van der Heijden, F., Dillingh, G., Bakker, A., & Prins, J. (2008). Suicidal thoughts among medical residents with burnout. Archives of Suicide Research, 12, 344–346. doi:10.1080/ 13811110802325349

Van der Heyden, J. H. A., Gisle, L., Hesse, E., Demarest, S., Drieskens, S., Tafforeau, J., & Van Der. Heyden, J. H. A. (2009). Gender differences in the use of anxiolytics and antidepressants: A population based study. Pharmacoepidemiology and Drug Safety, 18(11), 1101–1110. doi:10. 1002/pds

Vela-Bueno, A., Moreno-Jiménez, B., Rodríguez-Muñoz, A., Olavarrieta-Bernardino, S., Fernández-Mendoza, J., De la Cruz-Troca, J. J., … Vgontzas, A. N. (2008). Insomnia and sleep quality among primary care physicians with low and high burnout levels. Journal of Psychosomatic Research, 64(4), 435–442. doi:10.1016/j.jpsychores.2007.10.014

Webb, R. T., Kontopantelis, E., Doran, T., Qin, P., Creed, F., & Kapur, N. (2012). Suicide risk in primary care patients with major physical diseases: A case-control study. Archives of General Psychiatry, 69(3), 256–264. doi:10.1001/archgenpsychiatry.2011.1561

Williams, E. S., Manwell, L. B., Konrad, T. R., & Linzer, M. (2007). The relationship of organiz- ational culture, stress, satisfaction, and burnout with physician-reported error and suboptimal patient care: Results from the MEMO study. Health Care Management Review, 32(3), 203– 212. doi:10.1097/01.HMR.0000281626.28363.59

Wilsnack, R. W., Vogeltanz, N. D., Wilsnack, S. C., & Harris, T. R. (2000). Gender differences in alcohol consumption and adverse drinking consequences: Cross-cultural patterns. Addiction, 95(2), 251–265. doi:10.1046/j.1360-0443.2000.95225112.x

Wool, C. A., & Barsky, A. J. (1994). Do women somatize more than men? Gender differences in somatization. Psychosomatics, 35(5), 445–452. doi:10.1016/S0033-3182(94)71738-2

Zwack, J., & Schweitzer, J. (2013). If every fifth physician is affected by burnout, what about the other four? Resilience strategies of experienced physicians. Academic Medicine: Journal of the Association of American Medical Colleges, 88, 382–389. doi:10.1097/ACM.0b013e31828

192 F. LHEUREUX ET AL.

Copyright of Work & Stress is the property of Taylor & Francis Ltd and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.