An International Journal of Work, Health & Organisations Work& Stress
�
Work & Stress
An International Journal of Work, Health & Organisations Work& Stress
___— …..,.._
ISSN: 0267-8373 (Print) 1464-5335 (Online)Journal homepage: https://www.tandfonline.com/loi/twst20
Suicidal tendency, physical health problems and addictive behaviours among general practitioners: their relationship with burnout
Florent Lheureux, Didier Truchot & Xavier Borteyrou
To cite this article: Florent Lheureux, Didier Truchot & Xavier Borteyrou (2016) Suicidal tendency, physical health problems and addictive behaviours among general practitioners: their relationship with burnout, Work & Stress, 30:2, 173-192, DOI: 10.1080/02678373.2016.1171806
To link to this article: https://doi.org/10.1080/02678373.2016.1171806
a Published online: 12 Apr 2016.
�� Submit your article to this journal 8′
W!l_ Article views: 628
….� View related articles 8′
(I) View Crossmark data 8′ CrossM.irk
eeJ Citing articles: 15 View citing articles 8′
Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journal1nformation?journa1Code=twst20
� WORK & STRESS, 2016 VOL. 30, NO. 2, 173-192 l
i � Routledge Taylor & Francis Group http://dx.doi.org/10.1080/02678373.2016. l 171806
Suicidal tendency, physical health problems and addictive behaviours among general practitioners: their relationship with burnout
Florent Lheureux G, Didier Truchot and Xavier Borteyrou
Laboratoire de Psychologie (EA3188), Universite de Franche-Comte, UFR SLHS, Besarn;on, France
ABSTRACT ARTICLE HISTORY 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: (Hl ): 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.
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
CONTACT Florent Lheureux Q florent.lheureux@univ-fcomte.fr
© 2016 lnforma UK Limited, trading as Taylor & Francis Group
174 9 F. LHEUREUX ET AL.
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 (Hl); 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, Makikangas, Kinnunen, Mauno, & De Witte, 2012).
WORK & STRESS (9 175
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 (Schonpflug, 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. Cathe 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. Cathebras et al. , 2004; Hughes et al., 1992; Sebo, Bouvier Gallacchi, Goehring, Kiinzi, & 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: Cathebras 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, & Vaananen, 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) (Hl). Hypothesis 2: burnout of GPs will be positively linked to their addictive behaviours. More exactly,
176 9 F. LHEUREUX ET AL.
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; Toyry 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, & Vaananen, 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 ofelasting 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
(H4c) and antidepressants (H4d), will be both direct and indirect (via the decrement of physical functioning and abilities).
Figure 1 swnmarizes all the hypotheses under study.
———————
·· …………………………………………. � :· :::�==========::..
Antidepressants symptoms
‘ /….••’ ·• ….\.Addiction to …
Alcohol
Burnout Physical health Anxiolytics Number of physical Emotional
Exhaustion \
��,�—————– 4 Depersonalization Lasting health / ··• ..
‘ Suicidal tendency Iproblems. __.. 1Ideations, plans
-> and::•=�� :,,’�-____________HI __________
J
WORK & STRESS (9 177
Figure 1. Summary of hypotheses investigated.
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
178 9 F. LHEUREUX ET AL.
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 a= .81; DP a = .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
WORK & STRESS (9 179
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 (frequencytquantity). 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 (L’1R2
) . 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
180 9 F. LHEUREUX ET AL.
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% CL
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).
Resu lts
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 O 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 (Cathebras 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
�
Table 1 . Descriptive stati stics (means and sta ndard deviations) and zero-order corre lat ion matrix of var iables u nder study.
Variab les M SD 2 3 4 5 6 7 8 9 1 0 1 1
1 . Age 2. Gender: % men (coded men = 3. Emotional exhaustion 4. Depersona l ization 5. Body Mass Index 6. Last ing hea lth problems 7. No. of physica l symptoms 8. Alcohol consumption 9. Tobacco consumption 1 0. Anxio lytic consumption 1 1 . Antidepressant consumption 1 2. Suic idal tendency
*p < .01 .
1 , women = -1 ) 50.6
74% 27.45 1 1 . 1 5 24.25 0.5 1 1 .58 2 .45 0.33 0.42 0.1 8 0.36
7 . 1
1 0.08 1 1 .40 3.09 0.78 1 .55 1 .82 1 . 1 6 0.94 0.84 0.69
.27* -.06
.00
. 1 8*
.22* -.01
.23*
.04
.03 – .01
.01
-0.02 . 1 1 * .28**
.04
-. 1 1 * .26* .03
-.05 -.07* -.02
.37*
.08*
. 1 3*
.44* -.02
.03
.22*
. 1 3*
. 1 7*
.09*
.04
.09*
.05 -.01
.01
.00
.03
. 1 3*
. 1 0*
. 1 2* -.03 -.02
.03 -.01
.29*
.09*
.01
. 1 4*
. 1 6*
. 1 4*
.01
.06*
.25*
. 1 9*
.26*
. 1 0*
.00
.02
.05
.03
.07*
.08* .29* . 1 9* .26*
0 :xi ;,,::
QO
:xi m V, V,
Table 2. Resu lts of the six hierarchical linear regression analyses (beta coefficients and 95% confidence intervals in brackets) . Alcohol consumption Anxiolytic consumpt ion Antidepressant consumption
Step 1 /3 [Cl]
Step 2 /3 [Cl]
Step 3 /3 [Cl]
Step 1 /3 [Cl]
Step 2 /3 [Cl]
Step 3 /3 [Cl]
Step 1 /3 [Cl]
Step 2 /3 [Cl]
Step 3 /3 [Cl]
Age
Gender”
Average hours/day
Body Mass Index
Tobacco consumpt.
Emotional exhaust.
Depersonalization
Lasting health prob.
No. phys. symptoms
R2
tiR2
•Men = 1 , Women = – * p < .05 . ** p < .01 .
1 .
. 18**
[. 1 2/.24] 9** . 1
[ . 1 3/.25] -.06*
[-. 1 1 /- .00] .OS
[-.01 /. 1 0] . 1 0*
[.05/. 1 6]
. 1 1 ** . 1 1
. 18**
[ . 1 2/.24] . 1 9**
[. 1 3/.25] -.06
[-. 1 1 /.00] .OS
[- .01 /. 1 0] . 1 0**
[.05/. 1 6] -.01
[-.07/.05] .03
[-.03/.09]
. 1 1
.00
. 1 6**
[. 1 0/.22] 9** . 1
[ . 14/.25] – .05
[-. 1 1 /.01 ] .03
[-.02/.09] . 1 0**
[.05/. 1 6] – .04
[-. 1 1 /.02] .04
[-.02/. 1 0] .07*
[.01 /.1 3] .04
[-.03/.1 0] . 1 2
* .01
.07*
[.01 /. 1 2] -.08*
[-. 1 3/-.02] .OS
[- .01 / .1 0] -.02
[-.08/.03] .01
[- .04/.07]
.01 * .01
.08**
[.03/. 14] -.04
[-. 1 0/.0 1 ] – .02
[-.07/.04] -.04
[-.09/.02] -.01
[-.06/.05] .30**
[.24/.36] 2** – . 1
[-. 18/-.07]
.08
.07**
.07*
[.0 1 /. 1 2] -.02
[-.08/.03] -.01
[-.06/.05] -.06*
[-. 1 2/.-0 1 ] -01 [-.07/.04]
** .21 [. 1 5/.27]
– . 10**
[-. 1 6/-.04] .OS
[-.01 /. 1 0] . 18**
[. 1 2/.24] . 1 1 .03**
.01 [-.04/.07]
-. 1 0**
[- . 1 6/-.04] .03
[-.03/.09] .06*
[.0 1 /. 1 2] .09**
[ .04/. 1 5]
.02
.02**
.02 [- .03/.08]
-.09**
[-. 14/- .03] -.00
[-.06/.05] .06*
[.00/. 1 1 ] .08**
[ .03/.13] . 1 7**
[. 1 1 /.23] -.07*
[- . 1 2/-.01 ]
.04
.02**
-.01 [- .06/.05] -.07*
[-. 1 2/- .01 ] .01
[- .05/.06] .03
[-.03/.08] .07**
[.02/. 1 3 ] .09**
[.02/. 1 5 ] -.05
[-. 1 0/.01 ] ** . 1 1
[.05/. 1 7] . 1 3**
[.07/. 1 9] .07 .03**
:-n :::c rn C :,::i rn C
� X
�
Table 2 (continued). Results of the six hierarchical l inear regression analyses (beta coefficients and 95% confidence intervals in brackets) .
Step 1 /3 [C l ]
Suicidal tendency
Step 2 /3 [C l ]
Step 3 /3 [C l ]
Number of physical symptoms
Step 1 Step 2 /3 [C l ] /3 [C l ]
Lasting health problems
Step 1 Step 2 /3 [C l ] /3 [C l ]
Age
Gender”
Average no. hours/day
Body Mass Index
Tobacco consumption
Emotiona l exhaustion
Depersonalization
Lasting health problems
No. of physical symptoms
.03 [-.02/.09]
– .03 [-.07/.03]
.03 [- .03/.08]
-.01 [-.07/.04]
.OS [-.00/. 1 1 ]
.01
.01
.OS [-.01 /. 1 0] -.01
[-.07/.05] -.02
[-.07/.04] -.03
[-.08/.03] .04
[-.01 /.09] .23**
[. 1 7/.29] -.06*
[-. 1 2/- .00]
.OS
.04**
.02 [-.04/.07]
.02 [- .04/.07] -.01
[- .06/.05] -.06*
[- . 1 1 /- .00] .03
[-.02/.08] . 1 3**
[.06/.1 9] -.04
[-.09/.02] . 1 0**
[.05/.1 6] . 18**
[. 1 1 /.24] .09 .04**
-.01 [-.07/.04]
-. 14**
[-.20/-.08] .OS
[-.00/. 1 1 ] . 14**
[.08/.20] .08**
[.02/. 1 3]
.03
.03**
.02 [-.03/.07]
-. 1 0**
[- . 15/-.05] -.04
[-.09/.0 1 ] ** . 1 1
[.06/. 1 6] .OS
[-.00/.09] .49**
[.44/.54] 1 2** -.
[- . 1 7/-.07]
.23
.20**
.23**
[. 18/.29] -.06*
[-. 1 2/-.00] -.02
[- .07/.03] 3** . 1
[.07/. 18] – .00
[-.06/.05 ]
.07
.07**
.24**
[. 18/.29] -.05
[-. 1 0/.01 ] -.05
[-. 1 0/.00] 2** . 1
[.06/. 1 7] -.04
[-.06/.04] . 15**
[.09/.20] -.04
[-. 1 0/.02]
.09
.02**
•Men = 1 , Women *p < .05. **p < .01 .
= – 1 . 0 :xi ;,,::
QO
:xi m V, V,
w 00
184 9 F. LHEUREUX ET AL.
variance (Step 2, !).R2 from .02 to .07, p < .01). EE was significantly and positively associ ated with anxiolytic (/3 = .30, p < .01) and antidepressant (/3 = .17, p < .01) consumption, as well as with suicidal tendency (/3 = .23, p < .01), the number of physical symptoms (/3 = .49, p < .01) and lasting health problems (/3 = .15, p < .01). However, the association with alcohol consumption was non-significant (/3 = -.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 (/3 = .21, p < .01, 1)./3 = .09), antidepressant consumption (/3 = .09, p < .01, 1)./3 = .08) and suicidal tendency (/3 = .13, p < .01, 1)./3 = .10), although each beta coefficient decreased (see 1)./3) .
The results obtained concerning DP were different. Overall, the standardized beta weights were lower (mean 1/3 1 = .07 at Step 2) than for EE (mean 1/3 1 = .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 (/3 = -.12, p < .01), antidepressant consumption (/3 = -.07, p < .05), suicidal tendency (/3 = -.06, p < .05) and the number of physical symptoms (/3 = -.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 (/3 = -.05 and /3 = -.04, respectively). Like for EE, DP was not significantly linked to alcohol consumption (/3 = .03).