AMERICAN JOURNAL OF INDUSTRIAL MEDICINE 58:795–806 (2015)
A Cohort Study of Psychosocial Work Stressors on Work Ability Among Brazilian Hospital Workers rio Dias de Oliveira Latorre, Maria Carmen Martinez, PhD,1 Maria do Rosa and Frida Marina Fischer, PhD3
2 PhD,
Background Hospital work is characterized by stressors that can influence work ability. The present study aims to assess the association between psychosocial work stressors and changes in work ability in a group of Brazilian hospital employees. Methods From 1,022 workers included in a 3-year cohort started in 2009, 423 (41.4%) returned the applied questionnaires in 2012. Changes in work ability were considered as the dependent variable and the investigated psychosocial work stressors as independent variables. Logistic regression models adjusted for potential con-founders (demographic, occupational features, social support, overcommitment, and situations liable to cause pain/injury). Results High levels of exposure to psychosocial work stressors were significantly associated with decreased work ability: job strain (OR ¼ 2.81), effort-reward imbalance (OR ¼ 3.21). Conclusion Strategies to reduce psychosocial work stressors should be considered to maintain hospital employees’ work ability. Such strategies have implications for institutional and social policies and might be included in quality management programs. Am. J. Ind. Med. 58:795–806, 2015. © 2015 Wiley Periodicals, Inc. KEY WORDS: aging; health care workers; psychosocial work environment; work organization; work strain; work stress; workers’ health; work ability; workload
INTRODUCTION Occupational stress at hospitals is associated not only with biomechanical and biological risk factors, but also with psychological stressors, including the responsibility for human lives, close contact with patients’ pain and suffering, complex patterns of relationships with patients,
Samaritano Hospital of S~ao Paulo, S~ao Paulo, Brazil Epidemiology Department, School of Public Health, University of S~ao Paulo, S~ao Paulo, Brazil 3 Environmental Health Department, School of Public Health, University of S~ao Paulo, S~ao Paulo, Brazil Correspondence to: Maria Carmen Martinez, PhD, Environmental Health Department, School of Public Health, University of S~ao Paulo, Avenida Dr. Arnaldo, 715, S~ao Paulo 01246-904, SP, Brazil. E-mail: [email protected] 1 2
Accepted 28 April 2015 DOI 10.1002/ajim.22476. Published online 24 May 2015 in Wiley Online Library (wileyonlinelibrary.com).
ß 2015 Wiley Periodicals, Inc.
their relatives, coworkers and managers, and lack of recognition of the work done [Jennings, 2008; Merisalu et al., 2011]. In past decades, the demands imposed on hospital staff increased as a function of the treatment of long-term inpatients, new technologies, complex work processes, and changes in labor relations [Williams et al., 2005; Jha et al., 2009; Grebenkov et al., 2011]. Exposure to psychosocial stressors became more frequent and stronger in unstable and mentally taxing environments [Danielsson et al., 2012]. Work stress in the hospital setting is associated with negative outcomes, including premature reduction of “work ability” [Hasselhorn et al., 2005; Pranjic et al., 2006; Coomber and Barriball, 2007; Camerino et al., 2008; Jennings, 2008; Simon et al., 2008; Grebenkov et al., 2011; Merisalu et al., 2011]. Work ability is defined as the capacity to perform tasks as a function of job demands and workers’ state of health, and physical and mental competence [Tuomi et al., 2005; Ilmarinen et al., 2008]. Work ability is,
796
Martinez et al.
thus, the result of multifactorial and complex interactions, such that physical and psychosocial stressors are some of its strongest predictors [Ilmarinen et al., 2008]. Two theoretical models are mainly applied to the study of the psychosocial factors associated with work ability, i.e., the demand-control model (DCM) and the effort-reward-imbalance (ERI) model. The DCM emphasizes the role of interactions between psychological job demands and control at work (skill discretion and decision authority). According to this model, psychological job strain occurs when demands are high and control is low, while a third variable, namely, poor social support at work, might further increase the risk of job strain [Karasek et al., 1998; Karasek & Theorell, 1990]. In turn, the ERI model is based on the notion of social reciprocity and prioritizes the workers’ perception of imbalance between efforts made (demands and obligations) and rewards earned (money, recognition, and job security) as a source of stress [Siegrist, 2005, 2008]. According to this model, overcommitment might mediate that relationship; individuals overcommitted to work might underestimate demands and overestimate their coping resources, while failing to perceive work conditions characterized by nonreciprocal exchange [Siegrist, 2008]. The psychological stress that arises from imbalance between efforts and rewards and between demands and control at work has been shown to induce negative effects on work ability [Conway et al., 2008; Bethge et al., 2009, 2012; Bostr€ om et al., 2012; Prochnow et al., 2013]. In the hospital setting, psychological stressors affect employees directly involved in patient care, as well as administrative staff. Several studies have assessed the association between work ability and exposure to physical and psychosocial stressors in hospital professionals [Hasselhorn et al., 2005; Pranjic et al., 2006; Fischer et al., 2007; Camerino et al., 2008; Rotenberg et al., 2008]. However, most such studies have cross-sectional design, which does not allow establishing causal inferences, and assessed specific professional categories (e.g., nurses or doctors only). The aim of the present study was to evaluate the association between exposure to psychosocial work stressors (job strain and effort-reward imbalance) and changes in work ability in a 3-year cohort of Brazilian hospital employees.
MATERIALS AND METHODS Study Design and Population A cohort study with 3-year follow-up (2009–2012) was conducted at a private hospital in S~ao Paulo, Brazil. It is a medium-size, 300-bed hospital in which high-complexity procedures are performed. At baseline (2009) ill employees and those on maternity leave at the time of data collection (n ¼ 128) were excluded from the study. The remaining 1,395 eligible employees were invited to participate in the
study. Among those, 169 did not reply to our invitation. A total of 1,226 (87.9%) participants remained, including nursing technicians, and nursing assistants (28.6%), registered nurses (11.6%), unskilled helpers (20.6%), administrative staff (16.7%), general technicians (11.3%), cleaners (6.6%), and waiters (4.6%). From that population, 204 individuals were further excluded, because their work ability was already impaired at baseline as assessed by means of the Work Ability Index (WAI). WAI is a seven-dimension questionnaire used to evaluate work ability; its total score varies from 7 to 49 (additional information in section “Measurements” and Online Supplementary Information Appendix 1). The cutoff point to define inadequate work ability was 40 for the younger employees (19–34 years old) and 37 for the older ones (>35 years old), according to the criteria formulated by Kujala et al. [2005] and Tuomi et al. [2005] respectively. The first wave (2009) thus comprised 1,022 participants with adequate work ability. Only 41.4% (n ¼ 423) of these employees completed the questionnaire during the second wave (2012). Employees who did not participate in the second wave (n ¼ 599) were excluded from analysis. Loss to follow-up was due to dismissal/resignation (60.3%), nonresponse (35.7%), and sick leave (4.0%); no participant died along the study period. The proportion of participants lost to follow-up was higher among males compared to females (67.1% vs. 54.0%; P < 0.001), older participants (40 years old: 60.6% vs. 2 days per week). Nursing staff (registered nurses, nursing technicians, and nursing assistants) accounted for 46.6% of the study population. The average age of entry into the workforce was 16.4 years old (SD ¼ 3.2) and 16.3% of the participants had more than 14 years into the workforce. The average length of time at the present occupation was 10.7 years (SD ¼ 7.50), and 40.2% of the participants had been at the present occupation for more than 11 years. The average number of working hours per week (at the hospital, second job, and/or performing household chores) was 55.5 (SD ¼ 16.4); 40.7% of the participants reported to work more than 51.0 hours per week; 13.9% reported a second job. The nursing staff was the group with the largest proportion of participants with a second job, corresponding to 23.6% of the nursing technicians, 20.0% of the nursing assistants, and 19.7% of the registered nurses. The average level of violence at the workplace reported by the participants was low (score 7.4 on a scale ranging from 7.0 to 21.0) (data not shown). The average score on WAI decreased from 44.1 (SD ¼ 2.8) at baseline to 42.0 (SD ¼ 5.0) points at the end of the study. The incidence of impaired work ability was 18.0% (data not shown). Relevant percentages of participants reported scores that were not in the best tertile to work stressors at baseline (2009), which increased in the last assessment in 2012: high job strain, from 63.6 to 76.4% (P < 0.001); low social support from 61.5 to 76.1% (P < 0.001); high effort-reward imbalance, from 61.7 to 71.9% (P ¼ 0.001); high overcommitment, from 60.0 to 65.5% (P ¼ 0.048); and high WRAPI, from 61.2 to 67.4% (P < 0.00). A large proportion of participants assessed in 2012 reported the exposure to work stressors had “worsened/ remained poor”: job strain (20.3 worsened and 53.2% remained poor), social support (22.7% worsened and 52.7% remained poor), effort-reward imbalance (18.7% worsened and 50.6% remained poor), overcommitment (18.4% worsened and 45.9% remained poor) and WRAPI (13.9% worsened and 49.4% remained poor) (data not shown).
Factors Associated With Work Ability Table II describes the results of bivariate analyses of sociodemographic variables and health behaviors. Decrease of work ability occurred more often among the women compared to the men (P ¼ 0.042) and among the participants younger than 30 years old compared to the older ones (P ¼ 0.039). No other variable exhibited significant association with decrease of the work ability (P > 0.050).
799
Job title was the only occupational variable significantly associated with work ability decrease (P ¼ 0.021). Impairment of the work ability was more frequent among the nursing technicians, nursing assistants, and waiters compared to the remainder of the occupational categories (Table III). Reports that work stressors worsened or remained poor were associated with decreased work ability: high job strain (P ¼ 0.004), low social support (P < 0.001), high effortreward imbalance (P ¼ 0.003), high overcommitment (P ¼ 0.002) and high WRAPI (P < 0.001) (Table IV).
Psychosocial Work Stressors (Multivariate Analyses) Table V describes the results of multiple logistic regression, which show that the following psychosocial work stressors were significantly associated with impaired work ability: (a)
(b)
Job strain: The final model (model 4) showed that the odds of work ability impairment were 2.81 times higher among the participants who reported high job strain as worsened/remained poor compared to the ones who reported it as remained favorable/ improved (P ¼ 0.019) after adjustment for sociodemographic and occupational variables and WRAPI. In model 5, which was adjusted for social support, the odds for worsened/remained poor high job strain decreased to OR ¼ 2.22 and was of only borderline statistical significance (P ¼ 0.077). Effort-reward imbalance: The final model (model 3) showed that the odds of work ability impairment were 3.21 times higher among the participants who reported high effort-reward imbalance as worsened/ remained poor compared to those who reported it as remained favorable/improved (P ¼ 0.006) after adjustment for sociodemographic and occupational variables. The odds ratio decreased and was no longer statistically significant following adjustment for WRAPI and/or overcommitment (models 4, 6, and 7).
In addition, we found significant associations (adjusted for sex, age, and job title) between potential confounders and work stressors: worsened/remained poor social support X worsened/remained poor job strain (OR ¼ 3.55; P < 0.001); worsened/remained poor overcommitment X worsened/ remained poor ERI (OR ¼ 5.05; P < 0.001); worsened/ remained poor WRAPI X worsened/remained poor ERI (OR ¼ 3.12; P < 0.001); worsened/remained poor WRAPI X worsened/remained poor overcommitment (OR ¼ 31.71; P ¼ 0.020) (data not shown).
800
Martinez et al.
TABLE II. Associations Between SociodemographicVariables, Health Behaviors and Changes in Work Ability, Hospital Workers, S~ao Paulo, Brazil, 2009^2012 Remained excellent or good Characteristic* Sex Female Male Age 29.9 years old 30.0^39.9 years old 40.0^49.9 years old 50 years Marital status Single Married/living with a partner Single/divorced/widowed Child-rearing responsibility No Sporadically/yes Educational level Elementary school Secondary school Higher education Smoking Never smoked Former smoker Smoker Alcohol intake 0 to1day/week 2 days/week Body mass index Normal Overweight Obesity Practice of physical activity Yes No Total
Shifted to moderate or poor
Total
N8
%
N8
%
N8
P**
243 104
79.7 88.1
62 14
20.3 11.9
305 118
0.042
78 160 87 22
72.9 86.0 83.7 84.6
29 26 17 4
27.1 14.0 16.3 15.4
107 186 104 26
0.039
132 164 47
80.5 85.4 74.6
32 28 16
19.5 14.6 25.4
164 192 63
0.130
171 165
81.4 82.1
39 36
18.6 17.9
210 201
0.862
14 155 166
82.4 79.9 83.4
3 39 33
17.6 20.1 16.6
17 194 199
0.664
271 46 27
83.6 75.4 79.4
53 15 7
16.4 24.6 20.6
324 61 34
0.280
320 18
82.5 75.0
68 6
17.5 25.0
388 24
0.355
181 120 40
82.6 79.5 87.0
38 31 6
17.4 20.5 13.0
219 151 46
0.477
132 204 347
84.6 79.7 82.0
24 52 76
15.4 20.3 18.0
156 256 423
0.211
Differences in totals are due to missing data. * Characteristics at baseline (2009). ** 2 x test.
We also found statistically significant associations (adjusted for sex, age, and job title) between potential confounders and work ability: worsened/remained poor social support X impaired work ability (OR ¼ 4.64; P < 0.001); worsened/remained poor overcommitment/impaired work ability (OR ¼ 3.01; P ¼ 0.001); worsened/ remained poor WRAPI/impaired work ability (OR ¼ 7.61; P < 0.001) (data not shown). The models are described in full detail in “Online Supplementary Appendix”: Appendix 3–Bivariate and
multiple logistic regression for changes in psychosocial work stressors associated with changes in work ability: Table A, B, and C.
DISCUSSION Our results showed that high levels of exposure to psychosocial work stressors (job strain and effort-reward imbalance) were associated with a decrease in work ability
Psychosocial Work Stressors and Effects on Work Ability
801
TABLE III. Associations Between Occupational Variables and Changes in Work Ability, Hospital Workers, S~ao Paulo, Brazil, 2009-2012
Characteristic
*
Age of entry into the workforce 18.0 years 14.0^17.9 years
A Cohort Study of Psychosocial Work Stressors on Work Ability Among Brazilian Hospital Workers rio Dias de Oliveira Latorre, Maria Carmen Martinez, PhD,1 Maria do Rosa and Frida Marina Fischer, PhD3
2 PhD,
Background Hospital work is characterized by stressors that can influence work ability. The present study aims to assess the association between psychosocial work stressors and changes in work ability in a group of Brazilian hospital employees. Methods From 1,022 workers included in a 3-year cohort started in 2009, 423 (41.4%) returned the applied questionnaires in 2012. Changes in work ability were considered as the dependent variable and the investigated psychosocial work stressors as independent variables. Logistic regression models adjusted for potential con-founders (demographic, occupational features, social support, overcommitment, and situations liable to cause pain/injury). Results High levels of exposure to psychosocial work stressors were significantly associated with decreased work ability: job strain (OR ¼ 2.81), effort-reward imbalance (OR ¼ 3.21). Conclusion Strategies to reduce psychosocial work stressors should be considered to maintain hospital employees’ work ability. Such strategies have implications for institutional and social policies and might be included in quality management programs. Am. J. Ind. Med. 58:795–806, 2015. © 2015 Wiley Periodicals, Inc. KEY WORDS: aging; health care workers; psychosocial work environment; work organization; work strain; work stress; workers’ health; work ability; workload
INTRODUCTION Occupational stress at hospitals is associated not only with biomechanical and biological risk factors, but also with psychological stressors, including the responsibility for human lives, close contact with patients’ pain and suffering, complex patterns of relationships with patients,
Samaritano Hospital of S~ao Paulo, S~ao Paulo, Brazil Epidemiology Department, School of Public Health, University of S~ao Paulo, S~ao Paulo, Brazil 3 Environmental Health Department, School of Public Health, University of S~ao Paulo, S~ao Paulo, Brazil Correspondence to: Maria Carmen Martinez, PhD, Environmental Health Department, School of Public Health, University of S~ao Paulo, Avenida Dr. Arnaldo, 715, S~ao Paulo 01246-904, SP, Brazil. E-mail: [email protected] 1 2
Accepted 28 April 2015 DOI 10.1002/ajim.22476. Published online 24 May 2015 in Wiley Online Library (wileyonlinelibrary.com).
ß 2015 Wiley Periodicals, Inc.
their relatives, coworkers and managers, and lack of recognition of the work done [Jennings, 2008; Merisalu et al., 2011]. In past decades, the demands imposed on hospital staff increased as a function of the treatment of long-term inpatients, new technologies, complex work processes, and changes in labor relations [Williams et al., 2005; Jha et al., 2009; Grebenkov et al., 2011]. Exposure to psychosocial stressors became more frequent and stronger in unstable and mentally taxing environments [Danielsson et al., 2012]. Work stress in the hospital setting is associated with negative outcomes, including premature reduction of “work ability” [Hasselhorn et al., 2005; Pranjic et al., 2006; Coomber and Barriball, 2007; Camerino et al., 2008; Jennings, 2008; Simon et al., 2008; Grebenkov et al., 2011; Merisalu et al., 2011]. Work ability is defined as the capacity to perform tasks as a function of job demands and workers’ state of health, and physical and mental competence [Tuomi et al., 2005; Ilmarinen et al., 2008]. Work ability is,
796
Martinez et al.
thus, the result of multifactorial and complex interactions, such that physical and psychosocial stressors are some of its strongest predictors [Ilmarinen et al., 2008]. Two theoretical models are mainly applied to the study of the psychosocial factors associated with work ability, i.e., the demand-control model (DCM) and the effort-reward-imbalance (ERI) model. The DCM emphasizes the role of interactions between psychological job demands and control at work (skill discretion and decision authority). According to this model, psychological job strain occurs when demands are high and control is low, while a third variable, namely, poor social support at work, might further increase the risk of job strain [Karasek et al., 1998; Karasek & Theorell, 1990]. In turn, the ERI model is based on the notion of social reciprocity and prioritizes the workers’ perception of imbalance between efforts made (demands and obligations) and rewards earned (money, recognition, and job security) as a source of stress [Siegrist, 2005, 2008]. According to this model, overcommitment might mediate that relationship; individuals overcommitted to work might underestimate demands and overestimate their coping resources, while failing to perceive work conditions characterized by nonreciprocal exchange [Siegrist, 2008]. The psychological stress that arises from imbalance between efforts and rewards and between demands and control at work has been shown to induce negative effects on work ability [Conway et al., 2008; Bethge et al., 2009, 2012; Bostr€ om et al., 2012; Prochnow et al., 2013]. In the hospital setting, psychological stressors affect employees directly involved in patient care, as well as administrative staff. Several studies have assessed the association between work ability and exposure to physical and psychosocial stressors in hospital professionals [Hasselhorn et al., 2005; Pranjic et al., 2006; Fischer et al., 2007; Camerino et al., 2008; Rotenberg et al., 2008]. However, most such studies have cross-sectional design, which does not allow establishing causal inferences, and assessed specific professional categories (e.g., nurses or doctors only). The aim of the present study was to evaluate the association between exposure to psychosocial work stressors (job strain and effort-reward imbalance) and changes in work ability in a 3-year cohort of Brazilian hospital employees.
MATERIALS AND METHODS Study Design and Population A cohort study with 3-year follow-up (2009–2012) was conducted at a private hospital in S~ao Paulo, Brazil. It is a medium-size, 300-bed hospital in which high-complexity procedures are performed. At baseline (2009) ill employees and those on maternity leave at the time of data collection (n ¼ 128) were excluded from the study. The remaining 1,395 eligible employees were invited to participate in the
study. Among those, 169 did not reply to our invitation. A total of 1,226 (87.9%) participants remained, including nursing technicians, and nursing assistants (28.6%), registered nurses (11.6%), unskilled helpers (20.6%), administrative staff (16.7%), general technicians (11.3%), cleaners (6.6%), and waiters (4.6%). From that population, 204 individuals were further excluded, because their work ability was already impaired at baseline as assessed by means of the Work Ability Index (WAI). WAI is a seven-dimension questionnaire used to evaluate work ability; its total score varies from 7 to 49 (additional information in section “Measurements” and Online Supplementary Information Appendix 1). The cutoff point to define inadequate work ability was 40 for the younger employees (19–34 years old) and 37 for the older ones (>35 years old), according to the criteria formulated by Kujala et al. [2005] and Tuomi et al. [2005] respectively. The first wave (2009) thus comprised 1,022 participants with adequate work ability. Only 41.4% (n ¼ 423) of these employees completed the questionnaire during the second wave (2012). Employees who did not participate in the second wave (n ¼ 599) were excluded from analysis. Loss to follow-up was due to dismissal/resignation (60.3%), nonresponse (35.7%), and sick leave (4.0%); no participant died along the study period. The proportion of participants lost to follow-up was higher among males compared to females (67.1% vs. 54.0%; P < 0.001), older participants (40 years old: 60.6% vs. 2 days per week). Nursing staff (registered nurses, nursing technicians, and nursing assistants) accounted for 46.6% of the study population. The average age of entry into the workforce was 16.4 years old (SD ¼ 3.2) and 16.3% of the participants had more than 14 years into the workforce. The average length of time at the present occupation was 10.7 years (SD ¼ 7.50), and 40.2% of the participants had been at the present occupation for more than 11 years. The average number of working hours per week (at the hospital, second job, and/or performing household chores) was 55.5 (SD ¼ 16.4); 40.7% of the participants reported to work more than 51.0 hours per week; 13.9% reported a second job. The nursing staff was the group with the largest proportion of participants with a second job, corresponding to 23.6% of the nursing technicians, 20.0% of the nursing assistants, and 19.7% of the registered nurses. The average level of violence at the workplace reported by the participants was low (score 7.4 on a scale ranging from 7.0 to 21.0) (data not shown). The average score on WAI decreased from 44.1 (SD ¼ 2.8) at baseline to 42.0 (SD ¼ 5.0) points at the end of the study. The incidence of impaired work ability was 18.0% (data not shown). Relevant percentages of participants reported scores that were not in the best tertile to work stressors at baseline (2009), which increased in the last assessment in 2012: high job strain, from 63.6 to 76.4% (P < 0.001); low social support from 61.5 to 76.1% (P < 0.001); high effort-reward imbalance, from 61.7 to 71.9% (P ¼ 0.001); high overcommitment, from 60.0 to 65.5% (P ¼ 0.048); and high WRAPI, from 61.2 to 67.4% (P < 0.00). A large proportion of participants assessed in 2012 reported the exposure to work stressors had “worsened/ remained poor”: job strain (20.3 worsened and 53.2% remained poor), social support (22.7% worsened and 52.7% remained poor), effort-reward imbalance (18.7% worsened and 50.6% remained poor), overcommitment (18.4% worsened and 45.9% remained poor) and WRAPI (13.9% worsened and 49.4% remained poor) (data not shown).
Factors Associated With Work Ability Table II describes the results of bivariate analyses of sociodemographic variables and health behaviors. Decrease of work ability occurred more often among the women compared to the men (P ¼ 0.042) and among the participants younger than 30 years old compared to the older ones (P ¼ 0.039). No other variable exhibited significant association with decrease of the work ability (P > 0.050).
799
Job title was the only occupational variable significantly associated with work ability decrease (P ¼ 0.021). Impairment of the work ability was more frequent among the nursing technicians, nursing assistants, and waiters compared to the remainder of the occupational categories (Table III). Reports that work stressors worsened or remained poor were associated with decreased work ability: high job strain (P ¼ 0.004), low social support (P < 0.001), high effortreward imbalance (P ¼ 0.003), high overcommitment (P ¼ 0.002) and high WRAPI (P < 0.001) (Table IV).
Psychosocial Work Stressors (Multivariate Analyses) Table V describes the results of multiple logistic regression, which show that the following psychosocial work stressors were significantly associated with impaired work ability: (a)
(b)
Job strain: The final model (model 4) showed that the odds of work ability impairment were 2.81 times higher among the participants who reported high job strain as worsened/remained poor compared to the ones who reported it as remained favorable/ improved (P ¼ 0.019) after adjustment for sociodemographic and occupational variables and WRAPI. In model 5, which was adjusted for social support, the odds for worsened/remained poor high job strain decreased to OR ¼ 2.22 and was of only borderline statistical significance (P ¼ 0.077). Effort-reward imbalance: The final model (model 3) showed that the odds of work ability impairment were 3.21 times higher among the participants who reported high effort-reward imbalance as worsened/ remained poor compared to those who reported it as remained favorable/improved (P ¼ 0.006) after adjustment for sociodemographic and occupational variables. The odds ratio decreased and was no longer statistically significant following adjustment for WRAPI and/or overcommitment (models 4, 6, and 7).
In addition, we found significant associations (adjusted for sex, age, and job title) between potential confounders and work stressors: worsened/remained poor social support X worsened/remained poor job strain (OR ¼ 3.55; P < 0.001); worsened/remained poor overcommitment X worsened/ remained poor ERI (OR ¼ 5.05; P < 0.001); worsened/ remained poor WRAPI X worsened/remained poor ERI (OR ¼ 3.12; P < 0.001); worsened/remained poor WRAPI X worsened/remained poor overcommitment (OR ¼ 31.71; P ¼ 0.020) (data not shown).
800
Martinez et al.
TABLE II. Associations Between SociodemographicVariables, Health Behaviors and Changes in Work Ability, Hospital Workers, S~ao Paulo, Brazil, 2009^2012 Remained excellent or good Characteristic* Sex Female Male Age 29.9 years old 30.0^39.9 years old 40.0^49.9 years old 50 years Marital status Single Married/living with a partner Single/divorced/widowed Child-rearing responsibility No Sporadically/yes Educational level Elementary school Secondary school Higher education Smoking Never smoked Former smoker Smoker Alcohol intake 0 to1day/week 2 days/week Body mass index Normal Overweight Obesity Practice of physical activity Yes No Total
Shifted to moderate or poor
Total
N8
%
N8
%
N8
P**
243 104
79.7 88.1
62 14
20.3 11.9
305 118
0.042
78 160 87 22
72.9 86.0 83.7 84.6
29 26 17 4
27.1 14.0 16.3 15.4
107 186 104 26
0.039
132 164 47
80.5 85.4 74.6
32 28 16
19.5 14.6 25.4
164 192 63
0.130
171 165
81.4 82.1
39 36
18.6 17.9
210 201
0.862
14 155 166
82.4 79.9 83.4
3 39 33
17.6 20.1 16.6
17 194 199
0.664
271 46 27
83.6 75.4 79.4
53 15 7
16.4 24.6 20.6
324 61 34
0.280
320 18
82.5 75.0
68 6
17.5 25.0
388 24
0.355
181 120 40
82.6 79.5 87.0
38 31 6
17.4 20.5 13.0
219 151 46
0.477
132 204 347
84.6 79.7 82.0
24 52 76
15.4 20.3 18.0
156 256 423
0.211
Differences in totals are due to missing data. * Characteristics at baseline (2009). ** 2 x test.
We also found statistically significant associations (adjusted for sex, age, and job title) between potential confounders and work ability: worsened/remained poor social support X impaired work ability (OR ¼ 4.64; P < 0.001); worsened/remained poor overcommitment/impaired work ability (OR ¼ 3.01; P ¼ 0.001); worsened/ remained poor WRAPI/impaired work ability (OR ¼ 7.61; P < 0.001) (data not shown). The models are described in full detail in “Online Supplementary Appendix”: Appendix 3–Bivariate and
multiple logistic regression for changes in psychosocial work stressors associated with changes in work ability: Table A, B, and C.
DISCUSSION Our results showed that high levels of exposure to psychosocial work stressors (job strain and effort-reward imbalance) were associated with a decrease in work ability
Psychosocial Work Stressors and Effects on Work Ability
801
TABLE III. Associations Between Occupational Variables and Changes in Work Ability, Hospital Workers, S~ao Paulo, Brazil, 2009-2012
Characteristic
*
Age of entry into the workforce 18.0 years 14.0^17.9 years
