Work psychosocial aspects and psychological distress among nurses



Tânia M AraújoI; Estela AquinoII; Greice MenezesII; Cristiane Oliveira SantosII; Lia Aguiar*

INúcleo de Epidemiologia. Universidade Estadual de Feira de Santana. Feira de Santana, BA, Brazil
IINúcleo de Estudos em Gênero, Mulher e Saúde. Instituto de Saúde Coletiva Universidade Federal da Bahia. Salvador, BA, Brazil





OBJECTIVE: To evaluate the association between psychological demand and job control and demand with psychological distress among nurses.
METHODS: A cross-sectional study included 502 female nurses working in a public hospital at the city of Salvador, state of Bahia, Brazil. The Demand-Control Model proposed by Karasek to evaluate the association between job control-demand and psychological distress was adopted. The SRQ-20 was used to measure psychological distress.
RESULTS: The prevalence of psychological distress was 33.3%, ranging from 20.0% among lady nurses to 36.4%, among nurse assistants. Strong dose-response gradients were observed between demand and psychological distress and the negative association between job control and psychological distress. Prevalence of psychological distress was higher (PR=2.6; 95% CI: 1.81-3.75) among professionals in high-strain jobs (high demand, low control) when compared to professionals in low-strain jobs (low demand, high control), after adjustment by potential confounders in a logistic multiple regression model.
CONCLUSIONS: Study findings reinforce the relevance of intervening in the organizational structure in order to increase control upon job and adjust the levels of psychological demands.

Keywords: Psychosocial aspects of work. Job strain model. Psychological distress. Gender. Nursing staff hospital. Occupational psychology. Mental health. Occupational health. Cross-sectional studies.




The study of the role of demands or environmental stimuli in stress responses largely dictates investigations in occupational stress. Stress theory is based on the evaluation of how the body responds to demands of the external environment, considering that stress is produced when demands exceed one’s individual capacities to respond to stimuli. When the available response mechanisms are not effective stress lingers. This may lead to negative health effects, such as high-blood pressure, depression and anxiety.8,12 With the development of these conceptual aspects – emphasizing the idea that any factor, physical or psychological, capable of affecting hormone levels could be considered a stressor – a large body of knowledge was developed highlighting the role of demands. Such studies, however, were limited to examining individual responses and almost always lead to motivation or adaptation capacity intervention measures.12

There are extensive studies on the harmful health effects caused by high levels of demand and excessive environmental stimuli.8 Work overload, especially among female nurses, has been widely consideredone of the main occupational stress factors.6

Another psychosocial dimension to be considered in evaluating the relationship between health and work is the level of job control. In regard to the impacts of job control on the health of workers, various aspects are evaluated and structured according to the subdivisions of the different schools of thought.

The concept of control, in the psychosocial perspective, was first developed by psychologists and was deeply marked by the emphasis placed on the ability of influencing life events and self-esteem or depression effects.17 In this perspective, job control was rarely studied as a descriptive element of the work environment, i.e., as an important factor of the perception of autonomy the individual builds based on himself and his life. From the early uses of the concept to the present, many changes have taken place in discussion that has been aimed at work locations, to the essence of productive processes, emphasizing their importance and their construction and deconstruction mechanisms.

Studies on job control have picked up in pace over the last twenty years and have become closely tied to the redefinitions of work processes in the new restructuring world economy context. On the other hand, such redefinitions were also, to a certain extent, the product of research on control, health and well-being.11,12

The importance given to control and demand in the different fields of knowledge, called for a simultaneous study of control and demand. In this perspective, Karasek11 developed the Demand-Control Model (DC), which divides job characteristics into four categories, based on combination of scores for psychological demands and control: high-strain jobs (high demand and low control), active jobs (high demand and high control), passive jobs (low demand and low control) and low-strain jobs (low demand and high control). The main predictor here is that most of the adverse reactions of psychological demands, such as fatigue, anxiety, depression and physical illness, occur when there is high job demand and low job control (high strain job).

According to the Demand-Control Model, this study aims at analyzing the association between job control and psychological demands, and psychological distress among female nurses.



This cross-sectional study is part of a wider investigation on health and work of nurses in a large public hospital in Salvador, Brazil.

The eligible population was all nurses employed by and currently working in the hospital.

The nurses were interviewed individually. An interviewer administered part of a questionnaire and the nurses were invited to answer a self-administered section. The questionnaire was divided into five sections: sociodemographics, occupation, housework, and general health conditions. The self-administered section included questions on the work process, emphasizing aspects of job demand and control.

Psychological distress was analyzed according to SRQ-20 scores. The cut-off for psychological distress was seven or more affirmative answers.15

Indicators were calculated to assess the variables. This was performed by grouping variables obtained from the self-administered questionnaire scores, based on Karasek11 indicators. For the worker job control variable an indicator was constructed by summing the scores obtained in the questionnaire items related to control. The results obtained from a scale indicate the level of worker control over her job according to the following categories: low; low-intermediate; high-intermediate, and high control.

For psychological demands an indicator was constructed by summing the scores obtained in the questionnaire items on demands, according to the Karasek model. The scale then showed the level of job demand in the following quartiles: low; low-intermediate; high-intermediate, and high demand.

To establish the DC Model categories, according to Karasek,11 psychological demands and control variables were dichotomized. Therefore different cut-offs were tested for both variables. The cut-off was set by observing the association between demand-control and psychological distress at the various cut-offs selected. Since there were no relevant differences between the psychological distress prevalence obtained from testing each cut-off, it was decided to establish an optimal cut-off in order to form numerically balanced groups. For control, the cut-off was set at high control, proceeding to joining the two remaining groups (high-intermediate, low-intermediate and low), which made up two categories: high control and low control. For demand dichotomization, the cut-off was set at low-intermediate demand, and high-intermediate demand and high demand were joined. Likewise, demand was divided into two categories: high (comprising high and high-intermediate demand) and low (comprising low-intermediate and low demand).

Based on the combination between levels of demand and control, the four Karasek model categories were established: low strain, passive job, active job, and high strain job.

Nurses exposed to a combination of high demand and low control (high strain) belonged to the higher exposure group. Nurses exposed to high demand, but to high control (active job) or to low control and low demand (passive job), belonged to the intermediate exposure group. And nurses with high control and low demand (low strain) were referred to as the reference group and formed the non-exposure group.

The effects of the main association were evaluated through the following covariables: aspects of work (occupation, years of service, weekly work load, shift, department, work years at this department, social support and job reward), aspects of housework (weekly work load, housework overload, help with housework) and sociodemographics (age, education, self-reported color of skin, head of the family, and migration past). The housework overload indicator was constructed according to procedures by Aquino,1 and corresponded to the total amount of time spent in four basic domestic activities (washing, ironing, cleaning and cooking – these activities were measured with a 0-5 scale) weighed by the number of people in the household, excluding the participant. Initially, for constructing this indicator, an ordinal variable was created and included three categories: low or null, intermediate and high. To dichotomize this variable the cut-off was set in the second tertile, proceeding to grouping low and intermediate overload, thus obtaining two categories: high and low housework overload.

Associations between demand- control and psychological distress were analyzed by prevalence rations and the respective 95% confidence intervals, according to the Taylor series model.

To evaluate effect modification or confounding at first a stratified analysis was carried out. The simultaneous effect of variables was determined by logistic multiple regression analysis, according to Hosmer & Lemeshow.10 All independent variables were treated as categorical variables. All the above-mentioned covariables were included in the selection stage of the logistic multiple regression analysis variables.

In stratified analysis and in logistic regression it was analyzed the interaction with covariables related to work, housework and sociodemographics.

To preselect the covariables that would be later included in the model it was considered covariables that had a p-value, obtained by likelihood ratio test, under or equal to 0.25, in univariate logistic regression analysis in which only the constant and one variable at a time were included in the model.10

In the logistic multiple regression analysis, it was adjusted the model including the main independent variable (combinations of demand and control from the Karasek model), and the covariables and pre-selected interaction terms by using the “back-to-front method”, reevaluating each stage, in order to select the variables. The test score was used to determine whether a variable excluded from the model should be re-included.

Considering that the estimated psychological distress prevalence in the target population was high, moving away from the estimated parameters for odds ratio (OR), prevalence ratio (PR) and the respective confidence intervals were estimated, according to the Delta Method.18



The study base consisted of 522 eligible workers with a very small non-participation rate of 3.9%. Therefore, the total number of participants was 502 women.

Most workers (73.2%) aged 30 to 49 years and were nurse assistants (81.1%). In respect to years of services, more than 64% of the participants reported more than 10 years in the current occupation.

It was found that 53.9% of the population had another paid job. In addition to double shift, another indicator of high demand for nursing professionals was the weekly work load. Workers reported an average workload of 45.7 (±19.5) hours per week.

Characterization of demand-control categories

In the high strain category there were a higher number of workers in the youngest age group under 35 (44.0%). In the remaining groups the distribution among age groups was similar (Table 1).

The variation between groups was not relevant in respect to the remaining characteristics, except for education. In the passive job category only 13.9% had university education; whereas among workers in active jobs it was 30.1%.

Considering work characteristics, there were significant differences between demand-control categories in regard to occupation, social support, job reward, and leisure time (Table 2).

In respect to occupation, the greatest differences were seen in passive and active jobs. In the latter, worker participation was considerably higher (29.9%), whereas in the first it was noticeably lower (12.4%).

The percentages of workers with low social support was very high (73.3%) in the high strain quadrant, in contrast to those found in low strain and active job workers, for whom this situation was the opposite and high support prevailed (69.2% and 68.5%, respectively).

There were high percentages of workers in high strain (47.0%) and passive jobs (48.2%) in the departments of emergency, surgery room and supply and equipment division. Among low strain workers, there was a lower percentage in these departments, 30.5%.

It was observed that high strain workers had a weekly work load of over 45 hours (57.5%), not enough leisure time (82.7%), and unrewarding job (59.7%). On the other hand, low strain workers had a weekly work load of less than 44 hours (54.2%), enough leisure time (52.5%), and job reward (70.1%).

In respect to work shift, only 20.4% of the workers were engaged in active night shift jobs.

Considering housework characteristics, high housework overload was reported by the majority in all quadrants in the DC Model. Noticeably high proportions were reported in high strain and passive job (70.1%).

Psychological distress patterns

The overall psychological distress prevalence was 33.3%.

The results indicate positive and statistically significant association at a p-value of 5% between work psychological demand and psychological distress (Table 3). The magnitude of the psychological distress variation was significant between low demand strata (18.9%) and high demand strata (56.1%). There was an almost three-fold difference between the latter and the first. There is a clear statistically significant dose-response gradient between demand and psychological distress, according to the Mantel-Haenszel test (c2=41,669; p£0.0001).

Job control was positively and statistically associated (p<0.05) to psychological distress. There is a statistically significant positive linear association with a dose-response gradient according to the Mantel-Haenszel test (c2=29.51; p£0.0001). Low control strata had the highest prevalence, 62.5%, and high control strata, the lowest, 23.9%.

Considering control and demand simultaneously, the low strain group showed lower psychological distress prevalence (16.9%) and the high strain group had higher psychological distress prevalence (57.5%); active job and passive job reported intermediate prevalence (Table 3). The differences found in the non-exposed low strain group were statistically significant, except when comparing passive job and low strain job.

The logistic regression model selection does not include any interaction term (Table 4). In addition to the main independent variable (demand-control), the following variables were included in the final model:

a) work: social support, job reward, and leisure time;

b) housework: housework overload;

c) sociodemographics: marital status and education.

There was an association between demand-control and psychological distress groups, except to passive job, even after adjustment for all confounder covariables. Logistic regression analysis indicated that workers exposed to both factors (high demand, low control) had higher psychological distress prevalence (2.60 times higher) when compared to workers not exposed to the same factors. Comparing the intermediate levels of exposure (active and passive job) and separately analyzing exposure to only one factor (low control or high demand) it was found that exposure to high demand had a higher prevalence than exposure to low control (Table 5).



Study limitations

Since cross-sectional studies provide a snapshot at one point in time or of a short period of time of the relation studied, they have the disadvantage of identifying only the survivors of the effect (prevalence bias) and their condition in the current exposure. This limitation is especially relevant in occupational studies because of the so-called healthy worker effect.

This study comprised all working nurses and reported a low non-participation rate (3.9%). Therefore, systematic errors in selecting participants seem minor. Besides, other aspects back the non-exclusion due to illness. The high psychological distress prevalence found, compared to findings in other occupational categories, may indicate that recruiting procedures and systematic evaluation, leading to lay-offs and leaves, may not have taken place; even if it is not possible to eliminate the hypothesis that recruitment processes, although less systematical and informal, may have played a significant role.

Since the study was carried out in a public hospital, where most workers enjoy job security and lay-offs are almost zero, a possible survival effect is minimal.

Although high overall prevalence and job security are not able to completely eliminate biases caused by the healthy worker effect, possible consequences do not significantly compromise the study findings because, if there was such bias, it would most probably tend to support the null hypothesis.

In a cross-sectional study, there are obvious difficulties in assessing whether psychologically distressed workers were selected for high strain jobs. However, it seems unlikely that relatively transitory psychological disorders, such as the ones identified by SRQ-20, would lead to recruitment for positions involving low control and high demand jobs. It is also unlikely that such disorders would lead to objective work conditions deterioration.

In respect to the evaluation of the assessment bias, it is worth noting that the SRQ-20 is considered a more specific rather than sensitive instrument.5,20 Therefore, potential distortions seem to indicate that the prevalence has been underestimated, with a higher proportion of misclassification of false negatives. On the other hand, if this was really a problem, it probably affected all groups alike and did not cause distortions in the association measures. Therefore, this would be a non-differential bias.

Most of the potential confounders according to the literature were evaluated by stratified and multivariate analysis.

The overall psychological distress prevalence (33.3%) indicates a serious mental health problem among the target population. The study prevalence is higher than the prevalence reported in similar studies of hospital workers. Pitta19 reported 20.8% prevalence, and Rego21 reported 20% overall prevalence.

The psychosocial dimensions of work proved to be relevant to the occurrence of psychological distress. High strain is an important psychological distress predictor. Workers who were simultaneously exposed to both factors had higher psychological distress prevalence than those not exposed to both or exposed to only one of the factors (high demand or low control).

The Karasek12 model argues that working in low control and high demand (high strain) situations is harmful to the workers’ health and that they are predictors of most adverse reactions resulting from work. This study has corroborated the positive association between high strain job and psychological distress. Other studies that have analyzed the association between high demand and low control at work and psychological symptoms2,3,13,16 support this study findings. Moreover this study is also consistent with other investigations that report the effect of high psychological demands on psychological distress and depressive symptoms.4

Considering the intermediate levels of exposure, it was observed higher psychological distress prevalence in active job workers rather than in passive job workers. High control seems not to have reduced the negative effects of high demand on mental health. On the other hand, low demand seems to have mitigated the negative effects caused by low control. Job demand in the target population was more closely associated to worker mental health rather than job control.

Landsbergis’13 studies on hospital workers in the US, and Bourbonnais’2 studies on nurses in Canada had similar findings. Both studies reported a greater burnout rate in active rather than in passive jobs. In addition to these studies, laboratory experiments carried out by Honig & Staddon (1977 apud Landsbergis, 1988)13 indicated that high demands are capable of blocking high control reinforcement, considering that high demands increase the response effort.

In this study, occupation-related factors may explain these findings to a certain extent. There were a greater number of nurses in the active job quadrant, whereas in the passive job quadrant the number of nurse assistants was greater. There was an evident difference in the roles of nurses and assistants in these categories.

The role of nurses in hospitals is considered highly stressful. Their work duties, despite their autonomy, bear the convergence of multiple tensions. According to Gray-Toft & Anderson,9 nurses tend to report higher occupational stress than nurse assistants and nursing aids.

Nurses were responsible for nursing personnel management and held higher positions within the institution, besides having more job autonomy when compared to nurse assistants. Their positions also included difficult responsibilities: they were responsible for nursing management, including quality and productivity. Because they had control over the work of other workers, they were also responsible for managing and solving conflicts and dissatisfactions, thus taking on control and exerting disciplinary power. On the other hand, they also had to deal with conflicts caused by two different working logics: healing and caring. Nursing is governed by the logic of caring but it is a subordinate of the healing logic that governs medicine. In this clash, it is the nurse who more directly takes over confrontations with the medical power; shaping up the link between the medical team and the nursing staff. She is responsible for the latter but a subordinate to the first. She is a target of tensions, conflicts and clashes of both sides.

Furthermore, it must be considered that control indicators were more directly designed for control analysis during work duties, and did not measure aspects of control directed at broader definitions of the organizational structure, which involves power relations. To this end, the limits of the control exercised in the performance of work duties are more evident to nurses, seeing that conflict situations with medical knowledge attest to their limited power within the hospital. Therefore, high control may lose, to a certain extent, the ability to mitigate the effects of high demand.

There was no significant difference in the psychological distress prevalence in the passive and low strain job categories. The hypothesis that the passive job quadrant would reveal intermediate psychological distress prevalence higher in low strain and lower in high strain jobs was only partially confirmed.

The hypothesis that social support and housework overload are effect modifiers of the association between demand and control and psychological distress was not supported. These variables are statistically associated to psychological distress, but did not alter the association studied. In regard to social support, similar results were obtained for nursing personnel and workers trained in Quebec.2 Most studies analyzing social support and mental health investigated the direct effects on mental health, however, only a few focused on whether social support modified the effects of the combination of high psychological demand and low control on mental health.3,12

Only few studies encompass housework overload when addressing female work.1,14 This aspect has been given a lower priority in studies related to work and stress, including investigations based on the Karasek model. The inclusion of housework overload in the final analysis model reinforces the need of including the analysis of the total workload in studies on health and occupation, taking into consideration both demands of work and home.

Ignoring the interaction between housework overload and the DC Model categories can also affect the demand and control indicators. One must consider possible misuse of indicators to evaluate the effects of work on male health, like the ones proposed by Karasek,12 in women studies. For instance, being monitored may be not as relevant to women as it is to men. Other aspects of work, such as being able to communicate with coworkers and emotional relationships, may be more important to women than to men, and these aspects were not considered by the indicators were used. According to Frankenhaueser,7 women in the work environment consider psychosocial elements such as support and communication very important. Since the indicators used to analyze demand and control did not include these aspects, it is possible that this study has underestimated both of these dimensions of work.

It can be concluded that although the DC model is extremely useful for studying occupational stress and mental health, it needs to be reformulated to include, within the operational proposal of demand and control constructs, aspects of power relations that structure and shape work processes and the specific characteristics of unqualified work. This has been reinforced by empirical evidence. Concerning that, it would be appropriate to construct a specific version of the model addressing control and demand measures to be used in human services organizations17 and to include more detailed items in contrast to the current general ones.



To Daniela Lopes and João Ricardo Lopes of Medical School of Federal University of Bahia, for their help in analyzing data.



1. Aquino EML. Gênero, trabalho e hipertensão arterial: um estudo de trabalhadoras de enfermagem em Salvador, Bahia [tese de doutorado]. Salvador: Instituto de Saúde Coletiva da UFBA; 1996.        

2. Bourbonnais R, Commeau M, Vézina M, Dion Gl. Job strain, psychological distress, and burnout in nurses. Am J Ind Med 1998;34:20-8.        

3. Der Doef M, Maes S. The job demand-control (-support) model and psychological well-being: a review of 20 years of empirical research. Work Stress 1999;13:87-114.        

4. Estryn-Behar M, Kaminski M, Peigne E, Bonnet M, Vaichere E, Gozlan C et al. Stress at work and mental health status among female hospital workers. Brit J Ind Med 1990;47:20-8.        

5. Fernandes SRP, Almeida Filho N. Validação do SRQ-20 em amostra de trabalhadores de informática. Rev Bras Saúde Ocup 1998;89/90:105-12.        

6. Fong CM. A longitudinal study of the relationships between overload, social support, and burnout among nursing educators. J Nurs Educ 1993;32:24-9.        

7. Frankenhaeuser M. The psychophysiology of sex differences as relate to occupational status. In: Frankenhaeuser M, Lundberg U, Chesney M. Women, work and stress: stress and opportunities. New York: Plenum Press; 1991. p. 39-61.        

8. Frankenhaeuser M, Gardell B. Overload and underload in working: outline a multidisciplinary approach. J Hum Stress 1976;2:35-46.        

9. Gray-Toft P, Anderson JG. Stress among hospital nursing staff: Its causes and effects. Soc Sci Med 1981;639-47.        

10. Hosmer DW, Lemeshow S. Applied logistic regression. New York: John Wiley & Sons; 1989.        

11. Karasek R. Job Demand, job decision latitude, and mental strain: implications for job redesign. Admin Sci Quar 1979;24:285-308.        

12. Karasek RA, Theörell T. Healthy work-stress, productivity, and the reconstruction of working life. New York: Basic Books; 1990.        

13. Landsbergis P. Occupational stress among health care workers: a test of job demands-control model. J Organ Behav 1988;9:217-39.        

14. Lennon MC. Women, work and well-being: the importance of work conditions. J Health Soc Behav 1994;35:235-47.        

15. Mari JJ, Williams P. A validity study of a Psychiatric Screening Questionnaire (SRQ-20) in primary care in city of São Paulo. Brit J Psychiatry 1986;148:23-6.        

16. Mausner-Dorsch H, Eaton WW. Psychosocial environment and depression; epidemiologic assessment of demand-control model. Am J Public Health 2000;90:1765-70.        

17. Muntaner C, O'Campo PJ. A critical appraisal of the demand/control model of the psychosocial work environment: epistemological, social, behavioral and class considerations. Soc Sci Med 1993;36:1509-17.        

18. Oliveira NF, Santana VS, Lopes AA. Razões de proporções e uso do método delta para intervalos de confiança em regressão logística. Rev Saúde Pública 1997;31:90-9.        

19. Pitta A. Hospital, dor e morte como ofício. São Paulo: Hucitec; 1990.        

20. Palacios M, Jardim S, Ramos A, Silva Filho JF. Validação do Self Report Questionnaire (SRQ-20) numa população de trabalhadores de um banco estatal no Rio de Janeiro, Brasil. In: Silva Filho JF. A danação do trabalho: organização do trabalho e sofrimento psíquico. Rio de Janeiro: Te Corá; 1998. p. 225-41.        

21. Rego MPCMA. Trabalho hospitalar e saúde mental: o caso de um hospital público no município do Rio de Janeiro [dissertação de mestrado]. Rio de Janeiro: Instituto de Medicina Social da UERJ; 1992.        



Correspondence to
Tânia M. Araújo
Rua Cláudio Manoel da Costa, 74 apto 1401 Canela
40110-180 Salvador, BA, Brazil
E-mail: araujotania@hotmail.com

Received on 23/1/2002.
Reviewed on 10/3/2003.
Approved on 4/4/2003.
Suported by Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq - Process n. 823765-88)



Based on the doctoral thesis presented to the Instituto de Saúde Coletiva da Universidade Federal da Bahia, in 1999.
This research was carried out in the Núcleo de Estudos em Gênero, Mulher e Saúde do Instituto de Saúde Coletiva da Universidade Federal da Bahia.
* Medical student of Medical School of the Faculdade de Medicina da Universidade Federal da Bahia. Salvador, BA, Brasil.

Faculdade de Saúde Pública da Universidade de São Paulo São Paulo - SP - Brazil
E-mail: revsp@org.usp.br