In this prospective study of Norwegian nurses' aides, most work factors were not predictors of the level of psychological distress. In a linear regression model of the level of psychological distress at follow-up, with baseline level of psychological distress, work factors, and background factors as independent variables, work factors explained 2 % of the variance, whereas the baseline level of psychological distress explained 34 %. The level of exposure to role conflicts, the level of exposure to threats and violence, working in apartment units for the aged, and changes in the work situation between baseline and follow-up that were reported to result in less support and encouragement were positively associated with the level of psychological distress. Working in psychiatric departments, and changes in the work situation between baseline and follow-up that gave lower work pace were negatively associated with psychological distress.
Comparisons with other studies
There was no difference between respondents working 19–36 hours per week and respondents working more than 36 hours per week with respect to the level of psychological distress 15 months later. Spurgeon et al. concluded that long working hours were a risk factor of mental health disorders, but most of the evidence in their review (up to 1997) was related to situations where working hours exceeded 50 hours per week . Later, studies of nursing personnel showed no difference in the prevalence of psychological distress between those who were working less than 35 hours and those who were working 35 hours or more , or between part-time and full-time workers .
The frequency of night shifts did not predict psychological distress in the present study. In contrast, earlier studies showed positive associations between night shift work and psychological distress in nurses [10, 25] and hospital workers . One study of registered nurses showed no association, though .
The frequency of heavy physical work tasks and the extent to which the respondents felt that their work required physical endurance did not predict psychological distress in the present study. Two earlier studies [19, 21], both with cross-sectional designs, gave contradictory results – one showed positive association  and the other showed inverse association  between physical work demands and psychological distress.
The baseline level of quantitative work demands did not predict psychological distress in the present study. On the other hand, changes in the work situation between baseline and follow-up that were reported to have given lower work pace were associated with reduced psychological distress. Many studies have shown an association between high work demands and psychological distress [3, 4, 9, 12, 19, 26, 29, 30, 32, 34], also in nurses or mixed hospital workers [12, 26, 29, 30, 32, 34], including one with prospective design . No significant association was found in a prospective study of nursing students, though .
Exposure to role conflicts at work was positively associated with the level of psychological distress 15 months later. The relationship between role conflicts at work and psychological distress has also been examined in other studies [9, 50–52]. However, none of these studies had prospective designs, none focused on nursing personnel, and the results are inconsistent. In a cross-sectional study of hospital workers, psychological distress was linked to "role difficulties", a composite measure that included role conflicts .
The level of control over work pace and the level of participation in important decisions at work did not predict psychological distress in the present study. Control at work was found to be inversely associated with psychological distress in many studies [3, 4, 12, 14, 19, 28, 29], also in two cross-sectional studies of registered nurses [28, 29] and in a prospective study of hospital workers . Control at work did not predict diagnosed depression, though .
The level of support from immediate superior did not predict psychological distress in the present study. On the other hand, changes in the work situation that were reported to have given less support and encouragement at work were associated with increased psychological distress. Social support at work has been found to be inversely associated with psychological distress in many studies [3, 4, 9, 16, 24, 25, 28, 29, 34, 53], also in prospective studies of nursing personnel [24, 25]. A prospective study of hospital workers showed no significant association, though . Some studies examined effects of social support from superiors, but with inconsistent results [9, 28, 33, 53].
Feedback about the quality of one's work did not predict psychological distress. Very few other studies, if any, have examined the relationship between feedback about quality of one's work and psychological distress.
Rewards for work well done did not predict psychological distress. Some studies – for example, a prospective study of civil servants  – have shown an association between effort-reward imbalance (high efforts combined with low rewards) and psychological distress. Effort-reward balance is not quite the same as the construct used in the present study, though.
The level of fairness in the immediate superior's leadership did not predict psychological distress. In contrast, several other studies showed inverse associations between relational justice (fairness of the supervisor) and psychological distress [11–13, 19, 22, 23], also in hospital workers [11–13], and with prospective design [12, 13]. However, these associations were relatively weak, in some studies only seen in women [11, 19], in one study only in men .
Social climate in the work unit (supportiveness, suspiciousness, and relaxedness) did not predict psychological distress in the present study. In earlier studies, psychological distress was positively associated with tense and prejudiced climate , and was negatively associated with coordinated and supportive climate in the work organisation . These were cross-sectional studies, though, and only one focused on nurses . In a prospective study of hospital workers, poor team climate predicted depression , but the measure of team climate (participation safety, support for innovation, vision, and task orientation) was very different from the measure of organisational climate used in the present study.
Bullying at work did not predict psychological distress. Other studies, however, have shown association between bullying at work and psychological distress in mixed healthcare personnel [20, 39]. One of these studies had prospective design .
Frequent exposure to threats and violence at work were positively associated with the level of psychological distress 15 months later. Exposure to threats and violence at work has earlier been linked to psychological distress in nurses or mixed healthcare workers, but only in cross-sectional studies [37, 38].
Working in apartment units for the aged was positively associated with psychological distress, whereas working in psychiatric departments was negatively associated with psychological distress. In a prospective study of nursing students, Parkes  found increased risk of psychological distress in male wards.
Explanations of the findings
As expected, exposure to threats and violence predicted psychological distress. Although most violent episodes do not result in serious physical injury, threats and violence are frightening events, and may have long-term psychological consequences, such as easily activated fear of recurring violence  and post-traumatic stress disorder.
Exposure to role conflicts predicted psychological distress. Nurses' aides interact with many people at work, including patients, patients' relatives, and professionals, who may all communicate role expectations. Role conflict occurs when role expectations are in conflict, as when the focal person receives incompatible requests from two or more people (intersender role conflict) or incompatible requests from one person (intrasender role conflict), or when there is conflict between the needs or values of the person and the expectations that this person receives from others (person-role conflict) . In the present study, the index was based on three items, each representing one of these types of role conflicts. The analyses showed that it was person-role conflicts (having to do things that one feels should be done differently) that had strongest effects.
The associations with changes in the work situation suggest that it may be possible to influence the level of distress by making changes in working conditions. The fact that variations in psychological distress between practice areas were also seen after adjustments for specific work factors suggests that there are unmeasured work factors of importance, or differential selection into jobs in different practice areas.
The social situation of singles, with more loneliness and less support, could explain the association between being single and psychological distress. The association between long-term health problems and psychological distress confirms what is well known, that health problems may be a psychological stressor.
The most striking finding, however, was the weak association with most of the work factors. For many of these work factors, the values of the coefficients suggested a direction of effects (positive or negative) in line with expectations, but the absolute values of the coefficients in the full regression model were in most cases so low, and the corresponding standard errors were so high, that the findings can hardly be said to represent trends. An exception is the association with the handling of heavy objects, which turned out to be of borderline significance. Handling of heavy material is a well-known risk factor of musculoskeletal disorders, which could, in turn, evoke psychological distress. These effects of material handling may have been underestimated in the present study because of methodological limitations (see below). The association with bullying at work was also of borderline significance, but the value of the coefficient suggested an unexpected inverse relationship, perhaps due to selection bias.
For several work factors, such as quantitative work demands, control at work, and exposure to bullying at work, the results of the present study are inconsistent with the results of earlier studies (see above). This inconsistency with earlier studies, and the fact that work factors explained such a modest part of the level of psychological distress call for an explanation. The question is whether work factors really have so little effects in nurses' aides, or whether the research methods were not able to disclose the effects.
Considering the design of the present study, with adjustments for baseline level of psychological distress, and with an observation period of only 15 months, one would not expect to find strong associations. Adjustment for baseline distress helps to determine the temporal order between exposure and outcome, but it may also, in part, adjust for effects, giving the analyses an element of over-adjustment . If work factors are relatively stable and have affected the level of distress already before baseline, further effects during the following 15 months may be so weak that they could be difficult to uncover. The fact that short-term fluctuations of symptoms between baseline and follow-up were not recorded may have given our design an indirect emphasis on long-term conditions. This may have made it even more difficult to find effects.
The study was based on a large, randomly selected, nationwide sample. The response rate at baseline was not optimal, though. We do not know whether there were differences between the eligible population and the participants, but one should take into account that selection bias due to non-response at baseline could be an important limitation in terms of generalisability. The number of dropouts between baseline and follow-up was low (20 %), but there were several differences between respondents and dropouts with respect to baseline characteristics. Hence, selection bias due to dropout may have influenced the results. A healthy worker selection, due to the fact that vulnerable or unhealthy persons may have avoided specific high-exposure jobs or changed to lower-exposure jobs before entering the study, may have resulted in underestimation of associations between work factors and psychological distress.
The SCL-5 seems to have good validity as measure of psychological distress . The questions used to assess the frequency of patient handling were found to have good validity in a British study . The questionnaire instruments that were used to measure psychological, social, and organisational work factors have been found to have good construct and predictive validity as well as good internal consistency and six-week test-retest reliability . However, the internal consistency of one of the indices (the index of control over work pace) was relatively low (0.57), and the validity of the question used to assess the frequency of handling heavy objects is unknown. Changes in the work situation between baseline and follow-up also represent an uncertainty in our assessment of the work factors. We recorded and adjusted for some types of changes in the work situation, but there may have been types of changes for which we could not control. Besides, information about changes in the work situation between baseline and follow-up was collected at follow-up, and may have been influenced by the respondents' health at this point of time. Hence, the associations between changes of work and psychological distress do not represent prospective relationships from a technical point of view.
The relative homogeneity of the participants in educational attainment and occupation, and the fact that we were able to control for a series of background factors, served to enhance the internal validity of the study. However, the results may have been influenced by background factors for which we were not able to control. Among the potential confounders are work factors other than the ones that were measured, such as predictability at work and job security. Psychological trait factors, such as neuroticism, may also have influenced the results, although much of their confounding effect was probably leveled out by the adjustments for baseline level of psychological distress.