Predictors of self-efficacy in women on long-term sick leave

Abstract

Self-efficacy has been shown to be related to sick leave and to be a predictor of return to work after sickness absence. The aim of this study was to investigate whether factors related to sick leave predict self-efficacy in women on long-term sick leave because of pain and/or mental illness. This cross-sectional study uses baseline data from 337 Swedish women with pain and/or mental illness. All included women took part in vocational rehabilitation. Data were collected through a sick leave register and a baseline questionnaire. General self-efficacy, sociodemographics, self-rated health, anxiety, depression, view of the future, and social support were measured and analyzed by univariate and multivariate linear regression analyses. The full multivariate linear regression model, which included mental health factors together with all measured factors, showed that anxiety and depression were the only predictive factors of lower self-efficacy (adjusted R²=0.46, P<0.001) and explained 46% of the variance in self-efficacy. The mean scores of general self-efficacy were low, especially in women born abroad, those with low motivation, those with uncertainties about returning to work, and women reporting distrust. Anxiety and depression are important factors to consider when targeting self-efficacy in vocational rehabilitation.

Introduction

The concept of self-efficacy comprises an individual’s belief in his or her own ability to perform a specific action (^{Bandura, 1997}). Self-efficacy is associated with health and related negatively to physical and psychological symptoms (^{D’Amato and Zijlstra, 2010; Roddenberry and Renk, 2010}). Individuals assess their self-efficacy on the basis of own experience of bodily and mental state, but social persuasion and past successes are also relevant and may enhance self-efficacy (^{Bandura, 1997}). Self-efficacy is related negatively to the length and frequency of sickness absence periods (^{Sommer et al., 2013}).

Decline in self-efficacy is shown to be an obstacle for returning to work after sickness absence (^{Jensen, 2013}). Self-efficacy has, moreover, been shown to be a predictor of return to work after sickness absence (^{Brouwer et al. , 2009, 2010Brouwer et al. , 2009, 2010; Huijs et al., 2012; Volker et al., 2014}).

Although sick leave is warranted by an inability to work, sick leave may in itself also contribute toward negative consequences such as depression, worsening economy, stress, risk of being trapped in a sick role, decreased opportunity and desire to join recreational and social activities, and impaired sleep and self-confidence (^{Alexanderson et al., 2003;Vingard et al., 2004}).

There is a sex difference in the utilization of sick leave. During the last 30 years, women have been on sick leave more frequently and for longer periods than men in Sweden and other Western European countries (^{Deutreux and Viksten, 2004; Angelov et al., 2011}). Among women, about 70% of the reasons for sick leave are related to chronic pain and tiredness, which, in addition to insomnia, anxiety, and depression, are common reasons for women to seek healthcare (^{Upmark, 2008}).

As self-efficacy is an important factor to address in vocational rehabilitation, there is a need to identify factors predicting self-efficacy among individuals on long-term sick leave. To our knowledge, such studies are scarce. The aim of the present study was to investigate whether factors related to sick leave, view of the future, social support, and health predict self-efficacy in women on long-term sick leave because of pain and/or mental illness.

Participants and methods

Study design

This is a cross-sectional study using register data and data through self-reported questionnaires.

Study population and procedure

This study used baseline data from participants of two randomized-controlled return to work trials conducted at a university hospital in Sweden. A total of 1305 individuals, who were expected to reach their maximum number of paid sick leave days, were identified by the Swedish Social Insurance Agency from late 2009 to late 2011. According to the inclusion and exclusion criteria, 1009 women were eligible for inclusion in the study. The inclusion criteria were as follows: women, being on sick leave for mental disorder and/or pain, and age 20–64 years. The exclusion criteria were as follows: presence of bipolar disorder type 1, schizophrenia, at current suicidal risk, ongoing substance or alcohol abuse (according to the sickness certificates), and taking part in psychotherapy or vocational rehabilitation programs.

An invitation letter was sent to the 1009 women who fulfilled the inclusion criteria. Of those invited, 422 provided informed consent to participate. The current study uses data from baseline questionnaires. Respondents who did not return the questionnaire (n=46) were excluded for medical reasons or ethics (n=21), and nonresponders to the General Self-efficacy Scale (GSE) (n=18) were excluded from the analysis, leaving 337 women in the final sample (response rate 84%) (see Fig. 1 for a flow chart of the recruitment procedure).

Fig. 1: Flow chart of the recruitment process of the study sample.

The study was carried out in accordance with the Code of Ethics of the Declaration of Helsinki and approved by the Regional Ethical Review Board of Uppsala (Dnr. 2010/088 and 2010/088/1). All participants provided written informed consent.

Measures

The data were collected through self-report questionnaires from April 2010 to January 2012.

Information on the extent of ongoing sick leave, work absence duration, diagnosis, and employment status was gathered through the Swedish Social Insurance Agency registers. A physician used the diagnosis stated on the sick leave certificates to classify participants’ main problem as psychiatric, pain related, or both. The distribution was as follows: pain (30%), psychiatric (39%), and pain and psychiatric combined (31%).

Information on age, educational level, and country of birth was collected through a study-specific questionnaire.

Self-efficacy was measured using the GSE (^{Schwarzer and Jerusalem, 1995}). GSE consists of 10 statements and is answered on a four-point Likert scale ranging from 1=‘not at all true’ to 4=‘completely true’. Means were calculated as long as no more than three statements were missing (^{Schwarzer, 2011}). The GSE is translated into Swedish and validated (^{Love et al., 2012}). The internal consistency (Cronbach’s α) was 0.94. There is no definite cut-off score for GSE, but in a general population the mean was 2.9 (^{Schwarzer, 2011; Love et al., 2012}). The distribution of self-efficacy less than 2.9 versus 2.9 or more was used to categorize women with lower versus higher level of self-efficacy. In the linear regression analyses, the full self-efficacy scale was used as an outcome variable, ranging from 10 to 40 points.

Three study-specific questions were used to investigate view of the future. The questions were answered on five-point Likert scales. Question (1) ‘I believe that my health in 6 months will be…’ with answer options from ‘very good’ to ‘very poor’; (2) ‘My health must be completely restored to return to work’; and (3) ‘I am motivated to return to work’, both with answer options ranging from ‘totally agree’ to ‘totally disagree’.

Social support was measured by study-specific questions based on the National Public Health Survey of Sweden (^{The Institute of Public Health, 2013}). Data were collected through three single questions: (1) ‘Do you have a close friend whom you can contact and talk to about anything?’ with answer options ‘yes’ or ‘no’; (2) ‘How many people are there in your surroundings that you can easily ask for things? For example, people that you know so well that you can ask for help with bringing in the post or watering the flowers?’ with answer options: ‘none’, ‘1–2’, ‘3–5’, ‘6–10’, ‘11–15’, and ‘more than 15’; (3) ‘One can trust most people?’ answered on a four-point Likert scale ranging from ‘do not agree at all’ to ‘totally agree’.

The health situation was assessed using Self-Rated Health with the single question ‘In general, how would you rate your health?’ answered on a five-point Likert scale from ‘very good’ to ‘very poor’ (^{Life and Health, 2008}). The Hospital Anxiety and Depression Scale (HADS) was used to assess anxiety and depression (^{Zigmond and Snaith, 1983}). HADS is responded to on a four-point Likert scale ranging from 0 to 3. The items were summed in two subscales with scores ranging from 0 (no distress) to 21 (maximum distress). A score of 0–7 indicates a ‘noncase’, 8–10 a ‘possible case’, and 11–21 a ‘probable case’ of anxiety and depression. HADS is translated into Swedish and validated by ^{Lundqvist et al. (1991)}. Missing values were handled by replacement of the individual’s mean scores when at least four questions for each subscale were answered.

Statistical analyses

Descriptive statistics were used for sociodemographic characteristics, one-way ANOVA was used for differences in group means, and χ² tests were used for differences in proportions. Univariate linear regression analyses were carried out to assess associations between the independent variables and self-efficacy (as a dependent variable). Multivariate linear regression analyses were carried out to test models of independent variables and their adjusted associations with self-efficacy. The multivariate models were as follows: model 1=age, country of birth, education level, employment status, time for ongoing sick leave and level of sick leave; model 2=model 1+view of the future (health in 6 months, must be restored to return to work, motivated to return to work)+social support (close friend, get support from someone, trust in people); and model 3=model 2+health (self-rated health, anxiety, and depression). Results are presented with 95% confidence intervals (CIs). A collinearity test showed no multicollinearity.

No differences were found in sociodemographic variables between responders and nonresponders other than a higher proportion of unemployment among nonresponders. There were a small number of unanswered questions in each measure, mainly about education and sick leave (internal missing values 5%). No differences were observed in self-efficacy for nonresponders of education (P<0.722) and sick leave (P<0.103) compared with responders to these questions.

Data were analyzed using the statistical package for the social sciences (SPSS, version 21; IBM Corporation, Armonk, New York, USA). The significance level was set at P value less than 0.05.

Results

Characteristics of the study sample and group differences

Demographic and clinical data are presented in Table 1. The differences in characteristics between women with higher versus lower self-efficacy were higher proportion of women born abroad (P<0.016), low motivated or uncertainties about returning to work (P<0.001), having no close friends (P<0.04), and reporting distrust in people (P<0.001) among women with lower self-efficacy. Women with lower self-efficacy reported worse self-rated health (P<0.001) and higher anxiety (P<0.001) and depression (P<0.001) compared with women with higher self-efficacy (Table 1).

Table 1: Characteristics of the study population by total, and categorized with lower versus higher level of self-efficacy

The mean scores for self-efficacy were lower in foreign-born women than in native Swedish women (1.90 vs. 2.40, F=33.4, P<0.001) and in unemployed women compared with employed women (2.17 vs. 2.37, F=6.38, P<0.012). Furthermore, the mean score for self-efficacy was significantly lower in women who did not have close friends to contact than in women who had close friends (2.14 and 2.40, respectively, F=11.8, P<0.001). These data are not shown in Table 1.

Linear regression analyses

The univariate analyses showed that self-efficacy was associated with several sociodemographic and health-related factors, view of the future, and social support. No associations were found between self-efficacy and age, educational level, and time for ongoing sick leave (Table 2).

Table 2: Linear regressions: associations between age, country of birth, education, employment, time for sick leave, level of sick leave, view of the future, social support, health and self-efficacy, B-values, and 95% confidence interval

The independent variables were also analyzed in stepwise multivariate linear regression models (Table 2). Model 1 showed that being born abroad and having full-time sick leave were predictors of lower self-efficacy. Model 2 showed that being born abroad, low valuations of expected health status in 6 months, and not having trust in people were all predictors of lower self-efficacy. The final model (model 3), which included mental health factors together with all factors in previous models, showed that anxiety (B=−0.42, 95% CI −0.60 to −0.25) and depression (B=−41, 95% CI −0.61 to −0.20) were the only predictive factors of lower self-efficacy and explained 46% of the variance in self-efficacy for the full model (adjusted R²=0.46, P<0.001) (Table 2).

Discussion

The aim of this study was to examine whether factors related to sick leave (sociodemographic characteristics, view of the future, social support, and health) predicted self-efficacy in women on long-term sick leave because of pain and/or mental illness. The main results showed that anxiety and depression were the only principal predictive factors and explained 46% of the variance in self-efficacy. Furthermore, the mean self-efficacy scores were generally low, especially for women born abroad, women who had low motivation or uncertainties about returning to work, and for women reporting distrust in people.

Several studies show that self-efficacy is a predictor of return to work after sickness absence (^{Brouwer et al. , 2009, 2010Brouwer et al. , 2009, 2010; Huijs et al., 2012; Volker et al., 2014}). However, we have not been able to find any studies reporting predicting factors for self-efficacy in the area of vocational rehabilitation. In our study, self-rated levels of anxiety and depression were the only significant predictors, which suggest that psychiatric health should be addressed in vocational rehabilitation aimed at increasing self-efficacy, at least for women on long-term sick leave because of pain and/or mental illness.

The study group reported high levels of anxiety and depression on HADS, with a mean score of 10.8 versus 9.1. This could be compared with a general Swedish population with a mean score of 4.6 for anxiety and 4.0 for depression (^{Lisspers et al., 1997}). Mental health problems such as depression are strongly associated with long-term sickness absence (^{Bültmann et al., 2006}), and the combination of physical and mental health problems is associated with longer duration until full return to work (^{Huijs et al., 2012}). Previous studies of patients with rheumatic diseases also showed associations between HADS and lower self-efficacy (^{Lowe et al., 2008; Meesters et al., 2014}). For patients with spinal cord injury, higher disease-management self-efficacy was associated with less severe depressive symptoms and lower pain interference, and explained 48% of the variance in disease-management self-efficacy (^{Pang et al., 2009}). Anxiety and depression is also correlated negatively with self-efficacy in women with breast cancer (^{Chang et al., 2014}). Furthermore, ^{Andersson et al. (2014)} showed that men and women with low self-efficacy more often reported mental illness than those with higher self-efficacy.

Self-efficacy was generally low, especially among foreign-born women, and they also reported overall poorer health than native Swedish women (^{Lytsy et al., 2015}). A literature review concludes that mental health is poorer in foreign-born populations, with a higher prevalence of depressive symptoms, and that migrants from refugee countries had a higher prevalence of anxiety than native Swedes (^{Gilliver et al., 2014}). These findings are in line with those of our study. The reason for poorer mental health in the foreign-born women may be related to migration in some way.

Women who were employed had higher self-efficacy mean scores than unemployed women. Having an employment could mean a sense of belonging to a working society and having a professional identity, which may have an impact on self-esteem. These factors could contribute toward the higher self-efficacy in employed compared with unemployed women. As work is known to have positive effects on well-being and self-esteem, these factors and a putative better health status may be related to the higher self-efficacy in women who were on part-time sick leave and worked part time in our study.

Having a close friend was associated with higher self-efficacy and women who did not have close friends had a lower self-efficacy mean score compared with women who had friends. Previous research showed a positive relationship between family support and self-efficacy in patients with chronic obstructive pulmonary disease (^{Kara Kasikci and Alberto, 2007}).

Strengths and limitations

The strengths of this study include the large study population, the use of official register data, and validated assessments of the outcome measure (self-efficacy). As this is a cross-sectional study, causal inferences are not possible, merely the study of associations between measured factors and general self-efficacy. Comparisons were made with studies measuring self-efficacy with other scales in different contexts, which could be considered a weakness (^{Pang et al., 2009; Chang et al., 2014; Meesters et al., 2014}).

Conclusion and implications

Considerable research suggests that self-efficacy is an important predictor of returning to work after sickness absence. Therefore, knowledge of factors influencing self-efficacy is important when improving vocational rehabilitation. Our findings imply that levels of anxiety and depression are associated strongly with self-efficacy. This indicates that mental health must be addressed in vocational rehabilitation and in interventions aimed at increasing self-efficacy. It is known that long-term sickness absence may have negative effects on an individual’s mental health. This underscores the need to look for the existence of psychiatric symptoms, such as anxiety and depression, in individuals taking part in vocational rehabilitation. Also, this presents the need for integration of psychiatric care in vocational rehabilitation to improve mental health, and thereby well-being and self-efficacy. Thereafter, actions can be taken to strengthen self-efficacy further if deemed necessary to support return to work.

Acknowledgements

This work was supported by REHSAM, under Grant RS2011/010, a research program financed by the Swedish Ministry of Health and Social Affairs, the Social Insurance Agency. The funding organizations played no role in the planning, execution, or analyses of the study.

Conflicts of interest

There are no conflicts of interest.