Long-term effect of a care manager on work ability for patients with depression – the PRIM-CARE RCT

1 Introduction

The collaborative care model (CCM) is widely used in primary care for people with common mental disorders (CMD). CCM is a multicomponent pragmatic strategy to deliver and coordinate care to patients with chronic disease and is based on Wagner’s Chronic Care Model [1]. CCM engages the patient in self-care by means of organisational leader support and decision-making systems [2]. It is found to be cost-effective in the care of patients with CMD, improves mental and physical outcomes [3] and is shown to be more effective than treatment as usual [4]. Although collaborative care models have shown greater improvements in depression in medium and long-term follow-up assessments, this has not been shown in the very long-term [5].

According to WHO, more than 300 million people in the world are affected with depression [6]. The core symptoms are low mood and/or loss of interest in combination with other symptoms such as fatigue, feelings of hopelessness, sleeping problems, impaired concentration and suicidal ideation [7]. Mild to moderate depression, anxiety disorders and exhaustion syndrome (i.e. CMD) are common reasons for both sick leave and disability [8]. In 2017, Pirbaloutia [9] showed a significant relationship between depression and musculoskeletal disorder. In addition, depression also causes economic burden in terms of reduced productivity and loss of workdays [10]. However, employment can be a protective factor against chronic course of the depression [11]. When left untreated, depression affects the daily functioning with major societal consequences and decreased work ability in terms of employee performance, not being engaged at work or missing days of work [7].To increase productivity in the workforce there is a need to enhance treatment for depression and anxiety [12]. One way to reduce sick leave among depressed people is a structured telephone outreach in combination with a care management programme and medication [13].

In a recent study, the PRIM-CARE RCT, a care manager organisation was evaluated in a Swedish primary health care context [14]. The study showed that implementation of CCM with a care manager at the primary care centre (PCC) for patients with mild/moderate depression reduced depression symptoms more extensively and resulted in a faster return to work (RTW) within a 3-month perspective compared to usual care. The intervention was also cost-effective in a 6-month follow-up [15]. Knowledge is still lacking regarding the effect of a care manager organization at the PCC on work ability, job strain and reduced sick leave duration among gainfully employed primary care patients with depression when examined in a 12-month perspective.

The aim of the present study was to investigate whether care manager support leads to improved work ability, decreased job strain, and reduced sick leave duration in a 12-month perspective among primary care patients with depression.

2 Method

2.2 Outcome measures

Collection of data included age, gender, children < 18 years living at home, alcohol consumption per week (< 2 times per week/ 2–4 times per week), smoking (yes/no), educational level, occupational class, employment and country of birth. For patient prediction of return to work (RTW), we used the single question “When will you be back at work?” with the following response alternatives: “within 1–4 weeks”, “within 1–6 months”, “never”, or “do not know”.

Occupational class was classified according to the Statistics Sweden socio-economic classification system, “Socioeconomic indexation” (SEI). Self-reported job title and work tasks obtained from the questionnaire were transformed using the SEI into five categories: high-level non-manual, medium non-manual, low non-manual, skilled manual, and unskilled manual work. The five categories were further merged into three categories: 1) High white collar, 2) Middle/low white collar, 3) Blue collar/students.

The primary outcome was work ability, as assessed by the Work Ability Index (WAI) [18]. WAI has been validated and has shown to have acceptable test-retest reliability, providing support for its applicability in occupational research [19]. For the purpose of the current study, a single WAI item was used, concerning the question “current work ability compared with the life time best”, with a possible score of 0 (“completely unable to work”) to 10 (“work ability at its best” [20]. We used Ilmarinen’s [21] definition of work ability, that is, the balance between human resources and the demands of work.

Several secondary outcomes were assessed. To assess psychological demands, decision latitude and social support in the workplace, we used the Swedish Demand-Control-Support questionnaire, as it is suitable to use when measuring outcomes related to public health problems [22, 23]. The job strain model was used to analyse work aspects related to demand and control, and its reliability and validity has been confirmed [24, 25]. Quality of life was measured using the EuroQoL-5D [26]. The MADRS-S was used to measure the severity of the depression. All outcome measures were collected at baseline and at 3, 6 and 12 months.

Information concerning sick leave for the 3-, 6-, and 12-month follow-up periods was collected from the electronic patient records and from patients’ questionnaires. Return to Work (RTW) was identified as the date when sick leave was reduced from 100%to 75%or less.

The Demand–Control–Support Questionnaire [27] contains 17 items: 5 for demands, 6 for control and 6 for social support. The response alternatives for demands and control were “yes often”, “yes rather often”, “no, seldom” and “no”. Values were given to each answer alternative, and the summary scores were calculated for each index and dichotomised using the median score as a cut-off point. The demand index ranged from 5–20 and was dichotomised into low demand (5–13 score) and high demand (14–20 score). The control index ranged from 6–24 and was dichotomised into low control (6–18 score) and high control (19–24 score). The social support index regarding support intensity was based on the following response alternatives: “agree, totally,” “agree, rather well,” “do not agree particularly well,” and “do not agree at all”. The support subscale ranged from 6–24 and was dichotomised by median score into low support (6–19 score) and high support (20–24 score).

To analyse the combination of demand and control, the job strain model was used [24]. Using median values, each index was dichotomised into high and low control and high and low demand, respectively. The demand and control variables were dichotomised and combined into the job strain index as follows: low-strain jobs (low demand, high control), high-strain jobs (high demand, low control), passive jobs (low demand, low control) and active jobs (high demands, high control).

2.3 Power calculation

The power calculation was based on achieving sufficient power for the analysis of the main outcome measure of the RCT, which was the reduction of depressive symptoms. In order to achieve a power of 80%and a significance level of 10%(two-sided) and to detect an effect size of 3 units in the difference between the two groups, around 200 patients were needed in each group.

2.4 Statistical analysis

Standard methods were used for descriptive statistics. Frequencies were compared by using Chi-square and mean ranks by Mann Whitney U-test, all two-sided tests. Statistical significance was set at p < 0.05. In order to compare change for job strain from baseline to 3-, 6- and 12-month follow-up with regards to age and gender, logistic regression analysis was used. Linear regression analysis was used when testing associations between WAI and MADRS-S, EQ-5D, and sick leave, respectively, controlling for age, gender, marital status, alcohol consumption, and group (intervention/control). Data analysis was performed with statistical software SPSS version 25 (IBM Corp., Armonk, NY, USA).

3 Results

In this 12-month follow-up, only patients with gainful employment were included. In the PRIM-CARE RCT, 376 patients (192 in intervention and 184 in CAU groups) with depression were invited to participate. The study ended when the number of participants was considered to be sufficient, with 376 individuals included. In total, 269 patients had a gainful employment and met the criteria for the present study, 142 (74%) in the intervention and 127 (69%) in the CAU group. In the 12-month follow up 107 patients in the intervention group (75.4%) and 101 in the CAU group (79.5%) participated.

The demographic, lifestyle, educational, occupational and ethnic characteristics, as well as perception of job strain for intervention and control groups at baseline, are shown in Table 1. There were no statistically significant differences between the groups except for a slightly higher percentage of participants with lower education in CAU group compared to care manager group.

There was a significant difference between intervention and control groups concerning MADRS-S at 3, 6, and 12 months and for EQ-5D at 0–3 and 7–12 months (Table 2). For net sick leave there was no significant difference (Table 2) and the same could be seen for WAI at any of the measurement points during the period 0–12 months (Fig. 1).

OPEN IN VIEWER

Fig. 1

WAI mean values in intervention and control group patients with gainful employment at baseline, 3, 6, and 12 months. No statistically significant difference between intervention and control group.

Test of association between WAI and baseline, 3-, 6-, and 12-month values of depressive symptoms measured with MADRS-S was significant, controlled for age, gender, civil status, alcohol consumption, and group (intervention/control), and this was also relevant for EQ-5D. For net sick leave, the association was significant for 0–3 and for 4–6 months, but not for the 7–12 month period (Table 3). In Table 3, B represents the change in WAI that is due to 1 unit of change of MADRS-S, EQ-5D and sick leave days at respective measurement point.

There was a greater proportion of the individuals in the control group compared to individuals in the intervention group who perceived higher social support throughout 0, 6, and 12 months (Fig. 2). However, this was not significant. The proportion who perceived higher social support was reduced in both intervention and control groups over time.

Perceived intra-individual change regarding job strain during 0–3, 0–6, 0–12 months, from worse/no change to better did not show any significant difference between intervention and control groups after controlling for age and sex.

OPEN IN VIEWER

Fig. 2

Percentage of individuals with gainful employment in intervention and control group at baseline, 3, 6, and 12 months who perceived high social support. No statistically significant difference between control and intervention group.

Number and percentage of individuals on sick leave and not on sick leave, respectively, were compared within the different job strain groups regarding both intervention and control groups at baseline, 0–3 months, 0–6 and 0–12 months. There was a statistically significant difference in sick leave (yes/no) in the different job strain groups for the intervention group for all periods 0–12 months (p = 0.004), but only at baseline for the control group (p = 0.013).

The patients on 100%sick leave were asked at baseline to estimate when they would be able to go back to work (within 1–4 weeks, within 1–12 months, never, don’t know). The patient’s own prediction was compared with the actual outcome within 1–12 months (n = 107). There was a statistically significant (p = 0.047) difference concerning assumption and actual outcome for the control group, with less individuals on return to work within assumed time (68%for the control group compared to 91%in the intervention group for return within 1–12 months).

4 Discussion

In the present study, the care manager contact did not seem to have any positive effect concerning work ability and job strain among the patients despite a long-term effect on depression symptoms and quality of life. There was a concordance between the patient’s prediction concerning when return to work might occur and the actual return. Having a care manager contact makes it possible to follow up and support patients with depression and helps to provide continuity and accessibility during the depression course. However, improved work ability and job strain for patients with depression could not be shown in this 12-month follow-up study. This may be due to the fact that the care manager’s support was only provided during the first three months of care, which may have been sufficient for the achievement of an effect on work ability and job strain in the short-term, i.e. six months, but not sufficient for the achievement of a more long-term effect, i.e. one that lasted for 12 months.

The care manager support did not have any effect on patients’ reported job strain. However, job strain has to do with demand and control in the workplace, and the PRIM-CARE study does not directly intervene in the workplace context. That type of intervention seems to be necessary according to a prior study [28]. One study shows that severity of depressive symptoms can be reduced using an occupational-based intervention addressing occupational functioning and work recovery [29].

There were no differences between the groups concerning social support. However, the measure social support only describes the individual’s perceived support from colleagues and employer. Nevertheless, it is included in the care manager role to support the patient. Therefore, we believe that the person-centred support from the care manager strengthens the individual’s work ability that possibly also leads to earlier return to work. It is of a great importance for health professionals to pay extra attention to the RTW process to reduce the suffering and the economic burden for the individual [30].

The greater concordance between the predicted return to work and the actual return to work suggests that the care manager has a positive effect on how patients perceive their situation. This may imply that when the care manager, together with the patient, creates a care plan, it not only strengthens the patient’s possibility to develop a realistic view of the situation, but also enables the patient to make adequate predictions concerning RTW [31]. Several studies have highlighted the difficulties surrounding return to work. For example, Sampere et al. [32] showed that during medical visits, the patient’s expectation concerning return to work is an important prognostic factor for professionals to consider. Other factors of importance are symptom severity, having no previous absenteeism, and younger age [33]. How workers perceive their ability to successfully return to work is important knowledge for health professionals and could be measured with the RTW-Self-efficacy scale, which detects risk for late RTW [34]. Prior studies also show that there is a need for tailored interventions in the workplace without any barriers [28, 35]. Shariat [36] suggests a mixture of clinical treatments focusing on workplace interventions concentrating on the individual's physical and psychological characteristics.

The care manager at the PCC has previously been shown to have several positive effects in addition to depression reduction, as evident in the 3-month outcome concerning return to work [14] and the 6-month outcome concerning depression-free days and health-economic effects [15]. In addition to the severity of depression, it is important to pay attention to quality of life as it is an important predictor of RTW [37].

Usual primary care in Sweden has high quality, and most PCCs offer several professional competencies, although continuity of care is low compared to other Nordic countries [38]. The implementation of a care manager at the PCC for patients with depression brings a more advanced organisation and competence into practice that should primarily aim to increase quality of care for the patient with increased quality of life and patient satisfaction. There are some indications that this was the result of the care manager intervention [14].

4.1 Strengths and limitations

This was a real-world complex intervention in clinical practice that used validated instruments. The study was conducted in primary care where depression is a common diagnosis among the patients. The long follow-up time was a strength that provided important information about the patients’ wellbeing and perceived work ability in the long run. Few RCTs have been performed in primary care that have evaluated the long-term effects of collaborative care organisation with a care manager for patients with depression in relation to work ability, job strain and return to work.

The total sample in the PRIME-CARE RCT consisted of 376 patients. In this study only those patients with a gainful employment were included, i.e. 269 patients, since the main outcome was work ability. The limited number of patients might have affected the results. The majority of patients were women, which may have been a potential bias as few men participated. The study used instruments that were not translated, which excluded individuals who did not understand the Swedish language. It was also a limitation to only consider the individuals’ support at the workplace and not in other contexts.

The results are generalisable to other PCCs where a care manager function has been established.

5 Conclusions

Improved work ability and reduced job strain for patients with depression compared to care as usual could not be shown in this 12-month follow-up study in spite of the fact that the group with care manager support showed statistically significant reduction of depressive symptoms and increased quality of life compared to the control group. This indicates that the patient's initial perception of work ability is an important factor to notice early in the rehabilitation process. Provision of care manager support for three months only may not be sufficient for the achievement of a long-term effect on improved work ability and job strain compared to usual care of depression in primary care.

Ethics approval

The study was approved by the Regional Ethi-cal Review Board in Gothenburg, Sweden, Etikprövningsmyndigheten (www.epn.se) (Dnr.903-13, T975-14). Trial registration: Identifier NCOT02378272, 2 February 2015.