Article Text

Worksite mental health interventions: a systematic review of economic evaluations
1. Heleen H Hamberg-van Reenen1,
2. Karin I Proper2,
3. Matthijs van den Berg1
1. 1Centre for Public Health Forecasting, National Institute for Public Health and the Environment, Bilthoven, The Netherlands
2. 2Centre for Prevention and Health Services Research, National Institute for Public Health and the Environment, Bilthoven, The Netherlands
1. Correspondence to Heleen H Hamberg-van Reenen, Centre for Public Health Forecasting, National Institute for Public Health and the Environment, PO Box 1, Bilthoven 3720 BA, The Netherlands; heleen.hamberg{at}rivm.nl

## Abstract

Objective To give an overview of the evidence on the cost-effectiveness (CE) and financial return of worksite mental health interventions.

Methods A systematic search was conducted in relevant databases. Included economic evaluations were classified into two groups based on type of intervention: (1) aimed at prevention or treatment of mental health problems among workers or (2) aimed at return to work (RTW) for workers sick-listed from mental health problems. The quality of the included economic evaluations was assessed using the Consensus Health Economic Criteria list (CHEC-list).

Results Ten economic evaluations were included in this systematic review. All four economic evaluations on the prevention or treatment of mental health problems found a positive cost-benefit ratio, although three of these studies had low to moderate methodological quality. In five out of six economic evaluation studies on RTW interventions, no favourable CE or cost-benefit balance was found. One study of moderate methodological quality reported on a positive CE balance.

Conclusions Due to a limited number of economic evaluations on worksite mental health interventions of which a majority was lacking methodological quality or lacking evidence, only a tentative conclusion can be drawn from the results of this systematic review. Worksite interventions to prevent or treat mental health problems might be cost-effective, while those RTW interventions that included a full economic evaluation aimed at depressed employees do not seem to be cost-beneficial. More high-quality economic evaluation studies of effective worksite mental health interventions are needed to get more insight into the economic impact of worksite mental health interventions.

## Introduction

The prevalence of mental health problems is high1 and rising.2 Worldwide, 450 million people suffer from a mental health problem.3 In The Netherlands, the lifetime prevalence of mental health problems is over 40%.4 Among the working population in developed countries, the 1-year prevalence rates of mental health problems range from 10% to 18%.5 ,6 Mental health problems can affect the individual's ability to work leading to absence due to sickness, and on the longer-term work disability. In 2008, mental health problems were one of the first three leading causes of work disability,7 but it has been predicted that by 2020, mental health problems, in particular depression, will rate as the leading cause of work disability.8 Figures from The Netherlands have shown that 19% of all absence due to sickness was caused by mental health problems.9 In 2005, 38% of disability benefits in The Netherlands was paid because of mental health problems.10 Due to medical consumption and lost productivity, mental health problems are associated with high costs.11 ,12 The costs of mental health problems in developed countries have been estimated at between 3% and 4% of the Gross National Product.3

Employers may benefit from interventions aimed at promoting employees’ mental health. There are several types of mental health interventions for (sick-listed) workers, varying from group interventions, to counselling by a general practitioner (GP), mental health coach or occupational physician, to medication, to cognitive behavioural therapy among others. Mental health interventions can either target the working population not (or short-term) sick-listed due to mental health problems (ie, prevention and treatment), or the working population at long-term absence due to mental health problems (ie, return-to-work (RTW) interventions). More in detail, primary preventive interventions target the entire workforce in order to increase mental health and to prevent mental health problems; secondary preventive interventions target high-risk workers, and aim to reduce mental health problems and to prevent sick leave. Treatment interventions target the working population with mental health problems either at short-term absence or not. RTW interventions, finally, are focused on improving RTW of workers who are sick-listed due to mental health problems.

Several interventions have effectively been used in the prevention or treatment of mental health problems, for instance, medication or cognitive behavioural therapy.13 However, when focusing on worksite mental health interventions, mixed results on mental health and work productivity have been shown in several reviews.14–20 To illustrate, Ruotsalainen et al14 found small but significant effects of person-directed interventions on the reduction of stress, burnout and anxiety. However, the evidence was limited due to lack of high-quality trials. In addition, Richardson and Rothstein15 found a medium to large effect of stress management interventions on work productivity. As to RTW interventions, Van Oostrom et al16 in their systematic review could not draw a conclusion on the effectiveness of worksite mental health interventions because of lack of randomised controlled trials (RCTs).

Besides the effectiveness of worksite interventions aimed at promoting mental health, it is of interest for employers whether investment in the programme is cost-effective (ie, the effects on mental health or quality of life (QALYs; quality adjusted life-years) present good value for the money invested) or cost beneficial (ie, the financial benefits, eg, due to decreased absence rates, exceed the investment costs, in other words, the ‘return on investment’ (ROI) is favourable). There are several reviews of economic evaluations on mental health interventions in the general population,20–22 but no firm conclusions could be drawn due to limited high-quality economic evaluations. As far as we know, there is no systematic overview of economic evaluations on worksite mental health interventions. Therefore, the objective of this systematic review is to give an overview of the evidence on the cost-effectiveness (CE) and financial return of interventions aimed at preventing or treating mental health problems, or to improve RTW of workers sick-listed from mental health problems.

## Methods

### Search

We searched for economic evaluations of worksite mental health interventions in different databases, that is, Medline, Scopus, NHS-EED/HTA and PsychInfo. A search strategy was drawn up according to ‘PICO’, that is, defining search terms on Population, Intervention, Comparator/Control intervention and Outcome. The search terms were related to the population (eg, workers, working population, occupational), different types of intervention (eg, prevention, treatment, RTW) and outcome (eg, mental health, stress, CE, cost benefits,  ROI, economic evaluation). No criteria were set for the content and format of the comparator or control intervention. The search was supported by a library specialist. The complete Medline search string can be found in the Appendix. The search strings for the other databases were based on this strategy.

We selected studies in English from 1 January 2000 to 14 June 2011. Inclusion criteria were: a working population (either sick-listed or not), an intervention on mental health problems (either prevention, treatment or an RTW intervention), and representing a full economic evaluation, with an outcome on CE (ie, comparing costs and effects in mental health), cost utility (ie, comparing costs and effects in quality-adjusted life-years (QALYs)), or cost benefits (ie, comparing costs and financial benefits, which are net benefits, or ROI). Articles which reported only on outcome measures or on costs (non-economic evaluations or cost studies) were excluded. Furthermore, studies on work resumption for psychiatric hospitalised patients were excluded, as well as economic evaluations on medication as solely interventional for mental health problems. Finally, economic evaluations including persons on sick leave with subgroup analyses regarding mental health problems were excluded. Only economic evaluations focusing on mental health interventions as primary target were included.

Included economic evaluations on mental health interventions were classified into two groups based on the target population of workers whether or not on absence due to sickness: (1) prevention or treatment for non-sick-listed, or short-term sick-listed workers, with or without mental health problems; (2) RTW interventions for (long-term) workers sick-listed due to mental health problems.

The selection of the studies was done by three reviewers (HH, KP, MvdB), and was first based on title and abstract. Moreover, the reference lists of the included economic evaluations, as well as related reviews were checked, and authors of related articles were contacted to ask for information on the performance of an economic evaluation. In case of uncertainty about eligibility, the publication was discussed in a joint meeting and, if necessary, the full text was retrieved and read for eligibility.

### Data extraction

From the selected economic evaluation studies, data were extracted using a predefined form on general study characteristics, characteristics of the economic evaluation and study outcome. General characteristics included authors, publication date, country, study population, intervention and control condition.

Characteristics of the economic evaluations included the design of the study (ie, economic evaluation alongside an RCT or a modelling study), the perspective from which the cost calculations were done (ie, the society including ‘all’ costs, or the employer), the intervention period and ‘time horizon’ (follow-up period), the measurement of effectiveness or financial benefits and measurement of costs, and finally, if sensitivity analyses or uncertainty analyses are carried out (ie, analyses estimating the robustness of the outcome, or indicating the variables that are the outcomes most sensitive).

Study outcome included effectiveness or financial benefits, costs and combined costs and effects (ie, incremental CE ratio, incremental cost-utility ratio (ICUR), cost difference (net costs or benefits), cost-benefit ratio, or  ROI: percentage, and sensitivity analyses or uncertainty analyses). See Drummond et al23 for further information on economic evaluations.

Data extraction was carried out by two sets of authors independently; half the studies by HH and KP, and the remaining by HH and MvdB. Discrepancies were resolved in a consensus meeting.

### Assessment of methodological quality

The quality of the included economic evaluations was assessed using the Consensus Health Economic Criteria list (CHEC-list) (see table 1).24 This checklist was developed based on a consensus procedure between international experts. The checklist contains 19 questions on different aspects of economic evaluations, for example, study design, time horizon, perspective, type of costs and effectiveness measures that are included, the way these are measured and valued, incremental analysis of costs and outcomes, discounting, sensitivity analyses, authors’ conclusions and generalisability of study results. Each question can be answered ‘yes’ or ‘no’. If the answer is ‘yes’, this means that the study either adequately performed the item of concern or reported the item in an appropriate way. The CHEC list was filled in by two sets of authors independently, half by HH and KP, and half by HH and MvdB. Discrepancies were resolved in a consensus meeting.

Table 1

Quality scores for criteria of the CHEC list; percentage of fulfilment included economic evaluations focused on prevention or treatment and on return to work (RTW), respectively

In this review, cut-off points to categorise studies into high, moderate and low quality were chosen, indicating low, moderate and high risk of bias, respectively. Studies that fulfilled >75% of the items were classified as high quality, fulfilment between >50% and ≤75% was classified as moderate quality, and ≤50% fulfilment was classified as low quality.34 ,35

## Results

### Study selection

The database search resulted in 3886 hits of which title and abstract were screened (see figure 1); 3868 articles were excluded for several reasons, such as: no working population, no intervention on mental health problems, or no full economic evaluation. The full texts of the remaining 18 articles were studied, of which 10 articles were included in this review. Checking reference lists of the included studies and related reviews (which were found in this search based on title and abstract), as well as inquiring with authors of related articles did not yield additional articles. Finally, four economic evaluations on the prevention or treatment of mental health problems were included,6 ,25–27 and six economic evaluations of RTW interventions for workers sick-listed due to mental health problems.28–33

Figure 1

Flow chart: Inclusion of studies.

### General study characteristics

Table 2describes the general characteristics of the studies that evaluated the economic impact of interventions aimed at preventing or reducing mental health problems, or aimed at RTW for workers sick-listed due to mental health problems.

Table 2

General characteristics of the included economic evaluation studies on mental health interventions aimed at prevention, treatment, or return to work among (sick-listed) workers

#### Prevention and treatment

Of the four economic evaluations that focused on the prevention or treatment of mental health problems, interventions consisted of enhanced care,6 ,25 a recreational music-making intervention,26 and a Critical Incident Stress Management (CISM) Programme.27 One of these interventions was applied to healthy long-term care workers (ie, without mental health problems),26 the other interventions were among mixed groups of blue collar and white collar workers with depression,6 ,25 or having experienced a critical incident at air traffic controller work.27 They all had a cost-benefit design, and one study additionally performed a cost-utility analysis.6 Two cost-benefit analyses from the USA reported on net benefit, or ROI, from the employers’ perspective.6 ,25 In these studies, the employers’ perspective was defined as follows: costs included intervention costs, and benefits were related to increased work productivity. The additional CE analysis was reported from the societal perspective6 in which the intervention and healthcare costs were compared with the effect on QALY. From the other two cost-benefit analyses from Denmark27 and the USA,26 no information was given on study perspective. Two of four economic evaluations on prevention or treatment carried out a sensitivity analysis,6 ,25

The six economic evaluations on RTW interventions were in the occupational health service setting,28 ,31 ,32 in a GP setting,33 ,29 or conducted by a psychiatrist.30 One study focused on sick-listed employees from an insurance company,30 one study was aimed at sick-listed police officers,31 and four studies contained mixed groups of absent workers. The study of Leon et al30 was carried out in the USA, the remaining five economic evaluations on RTW interventions were from The Netherlands. If stated in the article, the cost-benefit analyses were performed from the employers’ perspective, whereas the CE and cost-utility analyses were done from the societal perspective. In the latter, costs included healthcare costs (including intervention costs), and effectiveness was related to psychosocial functioning or QALY. For the cost-benefit analyses that were done from the employers’ perspective, intervention costs were compared with the reduced costs due to productivity loss. Thus, healthcare costs were not included. All included economic evaluations on RTW carried out a sensitivity or uncertainty analysis.

### Quality of the studies

Table 1 shows the quality of each study represented by the scoring of the CHEC list criteria. On average, 65% of the criteria were met, that is 45% for the economic evaluations on prevention or treatment,25 ,27 and 71% for the economic evaluations on RTW interventions.28 ,30 ,31

Four studies were classified as high quality, three as moderate and three as low quality. Two of the four economic evaluations on prevention or treatment were classified as low quality, one as moderate quality and one as high quality. For the economic evaluations on RTW interventions, three of six were classified as high quality, two as moderate and one as low quality.

Thirteen out of 19 items were fulfilled by more than 50% of the articles. For the articles on prevention or treatment and RTW interventions, these are four and 14 items, respectively. The items that were fulfilled by more than 50% of the articles were in the field of an appropriate study design, time horizon, identification of relevant outcome measures and the absence of conflicting interests. The items that were fulfilled by 50% or less of the articles were the chosen perspective, the valuation of costs, measurement of outcomes, discussion on generalisability and an ethical discussion.

### Results on the economic impact of the studies

Table 3 presents the results of the economic evaluation studies on the prevention or treatment of mental health problems, or RTW interventions among workers sick-listed due to mental health problems.

Table 3

Study outcome of the included economic evaluations on mental health interventions that were focused on prevention, treatment or return-to-work for (sick-listed) workers

## Discussion

Mental health problems among workers leads to productivity losses and high costs for employers. Investing in worksite mental health interventions may thus lead to financial benefits for employers, workers and members of organised labour organisations. This systematic review summarised results on the economic impact of worksite interventions to prevent or treat mental health problems, or to improve RTW of sick-listed workers. Only 10 economic evaluations on worksite mental health interventions were found, of which the majority was of low methodological quality, or evidence on effectiveness was lacking. Therefore, no firm conclusions can be drawn from the results of this systematic review. Nonetheless, based on the results of the few existing studies including a full economic evaluation, worksite interventions to prevent or treat mental health problems seem to be cost-effective, while RTW interventions aimed at depressed employees do not seem to be cost beneficial.

### Quality of the economic evaluations

In this systematic review, three studies were considered as low methodological quality, and four studies were classified as moderate quality. Due to the associated risk of bias, the results of those studies should be interpreted with caution. With regard to the different CHEC criteria, there were some quality items that were poorly met by the studies included. None of the studies on prevention or treatment (adequately) reported on the measurements of costs in physical units; for RTW interventions, 40% scored positively on this item. Moreover, only 25% and 40%, respectively, had the costs valued appropriately, or did not report on this. As to the measurement and valuation of outcomes, about 50% met this criterion. Because the way of measurement and valuation of both costs and outcomes is essential in order to understand and interpret the results of an economic evaluation, there is a strong recommendation for future researchers to adequately measure and value costs and outcomes. Future economic evaluations should pay further attention to the description of the study population, the competing alternatives, and also the generalisability of the results and ethical issues. These items were also poorly reported in the studies under review.

### Evidence for effectiveness

Half the included studies on the economic impact of RTW interventions were classified as high quality, the other half was of low to moderate quality. However, the studies with high quality did not prove that investment in worksite mental health programmes among sick-listed workers was either cost-effective or yielded monetary benefits. This was probably due to lack of evidence on effectiveness of the RTW interventions included in the economic evalutions.28 ,29 ,33 This raises the question about the value of an economic evaluation in case of a lack of a proven intervention effect. Since the intervention costs, and even the total costs, including healthcare costs and productivity-related costs, did not differ between the study groups, the intervention will not be cost-effective. Reasons for lacking evidence may lie in the complexity of the study population, that is, workers with probable severe mental health problems, who have been sick-listed for a longer period. From previous studies, it is known that different factors predict RTW for people with mental health problems, including work-related and non-work-related factors.36–38 More insight is needed in effective elements of interventions, as well as different types of effects of worksite mental health programmes.

### Study limitations

Although this review was performed in a systematic way using common and accepted criteria, there are some limitations that should be mentioned. First, because of heterogeneity in type of mental health interventions, study population and methodology of economic evaluations, no synthesis of the data was possible. This makes it difficult to compare results of different economic evaluations and give an overall conclusion on the results. Instead, we decided to extensively report on the studies, and have summarised the evidence qualitatively.

Further, in the evaluation of the methodological quality of the studies, we applied the CHEC list. Although this checklist has been used before in systematic reviews on the economic evaluation,34 ,35 there are some issues worth discussing. First, all criteria are counted with the same weightage, while not all criteria are independent. One can imagine that some items contribute to potential bias of results more obvious than other items. However, it is difficult to find reliable weightages for each of the items. Second, although the scoring was done by two sets of reviewers independently, the scoring is quite subjective. Moreover, the scoring is dichotomous. Third, the cut-off points to categorise studies into low, moderate and high quality were arbitrary, although these cut-off points have been used in previous systematic reviews of economic evaluations studies using the CHEC list.34 ,35 However, because all studies are scored in the same way by the same reviewers, the CHEC scores and the categorising can be used appropriately to mutually compare the risk of bias of the different economic evaluations that are included in this review.

Finally, in spite of the systematic search strategy, including the search in diverse electronic databases, reference check and contact with authors, we cannot exclude the possibility of having missed some economic evaluations on mental health interventions. For example, due to limitations in the language of publication and date of publication, economic evaluations on this topic might have been missed. By our focus on economic evaluations, studies that have examined the effectiveness of worksite mental health interventions, but did not evaluate the economic impact of the intervention were excluded. Therefore, from the results of his systematic review, we cannot draw conclusions on the general effectiveness of workplace mental health interventions.

## Concluding remarks

Due to a limited number of included economic evaluations, of which the majority was of low methodological quality, or evidence on effectiveness was lacking, only a tentative conclusion can be drawn from the results of this systematic review. Worksite interventions to prevent or treat mental health problems might be cost-effective, while RTW interventions aimed at depressed employees do not seem to be cost beneficial on the basis of those studies that included a full economic evaluation. As mentioned before, these tentative conclusions have to be taken with caution. Therefore, more high-quality economic evaluations of effective worksite mental health interventions are needed to gain more insight into the financial benefits of worksite mental health interventions. Before analysing this economic impact, there should first be ample evidence for the effectiveness of worksite mental health interventions.

View Abstract

## Footnotes

• Contributors All authors contributed to the work undertaken to generate this paper and have seen and approved the final manuscript.

• Competing interests None.

• Provenance and peer review Not commissioned; externally peer reviewed.

## Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.