Diabetes mellitus is a major chronic disease, which is connected to direct and indirect costs and productivity losses. However, its effects on labour market participation are not straightforward to identify, nor are they consistently included in cost-of-illness studies. First, this study aims to synthesise existing evidence regarding the impact of diabetes on labour market outcomes that imply a complete absence of work. Second, the analysis takes a particular look at relevant methodological choices and the resulting quality of the studies included.

Methods

We conducted a systematic literature research (PubMed, Embase, PsychINFO), by applying a standard screening, selection and results extraction process, which considered all types of studies including cross-sectional and longitudinal approaches. Risk-of-bias and quality within the studies were assessed and results were compared. We dedicated special attention to the modelling of potential reverse causality between diabetes and labour market outcomes and the consideration of comorbidities and complications.

Results

Overall, 30 studies satisfied our inclusion criteria. We identified four main labour participation outcomes: absence of employment, unemployment, early retirement, and disability pension. The studies reviewed show a negative impact of diabetes on the labour market participation outcomes considered. Disability and Workability with Diabetes Essay. However, only a few studies controlled for endogeneity, differentiated between type 1 and type 2 diabetes or modelled the impact of comorbidities. We report how modelling choices affect the directions and interpretations of the effects.

Conclusions

The available evidence mainly suggests a negative impact of diabetes on several outcomes indicating labour market participation. The methodological limitations identified can guide future research with respect to both outcomes and methods. This study provides therefore an empirical contribution to the discussion on how to model the economic impact of diabetes.

Diabetes mellitus is a major chronic disease with increasing public health relevance in high-, low- and middle-income countries. According to recent estimates, the number of people suffering from this condition worldwide will rise from 425 million in 2017 to 629 million by 2045 [1]. The progressing prevalence of this illness is especially due to type 2 diabetes, which constitutes 90–95% of diabetes cases, and the increasing average age of populations [2,3,4,5]. Due to this increase, total health care expenditures resulting from diabetes mellitus are estimated to rise from $727 billion in 2017 to $776 billion in 2045 [1]. Disability and Workability with Diabetes Essay.

Type 2 diabetes is closely linked to environmental and lifestyle risk factors, such as unhealthy diet, smoking and physical inactivity. Furthermore, the management of both type 1 and type 2 requires a high level of patient awareness and self-management [1]. For these reasons, many countries have established prevention and disease management programs to reduce incidence rates and to help affected people coping with the illness [6,7,8]. If poorly managed, both types of diabetes could lead to severe medical complications, which can affect an individual’s ability to work and may lead to lower productivity at work (presenteeism) or missing workdays (absenteeism) [9].

Existing systematic reviews suggest a clear effect of diabetes on economic costs [10, 11], work ability, work functioning, macroeconomic productivity and socio-economic consequences [9,10,11]. Despite this evidence, most cost-of-illness studies base their calculations of indirect costs on productivity losses due to short or long term morbidity (absenteeism, presenteeism and disability pension) and mortality [12]. However, as suggested by the American Diabetes Association [13], considering only these factors might result in a rather conservative approach, since individuals with diabetes might have lower workforce participation rates than the overall population, which would not be adequately captured simply accounting for such short and long term productivity losses. Although the underestimation caused by this flaw could be mitigated by adopting a friction cost approach, the effect remains of key importance in the correct computation of individual and general societal costs due to diabetes.

However, understanding and empirically estimating the effects of diabetes on workforce participation is not straightforward. The correct empirical strategy to examine the relationship between diabetes and workforce participation requires careful consideration of potential confounding, of reverse causality between the illness and workforce participation – otherwise termed “endogeneity” -, of different types of diabetes mellitus and of its associated complications.

Given the growing importance of diabetes, the complex assessment of its productivity losses, and the potential heterogeneity in the applied econometric methods to address this question, a careful pooling and critical assessment of the existing evidence regarding the impact of diabetes on labour market participation is needed. Therefore, the aim of the present review is twofold: First, we gather all existing evidence regarding the relation between diabetes and workforce participation outcomes (employment/unemployment, early retirement, and permanent disability pension). Second, we distinguish and assess methodological characteristics in existing studies. Disability and Workability with Diabetes Essay. Hence, this review contributes to the discussion on the appropriate modelling of diabetes impact, provides methodological guidance for future studies and, therefore, fosters informed decisions in health policy and research.

Search strategy

The review was conducted and reported following the PRISMA guidelines [14]. We applied a structured approach, combining keywords and Medical Subject Headings (MeSH®) or Embase Subject Headings (Emtree®) on diabetes and labour market outcomes. The full set of the search terms for one database is represented in detail in Additional file 1. We applied the search on three databases: PubMed, Embase and PsychINFO. All databases were accessed using our institutional login. Additionally, at the end of the selection process, eligible studies, but also economic modelling studies focusing on the impact of diabetes on the selected outcomes, were screened for references.

Inclusion and exclusion criteria

Included original studies had to be published in a peer-reviewed journal between 1st January 2000 and 28th March 2017 in any language and had to focus on the general population of adults aged 18–64. Papers focusing on women or specific ethnic groups were also considered eligible whereas studies, which only aim at specific subpopulations of patients suffering from other diseases were excluded. All articles screened by abstract had an English version of the abstract available, and for none of the eligible studies the use of a translator was necessary.

We focused on studies which evaluated the impact of diabetes or its biomarkers, such as hyperglycaemia or haemoglobin A1c (glycosylated haemoglobin) higher than 6.5% [15], on labour market outcomes indicating the complete absence of an occupation, i.e. employment, unemployment, early retirement or reception of a permanent disability pension, but not mortality or other measures of productivity covered in other reviews [9, 11]. In addition, studies were considered not eligible if diabetes appeared as a cluster of several conditions (e.g. metabolic disorders, cardiovascular risk factors) or if the outcome of interest could not be distinguished from other outcomes. Disability and Workability with Diabetes Essay.

We included both cross-sectional and longitudinal studies with the primary aim of estimating the impact of diabetes on the selected outcomes, while economic modelling studies (cost-of-illness studies and simulations) were not included.

Study selection process

After pooling the results in EndNote (Version X7) and eliminating duplicates, two authors (SP, KEF) carried out an independent three-step successive screening process of the articles regarding titles, abstracts and full-texts, by considering the predefined inclusion criteria described above. Disagreements were first discussed between the two authors and afterwards with the other authors (LS, ML).

Data extraction and synthesis

After the identification of all eligible studies, we collected the results by using a predefined extraction form based on the well-established Cochrane Consumers and Communication Review Group data template [16]. From each paper we extracted a standardised set of information including the general characteristics of the study, the data source and the study population, the outcome measure considered and its definition, the analysis method used, the type of results reported, and finally the magnitude of corresponding effects. For those studies, which take endogeneity into account, we also added the necessary information allowing the evaluation of their methodological rigor. In all cases, missing information was retrieved by consulting survey web pages, reading explanatory publications or contacting authors. Furthermore, we grouped the studies in four different outcome categories. Studies which analysed the impact of diabetes on a binary variable indicating the presence of an occupation were grouped under the term “employment”. Disability and Workability with Diabetes Essay. Other studies considered a binary variable indicating the absence of an occupation or the status “unemployment”, i.e. currently not employed but actively looking for an occupation. Furthermore, we identified two other clusters, i.e. studies which focused on “early retirement” and studies which focused on the full receipt of a permanent “disability pension”.

ORDER A PLAGIARISM -FREE PAPER NOW

Quality appraisal

We assessed the quality and the risk of bias of each eligible study based on the Newcastle-Ottawa Scale [17]. Two authors (SP and KEF) assessed each study independently and discussed disagreements with the other two authors. The scale entails three domains (selection, comparability, and exposure) with several sub-questions, focusing on representativeness of the dataset, measurement of exposure/outcome, and control variables included. Since the original scale is only available for cohort and case-control studies, we based our quality analysis on a modified version of the scale [9, 10]. Cross-sectional studies could be awarded a maximum of 6 points, while longitudinal studies had a maximum of 8 (see Additional file 2 for further details).

Due to the high heterogeneity of outcomes, we limited our analysis to a comparison of results based on their direction and level of significance. Furthermore, we focused our qualitative synthesis on methodological differences and how they influenced results in the studies. Finally, as a robustness check we focused our qualitative synthesis on studies which were awarded more than half of the maximum quality score indicating a low risk of bias. Disability and Workability with Diabetes Essay.