Including productivity in economic evaluations of healthcare interventions is a contentious issue from both methodological and practical decision-making perspectives. Some jurisdictions, for example The Netherlands following guidance issued by The National Health Care Institute (Zorginstituut Nederland, ZIN), recommend assuming the societal perspective and encourage the impact on productivity to be included in economic evaluations [
]. In contrast, some jurisdictions actively discourage the inclusion of productivity in economic evaluations. In England, the National Institute for Health and Care Excellence (NICE) explicitly states the impact of productivity should be excluded [
]. NICE’s discouragement is based on the premise that a healthcare budget is spent with the goal of increasing health and, therefore, only those costs and consequences that directly relate to the healthcare service should be included [
Normative arguments for the inclusion or exclusion of productivity are largely focussed on the distribution of consequences across different patient groups [
]. One view is how the inclusion of productivity may favour those interventions that predominately target the working population [
]. An alternative view suggests by neglecting to take account of productivity it may lead to unfavourable decisions for those interventions that help people who are struggling to stay at work or need help returning to work [
]. Methods developed to identify, measure and value the impact of presenteeism, suitable for economic evaluations, have focussed on quantifying the impact of presenteeism using costs. This approach means that wages must be used to value presenteeism which potentially introduces discrimination associated with factors such as gender, age, ethnicity, and educational attainment. Discrimination associated with wages forms the basis of arguments by Olsen and Richardson [
] against the inclusion of productivity in economic evaluations. An alternative conceptualisation of quantifying the impact of presenteeism is to view it as a non-monetary outcome, which would be of particular relevance for specific interventions that aim to improve both health status and subsequently ability to work (productivity).
A relevant example are workplace interventions (WPI), a set of activities, programmes or equipment that aim to help people with health conditions to remain or return to work by alleviating symptoms and/or improving functional ability [
]. WPIs, unlike many clinical interventions, may be funded by employers, or a healthcare service, with the goal of improving health and avoiding lost productivity [
Productivity is made up two distinct but related concepts: absenteeism and presenteeism. Absenteeism refers to the impact on productivity caused by being
from work because of poor health [
]. Presenteeism identifies the impact on productivity while
because of poor health [
]. Evidence suggests that the impact associated with presenteeism is far greater than that caused by absenteeism [
]. However, far less evidence exists that supports if, and how, presenteeism may legitimately be incorporated, measured and valued in economic evaluations of interventions, such as WPIs [
]. The apparent confusion over which methods ought to be used to capture the impact on presenteeism may have inadvertently discouraged researchers from collecting presenteeism-related data further limiting its availability to conduct further research [
]. Brouwer et al. [
] suggested it may be possible to predict presenteeism using health status data; two studies have begun to explore this by developing prediction models using regression methods; however, results are mixed and both conclude more research is required [
Mapping techniques (cross-walking; prediction modelling), describe a set of methods that are used to generate a quantitative link between a ‘source’ and ‘target’ measure to predict unavailable outcomes using existing data [
]. Mapping has generally been used to develop algorithms that use data to link disease-specific non-preference-based measures to predict preference-based scores such as the EuroQol Five Domains (EQ-5D) [
]. Mapping, while a second-best solution, allows researchers to estimate missing data keeping the burden on costs and time to a minimum. To date, mapping methods have not been used to develop an algorithm that links health status data with presenteeism; a potential method that may prove to be useful to predict levels of presenteeism associated with specific health states captured by generic measures of health status such as the EQ-5D. A number of measures of presenteeism are available [
] an example includes the Work Productivity Activity Index (WPAI) [
], a short survey designed to ask patients about their ability to work recording both absenteeism and presenteeism. The idea would be to use health status data, captured by the EQ-5D or Short Form Six dimensions survey (SF-6D) [
], to predict levels of presenteeism measured by the WPAI. This paper reports on the findings of a study to derive a preference-based measure of health from the SF-36 for use in economic evaluation. The SF-36 was revised into a six-dimensional health state classification called the SF-6D. A sample of 249 states defined by the SF-6D has been valued by a representative sample of 611 members of the UK general population, using standard gamble. Models are estimated for predicting health state valuations for all 18,000 states defined by the SF-6D. The econometric modelling had to cope with the hierarchical nature of the data and its skewed distribution. The recommended models have produced significant coefficients for levels of the SF-6D, which are robust across model specification. However, there are concerns with some inconsistent estimates and over prediction of the value of the poorest health states. These problems must be weighed against the rich descriptive ability of the SF-6D, and the potential application of these models to existing and future SF-36 dataset.
Conceptual validity is defined as the extent to which the ‘content of two different instruments reflect one another when used for mapping’ [
]. Studies conducted to assess the conceptual validity between two measures prior to the development of a mapping algorithm or prediction model are limited meaning that some existing algorithms may produce biased estimates and lead to incorrect decisions regarding resource allocation [
]. Following Round and Hawton [
] recommendations, this study aimed to understand the conceptual validity supporting the use of measures of health status (EQ-5D; SF-6D) to predict the degree of presenteeism. The findings from this qualitative study could be used to inform the potential development of a mapping algorithm for presenteeism using generic multi-attribute measures of health.
There are a number of long-term health conditions that are known to affect the ability of people to work. Inflammatory autoimmune conditions are one example and used as the focus for this study. Rheumatoid arthritis (RA), ankylosing spondylitis (AS), and psoriatic arthritis (PsA) are three inflammatory autoimmune conditions that have been previously shown to affect ability to work [
]. RA affects women more than men [
], AS affects men more than women [
] and PSA affect men and women equally [
]. There is substantial evidence to support that these conditions are responsible for a significant reduction in productivity in Europe, second only to mental health conditions [
]. These three conditions are the most common inflammatory autoimmune conditions in the United Kingdom (UK) affecting the body’s joints, tendons, muscles, and ligaments causing pain, stiffness, and fatigue of the joints [
]. If left untreated these conditions may cause permanent damage leaving the individual disabled [
]. Typically, the age of disease onset for all three conditions occurs before the age of 65 years old (the current retirement age in the UK meaning that individuals are affected during their working lifetime [
This study provided evidence using qualitative methods to support the conceptual validity of using two measures of health status (EQ-5D and SF-6D) to capture the impact on presenteeism caused by chronic health conditions, such as RA, AS and PsA. Overall, the qualitative evidence showed the domains within both multi-attribute measures of health status conceptually overlapped with the elements of poor health that resulted in presenteeism. This support for conceptual validity of the two existing measures of health status suggested they could be used as potential source measures in a subsequent study to produce a mapping algorithm to predict presenteeism. Directly comparing the conceptual validity of the two measures indicates that the SF-6D may be a more suitable measure to predict presenteeism because it includes two relevant constructs: ‘social interaction’ and ‘vitality’ (fatigue). A qualitative study of a working sample of employees with RA, AS or PsA found fatigue potentially has the greatest impact on presenteeism [
]. However, neither of the two measures of health status captured the impact of low ‘mental clarity’ on ability to work. Mental clarity, in the exemplar conditions, was a particularly important symptom preventing individuals from being able to think quickly and/or increasing the number of mistakes they made in their work.
This is the first qualitative study that has produced empirical evidence to understand the conceptual validity of ‘source’ and ‘target’ measures, in this case EQ-5D and SF-6D with presenteeism, prior to the development of a mapping algorithm. The study was conducted based on those recommendations published by Round and Hawton [
]. The results of this study provide evidence that supports the conceptual validity of using measures of health status, as measured by the EQ-5D and SF-6D, to predict levels of presenteeism, measured using for example the WPAI. These findings provide motivation to develop a subsequent mapping algorithm, which will be the next steps.
Two published prediction models for presenteeism that use data from a measure of health status, the EQ-5D-3L, provide some evidence that a quantitative relationship linking health status with the impact on presenteeism may exist; however, the evidence was insufficient to recommend the estimated prediction model [
]. Given these existing studies, the evidence from this qualitative study exploring how measuring health status could capture the impact on presenteeism provides additional insights to understand the identified ‘weak’ quantitative relationship linking health status and presenteeism. Using qualitative methods allows some explanation and interpretation for quantitative results derived from a subsequent mapping algorithm.
The results from this study only support a positive (descriptive) stance suggesting that the SF-6D, is potentially, more likely to be a suitable measure to capture presenteeism compared with the EQ-5D. A subsequent quantitative study that produces a mapping algorithm to measure the association between measures of health status and presenteeism could be used to provide further evidence to answer a normative question. The next step is to assess the predictive ability of the EQ-5D and SF-6D to levels of presenteeism using quantitative (regression) methods.
The inductive analysis identified ‘mental clarity’ as an additional symptom of RA, AS and PsA that impacts presenteeism. This finding can be viewed in two ways. It could be argued that this limits the generalisability of the findings from this study to other conditions. An alternative interpretation is that it is known that other chronic conditions, such as chronic fatigue syndrome, fibromyalgia, migraines [
], also report difficulties with mental clarity, which would suggests the findings maybe be generalisable to other conditions. A further study in other conditions is needed to provide evidence of the relevance of mental clarity on the impact of working with other health conditions.
An opportunity was missed to collect EQ-5D, SF-6D and WPAI scores after the qualitative interview was conducted. Collecting such data may have provided some interesting insights about how individuals describe how their health affects their ability to work compared with how they might report their health status on that day or level of presenteeism over the past 7 days (WPAI). However, on balance we felt the additional cognitive burden and potential for introducing research bias by asking respondents to complete these outcome measures outweighed the potential value of collecting a small sample of responses.
The participants interviewed for this study worked in manual and non-manual jobs. A relatively small proportion of the sample interviewed worked in manual jobs, which may limit generalisability of the results to those working in these occupations. However, the results, arguably, reflect the UK, and other developed countries, where a large proportion of working people predominantly work in non-manual jobs [
]. The study was only able to recruit and interview a relatively low proportion of men, which may limit generalisability of the impact inflammatory autoimmune conditions have on men and their ability to work. However, the responses from both genders within this study were similar and there is little reason to believe that the results would have been substantially different had more men been included.
The focus of this study was to understand which of the domains within existing outcome measures of health status are most relevant to capture the impact of inflammatory arthritis on ability to work. The degree of the impact, which is likely to be associated with the severity of disease experienced by the working individual, could potentially be captured by the levels attached to each domain within the outcome measures. We did not explore this issue which could be the topic of a subsequent study that aimed to understand how disease severity influences presenteeism.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.