Elsevier

Health Policy

Volume 48, Issue 1, July 1999, Pages 13-27
Health Policy

Productivity losses without absence: measurement validation and empirical evidence

https://doi.org/10.1016/S0168-8510(99)00028-7Get rights and content

Abstract

Productivity losses without absence are scarcely discussed in the literature. In this paper, the construct validity of three different measurement instruments for productivity losses without absence is investigated. The data were collected under employees of a Dutch trade firm, not in specific patient groups. On an average day, over 7% of the respondents were working with health problems, indicating that productivity losses without absence is quite a common problem. The amount of production losses related to these health problems are relatively small. However, for specific patient groups, the costs related to these productivity losses may be substantial.

Introduction

In economic evaluations of health care interventions, indirect non-medical costs or productivity costs often play an important role when the prevalence of the illness involved concerns people with paid or unpaid work. The incorporation of these costs in economic evaluations has been much debated. Two questions, whether they should be counted (for example, Refs. [3], [9]) and if so how they should be estimated (for example, Refs. [1], [4], [5]), have been extensively discussed in the literature. Recently, the consensus for incorporating these societal costs seems to be growing (for example, Refs. [2], [4]). Attention is usually focused on costs related to absence from paid work. However, it is obvious that absence from paid work is not the only situation causing production losses related to disease. Besides impaired ability to perform unpaid work, people may be at work while not being in optimal health. Mild (chronic) diseases or the onset of acute infectious diseases are examples of situations in which people may not function to their normal ability, yet are not impaired ‘enough’ to stay at home. These productivity losses without absence are potentially important, yet almost neglected in economic evaluations of health care.

Only two attempts to estimate the costs related to this productivity drop were found in the literature. One was developed by Osterhaus et al. [6] and was later adapted by Van Roijen et al. [8], and the other was developed by Van Roijen et al. [7]. The adapted Osterhaus method (O method) and that of Van Roijen (VR method) were compared in estimating productivity costs without absence due to migraine, and showed substantially different estimates, i.e. 968 versus 277 million Dutch guilders, respectively [7]. This shows these costs can be substantial for certain diseases and that more investigation is needed on how to realistically estimate these costs.

In this paper, we compare the O method, the VR method and an alternative experimental method, the QQ method, which aims at measuring the quantity and quality of work performed on a daily basis. We investigate the construct validity of these methods using data of employees of a Dutch trade firm. Section 2 presents a brief introduction of the three methods. In Section 3, the instruments and data used in this study are discussed. Section 4 discusses the results of the three methods, while Section 5 concludes the article.

Section snippets

Brief introduction of the three measurement methods1

The adapted O method for determining the efficiency losses due to illness, as used in the Dutch Health and Labour questionnaire, consists of two questions. We have chosen to use this ‘adjusted O method’, to be able to compare the outcomes in relation to the VR method with the earlier comparisons [7], [8]. The O method asks how many days during the past 2 weeks one went to work while suffering from health problems and, on a VAS scale, the average efficiency on these days. A difficulty with this

Methods

We aimed at establishing construct validity of the three measurement methods, basing our hypothesis on the expectations derived from Section 2 and previous experience. Construct validation is tested by examining whether different methods measure the underlying quantity (production losses) in expected ways. Here, it is expected that the VR method will yield the lowest results in terms of hours lost (with a relatively high number of ‘zero’ answers), the O method would yield the highest results,

Results from the diary ‘Consequences of Illness’

Of the 543 ‘Consequences of Illness’ diaries returned, there were 57 in which it was indicated that on at least 1 day during the week, the respondent felt ill while being at work. This means 10.5% of the diaries were (in principle) useful for this study. After investigation, it was concluded that 53 diaries were consistently completed and could therefore be used in the analysis below. Analyzing this subgroup of 53 diaries, it turned out that the average age (42.2 years) was significantly higher

Discussion

In this paper, we have investigated the construct validity of three measurement methods of productivity losses without absence. Statistical analysis indicated that our hypothesis, that Qt results would on average lie between VR results and O results, could not be rejected.

Compared with the Qt method and the O method, the VR method was a consequent outlier with a relatively high amount of zero answers. There are two possible explanations for these answers: making up for lost work during regular

Acknowledgements

The authors express thanks to J.J. van Dijk, M. Knotter and L.M. Lamers for their useful comments. This study was financially supported by Glaxo Wellcome, UK.

References (9)

  • M.A. Koopmanschap et al.

    The friction cost method for measuring indirect costs of disease

    J Health Econ

    (1995)
  • W.B.F. Brouwer et al.

    Productivity costs measurement through quality of life? A response to the recommendations of the Washington Panel

    Health Econ

    (1997)
  • M.F. Drummond et al.

    Methods for economic evaluation of health care programmes

    (1997)
  • K. Gerard et al.

    QALY league tables: handle with care

    Health Econ

    (1993)
There are more references available in the full text version of this article.

Cited by (194)

View all citing articles on Scopus
View full text