Examining productivity losses associated with health related quality of life using patient and general population data

Economic evaluation is used to inform decisions related to setting priorities in health care and whether health care interventions should be reimbursed by combining information on costs with benefits. The key focus when assessing the benefits from health care interventions is health which may be assessed in natural units such as life years saved or quality adjusted life years (QALY) for use in cost effectiveness and cost utility analysis respectively. QALYs focus on the health related quality of life (HRQoL) associated with different health states which are valued by members of the general public. Costs are the direct costs of providing health care but indirect costs (not funded directly through the health care system), which result from having poor health may also be included.

Indirect costs include productivity losses which refer to costs associated with time off paid and unpaid work due to illness. Measuring the productivity losses associated with specific health conditions has typically focused on self-reported or objective data on time off work (Zhang et al.2011). A different approach relies on estimating productivity directly from the health states. This allows productivity losses associated with different health states to be predicted where this information is not available. In order to assess the relationship between productivity and health, patient datasets containing information on health related quality of life (HRQoL) on a wide range of conditions measured using accepted HRQoLmeasures (such as EQ-5D) alongside productivity information are required. This would allow productivity losses, for example days off work, to be linked to particular health states described by these HRQoL measures. In addition to patient data, the recall period for the HRQoL measures should match the recall period of the productivity losses to minimise bias associated with mismatch due to different recall periods. For example, the recall period in the EQ-5D is today whereas measures of productivity such as the Health and Labour Questionnaire (van Roijen et al.1996) use a two-week recall period. Larger studies focusing on productivity tend to ask respondents to consider longer periods such as four weeks, three (six or twelve) months (Zhang et al.2011). Linking longer productivity losses to current HRQoL may either overestimate or underestimate the effect of conditions. This aggregate approach of estimating productivity has been used by Krol et al.(2013) using Dutch general public data, and Rowen et al.(2013) using UK patient data. Krol et al.(2013) used the EQ-5D and hypothetical time off work estimated by the respondents to develop their model. Rowen et al.(2013) used EQ-5D, International Classification of Diseases (ICD 10) codes and self-reported days of work to develop models to predict productivity losses. Models from Rowen et al.(2013) are 6 applicable in the UK setting as they use the recommended heatlh technology assessment measure, the EQ-5D, and have clinical diagnosis data based on ICD which is used by the Department of Health in the UK. However, Rowen et al.(2013) identified a number of limitations which may limit applicability of their research.

The patient dataset that was used represents individuals who had recently been hospitalised and on average, these patients are likely to be sicker than the typical patient treated by the National Health Service (NHS) in the UK. Sicker respondents are likely to have higher productivity losses and models derived from these data would overestimate the productivity effects in typical patients. There were also concerns that different recall periods were used for the HRQoL measure (EQ-5D) and the number of days off work. The EQ-5D recall period was today while productivity information related to the previous 6 weeks. Some individuals who reported full health (EQ-5D=1) also reported having a large number of days off work and this may have been a result of the mismatch in recall periods.

The work described in this report was commissioned by the Department of Health to inform its work on Value-Based-Pricing (VBP) (Department of Health, 2010), which is due to replace the current Pharmaceutical Pricing Regulation Scheme (PPS) in January 2014 for pricing medicines in the UK.

VBP will include additional payments to interventions that are deemed to provide benefit that is of greater social value instead of the current narrow focus on outcomes relevant to the NHS and Personal Social Services (PSS). This requires taking into account wider societal benefits of medicines beyond the health of the patient including productivity.