Estimating the marginal cost of a life year in Sweden’s public healthcare sector

The European Journal of Health Economics https://doi.org/10.1007/s10198-019-01039-0 ORIGINAL PAPER Estimating the marginal cost of a life year in Sweden’s public healthcare sector Jonathan Siverskog1 · Martin Henriksson1 Received: 15 November 2018 / Accepted: 12 February 2019 © The Author(s) 2019 Abstract Although cost-effectiveness analysis has a long tradition of supporting healthcare decision-making in Sweden, there are no clear criteria for when an intervention is considered too expensive. In particular, the opportunity cost of healthcare resource use in terms of health forgone has not been investigated empirically. In this work, we therefore seek to estimate the marginal cost of a life year in Sweden’s public healthcare sector using time series and panel data at the national and regional levels, respectively. We find that estimation using time series is unfeasible due to reversed causality. However, through panel instrumental variable estimation we are able to derive a marginal cost per life year of about SEK 370,000 (EUR 39,000). Although this estimate is in line with emerging evidence from other healthcare systems, it is associated with uncertainty, primarily due to the inherent difficulties of causal inference using aggregate observational data. The implications of these difficulties and related methodological issues are discussed. Keywords Opportunity cost · Threshold · Healthcare expenditure · Mortality · Life expectancy · Cost-effectiveness analysis JEL Classification C32 · C33 · C36 · I10 · I18 Introduction In practise, the decision to reimburse an intervention is often informed by judging its incremental cost-effectiveness ratio (ICER) against a cost-effectiveness threshold. Although imperative for resource allocation decisions and the interpretation of cost-effectiveness analysis, this threshold value has received remarkably little attention up until recently [1]. Sweden is no exception, and despite a long tradition of using cost-effectiveness analysis as an input into healthcare decision-making, the criteria for when an intervention is considered too expensive are vague. It has been argued that a threshold should represent the opportunity cost of healthcare resource use [2] and most commonly this is construed either Electronic supplementary material The online version of this article (https://doi.org/10.1007/s10198-019-01039-0) contains supplementary material, which is available to authorized users. * Jonathan Siverskog 1 Centre for Medical Technology Assessment, Department of Medical and Health Sciences, Linköping University, 581 83 Linköping, Sweden as private consumption forgone or health forgone. These two conceptions of opportunity cost are often referred to as the demand-side threshold (v-threshold), which tells us the consumption value of health gains, and the supply-side threshold (k-threshold), which indicates the marginal cost at which health could be generated if resources were not re-allocated to fund the evaluated intervention. Whether the demandside or supply-side threshold is deemed more appropriate depends, among other things, on the objective function and the constraints of the relevant authority. However, regardless of these aspects, there seems to be consensus in the literature that an estimate of the opportunity cost in terms of health forgone is often required [1, 2]. If resources are not readily transferrable between sectors, we cannot know whether reimbursement or approval decisions are expected to increase or decrease population health (by displacing other more productive healthcare services) without such an estimate. Furthermore, as noted by Brouwer et al. [1], even if resources are assumed (at least partly) transferable between sectors, an estimate of the supply-side threshold would be useful for understanding the discrepancy between what we would like to spend and what we are actually spending to gain health. Although estimates are emerging in the 13 Vol.:(0123456789) J. Siverskog, M. Henriksson literature [3–5], there is still no empirical estimate of the supply-side threshold for Sweden and many other countries where cost-effectiveness analysis is an important aspect of healthcare decision-making. The approach in available studies is to derive the threshold from the marginal effect of healthcare expenditure on mortality. There is, of course, already an abundance of studies on this relationship; see, e.g. Nolte and McKee [6] for a review or Gallet and Doucouliagos [7] for a meta regression analysis. As noted by Gravelle and Backhouse [8], however, it is important for such estimation to take into account that expenditure is unlikely to be exogenous with respect to mortality, which most past studies fail to do. Therefore, the relevant literature on this approach to supply-side thresholds is, so far, quite sparse. To the best of our knowledge, the study by Lichtenberg [9] is the first to express the relationship between healthcare expenditure and mortality as a cost per life year gained. It uses a time series on life expectancy at birth and public health expenditure in the United States and estimates a geometric lag model to derive a cost per life year of USD 9640. The issue of endogeneity is to some extent addressed by the model lag-structure and Granger causality testing. Martin et al. [10] estimate two disease-specific thresholds, one for cancer care at GBP 13,137 per life year and one for circulatory disease at GBP 7979 per life year. Their study considers a cross-section of years of life lost (YLL) and NHS expenditure for English primary care trusts. The relationship is estimated by two-stage least squares (2SLS), where the proportion of households that are lone pensioner households and the proportion of the population providing unpaid care act as instrumental variables for expenditure. Using data on EQ-5D scores for ICD-10 categories, average health-related quality of life (HRQoL) weights are calculated for cancer and circulatory disease to adjust the estimates to costs per quality-adjusted life year, QALY (GBP 19,070 and GBP 11,960, respectively). This approach is reapplied to ten programmes of care by Martin et al. [11]. Claxton et al. [3] build on this work to estimate an overall threshold for the English NHS at GBP 12,946, by also estimating the budget elasticity of expenditure for all programmes of care. This also attempts to include pure HRQoL effects by considering the QALY burden of disease. Claxton et al. [12] further consider a different set of instruments suggested by Andrews et al. [13] to re-estimate some mortality effects, but do not derive a new threshold. More recently, Edney et al. [4] estimate the effect of healthcare expenditure on YLL from a cross-section of small geographical areas in Australia. They also use the proportion of the population providing unpaid care as an instrument for expenditure and derive a threshold for the Australian health system at AUD 28,033 per QALY. YLL are adjust (...truncated)