Page 80 - Effective healthcare cost containment policies Using the Netherlands as a case study - Niek W. Stadhouders

Chapter 4
Spain, Australia, the US and Canada (Ariste and Di Matteo, 2017; Edney et al., 2018; Vallejo‐Torres et al., 2017; Woods et al., 2016).
This paper applies a novel approach for threshold estimation to hospital care in the Netherlands. We define opportunity costs as the health effect of a marginal change in spending for the average patient group. We restrict our analysis to the hospital sector, as this is where opportunity costs for new drugs and innovations are likely to fall. Other thresholds may apply if expenses are reduced in other sectors (e.g. primary care, tertiary care) to fund new technologies in the hospital sector. QALYs are constructed by combining gains due to lower mortality and gains due to increases in quality of life of all patients (Gheorghe et al., 2015). We define a production function with spending and the number of patients as inputs and QALYs as outputs for 11,000 patient groups based on gender, disease category and age category. We approximate the hospital production function using a translog specification, and estimate a fixed effects model on panel data covering 2012-2014. Threshold estimations are known to be sensitive to endogeneity (Martin et al., 2008). This is especially troubling when focusing on spending that aims to reduce mortality, as the health care costs involved with the last year of life are known to be substantial (Polder et al., 2006). Failure to account for these costs could underestimate the effect of health care on survival. As these costs are well studied and known for the Dutch situation (van Baal et al., 2011), we have the opportunity to correct for them. Furthermore, the translog specification accounts for exogeneous changes in health status that may confound the results, as increases in population health are likely to be reflected in reduced patient numbers. As robustness tests for omitted variable bias, we include general health trends (smoking, obesity, alcohol abuse). Estimation of the translog function renders the marginal effect of spending on the mean patient group, which can be interpreted as a supply-side cost-effectiveness threshold (Woods et al., 2016). This may provide information for Dutch policy makers in reimbursement decisions and strengthen the empirical base for using a threshold. Furthermore, we estimate patient group thresholds separately, which may point out inefficiencies in current spending allocation.
4.2 Data and methods
4.2.1 Data transformations
In the hospital sector, patients lose QALYs as a result of premature deaths (death-related QALY loss) and lower quality of life while being ill (morbidity-related QALY loss). Consequently, extra spending may add QALYs resulting both from prevention of premature deaths and increasing the quality of life of patients. In previous research, elasticities of spending on mortality were estimated, after which the outcome was transformed to QALYs
72