What is the value of explicit priority setting for health interventions? A simulation study

Abstract

Many countries seek to secure efficiency in health spending through establishing explicit priority setting institutions (PSIs). Since such institutions divert resources from frontline services which benefit patients directly, it is legitimate and reasonable to ask whether they are worth the money. We address this question by comparing, through simulation, the health benefits and costs from implementing two alternative funding approaches - one scenario in which an active PSI enables cost-effectiveness-threshold based funding decisions, and a counterfactual scenario where there is no PSI. We present indicative results for one dataset from the United Kingdom (published in 2015) and one from Malawi (published in 2018), which show that the threshold rule reliably resulted in decreased health system costs, improved health benefits, or both. Our model is implemented in Microsoft Excel and designed to be user-friendly, and both the model and a user guide are made publicly available, in order to enable others to parameterise the model based on the local setting. Although inevitably stylised, we believe that our modelling and results offer a valid perspective on the added value of explicit PSIs.

Keywords: Cost-effectiveness thresholding; Health technology assessment; Portfolio decision analysis; Priority setting institutions; Simulation.

Fig. 1

Distribution of total incremental expenditure per funding cycle for each decision rule. (a) Indicative UK data, (b) Indicative Malawi data

Fig. 2

Distribution of total incremental QALY gain per funding cycle for each decision rule. (a) Indicative UK data, (b) Indicative Malawi data

Fig. 3

Distribution of the difference in the total cost of funded interventions (Eq. (9)) against the difference in the total health benefit gain of funded interventions (Eq. (10)), per funding cycle for each decision rule. (a) Indicative UK data, (b) Indicative Malawi data

Fig. 4

Distribution of the difference in the total cost of funded interventions (Eq. (9)) against the difference in the total health benefit gain of funded interventions (Eq. (10)), per funding cycle for each decision rule, with budget limits. (a) Indicative UK data, (b) Indicative Malawi data

Fig. 5

Distribution of the difference in the total cost of funded interventions (Eq. (9)) against the difference in the total health benefit gain of funded interventions (Eq. (10)), per funding cycle for each decision rule, with limited application of threshold rule. (a) Indicative UK data, (b) Indicative Malawi data

Fig. 6

Distribution of the difference in the total cost of funded interventions (Eq. (9)) against the difference in the total health benefit gain of funded interventions (Eq. (10)), per funding cycle for each decision rule, with phased increase of the application of the threshold rule. (a) Indicative UK data, (b) Indicative Malawi data

Fig. 7

Comparison of the (log-transformed) indicative Malawi and UK data-sets with normally distributed data. For each country, the plots show the indicative data on the three aspects of an intervention: the incremental costs of administering each intervention per treated case (Cost), the incremental health benefits returned per case treated with the intervention (Health), and the number of cases treated by each intervention (Population). The indicative Malawi and UK data-sets are taken from Ochalek et al. [23] and Guthrie et al. [10], respectively, and are presented in Appendix 1

Fig. 8

${NHB(I_{\mathit{TH}},I}_{\mathit{CF}},t)$ as defined in Eq. (11), with portfolio $I_{\mathit{TH}}$ determined using threshold funding decision rule (5) and portfolio $I_{\mathit{CF}}$ determined using counterfactual rule (3), recorded as threshold $t$ varies in (5), for three levels of correlation ${\rho }_{\mathit{CQ}}$. (a) Indicative UK data: full distributions, (b) Indicative Malawi data: full distributions, (c) Indicative UK data: median values, (d) Indicative Malawi data: median values

Fig. 9

${NHB(I_{\mathit{TH}},I}_{\mathit{CF}},t)$ as defined in Eq. (11), with portfolio $I_{\mathit{TH}}$ determined using threshold funding decision rule (6) and portfolio $I_{\mathit{CF}}$ determined using counterfactual rule (4), recorded as threshold $t$ varies in (6), for three levels of correlation ${\rho }_{\mathit{CQ}}$. (a) Indicative UK data: full distributions, (b) Indicative Malawi data: full distributions, (c) Indicative UK data: median values, (d) Indicative Malawi data: median values

Fig. 10

${NHB(I_{\mathit{TH}},I}_{\mathit{CF}},t)$ as defined in Eq. (11), with portfolio $I_{\mathit{TH}}$ determined using threshold funding decision rule (6) and portfolio $I_{\mathit{CF}}$ determined using counterfactual rule (4), recorded as budget limit $l$ varies in (6), for three levels of correlation ${\rho }_{\mathit{CQ}}$. (a) Indicative UK data: full distributions, (b) Indicative Malawi data: full distributions, (c) Indicative UK data: median values, (d) Indicative Malawi data: median values

Fig. 11

${NHB(I_{\mathit{TH}},I}_{\mathit{CF}},t)$ as defined in Eq. (11), with portfolio $I_{\mathit{TH}}$ determined using threshold funding decision rule (7) and portfolio $I_{\mathit{CF}}$ determined using counterfactual rule (4), recorded as percentage $p_a$ varies in (7), for three levels of correlation ${\rho }_{\mathit{CQ}}$. (a) Indicative UK data: full distributions, (b) Indicative Malawi data: full distributions, (c) Indicative UK data: median values, (d) Indicative Malawi data: median values

Fig. 12

${NHB(I_{\mathit{TH}},I}_{\mathit{CF}},t)$ as defined in Eq. (11), with portfolio $I_{\mathit{TH}}$ determined using threshold funding decision rule (8) and portfolio $I_{\mathit{CF}}$ determined using counterfactual rule (4), recorded as percentage $p_Y$ varies in (8), for three levels of correlation ${\rho }_{\mathit{CQ}}$. (a) Indicative UK data: full distributions, (b) Indicative Malawi data: full distributions, (c) Indicative UK data: median values, (d) Indicative Malawi data: median values

Fig. 13

${NHB(I_{\mathit{TH}},I}_{\mathit{CF}},t)$ as defined in Eq. (11), with portfolio $I_{\mathit{TH}}$ determined using threshold funding decision rule (8) and portfolio $I_{\mathit{CF}}$ determined using counterfactual rule (4), recorded as the number of funding cycles $Y$ varies in (8), for three levels of correlation ${\rho }_{\mathit{CQ}}$. (a) Indicative UK data: full distributions, (b) Indicative Malawi data: full distributions, (c) Indicative UK data: median values, (d) Indicative Malawi data: median values