Welfare-based healthcare planning: methodology and application to thoracic surgical treatment of lung cancer in Germany

Background : Established approaches to healthcare planning often focus on geographic accessibility of healthcare services but do not systematically incorporate patient outcomes and preferences. Such approaches may lead to inadequate conclusions when centralization of healthcare would improve patient outcomes due to volume-outcome relationships. Methods : We developed the methodology of welfare-based healthcare planning and investigated its applicability by the example of thoracic surgical treatment of lung cancer (TSTLC) in German hospitals. Our Welfare-based Healthcare Planning Model (WHPM) included a volume-outcome model capturing the case-volume elasticity (CVE) of the patients’ one-year survival odds. We estimated the CVE using statutory health insurance data covering 10.3% of the German population in 2022. We predicted patient flows in 2035 in different healthcare planning scenarios (HPS) by a gravity model fitted to observed locations of patients and hospitals. The WHPM contained a utility function representing the preferences of individuals regarding outcome quality and geographic accessibility. We estimated this utility function by applying a random utility model (RUM) to data from a discrete choice experiment (DCE) and derived the individuals’ willingness to travel. Drawing on welfare economics, we aggregated patient utilities and used social welfare as criterion for comparison of HPS. We reported parameter estimates with 95%-confidence intervals (95%-CIs). Results : We included 1,449 patients with TSTLC treated in 189 hospitals and found a statistically significant CVE of 0.27 (95%-CI=0.07;0.46). The RUM fitted to the DCE data of 1,010 respondents indicated insensitivity of utility regarding travel time up to 60 minutes and increasing marginal disutility from travel time beyond 60 minutes. For an increase in the one-year survival probability from 90% to 91%, individuals would be willing to travel additional 66 minutes (95%-CI=45;93 minutes) when traveling 60 minutes and additional 23 minutes (95%-CI=18;33 minutes) when traveling 240 minutes. The top 1,000 HPS included between 15 and 22 hospitals. In the welfare-optimal HPS, which included 19 hospitals, the average travel time was 54 minutes (status-quo HPS: 40 minutes) and the one-year survival probability was between 90.5% and 93.6% (status-quo HPS: 89.1%). Conclusions : By accounting for trade-offs between outcome quality and geographic accessibility, the WHPM facilitates welfare-optimal healthcare planning.