A cost-effectiveness analysis of the effect of hospital variation in the probability of providing treatment with curative intent in potentially curable esophageal and gastric cancer patients.

For potentially curable esophageal cancer (EC) and gastric cancer (GC) patients, the probability of treatment with curative intent varies between hospitals and is associated with survival. This study examines the effect of this variation on health economics outcomes and cost-effectiveness. We performed a cost-effectiveness analysis from a societal perspective in potentially curable EC or GC patients selected from the Netherlands Cancer Registry. Resource use and costs were estimated for each treatment strategy from diagnosis until five years follow-up using a top-down costing method. Hospitals were divided into tertiles of low, medium, or high probability of treatment with curative intent using multilevel multivariable logistic regression. The primary outcome was the incremental cost-effectiveness ratio (ICER). Mean total costs per patient was not significantly different between low, medium, and high probability hospitals for EC (n = 9468) (€47,532 vs. €47,384 vs. €47,825), while for GC (n = 3085) costs were significantly lower in low compared to medium and high probability hospitals (€27,759 vs. €30,183 vs. €29,589, both P < 0.001). Costs per quality adjusted life year (QALY) were slightly lower in high probability hospitals for both EC and GC (EC: €29,181 vs. €28,646 vs. €27,659, GC: €25,003 vs. €22,505 vs. €20,495). ICERs were highest for high vs. medium probability hospitals for EC (€4900/QALY) and for medium vs. low probability hospitals for GC (€10,539/QALY). Variation in treatment with curative intent between hospitals affects health economics outcomes to a limited extent. Although all hospital comparisons were cost-effective, for the highest QALY gain, it is recommended to treat potentially curable patients as in high probability hospitals.