Net-benefit regression models with treatment interactions.

Multiple imputation by chained equation (MICE) estimates, 45 imputations (see Table 3). Interaction terms denoted by Intervention [.]. ISAR—Identification of Seniors at Risk.*Significant at 0.05 level.aHubert-White p-values and 95% CI corrected for heteroskedasticity.bCentred around mean.cIntervention coefficient for identical regression model with ISAR dummy reversed (moderate-risk dummy (ISARp = 0.29), indicates net monetary benefit (intervention vs. standard care) in the subgroup of high-risk patients.dIntervention coefficient for identical regression model with care-home dummy reversed is -£724 (95% CI: -1915, 468; p = 0.23).Net-benefit regression models with treatment interactions.