Objective or perceived health measure: which is a best cost predictor?

Background Several indices exist to describe the objective health and also to asses subjective health. The association between comorbidity, severity and subjective health with health expenditure is demonstrated, however it is not clear which of these aspects is more influent. We evaluated the role of comorbidity, severity of disease and subjective health to predict health costs in a non-hospitalized population. Methods In 2009/10 the questionnaire SF36 (8 scales: physical functioning-PF, physical role-PR, pain-PN, general health- GH, vitality-VT, social functions-SF, emotional role-ER, mental health-MH) and two summary scales (Physical-PCS and Mental-MCS) was distributed to 887 patients of general practitioners (GP) in Siena’s province, Tuscany-Italy. GPs calculate Charlson Index (CI) and Cumulative Illness Rating Scale (CIRS) for each patient. CI is an index summing diseases’ associated weights and adding one point for each decade after 40 years. CIRS produces a severity score as the mean weight of severity of coexisting illnesses. Information about gender, age, marital status, employment, BMI and smoking habits were collected. Data about pharmaceutical, diagnostic tests, specialistic visits and hospital admissions costs regarding each patient were obtained from Local Health Unit’ data warehouse. Association between SF36 scales, PCS, MCS, CI and CIRS with expenditure variables was studied using multivariate nonparametric quantile regression, adjusting for above confounders. Results Mean age 57,4, 63.5% females. At multivariate analysis resulted associated with diagnostic examination costs: CI (coeff. = 31.36, p < 0.001), CIRS (297.51, p < 0.001), PCS (-4, p < 0.001), PR (-0.53, p < 0.05), PN (-1, p = 0.001), GH (-1,5, p = 0.002); with pharmaceutical cost: CI (47.47, p < 0.001) SI (470, p < 0.001) PCS (-2.13, p < 0.05) PA (-0.75, p < 0.05) GH (-1.79 p < 0.001) VT (-0.082 p < 0.02). None of the health measures was associated with hospital admission expenditure. Conclusions Objective measures of health are strongly correlated with tested expenditures. Subjective measures are also correlated, although less of objective ones. Severity of disease is more correlated with health expenditures than comorbidity. Mental/psycholo- gical domains of subjective health have no effect on health expenditures. Key message The objective measures of health are a strong predictor of health expenditures Health expenditure is influenced not only by the objective state of health but also from a subjective.