Prospective cohort for estimating health expenditures at the disease level: the case of cancer in Colombia

Abstract

BackgroundEstimates of health expenditures at both aggregate and household level are rarely conducted by disease. Availability of suitable datasets is a major barrier for such measurement. Yet, being able to estimate expenditures by disease greatly enhances our understanding of the dynamics of health spending, as it reconciles epidemiological trends with economic outcomes. In this work, we use a population-based registry to estimate spending in cancer in Colombia. MethodsThe biggest social health insurance institution in Colombia, Nueva EPS, changed its network of hospitals and providers of cancer care in 2010. All enrolled patients diagnosed with cancer in the southwest of the country are now referred to one research hospital, Fundación Valle del Lili (FVL), which coordinates care. The latter has detailed data on date and stage of diagnosis, courses of treatment, utilisation of different services, and costs, severity of cases, and survival. Patients entering this system on or after June, 2011, are followed. Since there is a population-based denominator (enrolees of Nueva EPS), incidence is estimated and compared with other sources for the following cancers: breast, cervix, prostate, lung, leukaemias, lymphomas, and gastrointestinal. For these, cost estimates are produced from the point of view of the health system. Trends are decomposed according to the factors that contribute to cost: utilisation, technology, service mix, and prices. FindingsIn the first 20 months, 5624 cancer patients from Nueva EPS had received care at FVL. Mean age was 65·9 years (SD 13·7). 39% (2198/5624) were male and 61% (3426/5624) were female. The most frequent cancers were breast (26·7%), prostate (17·1%), and colorectal (7·3%). Among men, the incidence rate per 100 000 population was 182·0 for prostate cancer and 53·1 for colorectal cancer in Cali. Among women, the figure was 101·8 for breast cancer and 27·7 for colorectal cancer. After adjusting for age, figures came closer, though remained higher, in comparison with other sources and populations. Costs per case were largely determined by the choice of certain chemotherapy drugs. The programme has started to improve adherence and also stage distribution at diagnosis. InterpretationService mix and technology dominate the trends in costs. Future trends in costs can be decomposed according to separable factors: epidemiological, technological, and market drivers (prices). The second ones appear to be more important for cancer in Colombia, especially because of the cost of certain drugs. Streamlining the process of authorising services (on the part of the payer) and coordinating interdisciplinary care improves adherence and outcomes. The observation period is still short to evaluate the impact on costs per case. FundingThis research was conducted by PROESA and funded by an internal research grant from Fundación Valle del Lili, where the authors are directly or indirectly employed.