Context-Specific Estimation of Future Unrelated Medical Costs and Their Impact on Cost-Effectiveness Analyses

This study constructed and applied procedures for the estimation of unrelated future medical costs (UFMC) of women with breast cancer in Israel (as a case study) and examined the influence of including UFMC in cost-effectiveness analyses (CEAs).

Part I consisted of a retrospective cohort study based on patient-level claims data of both patients with breast cancer and matched controls during 14 years of follow-up. UFMC were estimated as (a) the annual average all-cause healthcare costs of the control subjects, and (b) as predicted values based on a generalized linear model (GLM) adjusted to patients' characteristics. Part II consisted of a CEA performed using a Markov simulation model comparing regimens of chemotherapy with/without trastuzumab, both excluding and including UFMC and for each of the UFMC estimates separately. All costs were adjusted to 2019 prices. Costs and QALYs were discounted at a yearly rate of 3%.

The average annual healthcare costs in the control group were $2328 (± $5662). The corresponding incremental cost-effectiveness ratio (ICER) was $53,411/QALY and $55,903/QALY, when UFMC were excluded or included, respectively. Hence, trastuzumab was not considered cost-effective compared with a threshold of willingness-to-pay of $37,000 per QALY, regardless of the inclusion of UFMC. When UFMC were estimated on the basis of the prediction model, the ICERs were $37,968/QALY and $39,033/QALY, when UFMC were excluded or included, respectively. Thus, in this simulation, trastuzumab was not considered cost-effective, independent of the inclusion of UFMC.

Our case study revealed that the inclusion of UFMC had modest effect on the ICERs, and thus did not alter the conclusion. Thus, we should estimate context-specific UFMC if they are expected to change the ICERs significantly, and transparently report the corresponding assumptions to uphold the integrity and reliability of the economic evaluation.