Health utility is a key input used to perform cost-utility analysis (CUA), which is increasingly used to inform resource allocation decisions. To identify the sources and elicitation methods of health utilities used in CUAs in oncology. We used the Tufts Cost-Effectiveness Analysis registry to identify oncology CUAs published in Medline between 1976 and 2021. Eligible CUAs had to include an oncology population (based on ICD-10 codes), report health utilities, and be published in English. The references of cited health utilities were traced to identify the original health utility study and the method of utility elicitation. Characteristic of included CUAs were summarized and the methods to derive health utilities were compared. A total of 1512 CUAs in oncology were identified. The majority of CUAs (n = 1428, 94.4%) were model-based. Malignant neoplasm of breast and female genital organs was the most common population considered (n = 424, 28.0%). Among these CUAs, 8714 health utilities were identified. Upon review, the sources of 2096 (24.1%) health utilities could not be traced. Of the remaining 6618 health utilities, 1718 (26.0%) were obtained from original health utility study embedded in CUA in which expert opinion (n = 547, 31.8%) or EQ-5D (n = 479, 27.9%) was most frequently used. The 4900 health utilities (74.0%) that were cited from external studies were most often derived using the standard gamble (n = 1258, 25.7%) or EQ-5D (n = 1190, 24.3%). Published health utilities are widely used in oncology CUAs, especially for model-based analyses. However, the identification, selection, and use of health utilities is suboptimal.
Read full abstract