OBJECTIVES: Guidance from the National Institute of Health and Care Excellence (NICE), Canada’s Drug Agency (CDA) and Zorginstituut Nederland (ZIN) on utility analyses is limited. A frequently used model for Health Technology Assessment (HTA) submissions is the repeated-measured mixed effects (RMME). Despite its theoretical limitations, like inability to handle ceiling effects and multimodality, previous analyses showed that RMME often outperformed other regression approaches. Its strength lies in accounting for subject-specific effects across repeated observations. This characteristic may become a weakness in the presence of latent variables. This study compared RMME with descriptive statistics where low baseline utility is prognostic for early progression and low utility drop post-progression.

METHODS: We simulated utility data resembling a randomized controlled trial using the Dutch EQ-5D-5L value set. Utility depended on treatment, progression, and severity (a latent variable influencing outcomes). Severe patients were more likely to progress, had lower baseline utility, and experienced minimal utility drop post-progression. An RMME model with baseline utility, treatment, and progression, as variables, was used to predict mean utilities, comparing results against the descriptive statistics and true values, based on treatment and progression.

RESULTS: Pre-progression, mean RMME utilities aligned with true values and descriptive statistics. Post-progression, the true utility dropped by 0.07 and 0.09 for treated and non-treated patients, respectively, with the descriptive statistics capturing 100% and 88% of the true drop, respectively. However, RMME estimated a smaller decrease, 71% and 56% of the true drop for treated and non-treated, respectively. In a scenario without baseline utility predicting post-progression drop, RMME estimates closely matched true values.

CONCLUSIONS: Although RMME is commonly used, seems to outperform other regression methods, and is accepted by HTA bodies, it remains prone to bias. Additional HTA guidance is necessary on conducting appropriate descriptive analyses and linking these to selecting health state utility estimation techniques to improve health economic evaluations.