How do HEOR studies handle missing data? – Healthcare Economist

That is the question answered in an article by Mukherjee et al. (2023). The authors define a “HEOR study” for this article as

…real-world evidence studies that performed secondary/post-hoc analysis using randomized methods.
controlled trial (RCT) data and a within-trial cost-utility analysis in which the outcome of interest was costs or PROs, including preference-based utilities (e.g., EQ-5D).

The most appropriate approach to impute missing data depends on assumptions about how the data is missing:

  • Missing completely at random (MCAR): The observed or unobserved values ​​of all variables in a study have no influence on the probability of a missing observation
  • Disappeared at random (MAR). The probability of missing data for a particular variable is associated with the observed values ​​of the variables (either observed values ​​of other variables in the data set or observed values ​​for the same variable at previous times) in the data set, but not with the missing data. You cannot check whether MAR holds on a data set.
  • Missing not at random (MNAR). In this case, the probability of missing data for a particular variable is related to the underlying value of that specific variable. MNAR can be ignorable (when missing values ​​occur independently of the data collection process) or nonignorable (when there is a structural cause of the missing mechanism that depends on unobserved variables or the missing value itself).

To address missing data, several techniques are available including: complete case analysis (CCA), available case analysis (AC), multiple imputation (MI), multiple imputation by chained equation (MICE), and predictive mean comparison.

To better understand what approaches are commonly used in health economics and outcomes research (HEOR), the authors conducted a systematic literature review in PubMed and examined what type of statistical methods were used to address missing measures of costs, utility or patient-reported outcomes.

The authors found that multiple imputation, multiple imputation by chained equation, and complete case analyzes were the most commonly used:

Of 1433 records identified, 40 articles were included. Thirteen studies were economic evaluations. Thirty studies used multiple imputation, 17 studies used multiple imputation using chained equation, while 15 studies used complete case analysis. Seventeen studies addressed missing cost data and 23 studies addressed missing outcome data. Eleven studies reported a single method, while 20 studies used multiple methods to address missing data.

The authors note that while they found a large amount of HEOR methodological literature on how to handle missing data in an RCT context; however, there were very few studies that attempted to actually implement these recommendations and impute missing data. You can read the full article. here.

We will be happy to hear your thoughts

Leave a reply

Register New Account
Compare items
  • Total (0)
Shopping cart