Handling Missing Data in Health Economics and Outcomes Research (HEOR): A Systematic Review and Practical...

Area of Science:

• Health Economics and Outcomes Research (HEOR)
• Biostatistics
• Data Science

Background:

• Missing data in HEOR studies can lead to biased inferences and flawed health policies.
• Existing literature often focuses on randomized controlled trials, not always applicable to HEOR data.

Approach:

• Systematic literature review of studies published up to 2020 in PubMed.
• Included studies using statistical methods for missing cost, utility, or patient-reported outcome data.
• Extracted data on study context, missing data type, and handling methods.

Key Points:

• 40 papers were included, with 13 economic evaluations.
• Multiple imputation (30 studies) and complete-case analysis (15 studies) were common methods.
• Most studies addressed missing cost or outcome data, with many using multiple methods.

Conclusions:

• Many HEOR studies fail to justify their missing data handling approaches.
• A single method without sensitivity analysis is often insufficient.
• Considering missingness mechanisms and performing sensitivity analyses are crucial for robust HEOR studies.