Noninterventional studies in the COVID-19 era: methodological considerations for study design and analysis

Area of Science:

• Epidemiology
• Health Services Research
• Real-World Data Science

Background:

• The COVID-19 pandemic caused significant global morbidity, mortality, and healthcare system disruptions.
• Changes in healthcare utilization (e.g., medication use, hospitalizations) are evident in real-world data (RWD).

Purpose of the Study:

• To examine how pandemic-related healthcare disruptions impact noninterventional studies using RWD.
• To identify potential threats to study validity (internal and external) introduced by these disruptions.

Main Methods:

• Discussion of hypothetical noninterventional study designs.
• Analysis of pandemic effects on participant selection, exposure/outcome ascertainment, and covariate data.
• Consideration of validity threats like confounding, selection bias, and missing data bias.

Main Results:

• Pandemic disruptions can alter participant selection, potentially threatening external validity.
• Changes in healthcare access and behavior can introduce confounding and bias, impacting internal validity.
• Vulnerable populations may experience amplified impacts, exacerbating health disparities.

Conclusions:

• A framework is proposed for designing and analyzing noninterventional studies using RWD during the COVID-19 era.
• Researchers must proactively address pandemic-induced changes in RWD to ensure study integrity.
• Careful consideration of validity threats is crucial for reliable real-world evidence generation.