Assessing Representativeness of Randomised Controlled Trials using Serious Adverse Events

Area of Science:

• Gerontology
• Clinical Trials
• Health Services Research

Background:

• Randomized controlled trials (RCTs) in older adults with frailty/multimorbidity face challenges in representativeness.
• Assessing trial representativeness is complex and crucial for applicability of findings.

Purpose of the Study:

• To evaluate trial representativeness by comparing rates of serious adverse events (SAEs) in trials to hospitalization/death rates in routine care.
• To assess if comorbidity count explains observed differences in SAEs.

Main Methods:

• Secondary analysis of 483 trials (n=636,267) from clinicaltrials.gov and routine care data (n=2.3M) from SAIL databank.
• Calculated observed/expected SAE ratios using hospitalization/death rates derived from SAIL data, adjusted for age, sex, and condition.
• Re-calculated ratios accounting for comorbidity count in a subset of 125 trials.

Main Results:

• For 12/21 conditions, observed SAEs in trials were lower than expected based on routine care rates (observed/expected SAE ratio <1).
• Median observed/expected SAE ratio was 0.60 (COPD), with an interquartile range of 0.44-0.88.
• Higher comorbidity count correlated with SAEs, but adjusting for it only partially explained the lower SAEs in trials.

Conclusions:

• Trial participants, particularly older adults, experience fewer SAEs than expected, indicating limited representativeness.
• The lack of representativeness is only partially explained by multimorbidity, highlighting the need for better trial inclusion strategies.
• The observed/expected SAE ratio can aid in assessing the applicability of trial findings to frail, multimorbid populations.