Triggered sampling could help improve longitudinal studies of persons with elevated mortality risk.

Area of Science:

• Clinical Research Methodology
• Longitudinal Study Design
• Biostatistics

Background:

• Longitudinal studies are susceptible to dropout bias, potentially skewing results.
• Advanced illness patient populations often face high mortality and dropout rates.
• Traditional study designs may not adequately capture critical data before participant attrition.

Purpose of the Study:

• To introduce and define "triggered sampling" (TS) as a novel method to mitigate dropout bias.
• To evaluate the utility and efficacy of TS in a real-world longitudinal observational study.
• To assess the impact of TS on capturing treatment preferences in patients with advanced illnesses.

Main Methods:

• Formally defined the triggered sampling (TS) methodology.
• Incorporated TS into a 2-year longitudinal observational study involving patients with cancer, congestive heart failure, or COPD.
• Analyzed treatment preference data (1-7 scale) using mixed-effects models and pre- vs. post-trigger comparisons.

Main Results:

• 148 out of 226 participants underwent at least one triggered interview.
• Patients who did not drop out post-trigger showed stable preferences (6.20 pre, 6.16 post).
• Patients who dropped out post-trigger exhibited a significant decrease in preferences (6.29 pre, 5.69 post; P=0.04).

Conclusions:

• Triggered sampling effectively captures data from participants at high risk of dropout.
• TS helps alleviate bias introduced by impending participant attrition in longitudinal studies.
• This method offers a valuable enhancement for longitudinal study designs, particularly in high-mortality populations.