0144 Assessing and addressing non-response at follow-up in the gulf study

    loading  Checking for direct PDF access through Ovid

Abstract

Background

The Gulf Long Term Follow-up (GuLF) STUDY is the first investigation to assess coronary heart disease (CHD) among oil spill workers. Participant non-response to periodic follow-up interviews can impact the ability to measure CHD over time. Describing predictors of non-response and the impact on analyses will improve generalizability of results to the study population.

Background

Objective: Assess predictors of non-response at the first follow-up interview in the GuLF STUDY and use inverse probability (IP) weights to account for these in an analysis of oil spill cleanup work duration and CHD.

Methods

We examined covariate distributions between those who did (n=21,245) and did not (n=10,364) complete the follow-up interview, and constructed IP censoring weights to account for these differences. We applied the weights to estimate cohort-level hazard ratios (HR) and 95% confidence intervals (95% CI) for oil spill cleanup work duration and self-reported nonfatal CHD.

Results

Those who participated in the follow-up interview were more likely to be older, have higher income and education, and were less likely to be smokers. After applying IP censoring weights and controlling for confounding, work duration (>180 days vs. 1–30 days) was non-significantly associated with nonfatal CHD [HR (95% CI)=1.50 (0.92–2.43)]. The results were similar without censoring weights [HR (95% CI)=1.41 (0.87–2.27)].

Conclusions

Several factors were associated with participation in the GuLF STUDY follow-up interview, however results for work duration and CHD were robust to censoring. Weighting improved generalizability of results but did not change conclusions about the association between work duration and CHD.

Related Topics

    loading  Loading Related Articles