Steven Stellman - Evaluation of non-response bias in a cohort study of World Trade Center terrorist attack survivors

Version 1

      Publication Details (including relevant citation   information):

      Yu, S., Brackbill, R. M., Stellman, S. D., Ghuman, S., Farfel, M.   R. 8 (1) 42-

      Abstract: BACKGROUND: Few longitudinal studies   of disaster cohorts have assessed both non-response bias in   prevalence estimates of health outcomes and in the estimates of   associations between health outcomes and disaster exposures. We   examined the factors associated with non-response and the   possible non-response bias in prevalence estimates and   association estimates in a longitudinal study of World Trade   Center (WTC) terrorist attack survivors. METHODS: In 2003-04,   71,434 enrollees completed the WTC Health Registry wave 1 health   survey. This study is limited to 67,670 adults who were eligible   for both wave 2 and wave 3 surveys in 2006-07 and 2011-12. We   first compared the characteristics between wave 3 participants   (wave 3 drop-ins and three-wave participants) and   non-participants (wave 3 drop-outs and wave 1 only participants).   We then examined potential non-response bias in prevalence   estimates and in exposure-outcome association estimates by   comparing one-time non-participants (wave 3 drop-ins and   drop-outs) at the two follow-up surveys with three-wave   participants. RESULTS: Compared to wave 3 participants,   non-participants were younger, more likely to be male, non-White,   non-self enrolled, non-rescue or recovery worker, have lower   household income, and less than post-graduate education.   Enrollees' wave 1 health status had little association with their   wave 3 participation. None of the disaster exposure measures   measured at wave 1 was associated with wave 3 non-participation.   Wave 3 drop-outs and drop-ins (those who participated in only one   of the two follow-up surveys) reported somewhat poorer health   outcomes than the three-wave participants. For example, compared   to three-wave participants, wave 3 drop-outs had a 1.4 times   higher odds of reporting poor or fair health at wave 2 (95% CI   1.3-1.4). However, the associations between disaster exposures   and health outcomes were not different significantly among wave 3   drop-outs/drop-ins as compared to three-wave participants.   CONCLUSION: Our results show that, despite a downward bias in   prevalence estimates of health outcomes, attrition from the WTC   Health Registry follow-up studies does not lead to serious bias   in associations between 9/11 disaster exposures and key health   outcomes. These findings provide insight into the impact of   non-response on associations between disaster exposures and   health outcomes reported in longitudinal studies.

      Address (URL): http://www.ncbi.nlm.nih.gov/pubmed/25889176