Abstract

The World Trade Centre (WTC) disaster occurred on September 11, 2001 and resulted in 2,876 immediate deaths and approximately 6,000 injuries. Shortly after the disaster, the WTC medical screening program (MSP) was started for the responders. The program was initially designed as a one-time medical screening, but evolved into a longitudinal medical monitoring program allowing researchers to investigate a wide range of health outcomes. There have now been more than 2,000 WTC scientific publications, but none that have formally investigated potential impacts of biases that may have arisen from the challenging conditions under which the studies were designed and data collected.

We conducted a systematic assessment of the potential impacts of biases on health effects studies of the WTC MSP. Systematic review of the published studies was supplemented with targeted analyses of the primary data collected from WTC responders to identify and evaluate biases in qualitative and quantitative approaches.

Several potential biases were identified using questionnaire data collected by WTC MSP. Selection bias likely occurred at least through self-selection into the WTC MSP. Loss to follow-up and non-random sampling due to differential recruiting strategies across study periods also likely occurred. Significant exposure misclassification may have occurred through use of self-reported information without validation from objective real-time exposure assessment. For some health outcomes, misclassifications may have occurred due to low inter-rater reliability during clinical diagnosis of health conditions, as well by changes over time in which health conditions were recognised as WTC-related. A quantitative analysis estimated the magnitude of the impact of these biases.

Implementing a disaster based epidemiologic surveillance faces many challenges; among them is designing health surveillance in ways that increase reliability and reduce potential biases. Our study aimed to strengthen future disaster epidemiology studies by learning from the challenges of data collection and analysis for the WTC MSP.