Reliability of self-reported medical history and selected health characteristics in the environmental health study for Western New York cohort

We examined the reliability of self-administered questions used to assess medical and family history and other characteristics in the Environmental Health Study for Western New York (EHS4WNY), a prospective cohort study designed to investigate industrial pollution and health. Between 2018 and 2022, 13,306 adults completed a self-administered, either mailed or identical online questionnaire version. Two hundred and ninety-five participants completed duplicate baseline questionnaires. Reliability was assessed using Cohen's kappa (K), weighted kappa (Kw), percent agreement, and intraclass correlation coefficient (ICC). Common medical conditions, including hypertension (K = 0.80; 95 % CI: 0.71, 0.89) and high cholesterol (K = 0.76; 95 % CI: 0.66,0.86) yielded excellent reliability. A personal medical history of colorectal cancer was moderately reliable (K = 0.67; 95 % CI: 0.05, 1.00). Family history of breast cancer had the highest reliability (K = 0.81; 95 % CI: 0.72, 0.91) and non-melanoma skin cancers had the lowest (K = 0.47; 95 % CI: 0.28, 0.65). Recalled height (ICC = 0.98; 95 % CI: 0.97, 0.98) and weight at age 18 (ICC = 0.96; 95 % CI: 0.94, 0.97) showed excellent reliability. This questionnaire demonstrated fair-to-excellent test-retest reliability of self-reported medical and family history and selected factors. Although we did not assess validity, self-reported medical history collected via questionnaire can be used to generate reliable data in general population studies.