Agreement between contemporaneously recorded and subsequently recalled time spent outdoors: implications for environmental exposure studies.

PURPOSE: The aim of the study is to evaluate the agreement between contemporaneously recorded and subsequently recalled time spent outdoors during 1 week among members of an occupational cohort.
METHODS: One hundred twenty-five radiologic technologists from northern and southern geographic areas in the United States recorded time spent outdoors for 7 consecutive days in a daily diary. Six months later, study participants completed a mailed self-administered questionnaire of the number of outdoor hours during the same 7-day period. We tested the agreement between questionnaire responses and diary entries. Logistic regression models were used to identify variables significantly affecting agreement.
RESULTS: Time spent outdoors comprised one fifth of the total time recorded in the diaries. Agreement (weighted kappa [kappa(w)]) between reported outdoor time during weekdays (kappa(w) = 0.49; 95% confidence interval [CI], 0.39-0.59) was significantly (p < 0.05) higher than for weekends (kappa(w) = 0.23; 95% CI, 0.12-0.34). Similarly, agreement was lower for weekends compared with weekdays in multivariate analyses, reaching statistical significance (p = 0.05) in only the southern regions.
CONCLUSIONS: Although our investigation was carried out among volunteers from the US radiologic technologist cohort, we believe retrospective questionnaires may be more accurate in reporting time spent outdoors for weekdays compared with weekends in any group of indoor workers. These differences have implications for the wording in future questionnaires about time spent outdoors and level and sources of uncertainty characterizing estimated time spent outdoors on weekdays versus weekend days.