You are here:
COMPARISON OF RECALLED AND PROSPECTIVELY COLLECTED GASTROINTESTINAL SYMPTOM DATA WITHIN A WASHINGTON COHORT
Citation:
Peterson, C, Thomas, James C, T. Kunde, AND R L. Calderon. COMPARISON OF RECALLED AND PROSPECTIVELY COLLECTED GASTROINTESTINAL SYMPTOM DATA WITHIN A WASHINGTON COHORT. Presented at Society for Epidemiologic Research Annual Meeting, Atlanta, GA, June 12-14, 2003.
Description:
Comparison of Recalled and Prospectively Collected Gastrointestinal Symptom Data Within a Washington Cohort.
Christina A. Peterson1,2,3; James C. Thomas 1; Twila K. Kunde4; Rebecca L. Calderon2
1 Department of Epidemiology, University of North Carolina at Chapel Hill
2 Epidemiology and Biomarkers Branch, National Health and Environmental Effects Laboratory, U.S. Environmental Protection Agency, Research Triangle Park, NC
3 Department of Environmental Sciences and Engineering, University of North Carolina at Chapel Hill
4 Lovelace Clinic Foundation, Albuquerque, NM
SUMMARY:
Background
We present a comparison of recalled gastrointestinal symptom information with daily symptom diaries.
Methods
Participants recalled the number of episodes of soft stools, liquid stools, and vomiting experienced since a reference date approximately four weeks in the past once during each of two 6-month study phases. Throughout the study, participants completed daily diaries of health symptoms. We conducted agreement analyses using kappa statistics for each symptom individually and for the sum of all symptoms and calculated the sensitivity and specificity of the recall questionnaire relative to the daily symptom diaries.
Results
Agreement between the diary data and the recalled information was slight ranging from k = 0.27 [95% confidence interval (CI) (0.22, 0.31)] for soft stools during Phase 1 to k = 0.37 [95% CI (0.26, 0.48)] for vomiting during Phase 2. Single symptom sensitivity ranged from 53.6% for liquid stools during Phase 1 to 79.1% for soft stools during Phase 2, while the sensitivity for any symptom report for the entire study was 78.6%. Calculated specificities were higher, ranging from 77.2% for any symptom during the entire study to 96.0% for vomiting during the entire study.
Conclusions