Journal of Epidemiology (Sep 2024)
Validity of Self-reported Helicobacter pylori Eradication Treatment From Questionnaire and Interview Surveys of the JPHC-NEXT Study: Comparison With Prescription History From Insurance Claims Data
Abstract
Background: We aimed to evaluate the validity of self-administered questionnaire surveys and face-to-face interview surveys for the detection of Helicobacter pylori eradication therapy. Methods: Participants were a cohort, aged 40–74 years, living in three different locations of Japan, who took part in the baseline survey (2011–2012) of the Japan Public Health Center-based Prospective Study for the Next Generation (JPHC-NEXT). Five years after the baseline survey, a questionnaire and interview survey were independently conducted to determine the history of Helicobacter pylori eradication treatment over the 5-year period. Prescription of Helicobacter pylori eradication medications in national insurance claims data from the baseline survey to the 5-year survey was used as a reference standard. Results: In total, 15,760 questionnaire surveys and 8,006 interview surveys were included in the analysis. There were 3,471 respondents to the questionnaire and 2,398 respondents to the interview who reported having received Helicobacter pylori eradication treatment within the past 5 years. Comparison of the questionnaire survey to national insurance claims data showed a sensitivity of 95.1% (2,213/2,328), specificity of 90.6% (12,174/13,432), positive predictive value of 63.8% (2,213/3,471), negative predictive value of 99.1% (12,174/12,289), and Cohen’s Kappa value of 0.71. Respective values of the interview survey were 94.4% (1,694/1,795), 88.7% (5,507/6,211), 70.6% (1,694/2,398), 98.2% (5,507/5,608), and 0.74. Conclusion: Both the questionnaire and the interview showed high sensitivity, high specificity, and good agreement with the insurance claim prescriptions data. Some participants may have received eradication treatment without going through the public insurance claim database, resulting in a low positive predictive value.
Keywords