| Summary: | 碩士 === 國立陽明大學 === 公共衛生研究所 === 95 === Objective: To validate respondents’ answers to history of diabetes mellitus, and found the better algorithm by using claim data or questionnaire. Finally, we sought to the individual factors of disagreement between survey and claim data.
Study Design and Setting: 6,792 participants of National Health Interview Survey (NHIS) >= 40 years of age were include in this study which using linked data from NHIS and National Health Insurance (NHI) claim data.
Results: Agreement between survey and claim data was substantial (kappa 0.67-0.82) for diabetes regardless of any algorithms. To define most chronic diseases by using claim data that should require two diagnoses and using survey data should enquire about the subjects’ treatment that would be better algorithms. Factors associated with high agreement by multivariate analysis were younger, female, lower BMI, and self-perceived better health status.
Conclusion: No matter using survey or claim data, we could get a validate population of diabetes mellitus. Using the right algorithms to define chronic diseases can improve the agreement between survey and claim data. In the further, the research about diabetes mellitus should be careful of the effect on age, BMI and self-perceived better health status.
|
|---|
