종적자료의 발생률 산출방법론 비교 고찰

다국어 초록 (Multilingual Abstract)

It is very important to calculate the frequency of diseases in disease monitoring and investigating its cause. In particular, in order to determine the incidence rate according to time, the incidence density (hereinafter referred to as incidence rate) is used, which is calculated with the number of events divided by the total observation time. The indices for an incidence rate are clearly defined, but every researcher may use a different missing value processing range in the longitudinal data depending on how many follow-up failures occurred. Thus the number of events (especially, disease occurrence) may vary. In this regard, in this study we calculated the incidence rate in a number of ways using the actual longitudinal data by comparing and discussing the resulting differences.
The data used in this study were community-based cohort (Ansan and Anseong) and Korea Health Panel data. The community-based cohort (Ansan and Anseong) is a prospective cohort data based on the follow-ups conducted at a 2-year interval from 2001 to 2012. The Korea Health Panel data was based on follow-ups conducted every year from 2008 to 2012.
Incidence rates were reviewed by separating the numerator and denominator in the calculation formula. The numerator was defined as three ways (A, B, C) with regard to diabetes and hypertension, and the denominator was defined as three estimation methods (①, ②, ③) for the observation time to calculate the incidence rate.

When calculating and comparing the incidence rate in various ways, the incidence rates were found to differ depending on how to the events were defined differently. In addition, the differences in the observation time differed according to how long the failed follow up period was prior to the defined event.
Incidence rates may vary depending on how the event definition and the observation time, but in this study, we calculated the incidence rate in nine ways (numerator: 3 kinds, denominator 3 kinds) using two kinds of longitudinal data and compared the incidence rates. The causes of differences in the incidence rate were explained in detail. Therefore, for a study that aimed to investigate the incidence rates, it was suggested to describe clearly and in detail on the data processing and to record the version of concerned data to ensure that the same results can be obtained at reproduction. In addition, the results of this study are to be provided as the basic data for investigating the causes that may have effect on diseases, to be utilized for the prevention and early treatment of diseases and to be utilized for basic data for health policy development.

목차 (Table of Contents)

• Ⅰ. 서론 1
• 1. 연구의 배경 및 필요성 1
• 2. 연구의 목적 4
• Ⅱ. 이론적 고찰 6
• 1. 발생률의 이해 6
• (1) 종적 연구 6
• (2) 참여 그룹의 유형에 따른 종적연구 구분 20
• (3) 발생률의 개념 26
• 2. 발생률 산출에 관한 선행 연구 29
• (1) 발생률 분자 정의 방법에 대한 선행연구 29
• (2) 발생률의 분모(관찰시간)에 관한 선행연구 39
• Ⅲ. 연구 방법 44
• 1. 연구자료 44
• (1) 지역사회기반코호트(안산,안성) 44
• (2) 한국의료패널 자료 47
• 2. 연구체계 48
• Ⅳ. 연구결과 50
• 1. 원자료(지역사회기반자료, 한국의료패널)를 이용한 발생률 비교 50
• (1) 분자 정의방법에 따른 발생률 50
• (2) 분모(관찰시간) 정의방법에 따른 발생률 61
• (3) 분자와 분모 정의방법에 따른 발생률 비교 70
• (4) 생존곡선을 이용한 차수별 발생률 비교 77
• (5) 노출요인이 발생률에 미치는 영향 비교 84
• 2. 자료가공을 통한 발생률 검토(지역사회기반코호트) 86
• (1) 참여자 유형에 따른 고혈압 분포 현황 86
• (2) 참여자 유형을 가공한 발생률 비교 88
• (3) 추적차수를 가공한 발생률 검토 92
• 3. 한국의료패널 원자료를 이용한 고혈압 발생률 비교 97
• Ⅴ. 고찰 99
• 1. 연구방법에 대한 고찰 99
• 2. 연구결과에 대한 고찰 102
• 3. 연구의 제한점 104
• Ⅵ. 결론 105
• 참고문헌 107
• 감사의 글 114