A Systematic Review of Studies on Public Health Using Multilevel Analysis: Focused on Research Trends and The Assessment of Risk of Bias

??? ????????? ????????? ??????????????? ?????? ?????? ?????????????????? ????????? ????????? ????????? ???????????? ????????? ?????? ????????? ???????????? ????????? ?????? ????????? ???????????? ?????? ???????????? ?????? ????????? ??????????????? ???????????? ????????? ???????????? ????????? ???????????????. 2000??? 1??? 1??? ?????? 2014??? 12??? 31????????? ????????? ??? ?????? ?????? ????????????????????? ????????? 59?????? ????????? ????????? ????????? ????????? ??????. ??????, ????????? ???????????? ?????? ?????? ????????? ????????? 20.9%??? ?????? ?????????. ??????, ?????? ?????? ????????? ?????? ????????? ??????????????? ??????????????????, ????????? ?????? ?????? ?????? ?????? ??? 40.7%??? ????????? ???????????? ?????????. ??????, ?????????????????????????????? ?????? ?????? ????????? ????????? 34.4%??? ????????? ???????????????. ??????, 4?????? ????????? ??????????????? ?????? ?????? ????????? ????????? ?????? ???????????? ????????????. ???????????? ??????????????? ?????? ???????????? ???????????? ????????? (28???, 47.4%), 30??? ?????? ???????????? ??????(29???, 49.2%), ???????????? ??? ?????????????????? ?????????(36???, 69.2%), ???????????? ?????? ?????? ?????? ??????(15???, 25.4%)?????? ????????? ????????? ?????? ???????????? ???????????????. ????????? ?????? ????????? ??????, ????????? ??????????????? ??????, ?????? ??? ??????????????? ????????? ????????? ?????? ????????? ?????? ?????? ??????, ?????? ????????? ????????? ????????? ????????? ????????? ????????? ?????? ????????? ?????????????????? ????????? ????????? ????????? ?????? ????????? ???????????? ??????????????? ??????????????? ?????? ????????? ???????????? ????????? ??? ?????? ?????????. This study was conducted to identify the status of studies on multilevel analysis concerning Koreans’ health, validate the assessment of risk of bias to assess the quality of journals, and provide a basis for practicing multilevel analysis for future public health. We analyzed a total of 59 studies published from January 1, 2000 to December 31, 2014 in domestic and international journal databases. The findings include the following. First, only 20.9% of total had statistical significance. Second, the regional reputation was the most frequently used variable with a statistical significance on health. Third, 34.4% were on health behaviour, knowledge or attitude. Fourth, 4 studies (6.7%) were identified with a potential risk of bias caused by individual variables. There were journals with missing in ICC (28, 47.4%), the use of regional sample under 30 (29, 49.2%), missing in a correlation analysis between regional variables (36, 69.2%), and missing in basic information on region data (15, 25.4%). These would cause the risk of bias. It will be a useful analytical tool to understand the causes of health when executing reliable data, using diverse regional variables, doing pretest to prevent risk of bias such as correlation analysis between regional variables and specifying mechanisms of health affected by regional characteristics