Multilevel models approach on longitudinal studies for disease data

A study of a disease in an area is periodically measured. In this study, time is nested within the individual so it derives a data with nested structure. Ordinary regression model is not appropriate used because it does not consider the existence of nesting effects. Multilevel model usually used for hierarchical data, because it is not ignore the effect of the data structure. According to Sneijder and Bosker, nested data can be viewed as hierarchical data, so it can be analyzed using multilevel model. Diarrhea case in an area is affected by clean water, hand washing habits, healthy latrines and healthy house. This data is measured periodically from 2012 to 2016. Multilevel model is applied to diarrhea case data in Bandung city which has nested structure. The results obtained from the analysis is that the data used has a hierarchy structure so that it can be analyzed using a multilevel model. In addition, the random intercept model used is very fit with three significant variables. In general, the result for this case is that the occurrence of cases of diarrhea in Bandung influenced by clean water and hand wash habits with the variation between districts.