Pattern Mixture Modeling: An Application in Anti Diabetes Drug Therapy on Serum Creatinine in Type 2 Diabetes Patients

The handling of missing data is the difficult problem in longitudinal data analysis. There are broad ranges of methods to handle missing observation in longitudinal data. The comparison of different missing observation handling techniques by Last Observation Carry Forward (LOCF), EM algorithm, Missing at Random (MAR) and Missing not at Random (MNAR) with pattern mixture modeling has been applied on clinical trial data set, studying the effect of drug treatment of metformin with pioglitazone respect to pioglitazone with gliclazide in type 2 diabetes patients for three follow up. It has been found by using the MAR and MNAR approach, that pioglitazone with metformin is more effective to reduce the serum creatinine as compared to pioglitazone with gliclazide.