PREDICTORS OF READMISSION AMONGST ADULT PATIENTS WITH DIABETES MELLITUS

Abstract

Readmission, although preventable, is common in patients with diabetes. The overall aim of this thesis is to identify associated predictors to help mitigate unplanned readmission. To address this, three studies were conducted. First, a systematic review with meta-analysis was performed to identify predictors of 30-day readmission. Next, a 30-day readmission model was developed and validated using electronic records from a local tertiary hospital. The same data was used to identify predictors of the number of one-year readmissions. The systematic review cum metaanalysis identified ten significant predictors of 30-day readmission, including age, length of stay (LOS), heart failure and renal disease. In the second study, predictors of 30-day readmission included LOS, peripheral vascular disease (PVD), ischaemic heart disease and number of drugs. LOS, PVD and number of drugs, together with age, number of emergency department visits, renal disease and residential status were found to predict number of readmissions within a year.