Multistate Models for the Analysis of Time to Type II Chronic Diabetic Complications in Debre Markos Referral Hospital, Northwest Ethiopia

Abstract

Introduction: Diabetes is a chronic, non-communicable disease characterized by elevated blood glucose levels. The purpose of this study was to jointly model the transition of diabetic patients in a series of clinical states and to assess the relationship between each state and different patient characteristics. Methods: A hospital-based retrospective study was conducted on 524 patients with type II diabetes, aged 18 years or older, who attended their medication between January 1, 2005, and December 31, 2017. Multistate models with different assumptions were considered to explore the effects of different prognostic factors on the transition intensity of type II diabetes mellitus patients. Results: During a median follow-up time of 7.4 years (Inter-Quartile Range=4.01), 54.8% of diabetic patients developed either microvascular or macrovascular complications, and 10.5% of them experienced both microand macrocomplications, and 16.66% of diabetes patients died. The assumption Markov was assessed by using the likelihood ratio test showed that Markov assumption was not held just for the transition. The transition rate of patients from the macrovascular state to the death state was affected by the residence of the patients (P=0.05) and age at diagnosis (p=0.01). The transition rates of patients with microvascular complications to death were significantly affected by baseline triglyceride level (P<0.001), age at first diagnosis (P=0.01), baseline glucose level (P=0.03, and baseline serum creatinine level (P=0.04). Conclusion: The semi-Markov model fitted the data well and could be used as a convenient model for the analysis of time to diabetes-related complications or death.

Keywords

• Diabetes mellitus
• Vascular complications
• Multistate models