Modeling Chronic Kidney Disease in Type 2 Diabetes Mellitus: A Systematic Literature Review of Models, Data Sources, and Derivation Cohorts.

The clinical effects of new treatments and their costs are often evaluated over a longer time frame than is possible in clinical trials by using computer simulation models. As new treatments are becoming available to treat chronic kidney disease, including in patients with type 2 diabetes, chronic kidney disease models may be used to inform clinical and economic decisions regarding these new treatment options. In the present study, we identified 49 published simulation models of chronic kidney disease used in populations with type 2 diabetes, and reviewed their structures and the data sources they used. The models focused mostly on disease states and outcomes associated with albuminuria (a condition in which the protein albumin is found in the urine) and end-stage kidney disease. Model structures with five disease states, including a kidney disease-free state, three kidney disease states, and death, were the most common. Relatively few models used glomerular filtration rates (a common measure of kidney function) or captured the possibility of an improvement in chronic kidney disease. Important data sources for many models were patient registries, cohort studies, and clinical trials, most conducted several decades ago in high-income countries with a high proportion of White participants. Several models developed in the past 5 years, particularly for Asian countries, instead relied largely or exclusively on country-specific data. In parallel, several individual patient simulations were recently developed from large outcomes trials for new treatments, including from trial subgroups covering specific geographical settings or ethnicities, shortly after trial publication.