New Roche study uses real-world data to predict diabetes-related chronic kidney disease
8 January 2019
- Chronic kidney disease is one of the most severe secondary complications related to diabetes1
- Study demonstrates new predictive model based on real-world data to offer enhanced accuracy compared to traditional methods in prognosing the risk of this diabetes-related long-term complication
- The research, published in Nature Medicine, was the result of a collaboration between Roche, IBM Watson Health, and partners2
Roche (SIX: RO, ROG; OTCQX: RHHBY) today announced promising results from a new study which supports the use of real-world data, including electronic medical record data, to predict the risk of chronic kidney disease (CKD) in individuals with diabetes. The study, published in the journal Nature Medicine [https://rdcu.be/bfKPU], showed that the newly developed predictive model based on real-word data was more accurate than previously published predictive algorithms with a baseline at 0.7937 (95% confidence interval [CI] 0.790, 0.797).3
Chronic kidney disease is one of the most severe diabetes-related complications, which is why the research team first targeted this medically, socially, and economically important application.4 It is characterized by the progressive loss of the kidney function, beginning with a decline in the glomerular filtration rate and/or albuminuria, eventually resulting in end-stage renal disease. Experts estimate approximately 10% of people with diabetes to be affected by this complication within three years following the initial diabetes diagnosis. This severe complication often requires dialysis or renal transplant therapy. Early recognition of the CKD risk followed by the appropriate interventions can slow down the progression or, possibly, even prevent the onset of this diabetes-related complication. The researchers from Roche and IBM used real-world clinical data originating from more than 600,000 people with type 1 and type 2 diabetes to develop a new predictive model to help identify the risks for developing CKD within the first three years after the initial diabetes diagnosis.5
While conventional diagnostic procedures, therapeutic recommendations, and medical risk assessments are frequently based on clinical trial data, a plethora of real-world medical data exists in clinics and medical doctors’ offices which could potentially help evaluate patient’s risk level and tailor the most appropriate treatment. The accuracy of the new predictive model developed by Roche and IBM was compared with certain clinical trial data and showed to be better than the traditional methods of estimating the risk of newly diagnosed patients developing chronic kidney disease.5
“Our research team from Roche and IBM successfully invented this novel predictive model to identify those people with diabetes who are at a high risk of developing CKD in the near future,” says Wolfgang Petrich, Ph.D., Research Lead at Roche Diabetes Care, Mannheim, Germany.
A direct comparison between this new predictive model and already existing, similar models, which were solely derived from clinical trial data, revealed that the model developed by our team outperformed tested methods in a one-to-one comparison on real world data as well as for study cohorts selected a posteriori.”
“This study demonstrates the growing importance of real-world data and predictive analytics in diabetes care,” said Dr. Mark Davies, Chief Medical Officer (EMEA), IBM. “There is a growing need to improve screening performance and the decision-making processes in diabetes care, and this new data suggest that real-world data and analytics can be applied to help in early recognition of risk of CKD.”
A follow-up evaluation was performed in co-operation between Roche and the Indiana Biosciences Research Institute (IBRI), the Regenstrief Institute, and Eli Lilly and Company. This evaluation was able to confirm all prior findings, when applying the new model to independent real-world data originating from almost 100,000 additional people with diabetes. Although additional research is suggested by the research team, the initial results indicate that real-world data-based predictive models may provide an efficient tool for improved, personalised healthcare aiding clinical decision making, while offering the potential for a timely mitigation of the impact and severity of diabetes-related complications.
1,3,4 - Diabetes and Chronic Kidney Disease (2016). Retrieved from https://www.kidney.org/news/newsroom/factsheets/Diabetes-And-CKD
2 - Eli Lilly, Indiana Biosciences Research Institute, Regenstrief Institute
3,5 - S. Ravizza et al. Predicting the risk of early chronic kidney disease in diabetes patients using real world data. Nature Medicine (Nature Medicine | VOL 25 | JANUARY 2019 | 57–59); full text access: https://rdcu.be/bfKPU