OBJECTIVES

: Canagliflozin, a sodium-glucose cotransporter 2 inhibitor indicated for lowering glucose, has been increasingly used in diabetes because of its beneficial effect on cardiovascular and renal outcomes. However, clinical trial data have documented an increased risk of lower extremity amputations (LEA) associated with canagliflozin. Regardless of its severity, it remains unclear the underlying causes of LEA in patients taking canagliflozin and which patient characteristics increase the risk of LEA. The present study aimed to apply machine learning to predict LEA among diabetes patients treated with canagliflozin, and to determine the factors associated with LEA.

METHODS

: Using claims data from a 5% random sample of Medicare beneficiaries, we identified 13,904 diabetes individuals initiating canagliflozin between April 1, 2013 and December 31, 2016. The samples were split randomly and equally into training and validation sets. We identified 41 predictor candidates using information from the year prior to canagliflozin initiation, and applied four machine learning approaches (elastic net, least absolute shrinkage and selection operator (LASSO), gradient boosting machine and random forests) to predict LEA risk after canagliflozin initiation.

RESULTS

: The incidence rate of LEA was 0.57% over a median 1.5-year follow up. Among the four models applied, LASSO produced the best prediction and yielded a C- statistic of 0.81 (95%CI: 0.76, 0.86). In individuals in the top 5% of the risk score, the actual incidence rate of LEA was as high as 3.74%. Among the 16 important factors selected by LASSO, history of LEA [adjusted odds ratio (aOR): 33.6 (13.8, 81.9)] and loop diuretic use (aOR: 3.6 (1.8,7.3)) had the strongest associations with incidence of LEA.

CONCLUSIONS

: Our machine learning model efficiently predicted the risk of LEA among diabetes patients undergoing canagliflozin treatment. The risk score may support optimized treatment decisions for individual patients and thus improve the health outcomes of diabetes patients.