Behavioral Risk Score for Predicting Well-Controlled HbA1c Level in Diabetes Type 2 Patients

Abstract

Background: Diabetes Type 2 is chronic disease that can progress from simple hyperglycemia to severe complications. Many behavioral risks have been discovered for blood sugar prediction. Objective: To develop a simple behavioral risk scoring to predict well-controlled HbA1c level in diabetes type 2 patients. Materials and Methods: A total of 140 diabetes type 2 patients were recruited. Patients were interviewed about behavioral factors affecting blood sugar in three months retrospectively. To develop the risk score, risk indicators measured at the time of recruitment were built by logistic regression. Regression coefficients were transformed into item scores and added up to a total score. A risk scoring scheme was developed from behavioral predictors: eating desserts and soft drinks, regular exercise and strict medication intake. The scoring scheme was applied in bootstrap internal validity test to test the model performance. Results: The scheme explained, by area under the receiver operating characteristic curve (AuROC), 91.6% (95% CI 0.87 to 0.96) of being good diabetic control (HbA1c ≤7%) with good calibration (Hosmer-Lemeshow χ²=3.61; p=0.61). The likelihood ratio of being good diabetic control (scores greater than or equal to 1) and poor diabetic control (score lower than 1) were 3.83 (95% CI 2.69 to 5.46) and 0.11 (95% CI 0.05 to 0.21), respectively. When applied in bootstrap internal validity test, the score showed good performance with AuROC 88.7% (95% CI 0.81 to 0.93). Conclusion: A simple and non-invasive scoring scheme of three predictors provides good prediction indices for being good and poor diabetic control patients. This scheme may help clinicians in order to take further appropriate action for diabetic control. Keywords: Diabetes type 2; HbA1c controlling level; Prediction; Risk scoring