Binary Regression With a Misclassified Response Variable in Diabetes Data

Objectives: The categorical data analysis is very important in statistics and medical sciences. When the binary response variable is misclassified, the results of fitting the model will be biased in estimating adjusted odds ratios.  The present study aimed to use a method to detect and correct misclassification error in the response variable of Type 2 Diabetes Mellitus (T2DM), applying binary logistic regression.  Methods: Data from the Diabetes Screening test in the Health Center of Zahedan City, Iran, were explored.  It included 819 Iranian adults with a binary response variable (T2DM). By a new method, the misclassification parameters and the estimated parameters in logistic regression were validated. Statistical analysis was performed using SAS, and P<0.05 were considered as statistically significant. Results are presented as Odds Ratio (OR) and 95% Confidence Interval (CI).  Results: Increased age (OR=1.04, 95% CI=1.02-1.06), hypertension (OR=3.06, 95% CI=1.80-5.21), and obesity (OR=1.99, 95% CI=1.26-3.15), all elevated the odds of T2DM.  Discussion: The method provided adjusting for bias due to misclassification in logistic regression, and using it is recommended.