Monitoring patients with diabetes mellitus: An application of the probit model using managed care claims data

The primary objective of this study was to estimate the likelihood of the use of either a glycosylated hemoglobin (HbA(1c)) test or an eye examination, or both, among a cohort of patients diagnosed with diabetes mellitus, A secondary objective was to provide a step-by-step discussion of the applicability of an econometric model to managed care organizations, The study used medical and pharmacy claims data from a managed care organization for the calendar year 1995, A probit regression model was specified to estimate the probability of occurrence for either an HbA(1c) test or an eye examination among patients with insulin-dependent, non-insulin dependent, or atypical/unclassified diabetes. Data were available only for patients under 65 years of age due to data truncation for patients covered by Medicare, resulting in a study sample size of 6,841. Results indicate that age, presence of hypertension, hyperlipidemia, multiple cardiovascular comorbidities, ophthalmic disease, and combinations of multiple commonly observed comorbidities were positively correlated with the probability of either HbA(1c) testing or eye examination. Gender and the type of benefit plan were not statistically significant as predictors of disease monitoring, A total of 1,860 patients with diabetes mellitus were predicted by the model to have undergone one of the two monitoring procedures; but in actuality, these patients were not monitored in 1995, They could be considered as high-risk patients who were not getting recommended monitoring. The probit model shows a predictive power of 64.48%.