Medicaid Managed Care: RESEARCH DESIGN AND METHODS
The study population includes patients enrolled in a Medicaid MCO who have at least one record of a paid prescription claim for metformin or a TZD between January 15, 2000 and June 15, 2002. Enrollment in Medicaid was continuous from receipt of the study drug through the end of the study timeframe. None of the Medicaid plans had tiered copay designs charging different amounts for brand and generic drugs during the study timeframe.
Patients were considered if they had both pharmacy and medical claims and at least one paid claim for metformin or a TZD. Those with claims for both metformin and a TZD during the study timeframe were excluded. Patients with a claim for a study drug prior to June 15, 2000 were excluded from the data analysis to maximize the chance that the study population would be composed of new utilizers of either medication. Additionally, inclusion required the presence of a diagnosis of diabetes mellitus (Actos tabletes is used to treat type II of diabetes) (ICD-9-CM 250.xx). The effective study population hence consists of all diabetic patients with new claims for TZDs or metformin between June 15, 2000 and June 15,2002. The study drug that was initiated was labeled the index drug. The date of the first prescription for the index drug was labeled the index date.
Variables that were analyzed included age, gender and race. The patient’s residential setting was defined as urban or nonurban. We also included a variable denoting the presence of common diabetes comorbidities: hypertension (ICD-9-CM 401.xx, 402.xx, 403.xx, 404.xx, 405.xx, 997.91), coronary heart disease (ICD-9-CM 410.xx, 41 l.xx, 412.xx, 413.xx, 414.xx), hypercholesterolemia (ICD-9-CM 272.0x) and stroke (ICD-9-CM 430.xx-438.xx). To account for the severity of diabetes (is used to control blood glucose levels in type 2 diabetes), occurrence of the following diabetes complications that occurred before initiation of the index drug were included: nephropathy (ICD-9-CM 250.4x, 58l.xx, 582.xx, 583.8x), neuropathy (ICD-9-CM 250.6x, 337.lx, 344.61, 354.xx, 355.xx, 356.9x, 357.2x, 357.9x, 536.3x, 593.7x, 596.4x, 596.5x, 607.84, 713.4x) or retinopathy (ICD-9-CM 250.5x, 362.0x, 362.lx). Similarly, we identified any use of oral (alphaglucosidase inhibitors, sulfonylureas and insulin secretagogues) and insulin prior to the index date.
The unit of analysis was the patient. Univariate procedure was performed to provide descriptive statistics and distributions of the demographics on the study population. The association between index drug (either metformin or TZDs) and other variables was explored using bivariate analysis, Chi-squared and t tests. Multivariate logistic regression analysis models were built to assess the independent effect of combined variables on the likelihood of being prescribed TZD or metformin. Predictor variables included race, gender, age, residential setting, preexisting comorbidities and diabetes complications, other oral diabetes drugs use and insulin use. All statistical analyses were performed with SAS version 8.2 software (SAS Institute, Cary, NC).