Metabolic Testing for Adults in a State Medicaid Program Receiving Antipsychotics

Abstract

Importance Medicaid quality indicators track diabetes mellitus and cardiovascular disease screening in adults receiving antipsychotics and/or those with serious mental illness. Objective To inform performance improvement interventions by evaluating the relative importance of patient, prescriber, and practice factors affecting metabolic testing. Design, Setting, and Participants A retrospective cohort study was conducted using Missouri Medicaid administrative claims data (January 1, 2010, to December 31, 2012) linked with prescriber market data. The analysis included 9316 adults (age, 18-64 years) who were starting antipsychotic medication. Secondary analysis included the subset of adults (n = 1813) for whom prescriber knowledge, attitudes, and behavior survey data were available. Generalized estimating equations were performed to identify factors associated with failure to receive annual testing during antipsychotic treatment (adjusted odds ratio [OR], <1 favor testing). Data analysis was performed from October 1, 2014, to February 18, 2016. Exposure Oral second-generation antipsychotics. Main Outcomes and Measures A medical claim for glucose or lipid testing occurring within 180 days before and after the antipsychotic prescription claim. Results The 9317 patients (mean [SD] age, 37.6 [12.0] years) initiated antipsychotic medication in a variety of prescriber specialty-settings: 24.3%, community mental health center (CMHC); 27.6%, non-CMHC behavioral health; 24.3%, primary care practitioners; and 23.8%, other/unknown. Annual testing rates were 79.6% for glucose and 41.2% for lipids. Failure to test glucose and lipids was most strongly associated with patient factors and health care utilization. To illustrate by using findings from glucose modeling (reported as adjusted OR [95% CI]), lower failure to receive testing was associated with older age (40-49 vs 18-29 years; 0.64 [0.55-0.74]), diagnosis of schizophrenia or bipolar disorder (0.55 [0.44-0.67]), cardiometabolic comorbidity (dyslipidemia, 0.28 [0.22-0.37]), hypertension (0.59 [0.50-0.69]), and greater outpatient utilization (>6 encounters vs none; 0.33 [0.28-0.39]). Analysis incorporating prescriber practice information found lower failure to receive glucose testing if the patient received care at a CMHC (0.74 [0.64-0.85]) or if the initiating prescriber was a primary care practitioner (0.81 [0.66-1.00]). However, the initiating prescriber specialty-setting was not associated with lipid testing. Conclusions and Relevance Compared with prior reports, progress has been made to improve diabetes screening, but lipid screening remains particularly underutilized. Medicaid performance improvement initiatives should target all prescriber settings and not just behavioral health.