Small area variation and geographic and patient-specific determinants of metabolic testing in antipsychotic users

The American Diabetes Association and American Psychiatric Association recommend metabolic monitoring for all patients using second-generation antipsychotic (SGA) drugs. We estimated glucose and lipid testing rates among SGA-users from three state Medicaid programs and investigated small area variation and patient and geographic determinants of testing.

A retrospective new-user cohort study using Medicaid claims data from California, Missouri, and Oregon was conducted among 30,563 patients in 207 counties starting SGA medication September 2004-December 2005. Adjusted odds ratios for state, county, and patient factors associated with testing were calculated from multivariable hierarchical logistic regression models.

Mean 6-month testing rates were 51.6% (glucose) and 26.2% (lipids). Screening rates were positively associated with the number of Type 2 diabetes risk factors (RF) present: glucose -39% (0 RF) to 82% (5 RF); lipids -13% (0 RF) to 66% (5 RF). A four-fold difference in glucose testing rates (21-85%) and a greater than six-fold difference in lipid testing rates (0-62%) were observed between counties. In the adjusted regression models, age, cardiometabolic co-morbidity (diabetes, dyslipidemia), serious mental illness, persistent use of SGAs, and frequency of non-psychiatric medical office visits were significant determinants of glucose and lipid testing. Lipid testing was more likely for children and adults in California, as was glucose testing for children.

Glucose and lipid testing among SGA-users varied significantly between states, counties, and by patient characteristics. More effort is needed to understand provider and system reasons for testing disparities in order to inform risk management quality improvement interventions.