Diabetes Drug Adherence Lower for African Americans

COLUMBUS, Ohio, Aug. 4 -- One reason African Americans with type 2 diabetes tend to have comparatively poor clinical outcomes may be cultural, investigators here reported.

African-Americans with type 2 diabetes appear to be less likely than whites to take prescribed medications, reported Rahul A. Shenolikar, M.S., of Ohio State University, and colleagues, in the July issue of the Journal of the National Medical Association?.

In a retrospective study comparing medication compliance among Medicaid-insured patients who were given first-time prescriptions for oral antidiabetic agents, African Americans had a 12% lower adherence rate than whites, they found.

"That's an unacceptable difference, particularly because African Americans tend to have higher rates of diabetes and disease-related complications," said Rajesh Balkrishnan, Ph.D., a professor of pharmacy, a co-author.

"Adherence rates for these types of medications should be better than 90% regardless of who takes them," he added. "Such low rates of adherence may be related to lower socioeconomic status and to lower levels of education."

A study published by the Institute of Medicine in 2002 found that African Americans, Hispanics, and Native Americans have a burden of illness and mortality from diabetes that are between 50% and 100% higher than among whites. Other studies haves shown that African Americans with diabetes have worse glycemic control than others, Shenolikar and colleagues wrote.

To see whether the disparities in outcomes could be due in part to medication compliance, Shenolikar and colleagues here and at Wake Forest in Winston-Salem, N.C., looked at Medicaid-insured patients with a new, first-time prescription for oral diabetes medications.

The cohort included 1,527 African Americans, 1,128 white patients, and 514 patients of other racial/ethnic backgrounds. They determined medication adherence by looked at the ratio between prescriptions and refills, working under the assumption that a refill implied that the patient had taken the drug prescribed. The outcome was expressed as a medication possession ratio, which was calculated as the number of days that the patient possessed a prescription divided by the number of days between refills.

The authors used multivariate regression analyses to determine the difference in adherence rates adjusting for other covariates, including demographic characteristics (such as age and gender), clinical confounders (such as use of healthcare services over the previous 12 months), type of therapy, total number of medications, and number of comorbidities.

They found that whites had a significantly higher adherence rates than blacks, at 0.59 (standard deviation, 0.31), compared with 0.54 (SD 0.31, P<0.05). That is, whites took their medication 59% of the time, compared with African Americans, who took their medications 54% of the time.

In multivariate analyses, the difference translated into a 12% lower compliance rate among African American after adjustment for covariates.

The investigators also found that compliance was dependent on the type of medication prescribed, with those assigned to Glucophage (metformin) having a 62% lower compliance rate than patients assigned to sulfonylureas (P<0.05), and a 63% lower rate compared with patients assigned to thiazolidinediones (P value not given).

The investigators called for research into the factors that affect medication compliance in patients with type 2 diabetes.

"Many commercial insurers pay for educators to teach patients the importance of taking their medications as prescribed," Dr. Balkrishnan said. "Medicaid needs to do the same thing. While it invests a lot of money in providing services, it does little to educate its recipients about those services and how to use them. People need to understand the importance of taking their medications."

The authors noted a limitation of the study. "We excluded the elderly, non-continuously eligible, those on combination therapies and those who were institutionalized," they wrote. "Due to such exclusions, this study may not be generalizable to a large insured population."