Contrary to one of our hypotheses, we did not find differences in the odds of taking an intensified medication regimen at follow-up for English speakers and Spanish speakers with A1c
8% at baseline. However, we did find that among patients with suboptimal glycemic control, higher patient income was associated with an increase in the odds of intensification at follow-up. These results differ from a prior TRIAD analysis that found a difference of 3 percentage points in intensification rates between patients with annual incomes <$15,000 versus >$75,000.13
That study limited the analytic sample to patients who completed a 2-year follow-up survey. The current study had nearly twice the sample size of the previous study and a much larger proportion of low income participants. As shown in a supplementary online appendix
comparing results of the two studies (Supplementary Online Appendix
), the current analysis found that the likelihood of intensification was 13 percentage points higher among patients with annual incomes of >$75,000 compared to those with annual incomes <$15,000. The marked difference in results between two studies from the same dataset underlines the importance of 1) clearly identifying the preferred analytic sample to test a given hypothesis, and 2) considering how the application of exclusion criteria will affect this sample.
There are several potential explanations for the association between income and intensification seen in the current study. Providers may hold unconscious stereotypes about low-income patients, and/or be less likely to engage in patient-centered communication with low-income patients.19–21
These or other factors may dissuade providers from discussing an intensified medication regimen with low-income diabetic patients. Alternatively, because low-income patients tend to express more frequent concerns, symptoms, and psychosocial stressors compared to higher-income patients, less time may be available to discuss chronic disease management.19,22–24
A recent study of observed visits for diabetic patients in poor control found that each additional concern raised by the patient was associated with a 50% lower likelihood of a change in antiglycemic medication.25
Cost is often an important issue when medications are newly prescribed, and low-income patients are less able to afford copayments for additional medication/s beyond their existing regimen.26–28
Prior work using TRIAD data has shown that intensification is associated with an average decrease in A1c of 0.49% among patients with poor glycemic control.13
Furthermore, this effect was primarily observed among patients with incomes <$15,000.15
Together with the findings in the current study, this suggests that the specific low-income patients who benefit the most from intensification are the least likely to actually be intensified.
To our knowledge, this is the first study comparing intensification of anti-glycemic medication regimens among English-speaking and Spanish-speaking patients with suboptimal glycemic control. Our study included relatively few Spanish speakers. However, examination of the observed confidence intervals suggests that markedly lower odds of intensification for Spanish speakers as compared to English speakers are unlikely, and the importance of any such differences on language-based variation in glycemic control are probably minimal. Using the approach published by Smith and Bates to extrapolate from our observed confidence intervals for the primary language variable,29
we estimate that the likelihood of the “true” OR for intensification among Spanish speakers being less than that for English speakers is approximately 5%. Further, the likelihood of the “true” OR for Spanish speakers being <0.75 as compared to English speakers is less than 1%. In other words, our results suggest that differences in factors other than medication intensification are the most likely explanations of poor glycemic control among Spanish speakers in managed care.
One explanation for the absence of an association between intensification and primary language, as well as the absence of differences in earlier analyses examining other processes of care,4
may be that managed care plans providing comprehensive care to large Latino populations invest in high-quality interpreter services despite the lack of reimbursement from third-party payers.30
Providers who speak Spanish are also more likely to work in managed care settings and care for patients with private insurance than to care for Medicaid or uninsured patients.31
Our study has several strengths, including its longitudinal design, large, randomly sampled cohort, and the inclusion of participants from 10 separate health plans. However, there are also limitations. First, we were unable to measure patient adherence to their anti-glycemic medication regimens, which may have resulted in some misclassification of intensification status if patients missed refills for the entire baseline or follow-up study window. Second, we did not analyze continuous claims data between baseline and follow-up and may have missed some patients for whom medication regimens were successfully intensified but subsequently deintensified before the follow-up measurement. Third, we were unable to measure intensification with behavioral strategies or over-the-counter herbal therapies. Finally, we could not measure the impact of intensification on long-term outcomes such as diabetes complications.
In summary, we found that low-income patients with suboptimal glycemic control are less likely to be on an intensified regimen of anti-glycemic medications at follow-up compared to higher-income patients. However, we did not find differences in medication intensification between Spanish speakers and English speakers in poor control, suggesting that other factors may be responsible for persistent language-based disparities in glycemic control. Future studies are needed to explain the reduced rate of intensification among low-income patients in managed care settings.