Abstract
Objective
Given persistent racial/ethnic differences in type 2 diabetes outcomes and the lasting benefits conferred by early glycemic control, we examined racial/ethnic differences in diabetes medication initiation during the year following diagnosis.
Methods
Among adults newly diagnosed with type 2 diabetes (2005–2016), we examined how glucose-lowering medication initiation differed by race/ethnicity during the year following diagnosis. We specified modified Poisson regression models to estimate the association between race/ethnicity and medication initiation in the entire cohort and within subpopulations defined by HbA1c, BMI, age at diagnosis, comorbidity, and neighborhood deprivation index (a census tract-level socioeconomic indicator).
Results
Among the 77,199 newly diagnosed individuals, 47% started a diabetes medication within 12 months of diagnosis. The prevalence of medication initiation ranged from 32% among Chinese individuals to 58% among individuals of Other/Unknown races/ethnicities. Compared to White individuals, medication initiation was less likely among Chinese (relative risk: 0.78 (95% confidence interval 0.72, 0.84)) and Japanese (0.82 (0.75, 0.90)) individuals, but was more likely among Hispanic/Latinx (1.27 (1.24, 1.30)), African American (1.14 (1.11, 1.17)), other Asian (1.13 (1.08, 1.18)), South Asian (1.10 (1.04, 1.17)), Other/Unknown (1.31 (1.24, 1.39)), American Indian or Alaska Native (1.11 (1.04, 1.18)), and Native Hawaiian/Pacific Islander (1.28 (1.19, 1.37)) individuals. Racial/ethnic differences dissipated among individuals with higher HbA1c values.
Conclusions
Initiation of glucose-lowering treatment during the year following type 2 diabetes diagnosis differed markedly by race/ethnicity, particularly for those with lower HbA1c values. Future research should examine how patient preferences, provider implicit bias, and shared decision-making contribute to these early treatment differences.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
INTRODUCTION
The prevalence of type 2 diabetes among US adults varies significantly by race/ethnicity, with the highest rates of the disease seen among racial/ethnic minorities.1,2,3 Beyond differences in type 2 diabetes prevalence, the risk of diabetes-related complications also varies by race/ethnicity, with non-White adults significantly more likely to experience microvascular complications like retinopathy and nephropathy.1,4,5,6 Mounting evidence demonstrates that tight glycemic control during the early period following diabetes diagnosis confers lasting microvascular benefits.7,8,9 Given the importance of pharmacologic diabetes treatment in achieving early glycemic control, racial/ethnic differences in the timely initiation of type 2 diabetes medications may contribute to existing disparities in long-term type 2 diabetes health outcomes.1,2,10,11
Prior studies have examined race/ethnicity-based differences in type 2 diabetes treatment; however, few have focused on differences in type 2 diabetes treatment among newly diagnosed adults.12,13,14,15 A 2009 study of US adults found no differences in initiation of oral medications (metformin and sulfonylurea) between African American and non-Hispanic White adults, but in unadjusted analyses did note higher rates of sulfonylurea initiation among African American adults.14 A 2020 study of adults in UK noted that South Asian and Black adults initiated non-insulin diabetes monotherapy sooner than White adults.15 In both studies, comparisons were restricted to select racial/ethnic groups (African American vs. White, South Asian vs. Black vs. White), limiting the generalizability of the research findings to more diverse patient populations.
In this study, we examined racial/ethnic differences in diabetes medication initiation among newly diagnosed adults treated in a large, integrated healthcare delivery system, where self-reported, granular race/ethnicity data were available for nearly all individuals.
METHODS
We identified an incidence cohort of all adult (age≥21 years) Kaiser Permanente Northern California (KPNC) members newly diagnosed with type 2 diabetes from 2005 to 2016 using a previously validated algorithm.16 In brief, a diagnosis of diabetes was determined using diagnostic coding data (ICD-9 250) from outpatient or inpatient visits, the presence of two or more diabetes-consistent lab results, or a prescription for insulin or an oral hypoglycemic medication (excluding those associated with non-diabetes conditions like polycystic ovary syndrome).16 We required continuous KPNC membership for 24 months before the diagnosis (wash-out period) and 18 months after diagnosis (follow-up period). All individuals had at least one HbA1c value and BMI measurement during the wash-out period. If multiple BMI and HbA1c values were available, the values closest to the date of diagnosis were used. We classified eligible individuals into 12 race/ethnicity groups using self-reported data collected at clinic visits, during health plan enrollment, from member surveys, or on intake for hospitalization (see Table 1).16 Early medication initiation was defined by the first dispensing of a diabetes medication (metformin, sulfonylureas, dipeptidyl peptidase-4 inhibitors (DPP-4), thiazolidinediones, meglitinides, alpha-glucosidase inhibitors, sodium-glucose cotransporter-2 inhibitors (SGLT-2), glucagon-like peptide 1 receptor agonists (GLP-1), or insulin) during the year following diagnosis. Of note, nearly all medications are filled internally and KPNC pharmacies dispense lower-cost generic diabetes medications as the default.17,18,19
To account for differences in data missingness, we estimated inverse probability weights from a logistic regression analysis predicting availability of complete BMI and HbA1c data; these models included quartic of age, race/ethnicity, sex, and all potential interactions between these variables. We included these weights in all analyses. We specified weighted modified Poisson regression models to estimate the relative risk (RR) of medication initiation between races/ethnicities.20 An individual’s primary KPNC medical care facility (i.e., facility where they received primary care) was included as a random effect to account for within-facility clustering.
Our conceptual framework is that race/ethnicity is an attribute present at birth, and, thus, we considered certain patient measures known to differ by race/ethnicity that also influence treatment initiation or access (e.g., age at T2D diagnosis, HbA1c, BMI, socioeconomic status) as mediators of medication initiation differences rather than confounders.3,21,22,23,24,25,26 Structural racism differentially impacts individuals’ access to healthy foods and safe spaces for physical activity, as well as their exposure to racial discrimination and stressors (e.g., violence, crime), and, thereby, contributes to established racial/ethnic differences in age at T2D diagnosis, obesity prevalence, and the degree of initial hyperglycemia.27,28,29 Based on this conceptual framework, inclusion of such mediators in our regression models would bias these models toward the null. Instead, we examined whether the racial/ethnic patterns of medication initiation varied across subpopulations defined by the following characteristics: age at diagnosis (<50 years, 50–69 years, ≥70 years); diagnosis HbA1c (<7.5% (58 mmol/mol), 7.5–8.9% (58–74 mmol/mol), ≥9% (75 mmol/mol)), BMI (<30, 30–34.9, 35–39.9, ≥40 kg/m2); comorbidity (defined by a Charlson comorbidity score of 0, 1, or ≥2)30; and socioeconomic status (defined by neighborhood deprivation index (NDI), a validated census tract-level indicator).31 Before we examined differences in medication initiation within each of the aforementioned subpopulations, we modeled the interactions between race/ethnicity and the subpopulations and found all interactions were statistically significant. We used Stata/MP 16.1 software for all analyses.
RESULTS
From a starting cohort of 144,677, there were 77,199 racially/ethnically diverse adults with newly diagnosed type 2 diabetes and available HbA1c and BMI values in our sample (Table 1). We excluded 67,478 individuals who did not have an HbA1c and BMI value during the wash-out period. After applying inverse probability weighting, included individuals’ age, sex, and race/ethnicity did not vary from individuals that were not included due to missing BMI or HbA1c. Overall, 47% of individuals initiated a glucose-lowering medication during the year following diagnosis, but there were significant differences in medication initiation between racial/ethnic groups. The prevalence of medication initiation was lowest among Chinese (32%) and Japanese (35%) individuals and highest among Hispanic/Latinx (56%) and individuals of Other/Unknown (58%) races/ethnicities (Table 1). Compared to White individuals (referent), medication initiation was less likely among Chinese (RR 0.78 (95% confidence interval 0.72, 0.84)) and Japanese individuals (0.82 (0.75, 0.90)), and more likely in Hispanic/Latinx (1.27 (1.24, 1.30)), African American (1.14 (1.11, 1.17)), other Asian (1.13 (1.08, 1.18)), South Asian (1.10 (1.04, 1.17)), Other/Unknown (1.31 (1.24, 1.39)), American Indian or Alaska Native (1.11 (1.04, 1.18)), and Native Hawaiian/Pacific Islander (1.28 (1.19, 1.37)) races/ethnicities (Fig. 1). There was no significant difference between White individuals and Filipinos (1.03 (0.99–1.06)) or Vietnamese individuals (1.00 (0.86, 1.16)).
In examining differences in medication initiation between subgroups defined by age, index HbA1c, BMI, comorbidity, and NDI, between-subpopulation differences were only consistently noted between subpopulations defined by index HbA1c value (Fig. 2). The findings for individuals with an HbA1c<7.5% (58 mmol/mol) mirrored the results for the entire population, with Chinese and Japanese individuals less likely to start a medication, and all groups, other than Filipinos and Vietnamese, more likely to initiate medications. Overall, racial/ethnic differences were mitigated by higher HbA1c levels at diagnosis. For individuals with a diagnosis HbA1c of 7.5–8.9% (58–74 mmol/mol), the only significant difference was a lower likelihood of initiation among African American individuals (0.95 (0.91, 0.99)). For people with a HbA1c≥9% (75 mmol/mol) at diagnosis, the only significant finding was a greater likelihood of initiation in the American Indian or Alaska Native group (1.05 (1.04, 1.05)).
The racial/ethnic patterns of medication initiation seen in the overall population were relatively preserved across the subpopulations defined by individuals’ age at diagnosis, BMI, comorbidity index, and NDI level (Supplemental Figures).
DISCUSSION
Among individuals with a known self-reported race/ethnicity and newly diagnosed type 2 diabetes, fewer than half of individuals started a diabetes medication during the year following diagnosis. Hispanic/Latinx adults were the most likely to start diabetes medications, while Chinese and Japanese were the least likely to start medications. Between-group differences in medication initiation largely dissipated among subpopulations with higher HbA1c values at diagnosis.
Overall rates of medication initiation were lower than expected based on current treatment type 2 diabetes treatment guidelines and the demonstrated benefits of early pharmacologic treatment (particularly with metformin).10,11,32 It is difficult to compare our findings with prior work given the differences in race/ethnic groups compared and our decision to not adjust for clinical and demographic characteristics. Still, consistent with the findings from Mathur et al., we did note differences in treatment initiation by race/ethnicity, with South Asians and African American adults more likely to start pharmacologic treatment early following type 2 diabetes diagnosis.15
Many factors that may increase the aggressiveness of initial type 2 diabetes treatment are closely associated with patients’ race/ethnicity, including age at diabetes onset, level of initial hyperglycemia, BMI, and comorbidity burden.3,21,22,23,24,25,26 In our cohort, and consistent with prior work, Hispanic/Latinx individuals were diagnosed at a younger age, had higher diagnosis HbA1c values, and had a greater prevalence of obesity.21,23,33 In contrast, Chinese and Japanese individuals were older at diagnosis, had less severe hyperglycemia, and had lower BMI values.34 Given these differences, some of the observed between-group differences in treatment may reflect clinicians’ consideration of patients’ clinical differences and risk for diabetes-related complications. However, the persistence of lower treatment likelihood among younger Chinese and Japanese individuals (age<50 years), a patient subpopulation at increased risk for disease-related complications, raises concerns that observed differences may, in part, reflect the undertreatment of certain individuals.35,36,37,38
We found that individuals with less severe initial hyperglycemia (HbA1c<7.5% (58 mmol/mol)) had greater racial/ethnic differences in early medication initiation. This finding suggests that, without the clear signal of a markedly elevated HbA1c value, patient and clinician factors may play a more significant role in treatment decisions and that these factors differ by racial/ethnic groups. For patients, previous literature supports racial/ethnic differences in patient perceptions of pharmacologic treatment (e.g., cultural attitudes toward Western medicine) and their openness and comfort with patient-provider shared decision-making (e.g., different value or deference is given to doctor’s recommendations).39 For clinicians, differences in their treatment of patients with less marked hyperglycemia may reflect different familiarity with treatment guidelines, varying practice styles, and assumptions, both founded and unfounded, regarding a patient’s capacity for type 2 diabetes self-management or treatment preferences. Prior research has established that healthcare providers experience similar levels of implicit racial/ethnic biases as the general population, and such biases are associated with differences in treatment-related decisions.40,41 For example, in one study, providers’ assessments of a patients’ likelihood of adhering to medical advice and engaging in behavior changes differed by patients’ race/ethnicity.42 In the case of diabetes medication decisions, subconscious perceptions that individuals from certain race/ethnic groups are more or less likely to enact substantial behavior changes or to refuse pharmacologic-based treatments could contribute to differences in medication initiation (both higher and lower likelihood of initiation) between groups of newly diagnosed adults.
Our results should be considered within the context of the study design. All included individuals were members of a single healthcare system, potentially limiting generalizability. The consistency in the direction of observed race/ethnic differences between subpopulations defined by the NDI (a neighborhood-level marker of socioeconomic status) may reflect the integrated care setting, where barriers to primary care and lower-cost diabetes medications are minimized. We lacked race/ethnicity data for 1.7% of our sample. We grouped all medications, as our focus was on treatment in general, not specific medications, although the initial agent was metformin for most. We only focused on medication dispensing and cannot comment on racial/ethnic differences in medication adherence or treatment intensification. However, prior work has demonstrated racial/ethnic differences in adherence to diabetes medications and treatment changes, with non-White individuals less likely to be adherent or have their treatments intensified.14,15,43,44 Finally, we did not specifically examine the impact of certain patient (e.g., health literacy, prior medication use) and provider (e.g., race concordance, years in practice) factors, so we can only speculate on the potential role of these factors in the observed differences.
During the year following type 2 diabetes diagnosis, medication initiation differed significantly by race/ethnicity. Future research on the post-diagnosis period should identify situations where treatment differences are more likely to occur (e.g., greater clinical uncertainty) and distinguish treatment differences that reflect individualized type 2 diabetes care (e.g., informed by patients’ differing needs, risks, and treatment preferences) from those that stem from the need for clinician education (e.g., recent updates to treatment guidelines), provider bias (implicit or explicit), or barriers to patient-provider shared decision-making (e.g., limited health literacy, language discordant providers, lack of provider cultural humility). Such research can contribute to ensuring that glucose-lowering medications are appropriately and equitably offered to all newly diagnosed patients.
Data Availability
The data generated and/or analyzed during the current study are not publicly available due to institutional policies but are available from the corresponding author on reasonable request and with the appropriate IRB approvals.
References
Spanakis, E.K. and S.H. Golden, Race/ethnic difference in diabetes and diabetic complications. Curr Diab Rep, 2013. 13(6): p. 814-23.
Rodriguez, J.E. and K.M. Campbell, Racial and Ethnic Disparities in Prevalence and Care of Patients With Type 2 Diabetes. Clin Diabetes, 2017. 35(1): p. 66-70.
Cheng, Y.J., et al., Prevalence of Diabetes by Race and Ethnicity in the United States, 2011-2016. JAMA, 2019. 322(24): p. 2389-2398.
Karter, A.J., et al., Ethnic disparities in diabetic complications in an insured population. JAMA, 2002. 287(19): p. 2519-27.
An, J., et al., Prevalence and incidence of microvascular and macrovascular complications over 15 years among patients with incident type 2 diabetes. BMJ Open Diabetes Res Care, 2021. 9(1).
Wong, T.Y., et al., Diabetic retinopathy in a multi-ethnic cohort in the United States. Am J Ophthalmol, 2006. 141(3): p. 446-455.
Holman, R.R., et al., 10-year follow-up of intensive glucose control in type 2 diabetes. The New England journal of medicine, 2008. 359(15): p. 1577-89.
Laiteerapong, N., et al., The Legacy Effect in Type 2 Diabetes: Impact of Early Glycemic Control on Future Complications (The Diabetes & Aging Study). Diabetes Care, 2019. 42(3): p. 416-426.
Lind, M., et al., Historical HbA1c Values May Explain the Type 2 Diabetes Legacy Effect: UKPDS 88. Diabetes Care, 2021.
Nichols, G.A., C. Conner, and J.B. Brown, Initial nonadherence, primary failure and therapeutic success of metformin monotherapy in clinical practice. Curr Med Res Opin, 2010. 26(9): p. 2127-35.
American Diabetes Association Professional Practice, C., et al., 9. Pharmacologic Approaches to Glycemic Treatment: Standards of Medical Care in Diabetes-2022. Diabetes Care, 2022. 45(Suppl 1): p. S125-S143.
Elhussein, A., et al., Racial/ethnic and socioeconomic disparities in the use of newer diabetes medications in the Look AHEAD study. Lancet Reg Health Am, 2022. 6.
Goonesekera, S.D., et al., Racial ethnic differences in type 2 diabetes treatment patterns and glycaemic control in the Boston Area Community Health Survey. BMJ Open, 2015. 5(5): p. e007375.
Trinacty, C.M., et al., Racial differences in long-term adherence to oral antidiabetic drug therapy: a longitudinal cohort study. BMC Health Serv Res, 2009. 9: p. 24.
Mathur, R., et al., Ethnic disparities in initiation and intensification of diabetes treatment in adults with type 2 diabetes in the UK, 1990-2017: A cohort study. PLoS Med, 2020. 17(5): p. e1003106.
Karter, A.J., et al., Elevated rates of diabetes in Pacific Islanders and Asian subgroups: The Diabetes Study of Northern California (DISTANCE). Diabetes Care, 2013. 36(3): p. 574-9.
Parker, M.M., et al., An algorithm to identify medication nonpersistence using electronic pharmacy databases. J Am Med Inform Assoc, 2015. 22(5): p. 957-61.
Delate, T., G. Albrecht, and K.L. Olson, Out-of-plan pharmacy use by members of a managed care organization. Perm J, 2012. 16(2): p. 15-21.
Karter, A.J., et al., Glycemic response to newly initiated diabetes therapies. Am J Manag Care, 2007. 13(11): p. 598-606.
Zou, G., A modified poisson regression approach to prospective studies with binary data. Am J Epidemiol, 2004. 159(7): p. 702-6.
Wang, M.C., et al., Age at Diagnosis of Diabetes by Race and Ethnicity in the United States From 2011 to 2018. JAMA Intern Med, 2021. 181(11): p. 1537-1539.
Balasubramanian, B.A., et al., Racial/ethnic differences in obesity and comorbidities between safety-net- and non safety-net integrated health systems. Medicine (Baltimore), 2017. 96(11): p. e6326.
Gopalan, A., et al., Initial Glycemic Control and Care Among Younger Adults Diagnosed With Type 2 Diabetes. Diabetes Care, 2020.
American Diabetes Association Professional Practice, C., et al., 6. Glycemic Targets: Standards of Medical Care in Diabetes-2022. Diabetes Care, 2022. 45(Suppl 1): p. S83-S96.
Marcondes, F.O., et al., Are racial/ethnic minorities recently diagnosed with diabetes less likely than white individuals to receive guideline-directed diabetes preventive care? BMC Health Serv Res, 2021. 21(1): p. 1150.
Lopez, J.M., et al., Characterization of type 2 diabetes mellitus burden by age and ethnic groups based on a nationwide survey. Clin Ther, 2014. 36(4): p. 494-506.
Golden, S.H., The contribution of structural racism to metabolic health disparities in the USA. Lancet Diabetes Endocrinol, 2021. 9(8): p. 478-480.
Egede, J.K., et al., Perceived Stress as a Pathway for the Relationship Between Neighborhood Factors and Glycemic Control in Adults With Diabetes. Am J Health Promot, 2022. 36(2): p. 269-278.
Assari, S., et al., Racial Discrimination in Health Care Is Associated with Worse Glycemic Control among Black Men but Not Black Women with Type 2 Diabetes. Front Public Health, 2017. 5: p. 235.
Charlson, M.E., et al., A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis, 1987. 40(5): p. 373-83.
Messer, L.C., et al., The development of a standardized neighborhood deprivation index. J Urban Health, 2006. 83(6): p. 1041-62.
Brown, J.B., C. Conner, and G.A. Nichols, Secondary Failure of Metformin Monotherapy in Clinical Practice. Diabetes Care, 2010. 33(3): p. 501.
Centers for Disease Control and Prevention. Hispanic/Latino Americans and Type 2 Diabetes. 2019 30, 2020]; Available from: https://www.cdc.gov/diabetes/library/features/hispanic-diabetes.html.
Kanaya, A.M., et al., Heterogeneity of diabetes outcomes among asians and pacific islanders in the US: the diabetes study of northern california (DISTANCE). Diabetes Care, 2011. 34(4): p. 930-7.
Li, L., et al., Prevalence of microvascular diseases among tertiary care Chinese with early versus late onset of type 2 diabetes. Journal of diabetes and its complications, 2015. 29(1): p. 32-7.
Hillier, T.A. and K.L. Pedula, Complications in young adults with early-onset type 2 diabetes: losing the relative protection of youth. Diabetes care, 2003. 26(11): p. 2999-3005.
Huo, X., et al., Risk of non-fatal cardiovascular diseases in early-onset versus late-onset type 2 diabetes in China: a cross-sectional study. The lancet. Diabetes & endocrinology, 2016. 4(2): p. 115-24.
Song, S.H. and T.A. Gray, Early-onset type 2 diabetes: high risk for premature diabetic retinopathy. Diabetes research and clinical practice, 2011. 94(2): p. 207-11.
McQuaid, E.L. and W. Landier, Cultural Issues in Medication Adherence: Disparities and Directions. Journal of general internal medicine, 2018. 33(2): p. 200-206.
Hall, W.J., et al., Implicit Racial/Ethnic Bias Among Health Care Professionals and Its Influence on Health Care Outcomes: A Systematic Review. Am J Public Health, 2015. 105(12): p. e60-76.
Green, A.R., et al., Implicit bias among physicians and its prediction of thrombolysis decisions for black and white patients. J Gen Intern Med, 2007. 22(9): p. 1231-8.
van Ryn, M. and J. Burke, The effect of patient race and socio-economic status on physicians’ perceptions of patients. Social Science & Medicine, 2000. 50(6): p. 813-828.
Fernandez, A., et al., Adherence to Newly Prescribed Diabetes Medications Among Insured Latino and White Patients With Diabetes. JAMA Intern Med, 2017. 177(3): p. 371-379.
Heisler, M., et al., Mechanisms for racial and ethnic disparities in glycemic control in middle-aged and older Americans in the health and retirement study. Arch Intern Med, 2007. 167(17): p. 1853-60.
Funding
The study was funded by the National Institute of Minority Health and Health Disparities (R01MD013420), the National Institute of Diabetes and Digestive and Kidney Diseases (K23DK116968, P30DK092924, P30DK092949), and the National Institute on Aging (R01AG063391). The funders had no role in the study’s design and conduct, the completed analysis, the interpretation of the data, or the content and preparation of the manuscript.
Author information
Authors and Affiliations
Contributions
All the listed authors have met the requirements for authorship. A.G. and A.N.W. oversaw the study design, data analysis, result interpretation, and manuscript preparation. A.J.K. and N.L. contributed to the study design, result interpretation, and review/editing of the manuscript. A.G. is the guarantor of this manuscript.
Corresponding author
Ethics declarations
Ethics Approval
This work was approved by the Kaiser Permanente Northern California Institutional Review Board.
Conflict of Interest
The authors declare that they do not have a conflict of interest.
Additional information
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Anjali Gopalan and Aaron N Winn are co-first authors.
Prior Presentations
None
Supplementary Information
ESM 1
(DOCX 880 kb)
Rights and permissions
About this article
Cite this article
Gopalan, A., Winn, A.N., Karter, A.J. et al. Racial and Ethnic Differences in Medication Initiation Among Adults Newly Diagnosed with Type 2 Diabetes. J GEN INTERN MED 38, 994–1000 (2023). https://doi.org/10.1007/s11606-022-07746-4
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11606-022-07746-4