Objective: To review existing data to determine whether ethnic disparities exist for diabetes-related preventive care among adults in the United States.

Study Design: Literature review.

Methods: We identified diabetes-related studies published between 1993 and 2003, using a reproducible search strategy. Studies were selected for review if there were ethnic comparisons or if data on a specific ethnic minority were reported. From these studies, we extracted data on commonly accepted diabetes-related preventive-care measures (testing for glycemia, eye examinations, foot examinations, lipid profile, influenza vaccination, nephropathy assessment, smoking-cessation counseling). The sources were US healthcare facilities, national survey samples, Veterans Affairs facilities, Medicare databases, and managed care data.

Results: Thirty-six studies met our search criteria. Data were extracted on glycemia testing (15 studies), eye examination rates (27 studies), foot examination rates (18 studies), lipid-profile assessment (15 studies), percentage of patients receiving influenza vaccinations (8 studies), nephropathy assessment (7 studies), and counseling referrals for smoking cessation (4 studies). The majority of the data indicated that the rates of diabetes monitoring are generally low regardless of the population being studied. The major ethnic differences reported were lower rates of eye examination, influenza vaccination, and lipid-profile testing among Hispanics and African Americans than among non-Hispanic whites.

Conclusions: Despite the availability of evidence-based guidelines, rates of diabetes preventive care are low, particularly for some measures in ethnic minority groups. Additional data are needed to further elucidate these disparities.

(Am J Manag Care. 2005;11:349-360)

Diabetes contributes significantly to morbidity, mortality, and healthcare expenditures; and its prevalence is increasing rapidly in the United States. Some ethnic minority groups, such as African Americans, Hispanics, and American Indians, are disproportionately affected by high prevalences of diabetes and diabetes-related complications.^1,2 Evidence-based guidelines for clinical management of diabetes are available to prevent or delay these complications, including those published annually by the American Diabetes Association.³ These guidelines include recommendations to measure diabetes patients' degree of glycemic control, their blood pressure, and their lipid levels; to conduct foot and dilated eye examinations; to assess patients' urinary protein; to ensure that patients receive influenza vaccination; and to provide patients with lifestyle counseling on diet, exercise and weight control, and smoking cessation.³ Other guidelines, such as those of the Diabetes Quality Improvement Project⁴ and those in the Health Plan Employer Data and Information Set,⁵ are used to assess the quality of care provided to patients with diabetes. Higher rates of physician performance of preventive-care measures have been shown to correspond to better control for diabetes management compliance.⁵

Despite the availability of such guidelines and quality-of-care measures, there is an overall lack of adherence to these guidelines by healthcare providers.⁶ Research has shown that certain populations, especially ethnic minority groups, are less likely to receive the highest level of healthcare, especially for preventive services such as influenza vaccination and cancer screening.^7-9 However, ethnic disparities in diabetes preventive care have not been adequately defined. This qualitative review synthesizes the body of research over the past decade on diabetes quality of care across ethnic minority populations in the United States, as that quality is related to well-established diabetes preventive-care measures. These measures also have been referred to as processes of care, quality indicators, and standards of care. For this review, preventive-care measures include tests for glycemia (glycosylated hemoglobin or A_1c), eye and foot examinations, lipid profile, influenza vaccination, nephropathy assessment, and smoking-cessation counseling.

METHODS

We conducted a MEDLINE search on PubMed for diabetes studies done in the United States and published from 1993 to 2003, as part of a systematic review for a larger study on ethnic disparities in diabetes quality of care. Other data sources searched included Web of Science, Education Resources Information Center, Cumulative Index to Nursing and Allied Health, Combined Health Information Database, and the Cochrane Library. Studies involving patients with pre-diabetes, patients younger than 18 years of age, patients with impaired fasting glucose, or women with gestational diabetes were excluded. We "exploded" the Medical Subject Heading (MeSH) term "diabetes mellitus" into its 2 components (diabetes mellitus, insulin dependent; and diabetes mellitus, non-insulin dependent) and used these terms along with free text to capture terms not found as MeSH headings, such as literacy, immigration status, locus of control, and insurance. To evaluate the quality of diabetes care in ethnic minority groups, we used search terms including diabetes, preventive measures, processes of care, and quality of care. We then extracted studies with a focus on ethnic minorities, as indicated by their use of the terms "race/ethnicity/ethnology," "ethnic groups," or "minority groups." We repeated the search process without restricting the search to MeSH headings in order to capture free-text articles.

We imported references into a database (Reference Manager, version 10, ISI Research Soft, Berkeley, Calif) and deleted duplicate references. Primary reviewers (JKK and RAB) reviewed reference lists from citations for additional studies that met inclusion criteria. All studies containing process measures for preventive care of diabetes were included as long as there were minority-group comparisons and/or a minority population with diabetes was studied. Observational and intervention studies were included to be comprehensive.

Literature Review Process

We identified 450 abstracts that met our initial criteria. We then assessed the articles summarized by these abstracts to determine whether they contained diabetes preventive-care measures. Studies were included if data specific to an ethnic minority group were reported or if an ethnic minority group was compared with a non-Hispanic white group. Ethnic minority populations were classified as African American, American Indian/Alaska Native, Asian/Pacific Islander, and Hispanic. The latter category included persons whose country of origin was Central or South America, the Caribbean Islands, or Mexico. Where indicated, we included the terms "Hispanic black," "Hispanic white," "non-Hispanic black," and "non-Hispanic white." After excluding articles that did not meet inclusion criteria, a total of 390 full-text articles were retrieved, of which 36 had data specific to ethnic minority populations regarding preventive-care measures.

Data on performance of preventive-care measures are reported in table format by ethnic group and compared with corresponding data for non-Hispanic whites (when provided). When evaluating multiple ethnic minority groups we compared each with the non-Hispanic white group. Information for non-Hispanic white groups was repeated in the table if there were multiple ethnic groups in a given study. We also documented whether data on preventive-care measures were based on patients' self-report or were obtained from medical chart reviews.

RESULTS

The data from 35 studies that reported annual (unless otherwise specified) assessment of diabetes preventive-care measures are listed in Tables 1 through 5. One study not included in the tables is described in the text because specific percentages were not provided for the preventive measure. Data were extracted on glycemia testing (15 studies), eye examination rates (28 studies), foot examination rates (18 studies), lipid-profile assessment (15 studies), percentage of patients receiving influenza vaccinations (8 studies), nephropathy assessment (7 studies), and counseling referrals for smoking cessation (4 studies). Eighteen of the 36 studies provided data on multiple preventive-care measures. Twenty-five studies used medical chart review data, 10 used self-reported information, and 1 study used both.

Glycemia Testing

Of the 15 studies that provided data on glycemia testing (Table 1), 10 included African Americans,^10-19 2 included American Indians,^20,21 and 4 included Hispanics.^16,22-24 One study had both African Americans and Hispanics.¹⁶ Ten of the 15 studies had a non- Hispanic white comparison group, and 5 of those conducted a statistical comparison by race/ethnicity. Excluding studies that provided rates other than annually, the range in glycemia testing rates was 19% to 87% for African Americans with a median of about 60%. When more frequent testing was expected for quarterly assessment of glycemia among African American insulin users, the rate was only 8.1%, while semiannual assessment of glycemia was 26.4% for African Americans not using insulin.¹⁹ The annual testing rate among American Indians was 79.6% in a national study and 64% semiannually in a regional study.^20,21 Among Hispanics, the rate of annual glycemia testing ranged from 37.6% to 88.4%, with an estimated median of about 63% for these studies.^22,24 Rates of glycemia testing in a 2-year period were 30.7% in one study, while the rate for receiving 2 glycemia tests in a 2-year period was 41.4% in another study.^16,23

Three studies showed lower rates of glycemia testing among African Americans compared with non-Hispanic whites.^10,13,17 The largest study of glycemia testing showed a 6% lower rate of annual glycemia measurement among African American (66%) compared with non- Hispanic white (72%) Medicare recipients (n = 80 532) in North Carolina.¹⁷

Eye Examination

For eye examination rates (n = 28) (Table 2), data were available from 19 studies that included African Americans,^{6,8,10-18,25-31,37} 4 with American Indians,^20,21,32,33 and 10 with Hispanics.^{6,16,22,24,28,29,34-37} Six studies had data for multiple ethnic minority groups.^{7,16,28,29,37,38} Studies included performance of a dilated eye examination by a healthcare professional or a visit to an eye-care professional. Of the 13 studies with a non-Hispanic white comparison group,* 9 conducted a statistical comparison by race/ethnicity.†Among studies reporting annual rates of eye examination, data were highly variable for each ethnic group, with a range of 7% to 82% (estimated median 50.4%) among African Americans, 14% to 75% (estimated median 57%) among American Indians, and 28% to 84% (estimated median 60.8%) among Hispanics.

Five studies found statistically significant differences (as large as 7.9%) by race/ethnicity,^{8,10,13,22,30} with all of them showing a lower rate of eye examinations in the ethnic minority group than among non-Hispanic whites. In one of these studies, Schneider and collegues,⁸ using national Medicare managed care data, showed significantly lower rates of eye examinations among African Americans (43.6%) than among non-Hispanic whites (50.4%). In another, Wang and Javitt³⁰ found significantly lower rates of use of eye-care services among African American (45.7%) compared with white (53.6%) Medicare beneficiaries. Another study showed rates of 71% for African Americans for self-reported eye examination in a 2-year period.³¹ One study showed rates of 82.8% for both African Americans and Hispanics for eye examination referral in a 2-year period.¹⁶

In an analysis of National Health Interview Survey data, Cowie and Harris³⁷ found no significant differences in rates of self-reported receipt of eye care for African Americans and for Hispanics. However, another study showed higher rates of eye exams among Hispanics (80.8% to 84%) than among non-Hispanic whites (76.3%).²² An analysis of Behavioral Risk Factor Surveillance System data from 1995 to 2001 showed an increase in self-reported annual eye examination of 5.9% among non-Hispanic whites (62.4% to 68.3%) and a 14% increase among Hispanics (46.1% to 60.0%), while rates for non-Hispanic blacks declined during the same time period (69.0% to 63.8%).⁶

A recent study among veterans (429 928 patients) evaluated ethnic differences in diabetes complications. Although eye examinations were reported, the frequency of the exams was not defined.³⁸ Among the ethnic groups considered in this study (African American, Hispanic, Asian, and Native American), rates of eye examination were 51.4%, 43.1%, 53.2%, and 47.6%, respectively, compared with a rate of 50.5% among non-Hispanic whites. Rates for Hispanics and Native Americans were significantly lower than the rate for non-Hispanic whites, while the rate for African Americans and Asians was significantly higher.

Foot Examination

Studies providing data on foot examinations (n = 18) (Table 3) included 11 with African Americans,^‡5 with American Indians,^{20,21,31,32,40} and 5 with Hispanics.^{6,16,22,24,29} Three studies had both African Americans and Hispanics.^6,16,29 We reviewed studies reporting any level of foot examination, including a complete or partial foot examination (visual inspection, pedal pulse examination, and a sensory examination). Rates varied extensively based on the type of foot examination reported. In 6 studies comparing foot examination rates among African Americans with those among non-Hispanic whites, no statistically significant differences were found. None of the studies we reviewed compared foot examination rates among American Indians with those among other ethnic groups. One report comparing rates in 1995 and 2001 showed that the foot examination rates for Hispanics declined 8.2% (62.9% to 54.7%) during this period, while rates for African Americans increased by 16.1% (54.1% to 70.2%) and rates for non-Hispanic whites increased by 8.7% (52.7% to 61.4%) during the same period.⁶

Lipid Profile

Studies reporting rates of lipid profiles (n = 15) (Table 4) were evaluated from 11 studies with African Americans,^11-19,28,29 3 with American Indians,^20,21,33 and 4 with Hispanics.^16,22,28,29 Three contained data for both African Americans and Hispanics.^16,28,29 There was wide variation in the percentage of ethnic group members among whom lipid profiles were done (37% to 86% with a median of about 65%). Of the 9 studies that compared rates with those among non-Hispanic whites, 6 found that minority populations received lipid profiles less frequently than their white counterparts, with the differences ranging from -3.4% to as much as -19%.^{13,15,17,19,28,29} Three of these studies showed statistically significant differences.^15,19,28 Harris and colleagues, using data from the National Health and Nutrition Examination Survey III, found an annual lipid-profile rate of 68.1% among African Americans and 61.8% among Hispanics, compared with 80.8% among non-Hispanic whites, a difference of -12.7% and -19%, respectively.²⁸

More recent data showed that the rate of low-density lipoprotein laboratory testing among Department of Veterans Affairs enrollees was 68% among African Americans compared with 83% among non-Hispanic whites, a -15% difference.¹⁵ Three studies reported that lipid-profile measurement was done more frequently among a minority population compared with non- Hispanic whites, but not at a significantly higher rate.^11,16,22 Wisdom and colleagues¹⁹ split the sample being studied into 2 groups: insulin users and non-insulin users. They found that African American insulin users were more likely to get a lipid-profile assessment compared with non-insulin users; however, among non-insulin users, there was no significant difference compared with non-Hispanic whites (59.4% vs 62.8%, respectively).

Influenza Vaccination

Studies assessing rates of influenza vaccination (n = 8) (Table 5) included 5 among African Americans,^{6,13,17,41,42} 2 with American Indians,^32,33 and 3 with Hispanics.^6,22,42 Two studies had both African Americans and Hispanics.^6,41 All of the studies with African Americans and Hispanics reported comparative data with non-Hispanic whites. Of the 5 studies with African Americans, 3 showed significantly lower rates among African Americans (between 8.1% and 29%) than among non-Hispanic whites.^13,41,42 The percentage of patients receiving influenza vaccination varied widely (from 16% to 73%) with an interval midpoint of approximately 37% (33% to 41%).

In the largest study, Massing et al¹⁷ reported vaccination rates among North Carolina Medicare beneficiaries to be 33% for African Americans and 54% for non-Hispanic whites, a difference of -21%. One of the 3 studies among Hispanics in which statistical tests were performed showed significantly lower rates of influenza vaccination among Hispanics than among non- Hispanic whites, as much as a -13% difference.⁴¹ A wider range of influenza-vaccination rates was observed in these studies among Hispanics (38% to 66%) than among non-Hispanic whites (46% to 68%).

Nephropathy Assessment

Studies on nephropathy (n = 7) assessment are not included in table format. There were 4 with African Americans,^11-13,17 2 with American Indians,^20,21 and 1 with Hispanics.²² None of these studies had multiple ethnic minority groups. Four studies^11,15,17,22 had a non- Hispanic white comparison group, and 3 conducted tests of statistical significance across ethnic groups.^11,15,22 Of these 3 studies 2 showed higher rates of nephropathy assessment among African Americans compared with non-Hispanic whites (7.1% and 11% difference).^11,15 Heisler and colleagues¹⁵ found the 2-year nephropathy-assessment rates among African Americans (63%) to be higher than among non-Hispanic whites (52%) receiving care in US Veterans Affairs facilities. However, this difference was not statistically significant after adjustment for patient characteristics (patient's diabetes self-management, age, education, income, diabetes duration, severity of comorbidities, insulin usage or receiving more than 80% of care at a Veterans Affairs facility, or more than 2 visits in the last year). Massing and colleagues¹⁷ showed a 4% absolute lower rate of nephropathy assessment among African American Medicare recipients (49%) compared with their non-Hispanic white counterparts (53%).

Smoking-cessation Counseling

Studies on smoking-cessation counseling (n = 4) included 4 with African Americans^14,16,22,43 and 2 with Hispanics.^16,43 Two studies had both African Americans and Hispanics.^16,43 The data for this preventive-care measure varied from documentation of healthcare providers offering smoking-cessation counseling to documentation of rates of patient smoking cessation. Sample sizes ranged from 42 to 1453 for minorities and 119 to 294 for non-Hispanic whites. None of these studies showed significant ethnic differences in counseling rates.

DISCUSSION

The purpose of this review was to examine the available data to determine the extent to which ethnic disparities exist regarding diabetes preventive-care practices in the United States. We found that the largest ethnic disparities reported were in the overall rates of eye examination, lipid-profile testing, and influenza vaccination among African Americans and Hispanics. The studies we evaluated in this review also illustrate wide variation in the performance of these routine preventive-care measures for patients with diabetes.

One limitation of the studies in this qualitative review is the reliance on medical charts. With the advent of computerized medical records, these data may be more easily obtainable. Another limitation is that literature could have been omitted if author indexing of key words and titles was not descriptive of quality-of-care measures. There also were variations across studies in the definition of measures (eg, visual foot examination vs complete foot examination) and how frequently the measures were performed (eg, semiannually, annually, biannually). In reviewing lipid-profile assessment, studies did not consistently clarify whether cholesterol testing included a complete profile (total cholesterol, low-density lipoprotein, high-density lipoprotein, and triglycerides) or just 1 component of the profile.

Our ability to provide pooled estimates was restricted by the wide range in the ages of the patients studied and heterogeneity in study design. The patient populations also were not homogenous; therefore, it is difficult to adjust for the case mix. These confounders limited the possibility of making comparisons across studies. Some ethnic minority groups were represented in only a few studies, and no data were found for diabetes monitoring among Asian/Pacific Islanders. As most of the data for American Indians came from the Indian Health Service, no comparison data are available. With regard to ethnicity, the term "Hispanic" is not well defined in some of the studies cited and could refer to Latinos from Central and South America or the Caribbean. Some literature also cited Mexican Americans. With this nomenclature in mind, there are pitfalls in adequately describing and differentiating these groups. Finally, there was no information on socioeconomic factors, which might influence receipt of preventive care.

Despite these limitations, we found ethnic differences in rates of obtaining certain routine preventive-care measures among patients with diabetes. Some earlier reports explain some of the variability we found in this review. In a study of racial disparity in influenza vaccination rates, Schneider et al⁹ reported that common reasons for patients not receiving the vaccine included not knowing it was needed, fearing it would cause either the flu or other adverse effects, and not believing that the vaccine would prevent flu. Other literature indicates that minority groups tend to have less access to healthcare services.^13,26 After Heisler and colleagues¹⁵ adjusted for patient age, education, income, insulin use and diabetes self-management, the duration and severity of the diabetes, comorbidities, and health services utilization, they still found racial disparities in the rates of assessment of low-density lipoprotein.

Many potential barriers to care have been identified. Language may be one of these barriers, as nearly 8 million Hispanics in the United States do not speak English well and more than 1 in 4 households of Hispanics are linguistically isolated.⁴⁴ In a study of English-speaking versus Spanish-speaking Latinos, 32% of Spanish-speaking Latinos reported that they needed a translator and 23% reported language as a barrier with their primary care provider.²² In a nationally representative sample of 4811 African Americans, 3379 Hispanics, and 33 737 non-Hispanic whites with private or public health insurance, African Americans and Hispanics had greater access to primary care in managed care plans.⁷ In this study, however, Hispanics and African Americans had less access to specialty care than non-Hispanic whites.⁷ Lack of insurance has been cited as the most significant barrier to care in rural and ethnic minority Americans, as they are less likely to have health insurance than non-Hispanic whites, with the highest percentage of uninsured individuals among Hispanics (estimated at 35%).⁴⁴

Healthcare provider bias has been shown to influence the quality of care. A recent cross-sectional analysis of 150 391 visits of black and white Medicare beneficiaries revealed that physicians treating black patients had more difficulty obtaining subspecialists, diagnostic imaging, and nonemergency admission to the hospital.⁴⁵

CONCLUSION

Diabetes greatly increases the risk for death, cardiovascular disease, end-stage renal disease, diabetic retinopathy, and lower-extremity amputation.⁴⁶ To prevent these complications, diabetes patients are recommended to have regular monitoring and evaluation of A_1c levels (ie, glycemic control), blood pressure, lipids, and kidney, eye, and peripheral nerve functioning.³ Other preventive measures, such as influenza vaccination and smoking cessation, are also stressed to the patient on a regular basis.

Data from studies of other chronic diseases indicate that ethnic minorities are less likely than non- Hispanic whites to receive appropriate screening, secondary prevention care, and monitoring of high-risk patients.^47-52 Reasons that patients do not obtain these services are not clear, but are believed to be associated with both physician and patient characteristics.^44,45 For primary care providers, access to clinical resources, physician training, and clinical inertia may play a role.⁵³ Empirical evidence suggests that the presence of bias or prejudicial attitudes among healthcare providers along with clinical uncertainty and beliefs held by the provider may contribute to disparities in minority populations.^44,54 Patient-level variables that may contribute to racial disparities include mistrust of the medical system, refusal of recommended services, poor adherence to treatment regimens, and delay in seeking care.

More aggressive adherence to preventive-care measures is needed among all ethnic groups. Over the 10-year period evaluated in this review, the overall percentage of patients receiving diabetes-related preventive- care services remained suboptimal. Our findings further indicate that ethnic disparities in receiving diabetes preventive-care services were most pronounced for eye exams, lipid profiles, and influenza vaccination in both African Americans and Hispanics. These results stress the need for more emphasis on quality improvement among primary care providers and for a better understanding of barriers that prevent patients, especially minority patients, from receiving optimal diabetes monitoring and care.

Acknowledgment

We thank Ms. Carol Hildebrandt for her expertise in completing literature searches, assembling the references, and editing this manuscript.