An Assessment of Racial Differences in Clinical Practices for Hypertension at Primary Care Sites for Medically Underserved Patients


  • Daniel T. Lackland DrPH,

    1. From the Department of Biometry and Epidemiology1 and Departments of Medicine and Pharmacology,2 Medical University of South Carolina, Charleston, SC
    Search for more papers by this author
  • 1 Yan Lin MD,

    1. From the Department of Biometry and Epidemiology1 and Departments of Medicine and Pharmacology,2 Medical University of South Carolina, Charleston, SC
    Search for more papers by this author
  • 1 Barbara C. Tilley PhD,

    1. From the Department of Biometry and Epidemiology1 and Departments of Medicine and Pharmacology,2 Medical University of South Carolina, Charleston, SC
    Search for more papers by this author
  • and 1 Brent M. Egan MD 2

    1. From the Department of Biometry and Epidemiology1 and Departments of Medicine and Pharmacology,2 Medical University of South Carolina, Charleston, SC
    Search for more papers by this author

Daniel T. Lackland, DrPH, Department of Biometry and Epidemiology, 135 Cannon Street, Medical University of South Carolina, Charleston, SC 29425


Ethnic disparities in hypertension-related outcomes may relate to differences in medical care. This study assessed primary care sites serving low-income patients to determine if differences in process and treatment indicators might contribute to disparities in outcomes. Eight sites were enrolled with approx100,000 patients, collectively. Trained nurses abstracted a random sample of medical records for diagnoses, laboratory data, medications, and demographic variables. Data were obtained on 1250 white and 2786 African-American adults. African Americans were more likely (p<0.01) to be hypertensive (44% vs. 23%) and diabetic (16% vs. 8%) than whites. African Americans were more likely to have serum creatinine, potassium, lipid, and glycosylated hemoglobin values recorded in the medical record than whites (p<0.01). African-American hypertensives were more likely (p<0.05) than white hypertensives to receive calcium channel blockers, angiotensin-converting enzyme inhibitors, diuretics, and combination therapy. Thus, there appeared to be little difference in the level of care between the two groups. Based on the sites examined, ethnic variations in important process and treatment indicators do not explain racial differences in cardiovascular and renal outcomes.

Hypertension-related complications such as stroke and end-stage renal disease occur two to five times more frequently in African Americans than in whites.1 Moreover, the prevalence of hypertension is higher and target organ complications begin 10–20 years earlier in life among African Americans compared with whites.2 Although the explanation for the excess burden of hypertension-related risk and disease among African Americans is unclear, evidence suggests that failure to control blood pressure is a major contributor.3

Clinical trials demonstrate that effective antihypertensive therapy reduces racial disparities in the complications of hypertension such as stroke and total mortality.3 Therefore, the failure to reduce the relative disparities in hypertension-related outcomes may reflect a lower quality or quantity of services and/or less intensive therapeutic attempts for African-American than white hypertensives.1,4

Ethnic disparities in health have been observed for several key process-of-care and disease indicators.5,6 Because hypertension and related cardiovascular risk factors, including diabetes, affect a large proportion of the adult population, such patients are usually managed in primary-care settings.7 Therefore, we examined key process and treatment indicators in hypertensive patients from practices that served disproportionately more ethnic minority and low-income patients.


Primary-Care Practice Settings and Patient Population

The study protocol was reviewed and approved by the Office of Research Protection (ORP) at the Medical University of South Carolina. Approximately 17 primary-care sites for the medically underserved in South Carolina were invited to participate through an onsite visit by members of the investigative team. These sites were eligible for participation based on a greater proportion of ethnic minorities patients compared with the state average. Eight clinical practices in South Carolina encompassing both rural and suburban locations agreed to participate by signing documents approved by the ORP. The eight practice sites represented three types of primary-care settings: community health centers, private solo/group practices, and residency training programs.

Instrument for Chart Abstraction

A medical record abstract form was developed and field-tested in the pilot phase of the study. The instrument included patient demographic information, height and weight, blood pressure readings, diagnoses of hypertension and related cardiovascular risk factors and complications, selected routine and optional laboratory data recommended in the evaluation of hypertensive patients by the sixth report of the Joint National Committee on Hypertension (JNC VI),8 and all prescription medications. The routine laboratory data entered included urinalysis, serum potassium, sodium, creatinine, fasting glucose, total cholesterol, and high-density lipoprotein cholesterol. Optional data included microalbumin, 24-hour urinary protein, fasting triglycerides, low-density lipoprotein (LDL) cholesterol, and glycosylated hemoglobin (HbAlc) values.

Chart Abstraction and Data Management Procedures

The project database manager obtained a database for each practice with a unique number for each adult patient and generated a random sequence of these numbers. The randomized sequence of unique patient numbers was returned to the respective practices, which then pulled these records for review by the chart abstractors. Each chart abstractor, a licensed registered nurse, signed a confidentiality agreement approved by the ORP and reviewed by legal counsel for the Medical University of South Carolina. The abstractor reviewed the medical records and entered the data on the abstraction form that contained another unique number linked to the original identifier. No identifying information such as name, address, phone number, social security number, e-mail address, or medical record number was entered on the abstract form. A second certified registered nurse abstracted a random sample of 10% of charts reviewed by the first nurse. A member of the investigative team (BME) examined each chart abstraction form for completeness and clarity, including both the original and verification reviews. Coding of the forms was adequate to ensure that the data analyst could match the two abstract forms from the original and verification session on an individual patient. After all questions on the chart abstract forms were resolved, the forms were forwarded to the data management and analysis core for data entry and analysis.

Data Analysis

The agreement between the original and verification abstraction for key variables such as hypertension and diabetes and medications was consistently ≥90%. Descriptive analyses were generated for key laboratory data and antihypertensive medications. Blood pressure medications were classified as α1 antagonists, αβ blockers, angiotensin-converting enzyme (ACE) inhibitors, angiotensin receptor blockers, β blockers, calcium channel blockers (dihydropyridine and nondihydropyridine), diuretics, sympatholytics, and vasodilators. Comparisons between African Americans and whites were performed using the chi-square test. Hispanic patients (n=417) and “other” patients (n=97) were excluded from these analyses given the relatively small sample size and significant demographic differences, especially for the former, compared with the two major ethnic groups in this sample.


Chart abstracts were performed on a random sample of 4906 adult patients from the eight selected primary-care sites. From the random sample of medical records abstracted, African Americans comprised 61.2% of the sample, whites 27.5%, Hispanics 9.2%, and other 2.1%. The data provided in Table I is limited to the 2786 African-American and 1250 white patients because the 417 Hispanics were significantly younger (mean age 33 years) and had a higher proportion of men (53%). The other ethnic group included too few patients (n=97) for meaningful analysis. African Americans comprised a greater proportion of patients at the Community Health Centers, whereas whites comprised a larger proportion of patients in the residency training programs. As shown in Table I, African Americans were twice as likely to have a diagnosis of hypertension and diabetes than whites.

Table I.  Patient Characteristics as Determined From Chart Abstraction by Ethnic Group
 African Americann=2786Whiten=1250p Value
Male/female (%)32/6838/62<0.0001
Age (yr, mean±SD)46±1845±170.0007
18–35 yr(%)3032 
35–50 yr (%)3033 
Hypertension (%)4423<0.0001
Diabetes (%)168<0.0001
Practice setting  <0.0001
Community health center (%)5936 
Private solo/group (%)2726 
Residency training (%)1439 

The frequency of selected laboratory tests identified as routine and optional in the JNC VI report, which was in effect throughout the study period, showed that African Americans were more likely than whites (p<0.01) to have recorded measurements of serum potassium, creatinine, lipids, and HbAlc on their medical records (Figure). Conversely, whites were somewhat more likely (p<0.05) than African Americans to have some assessment of urine protein documented in their medical records. However, regardless of ethnicity, assessment of urine protein was unusual, and urine microalbumin was not identified in any of the charts abstracted for this study.

Figure Figure.

Percentage of patients with selected laboratory measurements documented on their medical record. The percentages of African Americans (AA) and whites (C) with selected laboratory tests documented in their medical records are shown for all patients and those with hypertension (HTN) and/or diabetes. The percentages with a record of having these selected tests were generally lowest for all patients combined and greatest in those with diabetes. A symbol within the bar indicates an ethnic difference in the percentage of patients with that laboratory test for the group (all patients, hypertensives, diabetics) indicated by the shading ( + =p<0.05, *=p<0.001). Although not shown, among hypertensives, C were more likely than AA (p<0.05) to have urine protein (uP) measured. sCr=serum creatinine, K+=serum potassium; chol=total cholesterol; HDL=high-density lipoprotein cholesterol; LDL=low-density lipoprotein cholesterol; TG=triglycerides; A1c=glycosylated hemoglobin

Patients with hypertension and/or diabetes were more likely than the overall sample of adults to have laboratory values for the measurements included on the abstract form with the exception of urine protein (Figure). Among hypertensive patients, a similar pattern of ethnic differences was observed, although only the values for urine protein, serum creatinine, potassium, and HbAlc remained significant. Among diabetic patients, only LDL cholesterol, high-density lipoprotein cholesterol, and HbAlc values were significantly more likely to be documented on the medical records of African Americans than whites.

Although the treatment of hypertension with α1 antagonists and β blockers were similar for African Americans and whites, ethnic differences were detected for several other agents (Table II). More specifically, hypertensive African Americans were more likely to be prescribed ACE inhibitors, calcium channel blockers, and diuretics than hypertensive whites (Table II). African-American hypertensives were more likely than white hypertensives to receive two or more medications (combination therapy) for hypertension (46% vs. 38%). Conversely, white hypertensives were more likely than African-American hypertensives to receive a single blood pressure medication.

Table II.  Categories of Antihypertensive Medications Determined From Chart Review for African-American and White Hypertensive Patients
 African Americann=1233Whiten=284p Value
Category of medications (%)   
Alpha-1 antagonist640.15
ACE inhibitor34270.01
Beta blocker15180.17
Number of antihypertensive medications (%)   
ACE=angiotensin-converting enzyme inhibitor; DCCB=dihydropyridine calcium channel blocker;
NdCCB=nondihydropyridine calcium channel blocker


These results demonstrate that primary-care sites providing services for a large proportion of ethnic minorities can be successfully recruited and utilized in clinical studies. Moreover, useful data on the prevalence of hypertension, diabetes, and related cardiovascular complications, as well as key process-of-care and intensity of medical therapy indicators, can be obtained in these health care settings by chart abstraction.

The findings from the chart review are consistent with previous clinical epidemiological reports and provide some potentially important insights. First, the prevalence of hypertension and diabetes among adults is much higher among African Americans than in whites receiving care at the sites selected for this study (Table I). The ethnic differences in prevalence of hypertension and diabetes are directionally consistent with previous reports.9–12 The lower rates of diagnosed hypertension and diabetes for Hispanic patients are likely due to the relative young age of the patients. Nonetheless, the significant proportion (9.2%) of patients of Hispanic origin identifies this at-risk population as an emerging consideration for medical practices in the Southeast, which historically have not served this ethnic group.

Although the prevalence of major cardiovascular risk factors was greater in African Americans than in whites, African Americans were, if anything, more likely and not less likely to receive indicated testing and treatment for high blood pressure (Figure). Nevertheless, the medical sites evaluated have a relatively low rate of obtaining laboratory tests recommended as routine in the JNC VI report. Nearly one third of the hypertensive and diabetic patients did not have a lipid measurement documented in their medical records. More than 30% of the patients with diabetes did not have a measurement of HbAlc in their medical records. However, African Americans were more likely to have the measurement than white patients with diabetes. African Americans were more likely than whites to have a measurement of LDL cholesterol documented in their medical record.

Of note, assessment of urine protein, even on a routine urinalysis (dipstick), was uncommon. Assessment of urine protein was unusual even in the hypertensive and diabetic patients, despite the fact that the test is recommended as part of routine care by JNC VI.8 Moreover, nearly 30%–40% of patients with hypertension did not have a measurement of serum creatinine. The comparatively low rates of renal functional assessment are of particular concern for African Americans with hypertension and/or diabetes given their much higher relative risk for end-stage renal disease than whites.1 To magnify this concern, an assessment of urine protein was absent over the previous 5 years in more than 95% of African Americans with hypertension or diabetes. The results suggest that the consensus guidelines for hypertension and diabetes mellitus are not followed in many patients.8,13,14 However, ethnicity in our study does not emerge as a major deterrent to laboratory testing recommended as routine or optional in the evaluation of hypertensive patients in JNC VI. In fact, African Americans are as or more likely to have key tests than whites with the exception of urine protein, which was rarely done in either ethnic group.

The results of the random review of charts on adult patients in these practices for medically underserved patients suggests there are barriers to obtaining routine laboratory tests but that African-American minority status is not one of those barriers. These findings are at variance with previous reports indicating that African Americans received fewer diagnostic services,15,16 although similar to a report from a large managed care organization comparing care for African American and whites with diabetes.17

Ethnic differences were also observed in the frequency that several classes of antihypertensive medications were prescribed to hypertensive patients based on a review of the medical records. African Americans were more likely than whites with hypertension to receive diuretics, ACE inhibitors, and calcium channel blockers and to receive two or more medications to treat high blood pressure (Table II). These findings would indicate that lesser intensity of pharmacotherapeu-tic effort on the part of providers is not the explanation for poorer outcomes in African-American vs. white hypertensive patients. Formularies can vary by health care plan, thereby limiting the availability of antihypertensive medications among different practice settings. However, the formularies were similar and essentially unrestricted (i.e., open) in the practices selected for this study. Based on the availability and use of the various classes of medications at the different sites, a relative open formulary was evident for each of the settings.

The findings in the primary medical care sites selected for serving a high proportion of ethnic minorities may generally not be reflective of primary care. The medical and administrative staff at these clinics may be more attuned (i.e., culturally sensitive) to serving and meeting the health care needs of minority patients than staff at other health care sites. Further studies are required to explore this possibility. Moreover, the key process and treatment indicators do not suggest that optimal care is always received. Many patients do not have key laboratory data recorded on their medical records (Figure). The proportion of patients with high blood pressure that are either untreated or managed with antihypertensive monotherapy (Table II) is relatively high in view of the fact that two or more agents are usually required to control readings to <140/90 mm Hg.18

Of note, many studies on indicators of the quality of medical care focus on the hospital record and chronic care databases19,20 or on limited data in outpatient settings.21 However, the treatment and control of hypertension and diabetes are most often addressed in primary-care settings,7 where data are typically more difficult to obtain. The results presented here describe a model system for assessing quality indicators of the health care process and treatment patterns in the primary-care setting. Furthermore, this model is successful in reaching health care sites that could be categorized as a medical home for underserved ethnic minorities.

Although the quality indicators measured in this study were similar for both race groups, the practice settings utilized were significantly different, with African Americans more likely to use federally funded community health centers and whites more likely to be seen at residency training program clinics. The sample size in this set of analyses was inadequate to stratify by practice setting, but it would be reasonable to consider that the racial differences seen in population-based quality indicators may be reflective of variations in medical care at different sites.

The results of this study demonstrate that outpatient primary-care clinical sites providing medical services for a large proportion of ethnic minorities can be enlisted to participate in a study to evaluate key process indicators. Moreover, data on key process and treatment indicators in patients with hypertension and diabetes mellitus can be obtained. The data indicate that recommended laboratory data are not documented in a substantial subset of patients, but that African Americans are somewhat more likely to receive recommended laboratory testing. Moreover, blood pressure treatment appears to be more aggressive among African American than among white hypertensives. The data suggest that other factors such as ethnic differences in the use and access to primary medical care, as well as the prevalence and perhaps the severity of diabetes and hypertension, are major contributors to disparities in major target organ complications. However, process of care and intensity of the therapeutic effort do not appear to contribute significantly in the healthcare settings assessed.

Acknowledgments and disclosure: The authors thank Donna Jordan and Kim Edwards for superb administrative assistance, Sheryl Mack for excellence in medical site recruitment and management, Wenle Zhao, MD, for superb guidance and oversight in database management, and the providers and office staff at the participating medical sites for their active cooperation in the EXCEED Blood Pressure Project. This work was supported in part by grant P01HS1087-03 (EXCEED) from the Agency for Healthcare Research and Quality, K24HL04290 and P60-MD00267 (EXPORT) from the National Heart, Lung, and Blood Institute; the Duke Foundation; and the South Carolina Department of Health and Environmental Control.