INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In the United States, African Americans have higher rates of asthma-related morbidity and mortality than Caucasians (1- 3). The population at highest risk are African American residents of impoverished areas in large cities (4). Because a large proportion of inner-city residents are African American, it is difficult to distinguish whether ethnic factors or socioeconomic factors are the major contributing factor to these poor asthma outcomes (7, 8).

The causes for poor asthma outcomes in African Americans remain unknown. However, distinctive patterns of asthma medication use and health care resource utilization have been observed in African Americans which are different from those patterns observed in Caucasians. When compared with affluent areas of Philadelphia, citizens in impoverished areas frequently fill inhaled bronchodilator prescriptions, but disproportionately less inhaled corticosteroids (9). The residents of these impoverished areas are predominantly African American, and this pattern of asthma medication use suggests that the focus of treatment is on acute symptom control versus suppression of chronic airway inflammation. In addition, urban African Americans frequently use emergency facilities as a source of primary care for their asthma (10) and other medical problems (11). Although financial barriers that limit access to routine medical care are often hypothesized as a likely reason for these patterns of care, it has recently been suggested that other factors may be involved. For instance, African Americans and Caucasians have culturally based differences in perception of well-being and different levels of trust regarding organized medicine's ability to meet their needs. These cultural differences are not fully attributable to differences in income level (12). In addition, a comparison of African American and Caucasian children with asthma on Medicaid concluded that high rates of asthma-related emergency department use and hospitalization in the African American population could not be fully explained by poverty (13).

In order to assess whether racial differences in patterns of asthma care persist in a managed care environment, where financial barriers to medical care access are minimized, we compared the medication use and health care facility utilization of African Americans and Caucasians who were enrolled in a large Detroit metropolitan area health maintenance organization (HMO) which operates under managed care principles. In this setting, limited out-of-pocket expense is required for physician visits or prescription medications. (A typical out-of-pocket copay is $0 to $10 for physician visits and $2 to $10 for prescriptions.)

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Design

A cross-sectional analysis was performed for a time period encompassing calendar year 1993. The locations and specialty of physicians seen during asthma-related visits were determined as well as the medication type and number of asthma prescriptions filled by African American and Caucasian subjects enrolled in the HMO. The study population included every member of the HMO meeting the following criteria: (1) continuous enrollment in the HMO for the entire calendar year 1993; (2) age of 15 to 45 yr (this criterion was chosen to minimize the inclusion of subjects at ages when diagnoses commonly confused with asthma such as chronic obstructive lung disease, or a single viral episode with wheezing, are common); (3) self-identified as African American or Caucasian ethnicity upon HMO enrollment; (4) at least one physician encounter in an outpatient setting (emergency department, primary care, or asthma specialty care) occurring in calendar year 1993, with a diagnostic code for asthma (493.0-493.9) listed as the primary reason for the encounter.

Description of the HMO and Health Care Setting

The base HMO population consisted of approximately 230,000 HMO members whose medical care was delivered by a multispecialty group practice contracted to provide medical care to the managed care HMO members. Low-income subjects often join the HMO through employer-provided health plans or Medicaid-sponsored enrollment.

The multispecialty provider group for the HMO consisted of over 800 physicians in 1993 with approximately one-third of them practicing primary care medicine (pediatrics, internal medicine, or family practice). All physicians are required to be board eligible in their specialty and become board certified within 3 yr of being hired. In 1993, there were 17 board certified/eligible allergists and pulmonologists in the network who are designated as "asthma specialty providers" for the purposes of this study. Seven of these asthma specialty providers practiced within the specialty of allergy and 10 in pulmonary medicine. Access to the specialty provider required a referral from a primary care physician.

The physical facilities included 27 ambulatory care clinics, providing a mix of primary and specialty care. Four of these facilities have onsite emergency departments, with one of these sites being physically connected to a 900-bed tertiary care facility. Asthma specialists provided outpatient services at the tertiary care facility and five additional centers distributed throughout the Detroit metropolitan area.

HMO members receive health care services, including outpatient, emergency department, and inpatient visits, as well as prescription medications which are provided to enrollees with limited out-of-pocket expense.

Determination of the Location of Asthma Care Encounters and Patterns of Medication Use

A system-wide database was used to gather information on all continually enrolled Caucasian or African American HMO members who received any asthma-related outpatient health care during 1993 as determined by a primary diagnostic code for asthma (ICD 493.0-493.9) associated with an outpatient encounter. The information retrieved from this database included medical record number, ethnicity (self-described as Caucasian or African American), sex, date of birth, street address, and number and location of asthma-related visits. These visits were categorized as primary care clinic, asthma specialty clinic, emergency department, or inpatient. Only self-described Caucasians and African Americans, as indicated upon HMO enrollment, were included in the analysis.

A separate database containing HMO billing information was used to estimate possible "out-of-plan" health care for urgent asthma episodes such as in emergency departments, or hospitalizations. We found no difference in the rate for out-of-plan emergency department encounters for Caucasian (5.5%) and African American (4.6%) HMO members. We found only one subject (Caucasian) who was hospitalized with asthma outside the health care system.

The number of asthma prescription medications filled at any pharmacy by the study subjects was also analyzed. This was accomplished by retrieving billing data submitted to the managed care HMO and identifying medications by unique National Drug Code (NDC). Seven classifications of asthma drugs were analyzed, including inhaled corticosteroids, inhaled cromolyn/nedocromil, inhaled beta-agonists, inhaled anticholinergics, oral theophylline-containing medications, oral beta-agonists, and oral corticosteroids. All asthma medication prescriptions filled in 1993 were used for analysis. In the managed care setting, each prescription fill represents what is estimated to be a 1-mo supply of medication for those drugs intended for continuous use.

Socioeconomic indicators were obtained for each subject by using street address and geocoding software. Each street address was mapped to a census block group and median household income. Median family income and average household size for that block group were obtained on the basis of 1990 Census data (14) and adjusted for 1993 income levels. Each census block in the Detroit area contains an average of 970 citizens and approximately 400 households living in a distinct geographic area. We determined average income per occupant by dividing median household income by average household size and used this as an indicator of socioeconomic status. For purposes of determining a low socioeconomic group, we arbitrarily chose a level of income reflecting 150% of the established poverty guideline for a single-person household (i.e., annual income of less than $10,450 per occupant) as a breakpoint to determine the subjects assigned to the "low-income group."

Statistical Methods

Mean age and income for African Americans and Caucasians were compared using Student's t test. Gender and marital status were compared between the ethnic groups using a likelihood ratio chi-square test. Number of visits and prescriptions filled was compared using Poisson regression. Where Poisson regression was not appropriate, Wilcoxon rank sum tests were used. When the covariates were included in the model, either Poisson regression or least squares regression on the transformed visit or prescription filled variable was employed. All statements of "adjusted analysis" refer to a regression including the four covariates of age, gender, marital status, and income along with the main factor of interest, ethnicity. Pairwise testing of prescriptions filled among the two copay levels was performed at the 0.017 level to adjust for multiple testing within each prescription drug category.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Demographic and Socioeconomic Characteristics of the Study Population

Entry criteria were met by 2,073 subjects including 464 African American and 1,609 Caucasians. A comparison of the demographic and socioeconomic characteristics of the two ethnic groups is shown in Table 1. The groups were of similar age but compared with Caucasians, more African American subjects were female (69.8% versus 63.6%), single (64.7% versus 50.3%), and had lower income levels.

Analysis of the demographic characteristics of a subgroup of "low-income" African American and Caucasian subjects (those living in census blocks where the median income was less than $10,450 per occupant) is summarized in Table 2. The African Americans and Caucasians comprising the low-income subgroup were similar in age and gender, but the African American low-income subjects were more likely to be unmarried and lived in areas where the median income per occupant was still lower than the Caucasians within this subgroup.

We identified 60 patients, including 52 African Americans (11% of AA study population) and eight Caucasians (0.4% of C study population) that enrolled in the managed care HMO through Medicaid programs. Exclusion of these subjects from analysis of the low-income subgroup did not change the demographic analysis except that marital status was no longer significantly different between the ethnic groups (data not shown).

Location of Asthma Care Encounters

The frequency and types of physician encounters observed for African Americans and Caucasians are compared in Table 3. An "average" African American HMO member visited a primary care physician for asthma as frequently as an "average" Caucasian member (0.95 ± 1.24 encounters versus 0.93 ± 1.21, p = 0.81). However, African Americans were more frequently seen in the emergency department (0.71 ± 1.33 encounters versus 0.28 ± 0.64, p < 0.001) and were hospitalized more often (0.08 ± 0.31 hospitalizations versus 0.03 ± 0.28, p = 0.002), while asthma specialist visits were significantly less frequent (0.32 encounters ± 0.93 versus 0.5 ± 1.12, p = 0.002) than Caucasians.

Physician encounter data for the low socioeconomic subgroup is shown in Table 4. Compared with Caucasians within this subgroup, African Americans were still more frequently seen in the emergency department and more likely to be hospitalized with asthma, whereas visits to primary care physicians and asthma specialists were similar between the ethnic groups. In addition, low-income Caucasians were less likely to visit an asthma specialist as compared with non-low-income Caucasians (0.29 ± 0.73 visits/yr versus 0.53 ± 1.16, p = 0.006) Otherwise, the low-income subgroup data for African Americans and Caucasians were similar to that obtained for the ethnic profiles for the entire study population.

Adjusted analysis on physician visit profiles revealed that African American ethnicity was still significantly associated with higher emergency department visits for asthma (p < 0.001), asthma hospitalizations (p = 0.023) and few asthma specialist visits (p = 0.027) even after controlling for income, marital status, gender, and age differences. Additionally, although there was a strong relationship between African American ethnicity and higher emergency department visits, this relationship was significantly stronger among males compared with females (race by gender interaction p value = 0.02). Adjusted analysis also revealed that lower income level was associated with both increased emergency department use among the females (p = 0.013) and increased asthma hospitalizations among both genders (p = 0.038). Higher income level and female gender was associated with more asthma specialty visits (p = 0.002 and p < 0.001, respectively). Older age was associated with more primary care visits, specialty visits, and hospitalizations for asthma (all significant at the p < 0.05 level).

Within the low-income subgroup, adjusted analysis confirmed that African Americans were more likely to visit the emergency department (p < 0.001). However, ethnic differences in asthma specialty visits disappeared (p = 0.45) and the difference in hospitalization lessened to borderline significance (p = 0.07) after adjusting for age, marital status, income, and gender.

Fifty-two African Americans (11% of AA study population) and eight Caucasians (0.4% of C study population) enrolled in the managed care HMO through Medicaid programs. Excluding these patients from the low-income subgroups did not change the results of the analysis in any way.

Patterns of Prescription Medication Usage

Prescription medication profiles comparing African American and Caucasian drug utilization are shown in Table 5. African Americans were more likely to fill prescriptions for oral corticosteroids (p < 0.001) while paradoxically using less inhaled corticosteroids (p = 0.038). Other commonly used classes of prescription asthma medications were used at similar rates by the two ethnic groups.

Table 6 displays the prescription fill data for the "low socioeconomic group" of African Americans and Caucasians. The pattern of medication use is indistinguishable between the two groups except for a slightly increased rate of inhaled anticholinergic use (p = 0.04) in the low-income, Caucasian population. Exclusion of the 60 Medicaid subjects from the low-income subgroup resulted in no changes except to slightly increase the ethnic difference in the use of anticholinergics (increased in Caucasians).

After adjustment for the covariates of age, income, gender, and marital status, inhaled corticosteroid was marginally higher in Caucasians compared with African Americans (p = 0.055) which disappeared in the low-income subgroup analysis (p = 0.26). Similarly, the use of oral corticosteroids was higher in African Americans in the adjusted analysis (p = 0.026) but was not apparent in the low-income group only (p = 0.64). The use of oral corticosteroids was associated with low income throughout the study population (p = 0.032) in adjusted analysis. Inhaled beta-agonist prescription fills were higher among male African Americans compared with females (gender by ethnicity interaction p = 0.002) but this difference was not apparent in the low-income group (p = 0.53). Increased anticholinergic prescriptions were associated with Caucasians ethnicity (p = 0.016), low income (p = 0.009), and advanced age (p < 0.001).

Table 7 shows that the use of all medication types were significantly higher in subjects who had at least one visit with an asthma specialist. The 78 African Americans and 448 Caucasians who had seen an asthma specialist had similar rates of use for all classes of medications except for increased use of oral corticosteroids prescriptions (1.65 ± 3.1 versus 0.99 ± 2.2, p = 0.021) in the African American subjects.

In addition, patients in both ethnic groups who were not seen by asthma specialists had similar rates of use for all medication types (including inhaled corticosteroids) except for increased use of oral corticosteroids and inhaled beta-agonists in African Americans.

We also wished to evaluate whether the amount of required copayment for prescription medications would alter medication use profiles. We were able to confirm copayment amounts on 1955 (94%) subjects. Fifty-three patients, including 1.8% of African Americans and 3.0% of Caucasians, had no prescription coverage. Our analysis showed that on average African Americans had lower copay amounts than Caucasians (76.7% of African Americans with a $2 to $3 copay compared with 70.6% of Caucasians, p = 0.04). In regard to each medication class, the prescription fill rates were greater in the lower copayment group but statistical significance was achieved only for the oral corticosteroids and oral beta-agonists when a $5 to $10 copay level was compared with a $2 to $3 dollar copay (Table 8). Our methods did not allow us to track prescription fill rates for the "no coverage" patients since billing records would not be generated by the HMO.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Our findings indicate that even when financial barriers to asthma care are reduced by participation in an HMO, there are demonstrable differences in the patterns of asthma care between African Americans and Caucasians which appear to be partially explained by ethnic differences in socioeconomic status.

African Americans and Caucasians in our study were seen for asthma care at similar rates by primary care physicians. This is contrary to the reports which have shown that urban African Americans typically rely on emergency departments as a surrogate for other outpatient services for asthma (10, 15) and for medical care in general (11). Our findings likely reflect the fact that our subjects were all enrollees of an HMO that promotes primary care encounters for all chronic diseases and participate in a system where financial barriers to receiving such care are minimal.

In spite of the similar rates of primary care asthma visits, we found that African Americans were more likely to seek care in the emergency department and be hospitalized with asthma. Although not a totally reliable indicator of asthma severity, we believe that this visit profile suggests asthma in African Americans is comparatively less well controlled than the Caucasian population in our study. This finding is not surprising as many studies have been published describing increased numbers of hospitalizations and increased asthma mortality rates in African Americans (16). Because African American ethnicity is associated with increased asthma morbidity and mortality, it is logical that this population should be strongly encouraged to seek and employ intensive asthma management strategies including education, aggressive pharmacologic treatment, and asthma specialist consultation.

Paradoxically, the African Americans in our study had low rates of asthma specialist consultation. Although we were unable to analyze whether a referral to an asthma specialist was ever provided, we were surprised to find that the increased rate of hospitalizations and emergency department visits did not trigger a concomitant increase in asthma specialist visits. Our findings and those of other investigators suggest that asthma specialists, compared with generalists, take a more aggressive pharmacologic approach to asthma (19).

The relative contributions of African American ethnicity and low socioeconomic status to increased asthma morbidity and mortality continue to be a matter of debate (7, 8). Our analysis of a low-income subgroup of African Americans and Caucasians suggests that primary care asthma visit rates were not affected by income in either ethnic group. However, asthma specialist visits were decreased among the low-income Caucasians to a degree that rates were similar to the African American population. In spite of this, emergency department and hospitalization rates in the low-income subgroup revealed persistently higher utilization rates among African Americans. This suggests that there are likely to be some persisting barriers to specialty referral among the low-income population regardless of ethnicity. We hypothesize that these barriers may include geographic discrepancies in the availability of asthma specialists in the inner city due to a relative paucity of providers or due to a lack of transportation available to inner city residents. In addition, the presence of sociocultural factors may influence low-income patients to perceive their asthma as a less urgent problem than other concomitant life stressors. Finally, there is the possibility that primary care physicians perceive low-income patients as unlikely to maintain compliance with the multidrug regimens which are typically suggested by asthma specialists.

The finding that African American ethnicity and low socioeconomic status is associated with decreased asthma specialist evaluation suggests that the populations with the most severe asthma have inappropriately low rates of evaluation by physicians with the most expertise in treating asthma. It is difficult to establish a cause and effect relationship between specialist evaluation and favorable asthma outcomes because this analysis would require an adequate indicator of "asthma severity" which typically is not separable from medication usage and health care utilization. However, in our study, the ethnic and socioeconomic groups that had the highest rates of inhaled beta-agonist use, oral corticosteroid use, emergency department visits, and hospitalizations were the same groups that had low asthma specialist utilization. A recent report suggests that allergists within a large HMO setting care for patients with more severe asthma and more closely adhere to current management guidelines (22). Certainly, this more aggressive asthma management approach may be beneficial to the low-income and African American populations as therapy with inhaled corticosteroids is associated with decreased asthma morbidity and mortality (23, 24).

Our analysis revealed differences in medication usage patterns which appeared to be more a function of socioeconomic status and asthma specialist evaluation than ethnicity. However, the lower rate of inhaled corticosteroids by the low-income and African American populations appeared to be particularly inappropriate in view of this group's pattern of high oral steroid use. Because frequent oral corticosteroid "bursts" and inhaled beta-agonist use is likely an indicator of poorly controlled asthma, a similar (and slightly lower) rate of use of inhaled anti-inflammatory medications among these populations suggests that physicians continue to underutilize inhaled anti-inflammatory agents in this high-risk group. This finding becomes striking when taking into account the fact that African Americans in our study had high rates of emergency department use and frequent hospitalizations in spite of the high rates of oral steroid use. National and international guidelines suggest that the regular use of inhaled corticosteroids is appropriate for most asthma patients, excluding only those with the most infrequent and mild symptoms (25, 26). Our findings are consistent with the findings of Haire-Joshu and coworkers who found that patient attitudes among low-income African Americans who receive asthma treatment from acute care settings include emphasis on self-treatment of acute asthma symptoms and not preventative care (27). In contrast to the National Heart, Lung, and Blood Institute (NHLBI) guidelines and National Asthma Education Program which encourage early use of anti-inflammatory medications, the relative underuse of inhaled corticosteroids appears to have progressed, particularly among low-income populations (28).

The African Americans identified in our study differed from the Caucasians by having a greater proportion of female subjects, unmarried subjects and also had a significantly lower socioeconomic level. Adjusted analysis did not show any gender effects on medication use except for increased beta-agonist use among African American males. However, females were more likely to visit asthma specialists and among African Americans, emergency department utilization was significantly less common in females. Therefore, the ethnic disparities in gender noted in the study population suggest that our data would be even more striking regarding asthma encounter locations if the African American males had been "equally represented" in the study population. Adjusted analysis did not reveal any independent effects of marital status on medication use or asthma visit profiles.

There are several limitations to this study. Although the study was designed to analyze a population where financial barriers to medical care were minimized, some payment, in the form of copays for outpatient visits and prescriptions, was still required of many subjects. This may have been a more significant barrier for the lower socioeconomic group in our study. However, our copayment analysis did show that African Americans in our study did have lower average copayment levels, suggesting this limitation was not a major factor in our study population.

We also attempted to adjust for socioeconomic discrepancies among African Americans and Caucasians by analyzing only those subjects living in households earning less than $10,450 per occupant. Our attempt at controlling for income levels was limited by the fact that African Americans in this subgroup still had lower median income than the Caucasians. However, when we attempted to analyze truly "income equivalent" groups, the low number of Caucasians precluded meaningful analysis.

The use of "number of prescriptions filled" is also a crude estimate of actual use. Physician-provided "samples" or use of previously prescribed medications kept in the home could lead to an underestimation of actual use. In contrast, filling of prescriptions does not necessarily mean the medications are being used and in this way, the "number of prescriptions filled" may overestimate true medication use. We were also unable to determine if low medication fill rates were due to failure of the subjects to fill a prescription written by their physician or whether the physician failed to prescribe the medications.

Finally, the use of a cross-sectional design eliminates the ability to evaluate cause and effect relationships between factors such as location of asthma visits and medication use, or the relationship between emergency department visits or hospitalization and subsequent referral to an asthma specialist. These issues would be best targeted for evaluation by a prospective study designed to increase our understanding of the factors that lead to ethnic differences in asthma treatment within a managed care setting.

We propose that the methods used in this study can be powerful tools to evaluate the effectiveness of interventions aimed at improving appropriate asthma management. For example, physician- and patient-specific medication use profiles can be monitored to analyze the effects of randomized intervention studies which aim to lower the threshold for providers to prescribe and patients to regularly administer preventative asthma medications. Subsequently, asthma visit profiles can be monitored to evaluate whether successful pharmacologic interventions actually result in an anticipated improvement in emergency department visits and hospitalizations.

In conclusion, within an urban health maintenance organization, African Americans exhibited patterns of asthma care including a relatively high rate of emergency department visits and hospitalizations compared with Caucasians, which could not be fully explained by racial differences in socioeconomic status. The fact that African Americans used less inhaled anti-inflammatory medications than Caucasians in our study appeared to be primarily an effect of lower socioeconomic status linked to lower rates of asthma specialist evaluation. These results suggest that novel approaches need to be developed that will notify physicians and patients when more aggressive asthma management is likely to be beneficial to high-risk asthma populations. This more intensive approach may include additional asthma education and advanced self-management techniques, more aggressive anti-inflammatory pharmacologic treatment, and expedient referral to an asthma specialist. Randomized studies need to be designed to evaluate which of these interventions are likely to most effectively modify the persisting pattern of asthma "crisis management" among inner city populations.