Asthma Care Quality for Children With Minority-Serving Providers

Objective  To compare asthma care quality for children with and without minority-serving providers.

Design  Cross-sectional telephone survey of parents, linked with a mailed survey of their children's providers.

Setting  A Medicaid-predominant health plan and multispecialty provider group in Massachusetts.

Participants  A total of 563 children with persistent asthma, identified by claims and encounter data.

Main Exposure  Whether the child's provider was minority serving (>25% of patients black or Latino).

Outcomes  Parent report of whether the child had (1) ever received inhaled steroids, (2) received influenza vaccination during the past season, and (3) received an asthma action plan in the past year.

Results  In unadjusted analyses, Latino children and those with minority-serving providers were more likely to have never received inhaled steroids. In adjusted models, the odds of never receiving inhaled steroids were not statistically significantly different for children with minority-serving providers (odds ratio [OR], 1.29; 95% confidence interval [CI], 0.63-2.64), or for Latino vs white children (OR, 1.76; 95% CI, 0.74-4.18); odds were increased for children receiving care in community health centers (OR, 4.88; 95% CI, 1.70-14.02) or hospital clinics (OR, 4.53; 95% CI, 1.09-18.92) vs multispecialty practices. Such differences were not seen for influenza vaccinations or action plans.

Conclusions  Children with persistent asthma are less likely to receive inhaled steroids if they receive care in community health centers or hospital clinics. Practice setting mediated initially observed disparities in inhaled steroid use by Latino children and those with minority-serving providers. No differences by race/ethnicity or minority-serving provider were observed for influenza vaccinations or asthma action plans.

Health care for minority patients has been shown to be concentrated among a subset of providers. Minority patients are more likely to receive care from community health centers, academic institutions, and providers who are foreign medical graduates, nonphysician clinicians, or not board certified.^1-4 Data from adult populations suggest that racial/ethnic disparities in health care quality are driven in part by differences between providers.^5,6 Data are mixed as to whether differences are due to factors associated with having a minority-serving provider such as provider characteristics or practice setting⁷ or to an independent association between quality and the proportion of minority patients in a practice.^8,9 Possible explanations for the latter are that providers who serve larger minority patient populations encounter more complex medical and social issues, language barriers, and lower health literacy.^4,10 While most studies in this area have been in adult populations, there is some evidence that minority infants are more likely to receive care at underperforming hospitals. However, data for pediatric care in the ambulatory setting are lacking.¹¹

Asthma is one of the most common chronic diseases of children, and racial/ethnic disparities have been well documented in childhood asthma treatment and outcomes.^12-14 Guidelines for the outpatient management of children with persistent asthma recommend daily inhaled steroid use, yearly influenza vaccination, and use of asthma action plans.¹⁵ However, minority children are often less likely to receive these services.^12,16-18 A number of studies have explored the relationship of race/ethnicity and other patient- and provider-level factors with asthma care^16,19-22 including one in which providers with large proportions of minority patients were less likely to report prescribing daily inhaled steroid use, even after controlling for practice type.²³ However, more data are needed to evaluate the role of having a minority-serving provider on quality of asthma care in the context of associated patient, provider, and practice setting characteristics.

Thus, this study's objectives were to examine whether children with and without minority-serving providers received asthma care of similar quality, as measured by receipt of inhaled steroids, influenza vaccination, and asthma action plans.

This was a cross-sectional study using linked parent and provider survey data on children with asthma in 2 health care systems, Harvard Vanguard Medical Associates, a large multispecialty group practice in the Boston metropolitan area with privately insured and Medicaid-insured patients, and Neighborhood Health Plan, a Medicaid-predominant health plan in Massachusetts. The study was approved by the institutional review boards of Harvard Pilgrim Health Care and Boston University School of Medicine.

The study included children aged 2 to 12 years with persistent asthma. Eligible children were identified from computerized claims and electronic medical records based on an algorithm modeled on Health Employer Data Information Set (HEDIS) criteria for persistent asthma.²⁴ Children were included if they had a clinical encounter with an International Classification of Diseases, Ninth Revision (ICD-9) code for asthma for an emergency department visit, a hospitalization, or 2 outpatient visits, plus any of the following during the prior 12 months: (1) 4 or more asthma medication dispensings; (2) 1 or more emergency department visits or hospitalization for asthma; or (3) 4 or more outpatient visits in which asthma was the primary diagnosis and 2 or more prescriptions for asthma medication. Children were excluded if we did not have a completed survey from their asthma care provider.

Parents of eligible children were surveyed by telephone in English or Spanish between March 2005 and March 2006 using a structured, closed-ended, 30-minute interview. If parents did not respond after 12 telephone attempts, a mailed version of the survey in English or Spanish was sent.

We sought to identify the nonspecialist physician or advanced practice clinician who was the child's primary asthma care provider by asking in the parent survey for the name of the provider most in charge of the child's asthma care. If this did not yield a valid response, we used the provider identified in computerized data as the child's primary care provider.

Eligible providers were mailed a structured, closed-ended survey in November 2005, with 2 additional mailing attempts for nonresponders. Providers were asked questions about their prescribing patterns for patients with asthma in addition to questions about themselves and their practices.

Data from each provider's survey were linked to parent survey data for each child for whom that provider was the primary asthma care provider.

The primary outcome variables were 3 measures of recommended care for children with persistent asthma that are consistent with national guidelines: receipt of inhaled steroids, asthma action plans, and influenza vaccination.¹⁵ We measured receipt of inhaled steroids and asthma action plans by parent report. Receipt of inhaled steroids was defined as report of current use of inhaled steroids or a “yes” response to the question of whether a provider had ever recommended that the child take a daily inhaled steroid. Receipt of asthma action plans was defined as “yes” responses to 2 questions: (1) whether the child's physician had given them a written treatment plan in the past 12 months for the child's asthma, and (2) whether this plan gave ways to determine how severe an asthma flare-up was such as green, yellow, and red zones. We measured receipt of influenza vaccination from parent report or electronic records indicating influenza vaccination in the prior season.

The primary predictors of interest in this study were the child's race/ethnicity as reported by the parent and whether the child received asthma care from a minority-serving provider. The child's race/ethnicity was categorized as white, black, Latino, or multi/other (children with more than 1 race or other races). If the parent described the child's ethnicity as Latino, the child was categorized as Latino regardless of race. The providers were categorized as minority-serving if they reported in the provider survey that more than 25% of their patient population was black or more than 25% of their patient population was Latino.

Other child-level covariates of interest included age, family income as a percentage of the federal poverty level, parental education, whether the child lived in a single-parent household, and insurance type. To adjust for asthma morbidity, we used computerized data from the prior year to measure whether the child had any asthma-related emergency department use or hospitalizations and to determine the number of bronchodilator and oral steroid prescriptions filled.

Other provider-level covariates included the provider's self-reported race/ethnicity, number of years in practice, practice type, and whether more than 25% of the provider's patients had Medicaid insurance or were uninsured.

We conducted bivariate analyses using χ² tests to evaluate whether child and provider-level characteristics were associated with the asthma care outcomes of interest (never receiving inhaled steroids, no influenza vaccination in the prior season, and no asthma action plan in the past 12 months).

We conducted separate multivariate analyses for each of the 3 outcomes of interest, modeling the odds of never receiving inhaled steroids, of not receiving an influenza vaccination, and of not receiving an asthma action plan. Models included patient and provider-level covariates that were associated with the outcome at P < .10 in bivariate analyses, with the primary predictors of interest (child race/ethnicity and minority-serving provider) forced into the models. When there was a significant bivariate association between race/ethnicity and one of the outcomes, we assessed the mediating effects of other covariates with an iterative, forced-entry modeling approach. The unadjusted first-stage model evaluated the association between race/ethnicity and the outcome of interest. Next, in the second stage model, patient-level covariates were added to the model if they were associated with the outcome at P < .10 in bivariate analyses. In the third stage model, minority-serving provider was forced into the model. Lastly, in the final model, provider-level covariates were added if they were associated with the outcome at P < .10 in bivariate analyses. All multivariate analyses were done using generalized linear mixed models to account for clustering of children within providers.

Of 1047 eligible parents, 754 completed the parent survey, for a response rate of 72%. Of these, 89% were completed by phone in English, 3% by phone in Spanish, and 8% by mail. Of 344 eligible providers, 251 completed the provider survey, for a response rate of 73%. Completed and linked parent and provider surveys were available for 563 children and 138 providers. The number of children per provider in the final study sample ranged from 1 to 32, with a median of 2 children per provider. Based on data from the parent survey, there were no statistically significant differences in terms of race/ethnicity, income, parent education, and single-parent household between children in the study sample and those excluded because they lacked linked provider data.

Of the 563 children in our study, 45% were white, 22% were black, 19% were Latino, and 14% were multiple/other races (Table 1). The 76 children with multiple/other races included children who were Asian (35) or Native American (1), or had multiple (31) or other races (9). Twenty-one percent had family incomes below the poverty level, 25% had parents with a high school education or less, 32% came from single-parent households, and 27% had insurance coverage from Medicaid. Fifty percent of children in our sample had minority-serving providers. Our study population differs somewhat from the general population of children in Massachusetts, of which 74% are white, 9% are black, 10% are Latino, 19% are below the poverty level, and 24% have Medicaid coverage.²⁵

Minority children tended to receive asthma care from different sources than white children. Care for minority children was concentrated among minority-serving providers, with 79% of black children and 69% of Latino children having a minority-serving provider, while only 29% of white children did (P<.001). Latino and black children were more likely to receive care from providers at community health centers (33% and 14%, respectively) than white children (4%), and Latino children were less likely to receive care at multispecialty group practices compared with black and white children (55% vs 77% vs 78%, respectively; P < .001).

Twenty-three percent of Latino children had never received inhaled steroids, compared with 11% of black children and 10% of white children (P = .004) (Table 2). Children with minority serving providers were more likely to have never received inhaled steroids than children whose providers were not minority serving (17% vs 11%, respectively; P = .03). Children had greater likelihood of never receiving inhaled steroids if their provider practiced in a community health center (36%) or hospital clinic (26%) than in a single-specialty or solo practice (12%) or multispecialty group practice (9%) (P < .001).

The percentage of children who did not receive an influenza vaccination in the prior year did not vary by race/ethnicity or by having a minority-serving provider. Likewise, the percentage who had not received an asthma action plan did not vary by race/ethnicity or by having a minority-serving provider.

We used a series of multivariate models to assess whether racial/ethnic variation in receipt of inhaled steroids was mediated by having a minority-serving provider and other patient- and provider-level factors (Table 3). In the unadjusted model, Latino children had significantly greater odds of never receiving inhaled steroids compared with white children (odds ratio [OR], 2.64; 95% confidence interval [CI], 1.42-4.94), a finding that persisted after adjusting for other patient-level covariates in the second-stage model (OR, 2.35; 95% CI, 1.09-5.09). However, when having a minority-serving provider was added as a covariate in the third-stage model, Latino children no longer had significantly increased odds of never receiving inhaled steroids (OR, 1.89; 95% CI, 0.84-4.28). In this model, the odds of never receiving inhaled steroids were not significantly different for children with minority-serving providers compared with children with other providers (OR, 1.90; 95% CI, 0.98-3.70). When remaining provider-level variables were added, the model would not converge because of correlation between variables for having a minority-serving provider and having a provider with more than 25% of patients with Medicaid or uninsured, so the latter variable was excluded. In the final model, with provider practice type added, the OR for having a minority-serving provider was reduced to 1.29 (95% CI, 0.63-2.64). Provider practice type was significantly associated with never receiving inhaled steroids, with increased odds for children whose provider practiced at a health center (OR, 4.88; 95% CI, 1.70-14.02) or hospital clinic (OR, 4.53; 95% CI, 1.09-18.92) compared with a multispecialty group practice. This sequence of results suggests that the initially observed associations of inhaled steroid use with patient race/ethnicity and minority-serving provider were largely mediated by the practice setting.

As in bivariate analyses, our multivariate models indicated no significant differences in receipt of influenza vaccinations or asthma action plans by race/ethnicity or by having a minority-serving provider. No significant association was observed between not receiving an influenza vaccination and either race/ethnicity or having a minority-serving provider in a model that controlled for age, single-parent household, and number of systemic steroid prescriptions dispensed (Table 4). We also did not see a significant relationship between not receiving an asthma action plan and race/ethnicity or minority-serving provider in a model that controlled for age, insurance type, and number of bronchodilator fills (Table 4).

Overall, this study did not find consistent differences in the quality of childhood asthma care delivered by minority-serving providers. Receipt of influenza vaccination and asthma action plans were similar for children with and without minority-serving providers. Children with persistent asthma were less likely to have ever been recommended inhaled steroids, based on parent reports, if they received care from community health centers or hospital clinics. The practice setting appeared to mediate variation in receipt of inhaled steroids by patient race/ethnicity and whether the provider was minority serving. Contrary to our original hypothesis, we did not find that having a minority-serving provider was independently associated with lower quality of asthma care.

Our finding that racial/ethnic disparities in receipt of inhaled steroids were attenuated after adjusting for provider-level factors is consistent with studies in other populations that have shown that racial/ethnic disparities are driven in part by where minority children receive care.^5,6,19,26 Our results suggest that for racial/ethnic disparities in inhaled steroid use, the setting of care plays a more important role than having large numbers of minority patients itself. We did not find an independent association between quality of care and having a minority-serving provider, in contrast to other studies that focused on adult populations receiving care from minority-serving hospitals rather than minority-serving outpatient providers.^8,9 Additionally, our results may differ because our analysis controlled for patient-level socioeconomic and other variables, was conducted at the patient rather than provider level, and used process measures of quality rather than health outcomes. Variation across studies may be seen because being a minority-serving provider may be associated with measured and unmeasured characteristics related to practice type that can affect quality such as organizational structure and staffing, provider training, payer mix, geography and residential segregation, work environment, and the availability of health information technology and other resources.^2,4,10,27,28 Practice-level factors such as having policies to promote access and continuity and the use of reports to clinicians have been shown to be associated with greater use of preventive asthma medications for children.²¹ Safety net practices serving large numbers of uninsured and Medicaid-insured patients may have fewer resources to devote to policies that promote adherence to guidelines.^29,30 Because of the correlation between having large proportions of minority and Medicaid patients, we were unable to assess the adjusted association between having large proportions of Medicaid patients and receipt of inhaled steroids. However, other studies⁹ that controlled for both found that having a minority-serving provider remained significantly associated with lower quality.

It is reassuring that we did not see differences by race/ethnicity and having a minority-serving provider for children's receipt of influenza vaccinations and action plans. As seen in other studies,³¹ racial/ethnic disparities may be present for some services and not others. Although differences by practice type were noted for receipt of inhaled steroids, health centers and hospital clinics appear to provide the same level of quality as other practice types for influenza vaccinations and action plans. Prescribing inhaled steroids may be more a more complex and nuanced task that is subject to multiple provider and patient factors that differ across practice types such as the extent of multiple competing medical and social issues, the availability of resources to support staff to conduct education and management about medications, the availability of translation services, or perceptions about adherence related to patient mix.

Our findings highlight a number of important policy issues related to quality improvement and the reduction of racial/ethnic disparities in asthma care for children. Our results show that disparities in quality may be not be present across all measures of asthma care, and that apparent disparities in some measures for minority children and those with minority-serving providers may be mediated by practice-site factors. This suggests that a targeted approach to improving quality and reducing disparities may be warranted. For example, focusing on improvement in inhaled steroid use in community health centers and hospital clinics might be a way to improve quality generally and for minority children in particular. Although most minority children receive care in private or group practices, 30% of black and 41% of Latino children receive care at community health centers,¹ and minority children in our study and others^1,32 are more likely than white children to receive care in health centers and hospital clinics.

Debate continues regarding whether improvements in quality of care for racial/ethnic minorities are best achieved by efforts specifically targeting these populations and the places where they receive care,³³ or whether broad-based quality improvement efforts will improve care for racial/ethnic minorities and reduce disparities.³⁴ Policies promoting pay-for-performance and public reporting will need to consider how rewards based on quality measures will affect providers in settings that serve minority children and how such policies may indirectly affect racial/ethnic disparities.^35-37 If baseline quality is lower for providers in settings that have disproportionate numbers of minority children, these providers could be penalized in pay-for-performance or public reporting programs through decreases in revenue or patient volume.

Our study population was drawn from a Medicaid-predominant health plan and a multispecialty provider group in Massachusetts, so our findings regarding quality of asthma care for minority children may not be generalizable to populations in other areas with different racial/ethnic composition, practice patterns, and health care systems. Although we controlled for many important patient and provider-level factors in assessing asthma care processes, we could not assess the effects of unmeasured provider-level factors such as training, cultural competence, technological resources, or availability of care management programs. Provider-level factors such as practice type may be correlated with unmeasured patient characteristics and provider-patient interactions in ways that could affect the quality outcomes studied. Our sample size was not large, leading to wide CIs in some of our findings. Despite this, we were able to detect statistically significant results, with CIs encompassing clinical and policy-relevant differences. Reliance on self-reported data is a limitation, and parents may not accurately recall when the provider had recommended or prescribed inhaled steroids or given them an asthma action plan. We attempted to capture provider behavior by asking about services that were provider initiated (such as having inhaled steroids recommended or given), rather than parent-initiated (such as adherence). Our use of claims-based HEDIS measures to identify children with persistent asthma may have led us to include some children with mild intermittent asthma for whom inhaled steroids might not be appropriate to recommend. If this were the case, we would expect higher rates of not receiving inhaled steroids for children whose asthma was erroneously identified as persistent. However, of the children in our study sample who did not meet more stringent criteria for persistent asthma (parent report of having >4 days or >1 nights with symptoms or >4 days of albuterol use in the past 2 weeks), rates of not receiving inhaled steroids were not significantly greater than for children who met these more strict criteria for persistent asthma (14.7% vs 12.2%, respectively; P = .4).

Children with persistent asthma were less likely to receive inhaled steroids if they received care in community health centers or hospital clinics. The practice setting appeared to mediate the initially observed disparities in inhaled steroid use by children who were Latino or received care from minority-serving providers. Differences by setting of care, minority-serving provider, and race/ethnicity were not observed for influenza vaccination and asthma action plans. Based on the limited set of asthma measures studied, our findings do not support the hypothesis that minority-serving providers deliver lower-quality asthma care. However, efforts to enhance the consistency of inhaled steroid-prescribing practices should consider focusing on settings such as community health centers and hospital clinics, which disproportionately serve minority children.

Correspondence: Allison A. Galbraith, MD, MPH, Center for Child Health Care Studies, Department of Population Medicine, Harvard Pilgrim Health Care and Harvard Medical School, 133 Brookline Ave, 6th Floor, Boston, MA 02215 (alison_galbraith@harvardpilgrim.org).

Accepted for Publication: July 13, 2009.

Author Contributions: Dr Galbraith had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Galbraith, Smith, Bokhour, Hohman, and Lieu. Acquisition of data: Bokhour, Miroshnik, Hohman, Gay, and Lieu. Analysis and interpretation of data: Galbraith, Smith, Bokhour, Miroshnik, Sawicki, Glauber, Hohman, and Lieu. Drafting of the manuscript: Galbraith, Miroshnik, Glauber, and Lieu. Critical revision of the manuscript for important intellectual content: Galbraith, Smith, Bokhour, Sawicki, Glauber, Hohman, Gay, and Lieu. Statistical analysis: Galbraith and Smith. Obtained funding: Lieu. Administrative, technical, and material support: Smith, Miroshnik, Hohman, and Gay. Study supervision: Smith, Glauber, and Lieu.

Financial Disclosure: Dr Sawicki reports having received grant support from Astra Zeneca for a separate asthma-related project. No other author reports any potential conflict of interest.

Funding/Support: This study was supported by National Institute of Child Health and Human Development grant R01 HD044070. Additional support for authors' individual efforts was provided by the Department of Veterans Affairs, Health Services Research and Development Service (Dr Bokhour); Agency for Healthcare Research and Quality grant T32 HS000063-13 to the Harvard Pediatric Health Services Research Fellowship Program (Dr Sawicki); and in part by Mid-Career Investigator Award in Patient-Oriented Research grant K24 HD047667 from the National Institute of Child Health and Human Development (Dr Lieu).

Disclaimer: The funding providers did not play a role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; or preparation, review, or approval of the manuscript.

Additional Contributions: We would like to thank Neighborhood Health Plan, Jack Lasche, MD, and Harvard Vanguard Medical Associates, and Pauline Sheehan, MD, for their collaboration and support of this work; and Ken Kleinman, ScD, from the Department of Population Medicine at Harvard Medical School and Harvard Pilgrim Health Care, for assistance with the statistical models and helpful comments on the manuscript. We also greatly appreciate the contributions of our study's research assistants.