Adverse asthma outcomes, such as hospitalization and death, are largely preventable, but rates are higher among some minority children compared to white children.(1-6) In 2004-2005, although black children had 1.6 times higher asthma prevalence compared to white children, they were 4 times more likely to be hospitalized and over 7 times more likely to die from asthma.(1) Federal efforts to reduce childhood asthma disparities include not only broad initiatives to track and address health inequities such as Healthy People(7) and the National Partnership for Action,(8;9) but also a coalition of federal agencies formed specifically to address childhood asthma disparities under the auspices of the President’s Task Force on Environmental Health Risks and Safety Risks to Children. This group released a Coordinated Federal Action Plan to Reduce Racial and Ethnic Asthma Disparities in 2012.(10)

Health disparities are measured to track progress toward their elimination, and to more effectively target interventions to improve outcomes in the most adversely affected groups.(10) Analytic methods are as important in measuring trends in disparities as the quality of the underlying data because different methods may lead to different conclusions.(11) For childhood asthma, a consideration that could influence interpretation of asthma outcome disparities is asthma prevalence differences between race/ethnic groups. Asthma outcome rates calculated using the entire population of children in a race/ethnic group as the denominator--population based rates (PBRs)--provide an overall measure of asthma impact for each group. However, because asthma prevalence varies between race/ethnic groups, PBRs may not provide specific information about the risks of poor outcomes for those with asthma in each group. Use of at-risk outcome rates (ARRs), which measure outcomes among those with asthma, removes the contribution of underlying prevalence differences between groups from the outcome rates. Limiting the denominator to only the group at risk for adverse asthma outcomes also acknowledges the nature of interventions to address asthma. There is no unified understanding of the factors driving the continued rise in asthma prevalence, and no widely accepted primary prevention interventions for asthma, but there are evidence-based interventions to control asthma once it develops. These secondary and tertiary prevention measures are directed toward reducing adverse outcomes among persons who have asthma. Therefore, analyzing ARRs in addition to PBRs may offer additional insight into racial disparities in asthma outcomes. We analyzed recent trends in asthma prevalence and outcomes among children by race (black and white) in national data sets from the National Center for Health Statistics (NCHS). We compared trends in disparities calculated using PBRs of asthma outcomes to those using ARRs.

Nationally representative data from the National Center for Health Statistics (NCHS) were used to calculate estimates of asthma prevalence and outcomes (asthma attack prevalence, health care use, and mortality) among children aged 0-17 years. Trends were assessed from 2001, the year in which the measure of current asthma prevalence was first used in the National Health Interview Survey (NHIS), to 2010, the most recent year of available data from all the included NCHS data sets. Race groups included in the study were limited to black and white race (single race without regard to ethnicity) because sample sizes do not always permit analysis for other race groups, and because Hispanic origin is not available across all the included data sets.

In 2001, an estimated 4.4 million white children had asthma, and in 2010, there was no statistically significant change in that estimate (4.5 million). In 2001, an estimated 1.2 million black children had asthma, and in 2010, an estimated 1.7 million were affected. From 2001 to 2010 (Figure 1), the difference in current asthma prevalence rates between black and white children increased (p-value<0.01). Among white children, the APC in the percent with current asthma from 2001 to 2010 was not statistically different from zero, indicating no trend over the period in current asthma prevalence (asthma prevalence during the period ranged between 7.4% and 8.2%). In contrast, current asthma prevalence among black children increased by an average of 3.6% per year from 2001 to 2010 (range 11.4% to 16.8%). The black/white rate ratio of asthma prevalence increased from 1.4 in 2001 to 2.0 from 2007 onward. The increased racial disparity in asthma prevalence among children impacts interpretation of disparities in outcome rates, as subsequent analyses show.

Figure 2 shows asthma attack prevalence PBRs and ARRs from 2001 to 2010 (see also Table I). The PBR is higher for black children than white children and did not change significantly for either black or white children from 2001 to 2010. There was also no change in the disparity between black and white children in PBR asthma attack prevalence over the period. In contrast, the ARR for asthma attack prevalence was initially similar for white and black children, and then declined significantly for both black and white children. Because the ARR declined more rapidly among black children (−2.3% per year) than among white children (−1.2% per year), white children had higher rates compared to black children at the end of the period and the disparity increased significantly.

For both PBR and ARR rates, black children had higher rates of ED visits compared to white children (Figure 3, Table I). Over the period, there were no significant changes in ED visit PBR trends for black or white children, and there was no change in the black/white disparity. In contrast, the ARR for asthma ED visits decreased significantly among black children, and the black/white disparity in ED asthma visit ARRs also decreased. When we performed a sensitivity analysis excluding records with imputed race, the results were similar with one exception. While there was a significant decline in the black/white disparity in ED visits ARRs in the main analysis, there was no significant change detected by the sensitivity analysis using the Joinpoint analysis. However, when AHRQ criteria were applied, the decline in the black/white disparity in ARRs was significant (see Online Repository for more detailed results).

Rates of hospitalization decreased among black and white children from 2001 to 2010 for both PBRs and ARRs (Figure 4, Table I)). There was no statistically significant change in disparity between black and white PBRs. In contrast, the black/white disparity in hospitalization ARRs decreased. In a sensitivity analysis, we estimated the most conservative (smallest) black/white disparity by assigning records with unknown race to the “white” category. The results were the same: all trends were in the same direction with the same statistical significance, and the same patterns in disparity trends were observed (see Online Repository).

For asthma deaths, PBRs did not change significantly for either black (insignificant increase) or white children (insignificant decrease). However, the trend for the racial disparity in PBRs increased significantly (Figure 5, Table I). In contrast, the ARR declined for both groups, by 3.4% per year for white children and by 3.6% per year for black children, but there was no change in racial disparity in ARR for asthma deaths over the period.

We used the AHRQ criteria to assess changing disparities over time as a sensitivity analysis. The results were the same as those presented above with the one exception already noted for the sensitivity analysis for unknown race for ED visits (see Online Repository).

The magnitude of disparities in 2010 in asthma outcomes are shown in Figure 6 as black/white rate ratios for PBRs and ARRs. For all 4 outcomes, the rate ratios for ARRs are lower than those for PBRs, although the 95% confidence intervals generally overlap. Among the general population, black children were 1.7 times more likely (95% CI 1.4, 2.1) to have ≥1 asthma attack in the past 12 months compared to white children. In comparison, the rate ratio for ARRs was 0.9 (95% CI 0.7, 1.2) showing a similar likelihood for black and white children with asthma to have ≥1 asthma attack. For asthma ED visits, the black/white rate ratio for PBRs was 4.9 (95% CI 2.9, 8.2) versus 2.6 (95% CI 1.5, 4.4) for ARRs. For asthma hospitalizations, the rate ratios were 3.3 (95% CI 1.2, 8.9) for PBRs versus 1.7 (95% CI 0.6, 4.7) for ARRs. The magnitude of disparities remains highest for asthma deaths: comparing the PBRs, black children were 7.1 times more likely to die from asthma than white children (5.2, 9.7) and when comparing the ARRs, black children were 4.1 more likely to die compared to white children (95% CI, 2.9, 5.8).

Different pictures of trends in racial disparities in childhood asthma are presented by PBRs, which reflect the general burden of a condition in a population group, and ARRs, which show the impact only among those at risk for the outcomes (Table II). The racial disparity in asthma prevalence increased in the early 2000s, and there are few broad, effective interventions to prevent asthma from developing. Although new research is identifying diverse areas to design primary prevention efforts,(24) and the Federal Action Plan to Reduce Racial and Ethnic Asthma Disparities included as one of the four main strategies: “accelerate efforts to identify and test interventions that may prevent the onset of asthma among ethnic and racial and minority children,”(10) most existing asthma prevention interventions are designed to decrease severity and improve control among those with asthma. Examination of ARRs may provide a more specific perspective of group differences in the effectiveness of those interventions than do PBRs. When racial differences in asthma prevalence are accounted for with ARRs, the racial disparities picture appears more favorable although disparities remain in 2010 for asthma ED visits and deaths regardless.

The one area where white children with asthma fared worse than black children was controlling symptoms, with ARRs showing a greater proportion of white children with ≥1 asthma attack in the past 12 months and a widening disparity over time. This pattern does not translate into greater rates of severe outcomes among white children, and it could be related to the observation in previous studies that among children with asthma, white children have higher outpatient visit rates.(1) During these encounters, milder symptoms might be recognized and treated before escalating into exacerbations requiring emergency care. Similarly, children in minority ethnic groups, low socioeconomic groups, and with greater disease severity have been shown to have lower accuracy in asthma symptom perception compared to referent groups.(25;26)

While ARRs account for different asthma prevalence between black and white children, they do not necessarily account for differences in asthma severity. Though we were unable to assess severity in this analysis, there is reason to believe that severity may differ by race. In addition to higher rates of poor asthma management among minority groups (e.g., underuse of preventive asthma medications),(27-31) past research suggests that black children on average have more severe underlying disease,(31;32) including a study of children of middle-class socioeconomic status.(33) Another study found that among children hospitalized for asthma, black children were in more severe condition, even after controlling for other demographic differences.(34) Furthermore, differential response to preventive asthma medications by race group has also been reported, including decreased responsiveness to glucocorticoids among black patients with asthma.(35;36)

This analysis has additional limitations. Asthma prevalence is based on proxy report of a history of an asthma diagnosis and if a child still has asthma. Although the questions are based on surveillance recommendations from the Council of State and Territorial Epidemiologists,(37) the performance of the questions has not been validated against medical records, and recall bias for a history of diagnosis may vary over the age range 0-17 years. Some asthma outcomes, particularly healthcare utilization measures, are dependent on factors other than need for care, including access to care, characteristics of the health delivery and financing systems, and perceived need. It is not clear from the available data how changes in any of these factors may have contributed to the patterns observed in this analysis. Race has a high rate of missingness in the NHAMCS and NHDS. Missing race was singly imputed for the NHAMCS by NCHS. Although these imputations are based on extensive methodological work, single imputation may underestimate variance compared to multiple imputation methods. No missing race values were imputed for the NHDS. An analysis of 1994 data concluded that hospitalizations with missing race were most likely to be for white patients.(16) If this pattern still holds, the disparities reported here may be overestimated. However, because no analysis of patterns by diagnosis was done, it is unknown whether underreporting of race varies by diagnosis. We performed a sensitivity analysis assessing the most conservative assumption that all hospitalizations with unknown race were for white children. The overall conclusions did not change. Finally, data on Hispanic ethnicity is either not available in the data sources used or has high rates of missing. Examining disparities by Hispanic ethnicity is important: asthma prevalence varies widely between Hispanic subgroups with Puerto Rican children having among the highest rates of asthma prevalence and Mexican children among the lowest.(4) While health surveys oversample Hispanic populations to obtain estimates on reported health outcomes, in healthcare utilization data sets, information on Hispanic ethnicity is missing to an even greater extent than race. Furthermore, although ethnicity has been imputed for emergency department visit data from 2003 onward, information is not available for Hispanic subgroups. Finally, mortality data is available by Hispanic subgroup, but the number of asthma deaths in these subgroups is generally too low among children to yield reliable estimates of annual death rates.

This analysis demonstrates the impact of racial differences in asthma prevalence on measured disparities in asthma outcomes. It is encouraging that asthma ED visit and hospitalization disparities based on ARRs have decreased, and that disparities in asthma deaths have not increased. These improved racial disparity results were observed during a period of decreasing adverse outcome rates for both black and white children with asthma. Nonetheless, concerns remain. Prevalence continued to increase among black children, and even after accounting for this factor, racial disparities remain in asthma ED visit and death rates.