Objective. Asthma hospitalization rates continue to increase nationally for children despite efforts by the National Institutes of Health and specialty organizations to improve outcomes through the dissemination of practice guidelines. To understand the generalizability of national trends to regional populations, we studied childhood hospitalizations over a 10-year period in four northeastern states.
Design. Longitudinal analysis of hospitalization rates by patient residence and patient characteristics using state hospital discharge datasets.
Population. Age <18 years residing in Maine, New Hampshire, Vermont, or New York state during the period 1985 to 1994.
Results. In multivariate analyses (controlling for age, sex, race/ethnicity, median household income, metropolitan status), we found that New York asthma hospitalization rates increased 3.8% per annum (95% confidence interval: 3.3,4.2), whereas in New Hampshire, rates decreased 5.8% (95% confidence interval: 7.6,4.1). Maine and Vermont rates did not change significantly during the study period. Increased asthma hospitalization rates were noted in black and Hispanic populations, in children residing in zip codes with lower median household incomes, and in those living in metropolitan areas. Hospitalization rates for nonasthma causes fell substantially. As a result, the proportion of hospital days attributed to childhood asthma increased in all population groups.
Conclusions. Asthma discharge rates measured by the state of residence or socioeconomic characteristic do not necessarily parallel national trends. None of the current hypotheses offered to explain national trends in asthma hospitalization rates (changes in disease severity, diagnostic substitution, or differences in the supply and character of medical care) can be the sole explanation of these regional trends. Efforts intended to improve asthma outcomes may benefit a greater number of children by redirecting resources toward specific populations identified through state hospital discharge datasets.
Asthma is the most common chronic illness of childhood and one of the most frequent causes of hospitalization. In the past 2 decades, pediatric asthma hospitalization rates have increased, along with a rise in prevalence and mortality.1-7 In comparison, the frequency of hospitalization for other common pediatric illness has declined.8 In response to these trends, the US Public Health Service has set a goal of a 20% reduction in pediatric asthma hospitalizations by the end of the decade.7 The National Heart, Lung, and Blood Institute and specialty organizations have made a coordinated effort through the National Asthma Education and Prevention Program to improve patient management with a goal of decreasing hospitalizations.9-11 Yet, data from the National Hospital Discharge Survey revealed that the rise in hospitalization rates observed during the 1980s has extended into the 1990s.7
To better understand the epidemiology of pediatric asthma hospitalizations, we conducted longitudinal studies in northern New England and New York state to answer the following three primary questions: 1) Are the national asthma hospitalization trends reflected within regional populations—population units that are more accessible to a focused public policy for asthma care? 2) Is there evidence that changes in asthma hospitalization rates are the result of substituting the diagnosis of asthma for lower respiratory illness that was formerly diagnosed as bronchitis or pneumonia? In this event, we would expect an inverse relationship between trends in asthma hospitalization rates over time and the trends for nonasthma lower respiratory discharges. 3) Do regional trends in asthma rates parallel the trends observed for other causes of childhood hospitalization?
Hospitalizations were studied in the entire age <18 years population residing in the four contiguous states of Maine, New Hampshire, Vermont, and New York for the years 1985 to 1994. Population counts by age, sex, and race/ethnicity at the zip code level were obtained from Claritas Corp (Arlington, VA) and served as denominators for rate calculations.
Hospital Discharge Data Sources
Hospital discharge datasets were provided by the Maine Health Care Finance Commission (Augusta), the New Hampshire Department of Health and Human Services (Concord), the Vermont Department of Health (Burlington), and the New York Department of Health (Albany). Data from the Massachusetts Health Data Consortium (Waltham) also were used to measure the hospitalizations of the study population to facilities in Massachusetts, which shares a border with all of the study states except Maine. Together, the datasets include standardized information about hospital discharges for all children residing in the study region to any region hospital and to hospitals in Massachusetts.
Discharge records included the patient's age, sex, zip code of residence, source of payment, and discharge diagnoses. Coding of race and ethnicity differed by state. Maine, New Hampshire, and Vermont did not code for race or ethnicity. We assigned these residents as white non-Hispanic, because in 1990 Census counts, nonwhites and Hispanics constituted 2% of the pediatric population for Maine and Vermont, and 3% for New Hampshire. New York discharges were assigned to one of four groups: white non-Hispanic, black non-Hispanic, Hispanic, or Other/unknown. Discharges assigned to the Other/unknown category constituted 7.3% of the total discharges during 1985 to 1993. This category was excluded from race/ethnicity analyses, because a corresponding census population is unknown. In 1994, the Other/unknown category constituted 25.2% of total discharges. Because this could bias rates substantially for the other race/ethnicity groups, crude rates by race/ethnicity were not determined for 1994, and the 1994 New York data was excluded entirely from all multivariate analyses.
Diagnostic Study Groups
We examined four causes of hospitalization based on the patient's principal discharge diagnosis: asthma (International Classification of Diseases, 9th ed, 493); lower respiratory illness excluding asthma (466, 480–486, 487.0, 490–492, 494–519); upper respiratory illness (460–519, excluding codes listed in lower respiratory illness and asthma); and nonrespiratory illness (all others). All hospital discharges were included in one of these groups.
Because we were unable to distinguish multiple hospitalizations for the same child, the unit of analysis was the age group-sex-race/ethnicity-zip code stratum for each study year from 1985 to 1994 (N = 374 718 strata). Age was grouped as 0 to 2, 3 to 4, 5 to 9, and 10 to 17 years. For each stratum and year, we computed the total number of hospitalizations, overall and by cause, and the population count. The median population size within strata in 1990 was N = 26 (range, 1 to 5649). Socioeconomic characteristics (median household income, metropolitan residence) were measured at the zip code level and assigned to each stratum.
To examine trends in asthma and nonasthma hospitalizations, we used a method for the analysis of longitudinal clustered count data consisting of an overdispersed Poisson regression model, weighting by the number of children in each stratum.12-14 The dependent variable was the discharge rate in each stratum and year, computed as the total number of hospitalizations divided by the total population age <18 years. Each model controlled for age group, sex, race, the age group by sex interactions, state, and zip code-specific socioeconomic characteristics (median household income, metropolitan or nonmetropolitan status). Year was included as a continuous variable to assess the linear trend in discharge rates over the study period. To study the differences in asthma discharge trends according to state, age, sex, race/ethnicity, income, and metropolitan status, we included interactions of each of these covariates with year in separate models.
We analyzed the proportion of total hospital days for asthma using methods for the analysis of longitudinal clustered binary data consisting of an overdispersed binomial regression model, weighting by the number of hospital days in each stratum.12-14 In these models, the dependent variable was the proportion of hospital days for asthma in each stratum, computed as the total number of hospital days for asthma divided by the total number of hospital days. Models were similar to those described above.
Point estimates and confidence intervals (CIs) for the rate ratios (Poisson models) and odds ratios (binomial models) were obtained by exponentiating the corresponding regression parameters. We incorporated variance overdispersion in the estimates of all SE values to account for multiple hospitalizations to the same child and clustering of discharges within strata.12 This increased the usual Poisson-based CIs for the discharge rate ratios by ∼70%, although the binomial-based CIs for the proportion of asthma days odds ratios were unaffected. We used the SAS statistical procedure GENMOD for all models.15 All tests were performed at the 5% level of significance and were two-sided.
Crude Discharge Rates
In 1994, discharge rates for asthma varied substantially by state of residence and population characteristics. New York state had the highest discharge rate (4.77 per thousand), 3.7 times greater than that of Vermont, the state with the lowest rate (1.30 per thousand) (Table1). Children living in low-income zip codes had hospitalizations 3.6 times higher (7.39 per thousand) compared with those in the highest income zip codes (2.04 per thousand). In metropolitan children, the discharge rate was 4.69 per thousand, more than twice that in nonmetropolitan areas (1.78 per thousand). In 1993, asthma hospitalizations were about sixfold higher in black, non-Hispanic children (12.83 per thousand), compared with white, non-Hispanic (2.04 per thousand). Hispanic children had discharge rates almost as high as blacks (10.09 per thousand).
Smaller differences in discharge rates across the study population groups were noted for lower respiratory, nonasthma causes (Table 1). Rates in 1994 were almost twofold higher for New York children than for those living in New Hampshire and, in 1993, threefold higher for black, non-Hispanics compared with whites. Children living in low-income zip codes were hospitalized with a 2.5 times higher frequency compared with those in high-income zip codes. There was little difference in discharge rates in metropolitan compared with nonmetropolitan children.
Crude Discharge Rates Trends
Over the 10-year study period, asthma discharge rates increased for New York children from 3.55 to 4.77 per thousand, whereas rates declined slightly in Maine (1.74 to 1.33 per thousand) and Vermont (1.40 to 1.30 per thousand). New Hampshire rates fell almost 50% from 1.98 to 1.09 (Figure)1.
An inverse relationship between the temporal trends for asthma and lower respiratory, nonasthma hospitalizations was not observed in any state. For lower respiratory causes of hospitalization excluding asthma, discharge rates decreased 39% in Maine, 29% in New Hampshire, and 8% in Vermont, but increased 8% in New York. Hospitalization for upper respiratory illness declined in all states, ranging from a 52% decrease in Vermont to an 89% decline in New Hampshire. For nonrespiratory illness, rates decreased 37% in Maine, 45% in New Hampshire, 31% in New York, and 40% in Vermont.
Multivariate Analysis of Hospitalization Trends
After adjusting for covariates (age, sex, race/ethnicity, median household income, and metropolitan status), New York asthma discharge rates between 1985 and 1993 increased 3.8% per annum (95% CI: 3.3,4.2), whereas New Hampshire rates decreased 5.8% per annum (95% CI: 7.6,4.1) between 1985 and 1994 (Table2). Rates during the period 1985 to 1994 did not change significantly in Maine or Vermont.
The greatest increase in asthma rates was observed in children residing in low-income zip codes (4.1% per annum; 95% CI: 3.5,4.8), with a moderate increase in those living in the middle- (2.2%; 95% CI: 1.5,2.9) and high-income zip codes (1.7% per annum; 95% CI: 0.9,2.4) (Table 2). Differences across race/ethnic groups (1985 to 1993) were observed, with increases noted in Hispanic children (6.8% per annum; 95% CI: 5.9,7.7) and black, non-Hispanics (4.5% per annum; 95% CI: 5.2,6.0). In white, non-Hispanic children, asthma discharge rates decreased 0.9% per annum (95% CI: 1.4,0.4). Children residing in metropolitan areas experienced a 3.7% per annum increase (95% CI: 3.3,4.2), whereas rates for those in nonmetropolitan areas decreased by 1.6% per annum (95% CI: 2.6,0.7).
The changes in asthma discharge rates by race/ethnicity were noted in all three strata of zip code income (Table3). Increase in asthma hospitalization rates were higher in black, non-Hispanics and Hispanic children living in high- compared with low-income areas, but these patients had a high proportion of Medicaid or self-pay source of payment, suggesting that the hospitalizations occurred in children of poor families living within affluent zip codes. For example, Hispanic children living in the high-income zip codes experienced a 10.1% per annum increase (95% CI: 7.3,13.0) in asthma hospitalizations, but 57.5% of these discharges had Medicaid or self-pay as the source of payment. In contrast, only 16.6% of the discharges for white, non-Hispanics living in the same high income zip codes were Medicaid or self-pay.
Although the direction and magnitude of annual changes in asthma discharge rates varied by population characteristics, the proportion of total inpatient days that were attributed to asthma increased in all study population groups (Table 2). For example, although New Hampshire children experienced a 5.8% per annum decline in asthma rates, the asthma day proportion increased 4.5% per annum (95% CI: 2.2,6.9), because hospitalization rates for nonasthma illness decreased at an even faster pace. Similarly, although discharge rates decreased in white, non-Hispanic children, the asthma day proportion increased 6.1% per annum (95% CI: 5.3,6.8), because nonasthma hospitalization rates decreased at a faster rate than hospitalization rates for asthma.
Most asthma hospitalizations for children are preventable through longitudinal disease management jointly undertaken by medical providers, the pediatric patient, and parents.16-18Guidelines detailing best practice for the diagnosis and management of asthma, with a particular emphasis on inner-city populations, have been widely disseminated9,10 Yet, lower hospitalization rates for pediatric asthma for the nation as a whole remain elusive.7 During the period 1987 to 1992, pediatric asthma hospitalizations (in those <14 years of age) increased nationally from 2.84 to 3.44 per thousand, then decreased in 1993 to 2.80 per thousand, only to rise in 1994 to 2.95 per thousand. These rates fall far short of the US government's goal of 2.25 per thousand in the year 2000.7
Our study shows that the national trends in pediatric asthma hospitalizations are not necessarily present in regional populations of children. Within Maine and Vermont, asthma hospitalization rates remained unchanged. In New York state rates increased, whereas New Hampshire rates fell. Irrespective of asthma hospitalizations, overall use of hospitals declined substantially for children in all four states.
The causes of rising or stable asthma hospitalization rates while the overall use of pediatric inpatient care has declined remain unsettled. Three explanations, separately or in combination, could account for these trends. First, the burden of illness from asthma may be increasing, or less effective medical care may be available to populations of high-risk children with asthma. These factors may be especially important19 in the poor and minority populations within New York state. Poor children with asthma have been shown in previous work to have fewer physician visits despite higher numbers of bed days and hospitalizations.20 Changes in medical care access or illness burden might explain rising hospitalization rates. It does not necessarily follow that the decline in New Hampshire's rates are the result of better access or lower illness burden in the pediatric population, because Maine and Vermont rates remained unchanged.
An alternative explanation for these secular trends is that measures of asthma hospitalization, and prevalence, are biased by a greater propensity of physicians to diagnose asthma in children whose illness was previously labeled bronchitis or viral pneumonia.21Diagnostic substitution could be a factor, although difficult to prove, in the stable rates seen in Maine and Vermont, where hospitalizations for lower respiratory illness (excluding asthma) have declined. The increased rates for lower respiratory illness, excluding asthma, observed in New York state are not consistent with this hypothesis for the rising asthma rates.
Beyond factors that alter the severity of asthma or the physician-diagnosed prevalence in populations, the use of hospitals is influenced by the supply and characteristics of medical care. These nonclinical factors are important predictors of regional variation in hospitalization rates and include the relative availability of hospital beds, the practice style of physicians, and the proximity of children to medical care.22-25 The penetration of managed care into the health care market also varies across the study states.26 These factors would be expected to similarly influence the common causes of medical hospitalization. As important as these factors may be in regional variation, the deviation of asthma rate trends from rates for other illness suggests that explanations specific to asthma also are responsible.
The lack of a single explanation for the changes in asthma hospitalizations is not surprising if one considers that asthma has all of the following characteristics: great diagnostic uncertainty in its mild and most common form, an illness severity that is sensitive to environmental and socioeconomic risk factors, and improvement in functional status requiring longitudinal monitoring, such as peak flow meters and medication use during asymptomatic periods. The likelihood of an asthma hospitalization, therefore, will be influenced strongly by factors that are unrelated to the underlying burden of illness. These include disease management and, just as important, the clinician's threshold for admission as influenced by hospital capacity.22,27 Even with uncertainty about the primary cause of these trends, initiatives designed to improve asthma outcomes can use regional hospitalization data to focus attention and resources to populations with high rates of hospitalization.
Programs intended to reduce asthma hospitalizations may be more effective if resources currently applied to children with declining rates were redirected to those populations with upward trends. Although higher hospitalization rates are known to occur in minorities and socioeconomically disadvantaged populations,2,22,28-30 our results demonstrate the value of measuring trends in smaller population groups. Each of the four states we studied has distinctive patterns of hospitalization. Identifying specific communities with unfavorable hospitalization trends would add additional specificity. It was not the goal of this study to measure the experience of the smallest subpopulation possible, but this extension of our approach may be even more useful. For example, do children living in poor communities in New Hampshire share the favorable trends seen across the state? Are blacks and Hispanics outside New York City experiencing rising asthma rates? These populations still are large enough for stable rate estimates and could bring even greater population specificity for those seeking to improve asthma outcomes.
Several limitations should be noted in this approach. First, hospitalizations could not be linked to detect multiple hospitalizations in the same patient. Therefore, we could not determine the extent that the reported trends reflect changes in the frequency of recurrent asthma hospitalizations. State hospitalization data rarely are available with unique patient identifiers to researchers, and linkage using available fields, such as date of birth and sex, is generally forbidden by regulation or statute. Although recurrent asthma hospitalizations are an important subject of study, previous reports show that they are infrequent enough to be primarily responsible for the trends we report.18,31,32
Second, calculation of population-based rates requires tabulation of hospitalizations independent of hospital location. To calculate accurately rates for children living in Maine, New Hampshire, and Vermont, we included data for the adjacent states. The reported New York rates are likely to be an underestimate, because we did not include Connecticut, Pennsylvania, and New Jersey data. The undercounting is likely to be small and does not affect our findings, but would pose a major problem if rates were calculated for communities along New York's western and southern borders.
Finally, we encountered underreporting of patient race and ethnicity characteristics in all four of the study states. A study of similar difficulties encountered in the National Hospital Discharge Survey concluded that discharges with unspecified race/ethnicity were likely to be in whites.33 We tested the sensitivity of our multivariate analyses to underreporting by assigning the Other/unknown race population to each of the three reported race/ethnicity categories in separate models. When compared with the reported trends, we found little difference, suggesting that any bias is likely to be very small. We are less certain that the zip code measure of median household income is an accurate measure of the socioeconomic status of the resident population. Although the attribution of area characteristics to individuals has been used widely in the study of pediatric hospitalizations,24,28,29,34 our results suggest that the hospitalized minority children residing in higher income areas are poorer than hospitalized white, non-Hispanics. Therefore, the median household income of a zip code should be understood as a characteristic of a neighborhood or community, but not of all the residents.
Regardless of the state we studied, asthma continues to be one of the most common causes of pediatric hospitalizations. Children hospitalized for asthma represent the failure of our social safety net as well as the failure of prospective disease management. Hospitalization for asthma is rarely inevitable, because the course of asthma is responsive to ambulatory care.16,35 Improved asthma outcomes and reduction in asthma hospitalizations continue to be an important indicator of the welfare of pediatric populations and of clinicians' skills in using limited health care resources toward more successful ends.
This work was supported by Grant R29-HL52076-01 from the National Heart, Lung, and Blood Institute.
- Received February 11, 1997.
- Accepted July 17, 1997.
Reprint requests to (D.C.G.) 211 Strasenburgh Hall, Dartmouth Medical School, Hanover, NH 03755.
This work was presented, in part, at the Ambulatory Pediatrics Association Annual Meeting, San Diego, CA, May 9, 1995.
- CI =
- confidence interval
- ↵Centers for Disease Control and Prevention. Asthma mortality and hospitalization among children and young adults—United States, 1980–1993. MMWR. 1996:350–353
- ↵National Center for Health Statistics. Healthy People 2000 Review, 1995–96. Hyattsville, MD: Public Health Service; 1996
- Gillum B,
- Graves E,
- Kozak L
- ↵National Asthma Education Program. Guidelines for the Diagnosis and Management of Asthma. Bethesda, MD: National Heart, Lung, and Blood Institute; 1991
- ↵National Asthma Education and Prevention Program. Expert Panel Report II. Guidelines for the Diagnosis and Management of Asthma. Bethesda, MD: National Heart, Lung, and Blood Institute; 1997
- ↵National Heart Lung and Blood Institute. International Consensus Report on the Diagnosis and Management of Asthma. Bethesda, MD: National Institutes of Health; 1992
- ↵McCullagh P, Nelder J. Generalized Linear Models. 2nd ed. New York, NY: Chapman and Hall; 1989
- Liang K,
- Zeger S
- ↵SAS Institute. Changes and Enhancements for Release 6.12. SAS/STAT software. Cary, NC: SAS Institute Inc; 1996
- Garrett J,
- Kolbe J,
- Richards G,
- Whitlock T,
- Rea H
- Finkelstein J,
- Brown R,
- Schneider L,
- et al.
- Halfon N,
- Newacheck PW
- Goodman DC,
- Fisher ES,
- Gittelsohn A,
- Chang CH,
- Fleming C
- McConnochie K,
- Roghmann K,
- Liptak G
- Homer C,
- Szilagyi P,
- Rodewald L,
- et al.
- ↵Interstudy Competitive Edge 5.2. St Paul, MN: Interstudy Publications; 1995
- ↵Beckett W, Belanger K, Gent J, Holford T, Leaderer B. Asthma among Puerto Rican Hispanics. Am J Respir, Crit Care Med. 1996;154:894–899
- Senthilselvan A
- To T,
- Dick P,
- Feldman W,
- Hernandez R
- ↵Kozak L. Underreporting of race in the National Hospital Discharge Survey. Adv Data. 1995:1–12
- McConnochie K,
- Roghmann K,
- Liptak G
- Copyright © 1998 American Academy of Pediatrics