 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
PEDIATRICS Vol. 104 No. 3 September 1999, pp. 454-462
,,,, and
From the * Albert Einstein College of Medicine, Bronx,
New York; and the Department of Pediatrics, University of Rochester
School of Medicine, Rochester, New York.
 |
ABSTRACT |
|---|
Top
Abstract
Methods
Results
Discussion
Conclusion
References
|
|---|
Background. Asthma morbidity and mortality has increased substantially in recent years, but asthma hospitalization rates among many geographic and sociodemographic groups have remained stable. Observations on asthma hospitalization rates and severity of acute episodes might provide valuable insight into the functioning of the health care system during this period of health care reform.
Objective. To analyze changes between 1991 and 1995 in childhood asthma hospitalization rates and severity of acute episodes.
Design and Methods. All 29 329 hospitalizations,
including 2028 for asthma, for the 198 893 children (<19 years of
age) in Monroe County (Rochester), New York, were studied during this
5-year period. Severity was determined by hospital record review on a
22% random sample. Using the worst oxygen saturation
(SaO2) during the first 24 hours of hospitalization as the primary index of severity, episodes were categorized as mild (0 to 
95), moderate (90 to 94), or severe (<90).
Results. Hospitalization rates are expressed as
hospitalizations per 1000 child-years. The overall asthma
hospitalization rate was 2.04 (95% confidence interval, 1.95-2.13).
The overall annual asthma hospitalization rate remained relatively
stable from 1991 (1.90) to 1995 (2.31), whereas the
hospitalization rates for severe asthma rose 270%
from 0.57 to 1.55during this period. Simultaneously, the hospitalization rates for
mild asthma decreased from 0.26 to 0.12. As a proportion of all asthma
hospitalizations between 1991 and 1995, severe episodes increased from
31.5% to 60.4%; conversely, mild episodes decreased from 14.1% to
4.7%.
Conclusions. Severity increased significantly among children hospitalized for asthma while the overall asthma hospitalization rate remained stable. It seems that the health care system in this community has responded to an increase in severity of asthma by raising the severity threshold for admission. Key words: asthma, secular variation, hospitalization, children, severity, outcomes, oxygen saturation.
Several recent studies determined that annual childhood
asthma hospitalization rates have stabilized among many geographic areas1,2 and sociodemographic groups2,3 suggesting that the severity of acute asthma has also stabilized. However, based on increasing frequency of adverse events among asthmatic children, including intubation,3,4 cardiopulmonary arrest,4 and death,4-9 other
studies suggest that the severity of acute asthma has increased.
These conflicting findings may reflect limitations of these measures.
Measures of utilization such as hospitalization rates are imperfect,
because, besides morbidity, they are influenced by factors such as
changes in hospital admission criteria, treatment methods, financial
incentives, and access to care.10 Asthma-related mortality
may be an inadequate proxy for severity throughout the entire
population, because with a national asthma mortality rate of 3.7 per
1 000 000 for people <24 years of age,7 these events
represent a minuscule proportion of all acute asthma exacerbations
among children. Moreover, if asthma, as claimed,11,12 is
sensitive to the quality of primary care, deaths attributable to asthma
may be more likely to represent inadequate primary care for isolated
episodes than an increase in severity throughout an entire population.
This study differs from previous studies because it uses an objective
clinical measure of acute asthma severity to examine changes throughout
time in severity of acute asthma exacerbations in a commu-nity-wide
population of children hospitalized. We were able to study essentially
all asthma episodes leading to hospitalization of children from Monroe
County (Rochester), New York, because virtually all admissions of
county residents for acute asthma episodes occur in Monroe County
hospitals. Two of the three pediatric inpatient facilities in Monroe
County account for >90% of acute pediatric asthma hospitalizations
(Rochester Health Care Information Group, unpublished data). Hospitals
in adjacent counties do not draw acute pediatric admissions. With a
long history of offering accessible care,13 Monroe County
providers are the point of first contact for essentially all the
community's most severe illness episodes.
Understanding changes in asthma severity throughout time is important
for several reasons. First, in recent years, an increase in
asthma-related utilization has been reflected in several different measures, including office contacts,2,14 use of
intubation,3 and asthma-related
expenditures.15 However, because some utilization is
inappropriate, whether this increased burden on the health care system
results from real increases in morbidity is unknown. In addition, if
asthma has become epidemic as some suggest, information on secular
variation in asthma hospitalization rates might provide valuable
insight into the functioning of the health care system. This is of
special interest during this period of health care reform.
The objectives of this study were: 1) to examine variation throughout
time in the annual asthma hospitalization rate among children residing
in Monroe County, New York; and 2) to measure changes in severity of
acute exacerbations among Monroe County children hospitalized for this
condition.
Study Population and Setting
Analysis was based on all hospitalizations for children dwelling
in Monroe County, New York, who were >1 month and <19 years of age
when admitted during the study period of January 1, 1991 through
December 31, 1995. Data were obtained from the Rochester Hospitals Data
Base. Information in this data set, developed and maintained
cooperatively by all Rochester-area hospitals, included medical record
number, birth date, admission date, discharge date, sex, primary
discharge diagnosis and up to 4 additional diagnoses, zip code of
residence, hospital, type of insurance, and discharge status. Among the
29 329 hospital admissions of children in Monroe County during the
study period, 2028 had asthma as the primary discharge diagnosis.
Medical Record Reviews
Detailed clinical information for characterizing illness
severity was obtained by medical record review. Because records were also reviewed as part of a study of variation in asthma hospitalization among socioeconomic areas,16 sampling varied by area. Random samples of asthma hospitalizations were drawn from three socioeconomic areas Sampling Strategy
The size of the sample was determined by our interest in
examining socioeconomic variation in acute asthma
severity,16 as well as temporal variation. General
pediatric and pediatric pulmonary colleagues decided, before data
collection, that a difference of 15 points between the percentage of
inner city and suburban children admitted with mild asthma was
clinically important. With the number of inner city and suburban
records reviewed, we were able to detect a population difference of 15 percentage points with a power of >90%. Because its socioeconomic
attributes fall between those of the inner city and the suburbs, a
smaller sample Random samples were drawn from two of the three hospitals in Rochester
that care for children with acute medical conditions. These two
hospitals accounted for 91.0% of suburban, 90.0% of other-urban, and
91.2% of inner city asthma hospitalizations. Emergency departments
(EDs) of these two hospitals were staffed by a single group of
residents and a closely-integrated group of attending pediatricians.
Definitions
Asthma hospitalizations were defined by a primary
discharge diagnosis coded between 493.0 and 493.91 according to the
International Classification of Diseases, Ninth Revision
(ICD-9).17 The primary measure of acute asthma
severity was worst oxygen saturation
(SaO2), defined as the lowest oxygen
saturation obtained in the first 24 hours of hospitalization at a time
when the child was receiving no supplemental oxygen. We chose worst
SaO2 as the primary measure of severity,
because it is reproducible, commonly used in clinical settings, and
directly reflects derangement in pulmonary physiology. Several studies
have concluded that SaO2 is useful in
predicting poor outcome.18-23 Other studies have found
that SaO2 is well correlated with other pulmonary function and clinical measures19,20,24 and that
it measures severity more accurately than other measures such as forced
expiratory volume in 1 second.19-21
Similar to previous studies,25 episodes were classified as
mild, moderate, and severe asthma based on worst
SaO2 values of 95% or greater, 90 to 94%, and
<90%. At least one SaO2 on room air was
recorded in the first 24 hours for 433 (98%) of the 443 episodes
reviewed. Among the remaining 10 episodes, 7, in which the worst
SaO2 was <90 despite supplemental oxygen, were assigned to the most severe category. Three episodes in which worst
SaO2 values were normal but high levels of
supplemental oxygen (50% to 100%) were being administered in the
critical care unit were also assigned to the most severe category.
Because of the large number of SaO2 values
obtained by the hospital staff in the first 24 hours for most children
admitted, only the best and worst values for both the ED and the first
three nursing shifts of the inpatient stay were abstracted from the record. For >90% of episodes, at least three different
SaO2 scores in the first 24 hours were
abstracted from the medical record.
The first 24 hours of hospitalization were defined as the entire ED
stay plus three nursing shifts on an inpatient unit. Nursing shifts
were used in analysis as the primary unit of time because of their
implications for hospital staffing. Nursing shifts were defined for
study purposes as 8-hour periods on an inpatient unit, beginning with
the time of arrival there. Periods of 4 hours or greater were counted
as a full nursing shift, because nurses generally work full shifts.
Additional measures of severity were used to assess the validity of the
primary severity measure, worst SaO2. The best
SaO2 in the ED and the number of nebulized
treatments administered in the first 24 hours of hospitalization were
used to assess changes throughout time in severity of asthma
exacerbations. Analysis of the best SaO2 values
was limited to those recorded in the ED, because
SaO2 almost always returns to normal within the
first 24 hours of hospitalization.26 Therefore, no
variation was expected among the best SaO2
values recorded during the first three nursing shifts on inpatient
unit. In addition, two other measures were used to assess the
relationship between SaO2 and a second
dimension of asthma severity, duration of illness episode. These
measures were the hospital length of stay, defined as the number of
hours from the first nebulized medication administration on the
inpatient unit until noon on the day of discharge, and the duration of
frequent nebulized medications, defined by the number of 4-hour periods
in which nebulized medication was administered 2 or more times. Data on
nebulized medication administration were available for 434 (98%) of
the 443 episodes.
Short-term rehospitalization, use of intubation, and asthma-related
in-hospital death were used to measure adverse events. Short-term
rehospitalization was defined as a hospitalization occurring within 31 days of a previous hospitalization. Use of intubation was determined by
medical record review. Information on asthma-related in-hospital death
was obtained from the Rochester Hospitals Data Base. Because the
majority of pediatric asthma deaths were previously shown to occur in
the hospital,27 asthma-related deaths among hospitalized
children were used as a proxy for all asthma-related deaths.
Other lower respiratory tract condition hospitalizations were defined
by a primary discharge diagnosis with an ICD-9 code of 466 (acute
bronchitis and bronchiolitis); 490 to 491 and 494 to 496 (chronic
obstructive pulmonary diseases); 480 to 486 (pneumonia); 471 to 473, 475 to 478, 500 to 508, or 510 to 519 (other respiratory disease).3
Analysis
Denominators for calculation of county-wide hospitalization
rates were based on the 1990 United States Census. All hospitalization rates were expressed as number per 1000 child-years. Hospitalization rates were not corrected to account for multiple hospitalizations of
the same child.
To determine the expected rate of admissions at each level of severity
for the entire population, samples were weighted according to
socioeconomic-area sample fractions. The calculation of standard errors
for rates was based on the Poisson distribution. Using data including
all children residing in Monroe County during the 5-year study period
(n = 994 465 child-years) and their hospitalization status (ie, hospitalized or not hospitalized) in each year, linear regression was performed to examine trends in hospitalization rates
throughout time. The dependent variable was hospitalization status, and
the independent variable was year. Because the sample size for analyses
of trends in hospitalization rates was large, a P value of
.001 or less was considered statistically significant. To examine
possible changes in coding practice, hospitalization rates for other
lower respiratory tract conditions were also examined.
In analysis of data extracted from the 443 randomly selected medical
records, proportions were compared using the Hospitalization for Asthma
During the 5-year study period, 2028 hospitalizations in Monroe
County were attributed to asthma, and the hospitalization rate for
asthma was 2.04 (95% confidence interval [CI], 1.95-2.13) per 1000 child-years. Asthma accounted for 6.9% of all childhood admissions.
Based on worst SaO2 values, the rate of mild,
moderate, and severe admissions were 0.21 (95% CI, 0.15-0.27), 0.85 (95% CI, 0.73-0.98), and 0.97 (95% CI, 0.84-1.10), respectively
(Table 1).
TABLE 1
METHODS
Top
Abstract
Methods
Results
Discussion
Conclusion
References
inner city, other-urban, and suburbanbut sampling strategy emphasized suburbs and inner city because of our
interest in comparing these two areas. Samples drawn from inner city
(n = 192) and suburban (n = 181)
represented 26% of admissions from these two areas. A smaller sample
(12%) was drawn for other-urban admissions (n = 70).
Altogether, 443 records were reviewed, constituting an overall sampling
fraction of 22%.
half the size of the other twowas drawn for
other-urban admissions to demonstrate a dose-response effect.
2
procedure. The t test was used to assess statistical
significance for differences between two means. The nonparametric
trends test developed by Cuzick28 for ordered data (such
as years) was used to determine the significance of time trends in
measures of severity, including SaO2 and
nebulized medications, by years. Conventional levels for 
(P < .05) were used to assess statistical significance
in this more modest sample.
RESULTS
Top
Abstract
Methods
Results
Discussion
Conclusion
References
Variation Throughout Time in Hospitalized Asthma Exacerbations
Rates varied substantially by age, gender, and socioeconomic status. For children <5 years (1 months to <5 years), 5 to 11 years (5 to <12 years), and 12 to 18 years (12 to <19 years), hospitalization rates were 3.48 (95% CI, 3.45-3.50), 1.71 (95% CI, 1.69-1.72), and 1.15 (95% CI, 1.15-1.16), respectively. Compared with children ages 12 to 18 years, the odds of hospitalization for asthma in children ages 5 to 11 years was twice as great (odds ratio, 2.05; 95% CI, 1.86-2.26), and the odds of hospitalization for asthma in children <5 years was threefold greater (odds ratio, 3.05; 95% CI, 2.72-3.41). Overall, hospitalization for asthma in boys (2.65) was almost twice as common as in girls (1.43). Non-Medicaid-covered children constituted 52.4% (vs 47.6% for Medicaid-covered children) in 1991 and 60.1% in 1995 (vs 39.9% for Medicaid-covered children) of all hospitalizations for asthma. There were no major shifts in the proportion of children covered by Medicaid during the observation period.
To determine if there was any important difference between the two hospitals in the administration of nebulized medications, the means of nebulized medications administered on the inpatient floor of each hospital were compared. When comparing the mean number of nebulized medications administered on the inpatient unit of each hospital, no statistically significant difference was found (12.4 vs 10.8; P = .25).
Stability of Hospitalization Rates Throughout Time
As shown in Fig 1, the annual
all-condition hospitalization rate for children varied little during
the study periodincreasing from 28.7 in 1991 to 29.2 in 1995 (P = .10). The annual asthma hospitalization rate also
remained stable (P = .06) from 1991 (1.90) to 1995 (2.31), and the annual hospitalization rate for other lower respiratory
tract conditions likewise remained stable (P = .01)
between 1991 (1.69) to 1995 (2.25).
|
Temporal Variation in Hospitalization Rates for Severe Asthma
In contrast to the overall rate of asthma hospitalization, the rate of hospitalization for severe asthma episodes rose in each year of the study period. During the 5-year study period, the hospitalization rate for severe asthma episodes increased from 0.57 to 1.55 (P < .001). As shown in Fig 2, among children <5 years and 5 to 11 years, the hospitalization rate for severe asthma episodes increased 340% (from 0.67 to 2.93; P < .001) and 160% (from 0.58 to 1.53; P < .001), respectively. For children ages 12 to 18 years, the hospitalization rate for severe asthma episodes remained almost unchanged (P = 1.000) between 1991 (0.48) and 1995 (0.36). The rate of hospitalization for severe asthma episodes increased 210% for males (from 0.64 to 2.00; P < .001) and 120% for females (from 0.50 to 1.10; P < .001).
|
Temporal Variation Among the Three Levels of Severity
During the study period, the rate of hospitalization for mild asthma episodes halved (from 0.26 to 0.12; P < .001), and, as a percentage of all asthma hospitalizations, mild episodes decreased (P = .001) from 14.1% in 1991 to 4.7% in 1995 (Fig 3). Between 1991 and 1995 the percentage of moderate episodes also decreased (P < .001) from 54.4% to 35.0%. Conversely, the percentage of severe episodes increased (P < .001) from 31.5% in 1991 to 60.4% in 1995. Consistent with these findings, the average best SaO2 in the ED decreased from 95.4 in 1991 to 94.1 in 1995 (P < .01). The average number of nebulized medication treatments administered during the first 24 hours of hospitalization increased from 7.6 in 1991 to 8.2 in 1995, but this difference was not statistically significant (P = .13).
|
Severity distributions for each age group followed that for the sample as a whole. Among children <5 years, severe asthma episodes increased from 24.5% to 60.3% (P < .001), and among children between 5 and 11 years, severe asthma episodes increased from 40.2% to 61.3% (P < .001). For children 12 to 18 years, the proportion of severe asthmatic episodes increased from 33.3% to 51.1% (P < .001). Notably for this oldest age group, in the last 2 years of the study no mild admissions were observed among records reviewed.
Temporal changes in severity distribution were not isolated to children from lower socioeconomic groups. As percentage of all children hospitalized for severe asthma episodes, non-Medicaid-covered children constituted 44.7% in 1991 and 59.9% in 1995.
Adverse Events
Short-term rehospitalizations as a percent of all admits remained stable (P = 1.000) between 1991 (2.9%) and 1995 (3.1%). Among our sample, no children were intubated. In addition, during the 5-year study period, no asthma-related deaths occurred among hospitalized children.
Agreement Among Measures of Severity
The three measures of severity were positively correlated with one another, although correlations between measures expected to reflect duration of the asthma exacerbation (length of stay, duration of frequent nebulized medication) were more strongly correlated with one another than with worst SaO2. The Pearson correlation coefficients between worst SaO2 (as a continuous variable) and length of stay, and worst SaO2 and periods receiving frequent nebulized medication were 0.16 (P < .001) and 0.21 (P < .001), respectively. The correlation coefficient between length of stay and periods receiving frequent nebulized medication was 0.63 (P < .001).
The relationship between these measures is also apparent when SaO2 is treated as a categorical variable. The mean length of stay for all asthma admissions was 50.3 hours. For episodes classified based on worst SaO2 as mild, moderate, and severe, the mean length of stay in hours was 38.9, 46.5, and 55.7, respectively (P < .01). The mean number of 4-hour periods with frequent nebulized medications for all asthma admissions was 3.15. For episodes classified based on worst SaO2 as mild, moderate, and severe, the mean number of periods receiving frequent nebulized medications was 1.78, 2.51, and 4.00, respectively (P < .01).
Clustering of Hospitalizations Among Patients
Among the 2028 hospital admissions for asthma, there were 1444 asthma-hospitalization patients, and they averaged 1.40 hospitalizations per child. During the 5-year observation period, 77.1% of patients were hospitalized for asthma only once and 91.3% were hospitalized for asthma two times or less.
Among the 443 medical records reviewed for the study, there were 411 children hospitalized for asthma at least one time. Among these 411 different children, medical record review identified only 10 (4.93%) who had been hospitalized more than once for a severe asthma episode, and no children who had been hospitalized more than twice for a severe episode. Among children hospitalized more than once for a severe asthma episode, only 3 (1.48%) had been hospitalized twice in the same year for a severe episode.
| |
DISCUSSION |
|---|
|
Top
Abstract
Methods
Results
Discussion
Conclusion
References
|
|---|
Trends in Hospitalization Rates
The hospitalization rate for severe asthma episodes (worst SaO2 < 90) increased threefold among Monroe County children hospitalized during the 5-year study period. Despite the dramatic increase in the hospitalization rate for severe asthma episodes, no increase in overall asthma hospitalization rate was observed. Stability of the overall rate was maintained, despite the increase in severe episodes, because the number (and proportion) of hospitalizations attributed to mild and moderately severe asthma episodes decreased. If the number of hospitalizations for mild and moderate episodes had remained constant between 1991 and 1995, the overall asthma hospitalization rate would have increased dramatically.
Increased severity was reflected in both males and females, as well as in the two youngest age groups (<5 years and 5-11 years). The greatest increase in severity was observed in children <5 years, a group shown consistently to be facing an increase in asthma-related illness burden.2,37-9,15 Although many authors consider diagnostic criteria for asthma in children <2 years to be less distinct than in children 2 to 5 years, we combined these two groups because very similar increases in severity were seen in each. Although increases were seen in the two younger age groups, no increase in the rate of hospitalization for severe asthma episodes was measured among the 12- to 18-year-old age group.
Consistency of Observations
Consistency of findings supports their validity. Along with an increase in the hospitalization rate for severe asthma episodes, we observed a statistically significant decrease in the average best SaO2 in the ED. Both of these observations suggest that severity of acute asthmatic episodes increased during the 5-year study period among hospitalized episodes. This apparent increase in severity is also consistent with previous studies suggesting an increase in childhood asthma morbidity based on increases in prevalence,27-9,29 patient-physician contacts,2,14 use of medication,15 asthma-related expenditures,15 proportion of asthma hospitalizations requiring intubation or cardiopulmonary resuscitation,2,3 and mortal-ity.24-9
No increase was observed in length of stay or number of periods with frequent nebulized medications. The lack of change in these measures of utilization is not surprising. Besides changes in severity, use of medical technology is influenced by changes in practice patterns, financial incentives, and reimbursement mechanisms.10
Alternative Explanations
Alternative explanations for the apparent increase in severe asthma admissions include improved access to care, changes in ICD-9 coding, and measurement bias. Improved access to care is an unlikely explanation for several reasons. First, other hospitalization rates measured, including rates for all conditions, asthma, and other lower respiratory tract conditions, remained stable throughout the study period. (The annual hospitalization rate for lower respiratory tract conditions varied significantly, P = .01, according to a conventional threshold for significance. However, because in 4 out of 5 years the rates are highly similar (Table 1), this P value seems to result from a combination of the large sample size and the high rate in the final year of the study.) Had improved access increased hospitalization rates, it is unlikely this increase would have been limited to asthma. Further, this community has a long history of easy access to care for all socioeconomic groups.13,30 Finally, we observed an increase in severe asthma episodes among suburban children, those with the fewest barriers to care, as well as among inner city children.
Changes in ICD coding practices also are unlikely to explain
observations. If children presenting with severe respiratory dysfunction were more likely to be coded as asthma in 1995 than in
1991, less severe episodes in 1995 would, to maintain a stable overall
asthma hospitalization rate, necessarily have been admitted under some
other lower respiratory tract condition diagnosis. This did not
happenwe found no important change in the annual hospitalizations
rates for other lower respiratory tract conditions.
Measurement bias associated with use of SaO2 as a measure of severity is also unlikely. First, measurement bias might have occurred if area physicians increasingly emphasized SaO2 in the decision to admit. With increasing emphasis during the 5 study years, a larger proportion of children with low SaO2 values might be hospitalized. Two observations suggest bias of this form did not occur. If the apparent increase in severity resulted from increased emphasis on SaO2, an increase in the number of children admitted with a SaO2 of <90 would be expected across all age groups. However, the rate for children between ages 12 and 18 years remained constant during the study period. In addition, our experience working as members of this group of clinicians suggests that SaO2 values have been used with a high degree of uniformity throughout the study period. Whereas SaO2 is measured as a continuous variable, providers in this community generally treat SaO2 as a categorical variable for purposes of clinical decision-making. SaO2 values of 95% or greater are considered normal. Values <90% almost always led to hospitalization, a threshold similar to those observed in other communities.18,20,23,24,31
A second type of measurement bias might result from an increase in the number of SaO2 values recorded in later years of the study. If this occurred, we should expect throughout time to see more SaO2 scores at the extreme ends of the distribution. However, both the average best SaO2 in the ED and the average worst SaO2 decreased during the study period.
Still another form of measurement bias might reflect changes in treatment. In theory, a fall in the average SaO2 throughout time might reflect increased use of bronchodilator medication. In some children with acute asthma episodes, SaO2 falls significantly (2 percentage points or more) with bronchodilator treatment,21,32,33 apparently as a result of an increase in ventilation-perfusion mismatch attributed to greater relaxation of vascular than bronchial smooth muscle.34 For several reasons, this effect on SaO2 is unlikely to have an important impact on findings of this study. Virtually all studies of bronchodilator effects on SaO2 in asthma episodes find that mean values improve with treatment.18,20,21,35,36 Tal et al,33 who observed a fall among 9 of 18 patients in whom SaO2 was measured at 5, 15, and 30 minutes, noted that the mean (92.1%) had returned to baseline by 30 minutes posttreatment. Such transient changes will often go undetected in a busy clinical setting such as that studied. Most observers found significant falls in SaO2 to occur in a small proportion (<20%) of their patients2136-38 Finally, inhaled bronchodilators were used extensively throughout the study period. It seems unlikely that the small and not statistically significant increase in the average number of nebulized medication treatments administered during the first 24 hours of hospitalization could account for the large increase in the proportion of asthma admissions with their worst SaO2 <90.
Increased Severity of Asthma in the Community Explains Increased Severity of Asthma in the Hospital
We believe the severity threshold for admission increased during the study period. This conclusion rests on assumptions about the severity distribution of acute asthma exacerbations in the population. Although severity was categorized for study purposes into discrete levels, asthma severity is distributed across a broad continuum. Most episodes are mild and do not result in hospitalization. Only a small fraction falls at the high end of the severity continuum. Observations, from another study, focused on the severity of chronic asthma in 11 pediatric practices that are dispersed among Rochester's socioeconomic areas are consistent with this distribution for acute asthma exacerbations (Yoos H, McMullen A, Kitzman H, personal communication, 1998). In 1997, 1590 children from these practices made one or more office visits for an acute asthma episode. Mild (1 acute asthma visit only), moderate (2 or 3 visits), and severe (>3 visits or 1 hospitalization) disease were found in 50%, 30%, and 20% of these asthmatic children, respectively.
By plotting frequency as a function of severity, the severity distribution for asthma exacerbations among the child population may be depicted as an asymptotic curve (Fig 4A); as severity increases, frequency approaches zero. Because no fundamental change in the shape of this distribution is expected, with an increase in the frequency of severe asthma exacerbations in the population we should also expect an increase at lower levels of severity. The increase in severity from time 1 to time 2 is represented by a rightward shift of the curve (Fig 4B). With this shift to the right, the total number of admissions can remain constant only if the severity threshold for admission is raised. In other words, the areas under these two curves representing admissions can remain the same only if the line indicating the average severity threshold for admission also shifts to the right.
|
If, in contrast, the severity threshold remained constant, the overall number of asthma hospitalizations could remain constant in face of increasing admissions in the most severe category only if the number of mild and moderate episodes hospitalized simultaneously decreased. This scenario, represented in Fig 4C by the shift from the asymptotic curve at time 1 to a curve that bends upward at the most extreme levels of severity at time 2, is unlikely because a frequency distribution with greater frequency at higher severity levels is implausible.
Limitations
A multidimensional construct such as severity of an acute asthma episode cannot be fully characterized by any single measure, such as worst SaO2. Our choice of SaO2 as a measure of risk and severity, although driven in part by its ease of use, widespread adoption, and retrieve-ability in record review, was based on substantial evidence supporting its validity.18-24 In this study, two observations tied more closely to the duration of the illness episode, length of stay and number of 4-hour periods with frequent nebulized medications, were used independently to validate worst SaO2 as a measure of severity. These measures supported worst SaO2 as a valid measure of severity; episodes with lower SaO2 had both increased length of stay and increased duration of frequent nebulized medications.
Analysis focused on hospitalization rates for asthma episodes rather than hospitalization rates for children with asthma. This is also true of the other hospitalization rates examined in the analysis (ie, hospitalization rates for all conditions and for all other lower respiratory tract conditions). We recognize that hospital admission for an individual child is associated with an increased likelihood of subsequent hospitalization. Clustering of events, such as hospitalization, violates the assumption of independence and yields standard errors that underestimate CIs for point estimates of hospitalization rates. However, it would be difficult to determine what period of time actually was spent in the community for each child residing in Monroe County at any time in the 5-year observation. Furthermore, because readmission for severe asthma episodes was uncommon, it seems highly unlikely that patient clustering had an important impact on these findings. It should be recognized that a case study of one community does not allow generalizations to other communities.
Implications of Increased Severity Threshold for Admission
Improved treatment methods (eg, oral corticosteroids), which allow physicians to care better for acute asthma exacerbations, may explain some of the decrease in mild and moderate asthma episodes hospitalized. Changing financial incentives intended to decrease hospital utilization, such as those exerted by managed care, may have also decreased the likelihood of hospitalization for children who could be treated outside of the hospital setting without high risk of an adverse outcome.
Regardless of its explanation, these finding may have important implications for Rochester and other communities. First, communities with stable asthma hospitalization rates cannot conclude that asthma morbidity burden is stable. Given nation-wide decreases in overall hospitalization rates, stable asthma hospitalization rates despite increasing severity in acute asthma may reflect a response to system-wide pressures to decrease use of inpatient services. Therefore, it is important to monitor changes throughout time in asthma severity as well as in hospitalization.
Second, our findings suggest that a substantial number of childhood asthma hospitalizations could be avoided in many communities. Perrin et al39 showed that pediatric hospitalization rates for all conditions in childhood, including asthma, are low in Rochester compared with Boston and New Haven. Further, using the same primary measure of asthma severity as used herein, their more recent study determined that mild asthma episodes in Rochester accounted for a substantially smaller proportion of asthma hospitalizations than in Boston or New Haven.25 Observations from this study indicate that, despite their historically high threshold for admission, Rochester-area physicians have treated an increasingly sick population without increasing use of inpatient services. Some evidence, including data on short-term rehospitalization, intubation, and in-hospital mortality, supports our clinical impression that this has been accomplished with no increase in adverse events. More detailed information relating to adverse events, family interviews focused on duration of symptoms for example, should be obtained through prospective studies.
| |
CONCLUSIONS |
|---|
|
Top
Abstract
Methods
Results
Discussion
Conclusion
References
|
|---|
Severity, as defined by worst SaO2, increased substantially during a 5-year period among children hospitalized for asthma in Rochester, NY. This increase in severity was attributable entirely to changes among children <12 years old. Because of an increase in admission threshold, the overall rate remained unchanged despite the increase in severity of acute episodes. The ability of Rochester-area physicians to care effectively for an increasingly sick group of children outside of the inpatient setting may have important implications for treatment of asthma in other communities and for understanding changes in practice patterns during health care reform.
| |
ACKNOWLEDGMENTS |
|---|
We thank David C. Goodman, MD, and Ann B. Flood, PhD, of the Center for the Evaluative Clinical Sciences, Dartmouth Medical School, for their comprehensive review and insightful criticisms of our manuscript. We also gratefully acknowledge the assistance of the Rochester Healthcare Information Group for provision of hospitalization data.
| |
FOOTNOTES |
|---|
Received for publication Sep 14, 1998; accepted Jan 26, 1999.
Reprint requests to (K.M.M.) Department of Pediatrics, Box 777, University of Rochester Medical Center, 601 Elmwood Ave, Rochester, NY 14642. E-mail: ken-mcconnochie{at}urmc.rochester.edu
| |
ABBREVIATIONS |
|---|
SaO2, oxygen saturation; ED, emergency department; CI, confidence interval.
| |
REFERENCES |
|---|
|
Top
Abstract
Methods
Results
Discussion
Conclusion
References
|
|---|
United States, 1960-1995.
MMWR Morb Mortal Wkly Rep.
1998;
47:1-28 [Medline]United States, 1980-1993.
MMWR Morb Mortal Wkly Rep.
1996;
45:350-353 [Medline]United States, 1982-1992.
MMWR Morb Mortal Wkly Rep.
1996;
51:952-955 United States, 1980-1987.
MMWR Morb Mortal Wkly Rep.
1990;
39:493-497 [Medline]This article has been cited by other articles:
 |
|
 |
|
  E. T. Naureckas, V. Dukic, X. Bao, and P. Rathouz Short-Acting {beta}-Agonist Prescription Fills as a Marker for Asthma Morbidity Chest, August 1, 2005; 128(2): 602 - 608. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. S. Mandell, W. W. Thompson, E. S. Weintraub, F. DeStefano, and M. B. Blank Trends in Diagnosis Rates for Autism and ADHD at Hospital Discharge in the Context of Other Psychiatric Diagnoses Psychiatr Serv, January 1, 2005; 56(1): 56 - 62. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. J. Akinbami and K. C. Schoendorf Trends in Childhood Asthma: Prevalence, Health Care Utilization, and Mortality Pediatrics, August 1, 2002; 110(2): 315 - 322. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D S Morrison and P McLoone Changing patterns of hospital admission for asthma, 1981-97 Thorax, September 1, 2001; 56(9): 687 - 690. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Bauchner, R. Vinci, and J. Chessare Milliman and Robertson---Going in the Wrong Direction Pediatrics, April 1, 2000; 105(4): 858 - 859. [Full Text]
|
|
|
|
 |
|
 J. R. Meurer, V. George, S. Subichin, J. Yauck, and P. Layde Asthma Severity Among Children Hospitalized in 1990 and 1995 Arch Pediatr Adolesc Med, February 1, 2000; 154(2): 143 - 149. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
||
|
||
Home | My Pediatrics | Journal Information | Current Issue | Past Issues | Subscriptions & Services | Contact Us Click here for faster international access © 1999 American Academy of Pediatrics. All rights reserved. |
||
|
||
