© 2003 American Thoracic Society
Hospital Readmissions for Childhood AsthmaA 10-Year Metropolitan StudyDivision of Allergy and Pulmonary Medicine, Department of Pediatrics, Washington University School of Medicine, St. Louis, Missouri Correspondence and requests for reprints should be addressed to Gordon R. Bloomberg, M.D., Division of Allergy and Pulmonary Medicine, Department of Pediatrics, Washington University School of Medicine, 1 Children's Place, St. Louis, MO 63110. E-mail: bloomberg{at}kids.wustl.edu
Studies of asthma admissions in the St. Louis metropolitan area have disclosed substantial numbers of children with readmissions. To determine the magnitude of readmissions and attributes of children with readmissions, a retrospective analysis of 8,761 children with 14,905 asthma hospitalizations for January 1, 1990 through December 31, 1999 at the two university affiliated children's hospitals in St. Louis was undertaken. Patient attributes of age, sex, race/ethnicity, residence, payor status, length of stay, and month of admission were compared between patients admitted once during that period and patients admitted multiple times. Main outcome measures were the total number of admissions and time to readmission during the study interval. A Lin, Wei, Yang, and Ying model of time to readmission showed that African-American children with Medicaid or no insurance are at higher risk of readmission (risk ratio 1.28) than are African-American patients with commercial insurance or white/other race/ethnicity patients regardless of insurance. Probability of readmission increased from 30% after a first admission, 46% after a second, and 59% after a third. Prior admission was a more specific indicator of readmission with greater positive predictive value than ethnicity or insurance status or their combination.
Key Words: asthma • child hospitalization • patient readmissions Asthma is a leading cause of hospitalization among children (1–4). Asthma together with infections contributes one-half of all hospitalizations in children 1 to 4 years of age and one-third of hospitalizations in children 5 to 9 years of age (5). The trend of pediatric asthma admissions in the U.S. peaked in the mid-1990s, reached a plateau, and has remained stable since then (6, 7). In these reports, only discharge events are counted and cannot account for individuals with repeat admissions. The magnitude of readmissions is an important component of hospital discharge data and has not been widely reported. Studies of the St. Louis metropolitan area show that the hospitalization numbers for childhood asthma increased threefold from 1983 to 1995 (8, 9). Estimating from one (St. Louis Children's Hospital) of the two children's hospitals in St. Louis, readmissions accounted for 15 to 25% of annual asthma admissions during that period (9). In these studies, readmissions did not account for the increasing asthma hospitalization rate. Over the past two decades, the specific issue of readmissions has drawn the attention of investigators in England, Canada, and New Zealand, as well as the United States. They have examined the readmission rates over various study periods ranging from 3 months to 4 years. For instance, readmission rates have varied from 24% readmitted within 6 months (10), to rates of 20 to 47% (11–14) in the following year from an index admission. Mitchell and coworkers (15) following patients from an index admission over a study period of 33 months found an increasing readmission rate of 23% readmitted within 3 months to 43% by 1 year and 51% by 2 years. Over an interval of 4 years, To and coworkers (16) reported that 29% of index patients were readmitted within 4 years. In a 1991 to 1995 CDC study of asthma, hospitalizations among Wisconsin residents under the age of 25 years (17), 18% were readmitted annually. In that study, 33% of all asthma-related admissions were readmissions with 26% of persons accounting for 51% of all asthma-related admissions. More recently, several reports of childhood asthma hospitalization now indicate a decrease in both new hospitalizations and readmissions (18–20). However, readmissions, in particular, remain as an issue of great concern. Studies of patients who represent readmissions have all examined the various characteristics that differentiate them from single admissions within the study interval. Age at admission (14–17, 21, 22), sex (14, 22), race/ethnicity (15, 17, 23), acute severity of asthma (14, 15, 24, 25), chronic asthma severity (13, 14, 22, 26), previous admissions (12, 15, 24, 27), socioeconomic status (23, 28), parental knowledge (22), and drug management (10, 12–15, 19, 22, 26, 29) have all been considered as risk factors associated with susceptibility to readmission for asthma. Cofactors associated with increased risk of readmission in these studies have included age at first admission, sex, race/ethnicity, and severity of asthma, drug management, and parental knowledge. Some have shown that a strong independent risk factor for readmission was a prior asthma admission. None, however, have examined the combination of socioeconomic status, race/ethnicity, and Medicaid insurance in characterizing the group of patients with readmissions. This study extends the findings of the previous literature with an examination of admissions at the two children's hospitals in St. Louis over a span of 10 years and identifying individual patients for all admissions. This comprehensive database is used to compare age at first admission, sex, and geographic location by zip code, ethnicity, insurance status, month of admission, and prior history of asthma admission, for all of those patients with a single admission within the 10-year span and those with multiple admissions. Our intention is to assess the magnitude of the readmission issue in our metropolitan area when readmissions of individual patients are counted over a very long time span. We then examine measurable risk factors that can be used to identify patients at risk for readmission, the purpose of which would serve to target intervention programs at an early admission number. We expect the characteristics of this patient population to direct our efforts in the design of such intervention.
Patients The sampling frame includes all hospitalizations with a primary discharge diagnosis of asthma (ICD-9 code 493) at St. Louis Children's Hospital and at SSM Cardinal Glennon Children's Hospital from January 1, 1990 through December 31, 1999. St. Louis Children's Hospital is a 235-bed teaching and community hospital of Washington University School of Medicine. SSM Cardinal Glennon Children's Hospital is a 190-bed pediatric hospital affiliated to St. Louis University School of Medicine. Both hospitals are located within the St. Louis city area and serve the metropolitan area of St. Louis and adjacent counties. Several community pediatric services at nearby general hospitals also serve pediatric patients with asthma from this region. In 1995, of the 22,338 pediatric hospital discharges from the St. Louis city and St. Louis county area, 19,641 or 87.9% were from the two hospitals studied in this report (30). At this same midpoint of the data period, 1995, the two study hospitals represented 85% of all pediatric asthma admissions within the city and county of St. Louis on the basis of a survey of all hospitals that admitted children in St. Louis city and St. Louis County. Thus, confining the data set to these two hospitals reflects a substantial portion of asthma admissions with no apparent bias in patient characteristics. Data for admissions of patients with asthma were extracted from the two hospitals' billing databases for the 10-year period. Medical record number, name, and date of birth identified a total of 10,637 patients representing 18,569 admissions. This sample represents individual patients accounting for each asthma admission at either hospital by cross-matching full name and date of birth between the two hospital data sets. To ensure that each patient's hospital admissions for asthma were counted as accurately as possible, patients with zip codes of residence outside of Missouri and those who may have received their asthma care from providers not included in data ascertainment were excluded from analysis. This excluded 1,876 patients (17.6% of all patients initially included), of whom 1,482 had non-Missouri zip codes for all of their admissions. The remaining 8,761 children comprised a sample for which capture of all admissions could be maximized while remaining sufficiently representative (82.4% of all those admitted) to provide a reasonably generalizable sample.
Analysis of Clinical Factors
Statistical Analysis
Patients Admitted and Readmitted in the 10-year Interval of Study In the study period of 1990 through 1999, there were 8,761 patients with 14,905 admissions (Figure 1) . Of the total number of patients, 6,142 were admitted only once and 2,619, or 30%, were admitted more than once. The 30% of patients who had readmissions accounted for 59% of all admissions. The readmissions (subsequent admissions after the first) represent 41.2% of all admissions for the 10-year study interval (Figure 1). The 2,619 patients who had readmissions had 6,144 readmissions, of which 3,525 were third or more admissions. Three or more admissions represented 23.6% of all asthma admissions.
Demographic Attributes of Those Patients Who Had a Single Admission during 1990 to 1999 Compared with Those Who Had Multiple Admissions The distribution of age groups was similar between single and multiple admissions. The largest number of admissions, for both single and multiple admissions, occurred among patients between 1 and 4 years old (Table 1) . For the two groups of patients, those admitted once and multiple times, there were more males than females (Table 2) . The mean age of admission for males (5.2 ± 4.4 years with a median of 4 years) was significantly less than for females (5.7 ± 4.8 years with a median of 5 years). Sex was not strongly related to multiple admissions, as 29.1% of females and 30.4% of males were admitted more than once (Table 2).
Among all patients admitted for asthma during the study period, 6,193 (71.5%) were African-American, 2,375 (27.4%) were white, and 91 (1.1%) were classified as of other ethnicity. African-American ethnicity was strongly associated with multiple admissions (Table 2). The ratio of African-American to all other patients was 2.16 for single admissions and 4.38 for multiple admissions (  2, p < 0.0001). The racial/ethnic distribution of children 0 to 17 years of age in the city and county areas from which these patients are drawn was African-American to white ratio of 1.71 to 1 in 1990 and 1.96 to 1 in 1999 for the city and 0.24 to 1 in 1990 and 0.28 to 1 in 1999 for the County. Analysis of zip code of residence at the time of first admission for asthma indicated that patients who had only a single admission and those with multiple admissions tended to live in the same zip codes. Because this distribution closely parallels the racial composition of the St. Louis area, zip code of residence does not emerge as a statistically significant predictor of multiple admission in models that also include ethnicity.
Insurance status is also statistically related to multiple admissions (Table 2). When these categories are stratified according to single and multiple admissions, the ratio of Medicaid or self-pay insurance to commercial insurance was 1.94 for the multiple admit group and 1.29 for the single admit group (
Clinical Characteristics of Asthma of Those Patients Who Had a Single Admission during 1990 to 1999 Compared with Those Who Had Multiple Admissions
Multivariable Model of Readmission
Pattern of Changes in Probability of Readmission The probability of readmission as a function of the number of previous admissions indicates increasing risk with each subsequent admission until a plateau is reached at the fifth admission (fourth readmission). Thirty percent of the patients with one admission were admitted a second time, 46% of patients admitted twice were admitted a third time, and 59% of patients admitted three times were admitted a fourth time. Thereafter, there is increasing variability due to decreasing numbers of patients. The general pattern is portrayed in Figures 2A and 2B .
Figures 2A and 2B also illustrate the proportion of patients readmitted by race/ethnicity and insurance status, respectively. The increase in likelihood of readmission from first to second to third admissions is greater for African-Americans than for other race/ethnicities. Similarly, it increases more steeply for Medicaid/self-pay cases than for those with private insurance. However, after the third admission, the probability of readmission is similar across these categories.
Sensitivity, Specificity, Positive and Negative Predictive Values of Criteria for Identifying Patients at Risk of Subsequent Admissions
Readmissions for asthma are a substantial portion of asthma admissions. In studies of asthma hospitalizations, readmission rates are often diluted or hidden (7, 16). In this study, we identified 30% of pediatric patients hospitalized over 10 years who had at least one readmission within that interval. These readmissions, in turn, accounted for 41.2% of the total asthma admissions. Furthermore, 23.6% of all admissions were third or more asthma admissions. This has significant implications concerning epidemiology, related socioeconomic issues, and particularly, the opportunity for targeted intervention programs designed to reduce childhood asthma hospitalizations. Starting with individuals most at risk for hospitalization using the history of a previous hospitalization, a large number of further asthma hospitalizations may be prevented by concentrating on the at-risk factors of a relatively small number of patients. In addition to readmissions as a large portion of asthma admissions in general, the probability of readmission over time increases with each subsequent admission. After a first admission, a child admitted for asthma is 30% likely to be admitted again for asthma and if admitted the second time has an even greater chance (46%) of a third admission. If a third admission takes place, the chance of continuing on to a fourth is 59%. Therefore, not only are children who are admitted for asthma even once likely to be readmitted again for asthma but these children are increasingly predisposed to hospitalization for asthma exacerbations with each subsequent hospital admission. Although race/ethnicity and insurance status relate to continued asthma hospitalizations throughout the duration of asthma in a child's life, the prediction of readmission from the number of previous admissions is specific and represents an impartial marker of the likelihood of subsequent admission. The measure of number of previous asthma admissions has a higher specificity and sensitivity than race/ethnicity, insurance status, or their combination. Any readmission should be used as a primary cue for intervention. Sensitivity and specificity of prior admission number in predicting a readmission suggests that the second admission is an effective time to fully reevaluate risk factors related to that patient. With the second hospitalization, 29.9% of the patients are represented (2,619/8,761) and of these 2,619 patients, 1,218 would go on to 3,525 additional admissions, which represents 23.6% of all admissions. Furthermore, this approach is unbiased and avoids characterization of the risk population by race/ethnicity or social-economic status. This may also be a marker for additional factors not accurately captured with available social and demographic information. It leaves open the possibility of identifying barriers to good care specific to the patient. Resources intended to prevent exacerbations can be directed to identified individuals rather than whole classes defined by minority or low-income status. With the above in mind, it appears that a second hospitalization is a reasonable signal for the increased investment of resources required for an individualized comprehensive management. It is apparent that there are significant disparities among the children hospitalized for asthma. A recent study of childhood asthma hospitalizations in the St. Louis metropolitan area found the risk of hospitalization to be 8.4 times greater for children living in lower socioeconomic zip code areas and 5.3 times greater in those zip codes with a higher percentage of African-American population (35). African-American and Medicaid/self-pay children are disproportionately represented in the readmission group. Among the patients admitted only once in the study interval, the ratio of African-American race/ethnicity to all other patients was 2.16 to 1, but among the patients with multiple admissions this ratio was 4.38 to 1. Insurance status also was significantly related to multiple admissions. Yet, in our model, neither race/ethnicity nor insurance type alone increases the likelihood of readmission. African-American race/ethnicity plus Medicaid/self-pay does significantly increase readmission (see Tables 3 and 4). The effect of race/ethnicity is different for patients with different types of insurance. Or, alternatively, the effect of insurance type differs depending on the race/ethnicity of the patient. Specifically, the data from this study support the conclusion that African-American patients with Medicaid/self-pay insurance are at significantly higher risk of readmission than are African-American patients with commercial insurance. Among white patients, there is not a clearly higher level of risk for patients with Medicaid/self-pay insurance. The principal conclusion is that African-American children with Medicaid/self-pay insurance are at higher risk of readmission than all white children and than African-American children with commercial insurance. The association of minority status and Medicaid/self-pay insurance associated with asthma hospitalizations has been amply reported and discussed (1–3, 28, 35–43). One study of asthma readmissions reported that the children who were readmitted did not differ from those with a single admission in terms of race/ethnicity or insurance status (13). A recent report using data from the National Center for Health Statistics found that it was not possible to determine from that data set whether the risk of readmission is influenced by race/ethnicity (7). The results of our study convincingly demonstrate that African-American children with Medicaid or no insurance are disproportionately predisposed not only to hospitalization for asthma, but also to repeated hospitalizations for asthma over a period of many years. However, race/ethnicity and socioeconomic status are not sufficient in themselves to explain child health problems (44). Factors that mediate the relationship that exists between asthma admissions in general and children of minority and low-income families may very well be the same factors that exist for readmissions. Yet, there may be some factors not considered within current paradigms. Issues that put children with asthma at-risk for increased health care use are physician prescribing practices, patient adherence to medications, and access to care. Lieu and coworkers (45) demonstrated that Black and Latino children had worse asthma status and less use of preventive asthma medications than white children within the same managed Medicaid populations (45). Underuse of appropriate controller medication either due to lack of prescribing by physicians or due to lack of adherence has been documented among minority and low-income children (46–51). Even children with multiple hospitalizations do not necessarily receive adequate therapy (49). Among inner-city children previously hospitalized, NHLBI guidelines for home management of asthma exacerbations are not being followed (52). Children who do receive inhaled anti-inflammatory therapy do have a reduced risk of hospitalization events (53), and it has been shown that patients treated regularly with an inhaled corticosteroid for 6 months after an asthma admission were 40% less likely to be readmitted for asthma (29). Continuous treatment in a randomized, controlled study resulted in a reduction in prednisone courses, emergency department visits, and hospitalizations for all groups (54). Yet, in disadvantaged low-income groups, children are often responsible even at an early age for taking their inhaled medications themselves (50). Children studied in the National Cooperative Inner-City Asthma Study were found to have identified primary care but the majority used a hospital-based pediatric clinic for acute symptoms (55). This is consistent with previous studies showing that poor children when ill were four times more likely to receive care in an emergency department than nonpoor children (56). There may be barriers to specialized care for asthma as there are for African-American children with frequent ear infections who have an identified usual place of health care and health insurance similar to white children (57). Inadequate management and lack of access may not be the primary cause of increased morbidity for these patients (58). Psychosocial factors may play also play a role. Chen and coworkers (59) found that individual caretaker characteristics such as a lower sense of mastery concerning asthma care predicted greater likelihood of future asthma hospitalizations and that a lifetime history of asthma hospitalizations was associated with family impacts of strain, conflict, and beliefs about not being able to manage one's child's asthma (59). Lozano and coworkers (60) noted that there might be specific problems in the successful management of chronic diseases such as asthma among African-American children with Medicaid insurance. They concluded that the higher use of emergency department and inpatient services among these children using Medicaid could not be fully explained by poverty or inadequate health insurance. Ray and coworkers (61) also noted in studying asthma hospitalizations that after controlling for socioeconomic status, notable differences by race/ethnicity still persist. Hisnanick and coworkers (23), reporting hospital admissions for American-Indian and Alaskan-Native children, found rates similar to those reported for white children despite having socioeconomic conditions similar to African-American children. Weil and coworkers (62), after controlling for baseline morbidity, noted that childhood hospitalization was correlated significantly with the caretaker's mental health and life stress. They emphasize the importance of nonmedical factors in asthma morbidity related to health care use in that hospitalization and unscheduled visits were not strongly associated to the day of wheeze or the functional status of the patient. Our results on the probability of readmission are consistent with previous reports. Schaubel and coworkers (27) reported that the probability of readmission to hospital for asthma within 2.5 years of the date of hospitalization was estimated at 42, 59, and 76% among children with one, two, and three previous hospitalizations, respectively. Mitchell and coworkers (15) concluded that the number of previous admissions was the best single predictor of readmission. They found that of patients with no previous admissions and one, two to three, and four previous admissions, 40, 50, 60, and 70%, respectively, were readmitted by 2.75 years. The similarity of these data from three countries, Canada, New Zealand, and this current U.S. report, strongly suggests that there is something unique about previous hospitalization for childhood asthma. Limitations inherent in our review include the evaluation of asthma severity and the breadth of our sampling frame. Although the extent of hospital coverage and longevity of this study allowed us to fully characterize the dimension of the readmission issue, it also limited our ability to fully evaluate disease severity as an independent measure. Examining relationships among events taking place over time and at several locations is likely to be more valid the broader the sampling frame. However, breadth of sampling frame increases difficulty of assuring accurate accounting of data. To obtain the most complete count of admissions for each patient, we limited our sample to patients with Missouri zip codes of residence in all records. Thus, children from Illinois who use St. Louis city hospitals for some of their asthma care are not represented. We limited our data to those from two urban hospitals. However, for the year corresponding to the midpoint of the study (1995), these two hospitals accounted for 85% of pediatric asthma hospitalizations in all St. Louis city and county hospitals. Consequently, it is unlikely that the omission of 15% of hospitalizations biases the present findings. In addition, some patients may have had admissions outside urban St. Louis. Again, this is likely to be a very small proportion of all their hospitalizations, the omission of which is unlikely to have biased our findings. Also, our model is limited to the first five readmissions. Readmission to hospital for asthma is a significant health care problem. Comprehensive intervention is costly and time consuming. Inpatient hospital services represent the largest direct medical expenditure for asthma. Forty-three percent of the economic impact of asthma is associated with emergency department use, hospitalization, and death (63). We have shown that readmissions are a substantial portion of the admission population and that there is a disproportionate association with African-American racial/ethnicity and low income as measured by insurance status. Furthermore, we demonstrate an increasing risk for readmission with each subsequent asthma admission. Although disease severity is a risk factor for hospitalization (64), additional cofactors may increase the risk for readmissions. Interventions to improve medication use, adherence to medication regimens, access to appropriate care and especially attention to psychosocial issues applied at a second hospitalization presents an opportunity to prevent a disproportionate number of predictable hospitalizations generated by a relatively small number of patients and thereby reducing the burden of morbidity and the cost of potentially preventable health care use. Research involving the outcomes of intervention triggered by a second hospitalization for asthma is anticipated and may potentially uncover additional factors not previously considered in the present paradigms of excess health care use.
Supported by grants from the NHLBI (HL45923, HL57232) and from the National Institute of Environmental Health Sciences (ES08711). Received in original form January 8, 2002; accepted in final form January 3, 2003
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
