Readmission and Ambulatory Health Care Use After Asthma Hospitalization Among Urban Minority Children

This retrospective cohort study reviewed charts of children discharged from an urban, tertiary-care, academic hospital between June 1, 2014, and May 30, 2016. We queried our clinical research database, Looking Glass Clinical Analytics (Clinical Looking Glass; Streamline Health, Atlanta, Georgia),¹⁶  to identify 2- to 18-year-olds discharged with an All Patient Refined Diagnosis-Related Group (APR-DRG) of asthma (141; n = 1591) and confirmed the cohort through chart review (Supplemental Fig 1). Children <2 years old were excluded because of the diagnostic overlap between asthma and bronchiolitis. The index admission was the first asthma hospitalization within the study period. Because this was a single-center study and ambulatory health record access was needed, only children receiving primary care at the hospital’s clinics were retained (n = 787). This was determined by an outpatient visit (health maintenance, vaccine encounter, or sick visit) within 18 months before the index hospitalization. Among these, children with coexistent complex chronic diseases (congenital heart disease, prematurity, sickle cell disease, neuromuscular disorders, structural airway malformations, or immunodeficiency) were excluded (n = 164) given their contribution to higher readmission risk.²  Children readmitted within 14 days of discharge (n = 10) were also excluded because they were readmitted before having the chance to attend a postdischarge visit. Children readmitted before the chance to follow-up at 3 months (n = 16) were also excluded from analysis of the routine visit. Readmission and follow-up of children at outside institutions were not measured. All ambulatory records 365 days from the index hospital discharge were reviewed. This study was approved by the institutional review board at the Albert Einstein College of Medicine.

Demographic variables retrieved included age (2–11 vs 12–18 years), sex, primary language spoken (non-English versus English), race and/or ethnicity (Hispanic, non-Hispanic African American, non-Hispanic other, or unavailable), and socioeconomic status (SES). Age was dichotomized given the higher readmission odds for adolescents.²  SES was calculated from small census tract and block data on the basis of home address and is a z score representing the deviation of this value from the mean of the New York state population.¹⁷  Index hospitalization variables included length of stay, need for PICU admission, need for magnesium, APR-DRG severity of illness (mild, moderate, severe, or extreme), controller medication prescription at discharge, and documentation of a scheduled postdischarge visit. Postdischarge variables included discharge date, date of first attended visit, documentation of routine (≤3 months) asthma follow-up at that visit, date of second attended visit addressing asthma symptoms, and number of ED visits and readmissions within 365 days of the index discharge.

Measures of baseline asthma disease severity 1 year before the index hospitalization included number of: ED visits, hospitalizations, asthma sick visits (acute exacerbation requiring bronchodilator treatment or systemic steroids), or asthma scheduled visits (asthma symptoms were addressed in the absence of exacerbation). A history of a PICU admission, visit to an asthma specialist, use of controller medications (inhaled corticosteroids, montelukast, and/or an inhaled steroid or long-acting bronchodilator combination inhaler), environmental exposures (tobacco, dogs and/or cats, mice, cockroaches, and/or mold), and presence of complicating comorbidities (obesity, allergic rhinitis, eczema, gastroesophageal reflux disease, and/or obstructive sleep apnea) were also collected.¹⁸  Asthma severity based on controller medication use in the year before the index admission was classified per the Expert Panel Report 3 guidelines as intermittent, mild, moderate, or severe persistent.⁷  All demographic variables and APR-DRG severity of illness were collected electronically (K.S.P.). All other variables were collected via manual chart review (D.E.R.), and the review was overseen and confirmed (K.S.P.).

Our study goal was to determine if there was an association between attending a postdischarge visit (within 14 days of discharge), a routine visit (within 3 months of postdischarge visit or 3 months of discharge, when asthma was addressed), or both visits and having at least 1 readmission for asthma within 365 days of the index discharge. We also examined the associations of demographic and clinical variables, including baseline asthma severity and severity of the index hospitalization, with attending postdischarge, routine, or both visits.

We examined the bivariate associations of readmission for asthma within 365 days with attendance at postdischarge, routine, or both visits and with demographic and clinical characteristics of the children using χ² or Fisher’s exact tests. We then developed 3 separate multivariable logistic regression models to determine if attending a postdischarge, routine, or both visits were independently associated with having an asthma readmission within 365 days when adjusting for covariates. A similar approach was taken with multivariable logistic regression analysis to determine the independent predictors of attending each type of visit. A manual backward selection procedure was used in model building, with full models including variables with P < .20 in bivariate testing. Age, sex, race and/or ethnicity, and asthma severity based on medication were included in each tested model but allowed to drop out if their contribution was not statistically significant. Covariates were removed in a stepwise manner until all remaining were significant at P < .05. After each step, a likelihood-ratio test to assess model fit was performed. Data were analyzed by using Stata 14.2 (Stata Corp, College Station, TX).

Among the 613 children retained in the analysis, 592 (96.6%) had another visit in our system within 1 year of the index discharge. Our population was 57.4% boys, 51.2% Hispanic, 36.9% non-Hispanic African American, and 10.8% non-Hispanic of other race. The median SES z score was −3.5, or 3.5 SDs below the average for New York state (interquartile range −6.7 to −1.4). The median age was 5 years (interquartile range 3–9). Ninety-two percent of the cohort identified as English speaking. On the basis of medication use, 43.2% of children had intermittent, 28.2% had mild-persistent, 21.2% had moderate-persistent, and 7.3% had severe-persistent asthma.

Table 1 shows the association of clinical and demographic variables with readmission. Children who attended a postdischarge visit were less likely to have 1 or more readmissions within 365 days of the index discharge (P = .03). Additional variables associated with readmission were an ED visit, asthma hospitalization, a scheduled visit for asthma in the previous year, history of PICU admission, disease severity, environmental exposures, a PICU stay during the index hospitalization, controller medication being prescribed at discharge, and index hospitalization APR-DRG severity of illness.

Having identified these clinical factors as well as follow-up visits as being significantly associated with readmission in bivariate analyses, we conducted multivariable analysis to identify independent predictors of readmission. Children who attended a postdischarge visit were less likely to be readmitted than children who did not attend (odds ratio [OR] = 0.65; 95% confidence interval [CI] = 0.43–0.97; Table 2). However, there was no association between attending a routine visit and readmission. Similarly, children who attended both a postdischarge visit and a routine visit were not less likely to be readmitted. In addition to attendance at a postdischarge visit, variables associated with readmission in all models include having an asthma hospitalization in the year before the index hospitalization and having a PICU stay during the index hospitalization (Table 2).

There were 416 (67.9%) children who attended a postdischarge visit (Table 3). Independent predictors of attendance included an asthma sick visit in year before the index hospitalization, higher index hospitalization APR-DRG severity of illness, and documentation of a scheduled postdischarge visit. Children with a history of PICU admission or those discharged on a controller medication were less likely to follow-up. English-speaking families also tended to be less likely to attend the follow-up, although this reached only borderline significance (OR = 0.46; 95% CI = 0.22–1.00).

Sixteen children were readmitted after their postdischarge visit before having a chance for routine follow-up and so were excluded from further analysis. More than half of the included children (332 of 597) attended a routine visit (Supplemental Table 4), and more than one-third (229 of 597) attended both a postdischarge and a routine visit (Supplemental Table 5). Children with an asthma sick visit in the year before the index hospitalization, asthma comorbidities, or higher index hospitalization APR-DRG severity of illness were more likely to attend these visits. English-speaking families were less likely to attend the routine visit (OR 0.46; 95% CI 0.24–0.89). Additionally, when providers documented the need for follow-up at the postdischarge visit, children were 2 times more likely to return for a routine visit (unadjusted OR = 2.3; 95% CI = 1.49–3.39). Although this occurred only 61% of the time, it was not associated with readmission (P = .97; Table 1).

To our knowledge, this is the first study to investigate the association of ambulatory follow-up with risk of long-term readmission among minority children with asthma. Children seen in the ambulatory setting within 14 days of discharge were less likely to be readmitted, but attendance at both a postdischarge and a routine visit was not associated with fewer readmissions. Children with higher ambulatory use before the index hospitalization and a more severe hospital course were more likely to attend all visit types. Additionally, documentation of ambulatory follow-up at both hospital discharge and the postdischarge visit was associated with higher attendance.

Our findings validate a recent single-center quality improvement initiative in Hawaii that found similar baseline rates of attendance at postdischarge visits (69%), which improved after interventions to comply with Children’s Asthma Care measures.¹⁹  The authors concluded that postdischarge visit attendance is associated with lower rates of asthma readmissions. However, they managed readmissions for 3 to 6 months postdischarge and do not report longer time intervals, which differs from our study. Although we found that children who attend a visit immediately postdischarge are less likely to be readmitted, those who attend routinely (suggesting good medical home access) are not less likely to be readmitted long-term. This suggests that long-term readmission rates may be determined primarily by disease severity. Furthermore, it is likely that multiple factors are necessary to prevent readmissions even with appropriate access to a medical home. Although we do not have details on medication adherence and other interventions, such as home-allergen exposure remediation, that decrease asthma burden,²⁰  we speculate that these additional factors may be more prevalent in families who attended a postdischarge visit. Furthermore, the decrease of ∼50% in the number of children seen for postdischarge visits and for routine visits identifies an area of improvement for our system and may have influenced the strength of the association between routine visits and readmissions.

Because ambulatory visit attendance may prevent readmissions, even if short-term, we investigated the factors associated with attending ambulatory visits after hospitalization, which is an understudied area of pediatric research. Children who were seen for an asthma sick visit in the year before the index hospitalization, had a complicating asthma comorbidity, or experienced a more severe index hospital course were more likely to attend visits in the year after hospitalization. Additionally, documentation of a scheduled postdischarge visit was associated with a likelihood of visit attendance that was 2 times higher. This result was replicated when we analyzed documentation of a need for a routine follow-up visit. However, providers appropriately documented the need for routine follow-up only 61% of the time, suggesting suboptimal provider compliance with National Heart, Lung, and Blood Institute guidelines on monitoring asthma control.⁷  Our observation extends findings from a previous study that linked a follow-up appointment made from the ED to greater likelihood of visit attendance within 4 weeks of ED discharge to a hospital admission for asthma.¹⁵  Scheduling and documenting follow-up visits before hospital discharge are likely key interventions for increasing follow-up, which may play a role in preventing readmissions and should be a focus of future quality improvement efforts.²¹ 

To date, most studies have focused on predictors of ambulatory follow-up after an ED visit rather than asthma hospitalization. Zorc et al¹⁵  found asthma severity, as measured by health care use (ED visits and hospitalizations) and controller medication use, was associated with ambulatory follow-up. Although we validate the association of previous asthma burden with ambulatory follow-up, our findings differ on the association of controller medication prescription with ambulatory visits. These differences may be due to ethnic differences because Zorc et al¹⁵  studied a primarily African American population who spoke English, and we studied a primarily Hispanic population with both English-speaking and non–English-speaking subjects. The limitation of caregiver recall and abstraction of chart documentation may also explain the divergent results.

Contrary to our prediction and findings from previous studies,⁸  children with a previous history of PICU admission and those discharged on a controller medication were less likely to attend a postdischarge visit. In our study, controller medications prescribed at or before discharge were associated with a decreased likelihood of attendance, although details on the timing of controller medication prescription were not available for the previous study.⁸  This, in addition to population differences (such as race and/or ethnicity and disease severity), may potentially underlie the disparate findings. On the basis of these findings and in addition to the association of previous PICU admission with future admissions, we speculate that a subset of children who suffer from asthma for a longer time or have more severe asthma may not follow-up with their pediatrician postdischarge because their caregivers either become accustomed to managing this chronic, episodic condition at home or are unaware of the chronic recurrent nature of the disease. These caregivers may fail to recognize the importance of routine follow-up in decreasing disease morbidity, highlighting a subset of caregivers who will benefit from additional asthma education by primary care providers during routine follow-up visits.

Our study fills an essential, recognized literature gap by examining the role of primary care follow-up after an inpatient asthma hospitalization in a pediatric population.¹²  We believe the quality of content covered, including adherence to controller medications, symptom monitoring, and early initiation of albuterol in the setting of an exacerbation, is just as, if not more, important than simply attending the visit. However, we found that more than one-third of providers did not document the need for routine asthma follow-up care during the postdischarge visit. A systematic analysis of the impact of ambulatory follow-up after an asthma ED visit or hospitalization also suggests that visit content is key to improving outcomes.²²  However, the studies reviewed focused on both adult and pediatric patients, were primarily ED focused, and did not address the association of routine follow-up visits with readmission, as was done in our study. We speculate that a lack of attention to visit content may explain the mixed results reported by previous studies that analyzed the benefit of ambulatory follow-up on health care use.^23,24  We also identify several areas of investigation to further elucidate the contribution of disease burden and that of ability to attend ambulatory visits to readmissions.

Our study has several limitations. Despite the large sample size, this is a single-center study of a low-income, minority population. These children received primary care at a tertiary-care center and thus had interprofessional care teams or providers and good access, which may differ from children who do not have this resource. In addition, we did not include children with complex chronic illnesses that may worsen asthma severity and contribute to higher health care use. These factors may limit the generalizability of our results. In addition, data were abstracted from the electronic medical record and chart review. Thus, they were subject to human entry error and limited by the reliability and presence of covariate documentation. To minimize this bias in the cohort formation, eligibility criteria were confirmed during chart review. To increase the accuracy of variables that were considered present on the basis of documentation (use of controller medications and environmental exposures), reviewers regularly reviewed data from both clinic notes and hospital discharge summaries. These variables, however, may be more likely to be documented in children who have more prehospitalization visits. Because analyses were limited to information that is available within our hospital system and patients may seek asthma treatment at outside institutions, variables including history of asthma hospitalization and ED visits as well as presence of postdischarge readmissions may be underreported. However, we believe this is minimal because we only included patients who were established in our system, and 97% of them had an ambulatory follow-up visit in our system. Finally, many psychosocial factors impact appointment attendance, including caregiver education and employment status, proximity to the office, and medical home outreach to confirm appointment status or assess postdischarge needs. These factors were not accounted for in our study and remain important considerations when designing future interventions aimed at ambulatory asthma follow-up.

We found that attending a postdischarge visit within 14 days of hospitalization is associated with a lower likelihood of long-term readmission. Additionally, those with a more severe exacerbation or previous asthma sick visits are more likely to attend ambulatory visits after discharge. Fewer children with higher disease burden attended the routine visit. Furthermore, documenting the appointment at hospital discharge improved attendance rates, highlighting the need for hospital resource allocation for this task. Children with more severe asthma may require extra encouragement to continually seek ambulatory care because caregivers may become accustomed to managing symptoms at home. Regular assessment of asthma control and medication adherence is critical to improve long-term outcomes. The next steps for this research include facilitating postdischarge follow-up among children hospitalized with asthma, evaluating the quality of these ambulatory asthma visits, developing alternative interventions to address key aspects of asthma management both during and after inpatient hospitalization, and assessing the long-term impact of these interventions on asthma outcomes.