Prospective Multicenter Study of Bronchiolitis: Predictors of an Unscheduled Visit After Discharge From the Emergency Department

Abstract

Objectives

There is little evidence about which children with bronchiolitis will have worsened disease after discharge from the emergency department (ED). The objective of this study was to determine predictors of post-ED unscheduled visits.

Methods

The authors conducted a prospective cohort study of patients discharged from 2004 to 2006 at 30 EDs in 15 U.S. states. Inclusion criteria were diagnosis of bronchiolitis, age <2 years, and discharge home; the exclusion criterion was previous enrollment. Unscheduled visits were defined as urgent visits to an ED/clinic for worsened bronchiolitis within 2 weeks.

Results

Of 722 patients eligible for the current analysis, 717 (99%) had unscheduled visit data, of whom 121 (17%; 95% confidence interval [CI] = 14% to 20%) had unscheduled visits. Unscheduled visits were more likely for children age <2 months (11% vs. 6%; p = 0.04), males (70% vs. 57%; p = 0.007), and those with history of hospitalization (27% vs. 18%; p = 0.01). The two groups were similar in other demographic and clinical factors (all p > 0.10). Using multivariable logistic regression, independent predictors of unscheduled visits were age <2 months, male, and history of hospitalization.

Conclusions

In this study of children age younger than 2 years with bronchiolitis, one of six children had unscheduled visits within 2 weeks of ED discharge. The three predictors of unscheduled visits were age under 2 months, male sex, and previous hospitalization.

The clinical course of children with bronchiolitis is variable. Most children have a mild severity of illness, but some children have a severe course manifested by apnea, intensive care unit (ICU) admission, intubation, or even death.^1–3 Even among children with apparently mild bronchiolitis, the clinical course is often unpredictable, making it difficult for physicians in the emergency department (ED) to determine the appropriate disposition for a child with bronchiolitis. This is illustrated by the varied rates of hospital admissions for children with bronchiolitis among pediatric ED attending physicians⁴ and between pediatric and general EDs.⁵ Furthermore, one study documented immensely different thresholds for ICU admission and intubation among 10 different hospitals.⁶ Despite, or perhaps because of, this variability in care, bronchiolitis is the leading cause of hospitalization in infants, accounting for 16% of all infant hospitalizations.⁷

Nevertheless, most children who present to the ED with bronchiolitis are discharged to home. Due to the prolonged, unpredictable 2- to 4-week recovery time for children with bronchiolitis,⁸ unscheduled visits for medical care following treatment and discharge from the ED remain problematic. One study of ambulatory patients found that 34% of patients with bronchiolitis relapsed to medical care, with a median time for relapse of 13 days.⁸

In an attempt to provide evidence-based data for the disposition of children with bronchiolitis, recent studies have found that patient demographics, medical history, clinical symptoms, and ED management are predictors of both severe disease^2,3,7,9–12 and safe discharge home from the ED.¹³ However, no evidence exists to predict which of those children discharged home from the ED may develop worsening bronchiolitis that results in unscheduled medical care visits. Identifying children at high risk of unscheduled visits may improve the ability of the emergency physician to devise a more personalized disposition plan and to provide more specific discharge counseling for parents and guardians.

We conducted a prospective multicenter study of over 1,200 children age <2 years who presented to the ED with bronchiolitis. The primary objective of the overall study was to identify factors associated with safe discharge to home from the ED. The objective of this secondary analysis was to determine among those patients with bronchiolitis discharged to home from the ED, which factors were associated with unscheduled visits for medical care visits for bronchiolitis.

METHODS

Study Design

We conducted a prospective cohort study during the 2004 to 2006 winter seasons, as part of the Multicenter Airway Research Collaboration (MARC). We retrospectively applied the American Academy of Pediatrics (AAP) definition to physician-diagnosed cases, and 98% of enrolled children satisfied these criteria. The institutional review board at each of the 30 participating hospitals approved the study, and informed consent was obtained for all participants.

Study Setting and Population

MARC is a division of the Emergency Medicine Network (http://www.emnet-usa.org). Using a standard protocol, investigators at 30 EDs in 15 U.S. states provided 18- to 24-hour-per-day coverage for a median of 2 weeks from December to March to coincide with high numbers of bronchiolitis visits. Inclusion criteria were attending physician diagnosis of bronchiolitis, age <2 years, and the ability of the parent/guardian to give informed consent. The only exclusion criterion was previous enrollment. The AAP, in its 2006 position statement, defines bronchiolitis as children having “rhinitis, tachypnea, wheezing, cough, crackles, use of accessory muscles, and/or nasal flaring.”¹⁴

Study Protocol

All patients were managed at the discretion of the treating physician. The ED interview assessed demographic characteristics, medical and environmental history, and details of the acute illness (including medications used during the week leading up to the ED visit and duration of symptoms). Median household income was estimated using the patients’ home ZIP codes.¹⁵ Children were considered to be premature if they were born at <35 weeks’ gestation.

ED chart review provided clinical data: respiratory rate from triage, clinical assessment of degree of retractions (combined for analysis into none/mild versus moderate/severe), O₂ saturation, management, and disposition.

Follow-up data regarding unscheduled visits were collected by telephone interview two weeks after the ED visit. The primary definition of an unscheduled visit was any urgent visit to an ED or clinic for worsening of bronchiolitis during the 2-week follow-up period. To further evaluate unscheduled visits among these children, a more restrictive definition also was evaluated. This more restrictive definition defined an unscheduled visits event as a worsening of bronchiolitis that led the parent or guardian to take the child for further evaluation, resulting in the doctor changing the child’s medications (e.g., adding nebulizers or steroids) or admitting the child, or the doctor added medications or made some treatment changes indicating worsened bronchiolitis during a routine medical visit (i.e., not a visit specifically for bronchiolitis). Unscheduled visits also were further classified as being admitted for worsening bronchiolitis and relapse to the ED. To examine the reproducibility of the unscheduled visits classification, two authors (JMM and CAC) independently reviewed the two-week follow-up data for 100 randomly sampled cases (kappa = 0.98). All forms were reviewed by site principal investigators, who are physicians, before submission to the EMNet Coordinating Center in Boston, MA. At the Coordinating Center, the data were further reviewed by trained personnel and underwent double data entry.

Data Analysis

All analyses were performed using STATA 9.0 (Stata-Corp, College Station, TX). Data are presented as proportions (with 95% confidence intervals [CIs]), means (with ± standard deviation [±SD]), or medians (with interquartile ranges [IQRs]). The association of factors with post-ED unscheduled visits was examined using the chi-square test, Student’s t-test, or Kruskal-Wallis rank test, as appropriate. All p-values are two-tailed, with p < 0.05 considered statistically significant.

Multivariable logistic regression was used to identify independent predictors of unscheduled visits within 2 weeks of ED discharge. The multivariable model was created using manual procedures. Factors associated with unscheduled visits at p < 0.2 were evaluated for inclusion in multivariable analysis. Age, sex, and race were included in the model because of their potentially clinical significance. The final model includes a complete case set with a sample size of 701 observations. Results are presented as odds ratios (ORs) with 95% CIs. Discrimination of the multivariable model was determined using the c-statistic. Model goodness of fit was evaluated using the Pearson chi-square goodness-of-fit test. We also reanalyzed the data excluding those children with a history of wheezing.

RESULTS

Of 1,456 enrolled patients, 722 (58%) were discharged to home and eligible for the current analysis. Unscheduled visit data were available for 717 (99%) of these children, and among them 121 (17%; 95% CI = 14% to 20%) had an unscheduled visit. None of the visits were for a routine or follow-up check. The median number of days to unscheduled visit was 2 (IQR = 1–4). The distribution of number of days to unscheduled visits is shown in Figure 1. Sixty-five percent of unscheduled visits occurred within 2 days of the ED visit. Using more restrictive criteria, 80 children (11%; 95% CI = 9% to 14%) had an unscheduled visit. Forty-nine children (6%) had an unscheduled visit leading to hospital admission, and 94 (13%) had an unscheduled visit to the ED.

Figure 1.

Number of days to unscheduled visit for children with bronchiolitis after discharge from the ED.

Comparisons of demographic and medical history between children with and without unscheduled visits are shown in Table 1. Children age younger than 2 months and males were more likely to have an unscheduled visit. No differences were observed with respect to race/ethnicity or having a primary care provider (PCP). Markers of socioeconomic status (e.g., estimated household income, insurance) were not found to be associated with an unscheduled visit (data not shown). The two groups also were similar with respect to most medical history factors, including disease or illness history, medication use, and health care utilization. Children experiencing an unscheduled visit were, however, more likely to have a history of hospitalization.

Table 1.

Demographic Characteristics and Medical History of Children Treated in the ED for Bronchiolitis and Discharged to Home, According to Unscheduled Visits Within 2 Weeks of ED Discharge

	No Unscheduled Visits (n = 596)	Unscheduled Visits (n = 121)	p-value
Demographic characteristics
Age <2 months	6	11	0.04
Male	57	70	0.007
Race/ethnicity			0.11
White	37	38
African American	38	28
Hispanic	21	29
Other	4	5
Has primary care provider	98	99	0.33
Medical history
Concomitant medical disorder	15	20	0.24
Premature*	7	11	0.19
History of wheezing	32	39	0.12
History of eczema	21	21	0.87
Ever hospitalized	18	27	0.01
Ever intubated	4	7	0.10
Medication use during past week
Inhaled beta-agonist	31	33	0.69
Antibiotic	15	15	0.88
Inhaled corticosteroid	6	9	0.19
Systemic corticosteroid	6	9	0.25
Healthcare utilization during past week
Number of PCP visits	0 (0–1)	0 (0–1)	0.64
Number of ED visits	0 (0–0)	0 (0–0)	0.45

Values are percentages or median (IQR).

IQR = interquartile range; PCP = primary care provider

^*Premature defined as gestation <35 weeks.

Emergency department presentation and clinical course are shown in Table 2. Initial measurements of respiratory rate and oxygen saturation were similar between the children with and without unscheduled visits, but children experiencing an unscheduled visit were more likely to have moderate or severe retractions. The two groups were equally likely to have presence of cough or wheezing while in the ED and received similar management. ED length of stay and medications prescribed at ED discharge also were similar for the children with and without unscheduled visits.

Table 2.

ED Presentation and Clinical Course of Children Treated in the ED for Bronchiolitis and Discharged to Home, According to Unscheduled Visits Within 2 Weeks of ED Discharge

	No Unscheduled Visits (n = 596)	Unscheduled Visits (n = 121)	p-value
Respiratory rate (breaths/min)	43 ± 13	46 ± 15	0.10
Respiratory rate < normal for age*	48	46	0.68
Retractions			0.04
None/mild	91	84
Moderate/severe	9	16
Oxygen saturation on room air	98 ± 2	97 ± 2	0.23
Lowest room air oxygen saturation	97 ± 2	97 ± 3	0.07
Presence of cough	83	87	0.35
Presence of wheeze	76	76	0.92
Number of inhaled beta-agonist  treatments over entire ED stay	1 (1–2)	1 (1–2)	0.73
Number of epinephrine  treatments over entire ED stay	0 (0–0)	0 (0–0)	0.35
Given corticosteroids	19	17	0.56
Given antibiotics	11	8	0.47
Any laboratory tests	64	63	0.74
Oral intake			0.11
Adequate oral intake	84	84
Inadequate oral intake	10	6
Unknown	6	10
Viral test results†
RSV positive	15	22	0.04
Influenza A positive	1	1	0.99
Influenza B positive	0.2	0	0.65
Adenovirus positive	0.2	0	0.65
Abnormal x-ray findings‡	60	46	0.06
ED length of stay (minutes)	165 (123–231)	180 (130–280)	0.78
Received discharge medication	93	89	0.11

Values are reported as mean (±SD), percentages, or median (IQR).

IQR = interquartile range; RSV = respiratory syncytial virus.

^*Normal values for age groups (in months): age 0–1.9, 45 breaths/min; age 2–5.9, 43 breaths/min; age 6–11.9, 40 breaths/min; and age 12–23.9, 40 breaths/min.

^†Among 255 children with viral testing.

^‡Abnormal x-ray findings include atelectasis, infiltrate, hyperinflated, or other findings (n = 330 children with chest x-ray).

Multivariable predictors of unscheduled visits within 2 weeks of ED discharge are shown in Table 3. Controlling for race, ethnicity, and prematurity because of their potential to be clinically important predictors, age younger than 2 months, male sex, and a history of hospitalization were independent predictors of a post-ED unscheduled visit. Analyzing the subset of children without previous wheezing who were discharged to home (n = 482), the results of the multivariable model did not materially change (data not shown).

Table 3.

Multivariable Logistic Regression of Factors Associated With Unscheduled Visits Within 2 Weeks of an ED Visit for Bronchiolitis

	OR	95% CI	p-value
Age <2 months	2.1	1.1–4.3	0.03
Male	1.7	1.1–2.5	0.02
Race/ethnicity
White	1.0	Reference	Reference
African American	0.6	0.4–1.1	0.09
Hispanic	1.3	0.8–2.1	0.37
Other	1.02	0.4–2.7	0.96
Premature*	1.6	0.8–3.2	0.16
History of hospital  admission	1.7	1.1–2.8	0.02

Area under the receiver operating characteristic curve: 0.64.

Pearson chi-square goodness-of-fit test statistic (33 degrees of freedom): 5.43 (p = 0.61).

^*Premature defined as gestation <35 weeks.

Using the more restrictive unscheduled visit classification, similar results were observed on unadjusted analysis (data not shown). Children <2 months were more likely to experience an unscheduled visit, while other sociodemographic factors were similar between the two groups. Similarly, children with a history of hospitalization were more likely to have unscheduled visits. ED presentation and management did not differ between the two groups when using the more restrictive unscheduled visit criteria. On multivariable analysis controlling for age, sex, race/ethnicity, prematurity, and history of hospitalization, significant predictors of unscheduled visits were age <2 months (OR = 2.5; 95% CI = 1.2 to 5.4, p = 0.02), prematurity (OR = 2.1; 95% CI = 1.03 to 4.4, p = 0.04), and history of hospitalization (OR = 1.8; 95% CI = 1.1 to 3.2, p = 0.03).

DISCUSSION

Although bronchiolitis is the leading cause of hospitalization for infants, the majority of children who present to the ED with bronchiolitis are discharged home.^4,5 If physicians could reliably identify those children who are most likely to have unscheduled visits, they would have the opportunity to provide more extensive discharge counseling and arrange for PCP follow-up for these high-risk children. In this prospective multicenter study, we found that roughly one of every six (17%) children who were discharged home from the ED after treatment for bronchiolitis had an unscheduled visit during the next 2 weeks. The median time to the unscheduled visit was 2 days, and 65% of children relapsed within 2 days of the initial ED visit. The three independent factors associated with unscheduled visits were age <2 months, male sex, and a previous history of hospital admission.

To our knowledge, this is the first prospective multicenter study to identify predictors of unscheduled visits in children age younger than 2 years with bronchiolitis. One retrospective case-control study compared children age younger than 12 months with bronchiolitis who returned and required admission to those who did not return.¹⁶ This study found that oxygen saturation and clinical assessment did not differentiate between these two groups. Previous studies have identified risk factors for severe bronchiolitis, some of which are similar to the factors we identified as predictors of unscheduled visits. Infants younger than 2 months have increased risk of apnea, ICU admission, prolonged hospital stay, and death.^2,3,9,10,12 Parallel to these data, our study found that infants younger than 2 months have higher risk of unscheduled visits. Prior studies also suggest that male sex may play a role in the severity of respiratory symptoms,^2,7 and we found that male sex was a predictor of post-ED unscheduled visits.

Two other published analyses did not find male sex to be associated with either increased risk of ICU admission¹⁷ or safe discharge home from the ED.¹³ Although the reasons are unclear, we postulate that these seemingly discrepant results might result from the unscheduled visit outcome representing an intermediate illness severity, while the previous two reported outcomes, admission to the ICU or safe discharge home, represented the extremes of the illness spectrum. Either way, we found that children with bronchiolitis who are either male or younger than 2 months are at high risk of both unscheduled visits and possibly more severe bronchiolitis. As a result, these children may benefit from additional discharge counseling if they are deemed safe for discharge.

Prior studies have examined the association of concomitant underlying illnesses with the severity of bronchiolitis. A multicenter prospective observational study showed that patients with congenital heart disease or underlying pulmonary disease had prolonged hospital stays and an increased risk of ICU admission and death.³ Analysis of a nationally representative sample found that most deaths from bronchiolitis were not among children with concomitant pulmonary or cardiac conditions, which may reflect their low frequency in the general population.² Our study also did not identify concomitant medical conditions as a predictor of unscheduled visits, but we did identify prior hospital admission as a predictor. The hospital admission variable is likely a more general proxy for children with severe underlying disorders. As with age younger than 2 months and male sex, children with bronchiolitis who had prior hospital admission may benefit from additional discharge counseling.

The children in this sample were deemed well enough for discharge to home at their initial ED visit. The most appropriate disposition for the children at high risk of an unscheduled visit is debatable. One could argue that if the child is going to return to the ED, it might be preferable to hospitalize the child at the initial visit. However, hospitalization of children with bronchiolitis is primarily used for close observation and supportive care for children at high risk of developing worsening respiratory difficulty or dehydration. Currently, hospitalization is not for therapeutic treatment, as most treatments have not been proven to be beneficial^18–21 or are only supported by small studies.²² Therefore, safely saving 1 or 2 days of hospitalization may be helpful for the family and should help reduce overall health care costs. We suggest providing strict anticipatory guidance instructions and close PCP follow-up for those children with bronchiolitis at high risk of post-ED unscheduled visits. Future research should determine if improvements in discharge counseling and scheduled PCP follow-up will decrease the rate of unscheduled visits in those children we found to be at high risk.

LIMITATIONS

The unscheduled visit rate of 17% in this study is lower than that of 34% found in a prior study on bronchiolitis;⁸ however, the duration of follow-up in the prior study was 28 days with a median time to relapse of 13 days. Thus, the 2-week duration of follow-up in our study may not have been long enough to identify all of the children who had unscheduled medical care visits. However, new data show that respiratory illnesses in this age group lasting longer than 2 weeks are likely to be due to new infection with a different virus strain rather than persistent infection from the same virus strain.²³ As a result of these data, our 2-week follow-up time frame should have been long enough for follow-up of the initial illness.

Second, no definite criteria for the diagnosis of bronchiolitis were specified. According to the AAP, bronchiolitis is a clinical diagnosis consisting of “rhinitis, tachypnea, wheezing, cough, crackles, use of accessory muscles, and/or nasal flaring.”¹⁴ The attending physician’s diagnosis of bronchiolitis satisfied the AAP’s definition of bronchiolitis 98% of the time, and although the current definition of bronchiolitis is admittedly broad and needs refining, attending emergency physicians from 30 hospitals, over 2 years, made the diagnosis of bronchiolitis based on their own clinical judgment. Therefore, from a clinician’s perspective, the children included in this study are exactly the children an ED attending would diagnose as having bronchiolitis.

Third, the primary definition of unscheduled visits was solely based on worsening of bronchiolitis that led the parent or guardian to take the child for an unscheduled visit to an ED or clinic within 2 weeks after initial ED discharge and did not require that the physician alter management or admit the child to the hospital. Although the reasons for a return are numerous (e.g., worsened disease, ongoing parental anxiety), the more restrictive definition of unscheduled visits that required the physician to alter management or admit the child to the hospital yielded similar results.

Fourth, the calculated ORs are modest. Nonetheless, these data may help increase the chances that an emergency physician would correctly identify those children with bronchiolitis who are more likely to return for an unscheduled visit. Finally, this study was performed in the ED of urban teaching hospitals, and may not be generalized to all patients with bronchiolitis.

CONCLUSIONS

We found three independent predictors of post-ED unscheduled visits by children with bronchiolitis who are discharged following their initial ED visit: age younger than 2 months, male sex, and previous hospitalization. Although further studies are required to confirm our results, these data may assist clinical decision-making in the ED, and may improve anticipatory guidance provided to parents and guardians of those children who are discharged home. Further studies are required to determine whether patients identified by these criteria who have close, scheduled follow-up with their primary care provider will have a lower rate of unscheduled visits.