Prospective Multicenter Study of Viral Etiology and Hospital Length of Stay in Children With Severe Bronchiolitis

Objective To determine whether hospital length of stay (LOS) for acute bronchiolitis is influenced by the infecting pathogen.

Design A prospective observational cohort study was performed during 3 consecutive years.

Setting Sixteen US hospitals participated in the study.

Participants Children younger than 2 years hospitalized with bronchiolitis were included.

Main Exposure The results of nasopharyngeal aspirate polymerase chain reaction pathogen testing served as the main exposure.

Main Outcome Measure Hospital LOS was determined.

Results Of 2207 participants, 72.0% had respiratory syncytial virus (RSV) and 25.6% had human rhinovirus (HRV); the incidence of each of the other viruses and bacteria was 7.8% or less. Multiple pathogen infections were present in 29.8% of the children. There were 1866 children (84.5%) with RSV and/or HRV. Among these 1866 children, the median age was 4 months and 59.5% were male. The median LOS was 2 days (interquartile range, 1-4 days). Compared with children who had only RSV, an LOS of 3 or more days was less likely among children with HRV alone (adjusted odds ratio [AOR], 0.36; 95% CI, 0.20-0.63; P < .001) and those with HRV plus non-RSV pathogens (AOR, 0.39; 95% CI, 0.23-0.66; P < .001) but more likely among children with RSV plus HRV (AOR, 1.33; 95% CI, 1.02-1.73; P = .04), controlling for 15 demographic and clinical factors.

Conclusions In this multicenter study of children hospitalized with bronchiolitis, RSV was the most common virus detected, but HRV was detected in one-quarter of the children. Since 1 in 3 children had multiple virus infections and HRV was associated with LOS, these data challenge the effectiveness of current RSV-based cohorting practices, the sporadic testing for HRV in bronchiolitis research, and current thinking that the infectious etiology of severe bronchiolitis does not affect short-term outcomes.

Bronchiolitis is one of the most common infectious respiratory conditions of early childhood¹ and the leading cause of hospitalization for infants.^1,2 The most common pathogen associated with severe bronchiolitis (ie, requiring hospitalization) is respiratory syncytial virus (RSV)³ and the second most common is human rhinovirus (HRV).^4,5 With the advent of molecular amplification techniques, however, it has become clear that a diverse group of pathogens is associated with severe bronchiolitis and that these pathogens may infect children in isolation or in combination as coinfections.^5-7

The clinical relevance of identifying the specific pathogen or combination of pathogens infecting a child with severe bronchiolitis remains unclear.^8-10 As a result, children with bronchiolitis, no matter the infecting pathogen, are considered to have essentially the same disease. Indeed, the 2006 American Academy of Pediatrics bronchiolitis clinical practice guideline⁸ recommends that clinicians limit viral diagnostic testing when caring for children with bronchiolitis. Children with HRV, however, may have different short-term and long-term outcomes than children with RSV.^5,11-14 Specifically, children with HRV bronchiolitis may have shorter acute clinical courses^5,11 and may be at increased risk of recurrent wheezing and asthma^12-14 compared with children with RSV bronchiolitis. To examine the clinical usefulness of identifying infectious etiology, we conducted a prospective, multicenter, multiyear study of more than 2000 children hospitalized with bronchiolitis. We hypothesized that children infected with HRV alone would have a shorter hospital length of stay (LOS) than would children infected with RSV alone.

We conducted a prospective, multicenter cohort study for 3 consecutive years during the 2007-2010 winter seasons as part of the Multicenter Airway Research Collaboration, a program of the Emergency Medicine Network (http:// www.emnet-usa.org). The number of participating sites varied during the 3 years: 13 sites in year 1, 16 sites in year 2, and 14 sites in year 3. Each month from November 1 until March 31, site investigators across 12 states used a standardized protocol to enroll a target number of consecutive patients from the inpatient units and the intensive care unit (ICU). Once the site reached its target enrollment for that month, the investigators would stop enrollment until the beginning of the next month.

All patients were treated at the discretion of the treating physician. Inclusion criteria were an attending physician's diagnosis of bronchiolitis, age younger than 2 years, and the ability of the parent/guardian to give informed consent. The exclusion criterion was previous enrollment. All consent and data forms were translated into Spanish. The institutional review board at each of the 16 participating hospitals approved the study.

The investigators conducted a structured interview that assessed patients' demographic characteristics, medical and environmental history, duration of symptoms, and details of the acute illness. Relevant comorbid medical disorders included respiratory, cardiac, neurologic, gastrointestinal, and immunologic diseases. Emergency department and daily hospital medical records provided further clinical data, including respiratory rates, daily respiratory rate trends, clinical assessment of degree of retractions (collapsed for analysis into none, mild, and moderate/severe), oxygen saturation, daily oxygen saturation trends, medical management, and disposition. These data were reviewed at the Emergency Medicine Network Coordinating Center, and site investigators were queried about missing data and discrepancies identified by manual data checks.

Nasopharyngeal aspirate collection was performed using a standardized protocol. Designated site personnel were trained using a lecture, written instructions, and video. All of the sites used the same collection equipment (Medline Industries) and collected the samples within 24 hours of a child's arrival to the medical unit or ICU. Once collected, the nasopharyngeal aspirate sample was added to transport medium, immediately placed on ice, and then stored at −80°C. Frozen samples were batch shipped on dry ice overnight to the central laboratory at Baylor College of Medicine, where they continued to be stored at −80°C.

All polymerase chain reaction (PCR) assays were conducted as singleplex or duplex 2-step real-time PCR. Real-time reverse transcriptase–PCR was used for the detection of RNA respiratory viruses, which included RSV types A and B; HRV; parainfluenza virus types 1, 2, and 3; influenza virus types A and B; the 2009 novel influenza A virus (H1N1); human metapneumovirus; coronaviruses NL-65, HKU1, OC43, and 229E; and enterovirus. Real-time PCR was used for the detection of DNA pathogens, which included adenovirus, Mycoplasma pneumoniae, and Bordetella pertussis. These tests are routinely conducted in the central laboratory of one of the investigators (P.A.P.), and details of the primers and probes have been described.^15-17

All analyses were performed using commercial software (Stata 11.2; StataCorp). Data are presented as proportions with 95% CIs and medians with interquartile ranges. Our primary analyses focused on RSV and HRV, the most commonly detected viruses in children with severe bronchiolitis. For the purposes of this analysis, we combined RSV-A with RSV-B because the clinical distinction between these subtypes of RSV was unremarkable. For analyses, we created a categorical variable that reflected the possible combinations of RSV/HRV status: (1) RSV only, (2) HRV only, (3) RSV in combination with HRV, (4) RSV in combination with non-HRV pathogens, and (5) HRV in combination with non-RSV pathogens.

We performed univariate analyses using the χ² test, Fisher exact test, and Kruskal-Wallis test, as appropriate. All P values were 2-tailed, with P < .05 considered statistically significant. Multivariable logistic regression analyses were conducted to evaluate independent predictors of a longer LOS (≥3 days; defined using the median value of 2 days) and other measures of severity: ICU admission and continuous positive airway pressure (CPAP)/intubation. Factors were selected for inclusion in the model if they were found to be associated with the outcome in unadjusted analyses (P < .20) or were potentially clinically significant. All regression models account for potential clustering by site. To further investigate independent predictors of LOS, a zero-truncated negative binomial model was also used to evaluate the relationship between demographic and clinical factors and LOS in days (continuous outcome). Children who were hospitalized for less than 1 day were assigned 0.5 days' LOS. Results of the zero-truncated negative binomial model are reported as incidence rate ratios (IRRs) with 95% CIs.

Of 3910 eligible children with severe bronchiolitis, 2207 children (56.4%) were enrolled. Enrolled and nonenrolled children were similar in age and sex (P > .05), but enrolled children were more likely to be white (61.4% vs 49.9%; P < .001) and Hispanic (36.3% vs 28.9%; P < .001). Enrolled children also were less likely to have an LOS of 3 or more days (44.4% vs 48.7%; P = .007).

Of the 2207 enrolled children, 1410 children (63.9%) had a single virus infection and 658 (29.8%) had 2 or more viruses; the remaining 139 children (6.3%) had no pathogen identified from our testing panel. Among all enrolled children, the pathogens detected were RSV-A (42.6%) and RSV-B (30.1%); HRV (25.6%); parainfluenza virus types 1, 2, and 3 (2.6%); influenza A, B, and the 2009 novel H1N1 (1.2%); human metapneumovirus (7.0%); coronaviruses NL-65, HKU1, OC43, and 229E (7.1%); enterovirus (4.9%); adenovirus (7.8%); M pneumoniae (0.9%); and B pertussis (0.2%). Children with RSV and/or HRV represented 1866 (84.5%) of the 2207 children. At least 1 other virus was detected in 32.3% of children who tested positive for RSV, 23.3% of those who had negative test results for RSV, 70.4% of children with HRV, and 15.9% of those without HRV.

Given the high frequency of RSV and HRV, we restricted this analysis to the 1866 children with RSV and/or HRV. Among these 1866 children, the median age was 4 months (interquartile range, 2-8 months), 59.5% were male, 62.0% were white, and 35.6% were Hispanic. The median LOS was 2 days (interquartile range, 1-4 days). Moreover, 1075 children (57.6%) had RSV only, 167 (8.9%) had HRV only, 287 (15.4%) had RSV plus HRV, 227 (12.2%) had RSV plus non-HRV pathogens, and 110 (5.9%) had HRV plus non-RSV pathogens. We examined the demographic and clinical characteristics according to these 5 groups (Table 1).

Unadjusted associations between various demographic and clinical characteristics and LOS (<3 days vs ≥3 days) are presented in Table 2. In general, younger children, white children, and those with gestational age less than 32 weeks were more likely to have a longer LOS. Furthermore, clinical factors, such as more severe retractions, lower oxygen saturation, apnea, inadequate oral intake, and ICU admission, were associated with a longer LOS. Two other unadjusted multivariable models were generated with ICU and CPAP/intubation as outcomes, but RSV/HRV status did not significantly predict either severity outcome (data not shown).

The multivariable logistic regression model for an LOS of 3 or more days is reported in Table 3. Controlling for 15 demographic and clinical characteristics as well as site, significant independent predictors for a longer LOS were age younger than 2 months, gestational age less than 32 weeks, presence of retractions, oxygen saturation less than 90%, ICU admission, and a viral etiology. Compared with children with RSV only, those with HRV only or HRV plus any other non-RSV pathogen were less likely to have a longer LOS (both P < .001), whereas children with RSV plus HRV infections were more likely to have an LOS of 3 or more days (P = .04). Even after restricting the analysis to the most common subset of children with bronchiolitis, those younger than 12 months and with a gestational age of 37 weeks or more, the results remained robust.

The zero-truncated negative binomial model, which examines LOS as a continuous outcome, showed similar results for the demographic and clinical factors presented in Table 3 with LOS as a dichotomous outcome (data not shown). This model also supported the virus findings. Compared with RSV only, the IRR for LOS was lower for children with HRV only (IRR, 0.73; 95% CI, 0.54-0.98; P = .04) and higher for children with RSV plus HRV (IRR, 1.18; 95% CI, 1.02-1.36; P = .03).

In this large, multicenter, multiyear prospective study of children hospitalized with bronchiolitis, we found that 29.8% of the children had multiple pathogen infections. The 2 most common viral causes were RSV (72.0%) and HRV (25.6%), and children with these viruses had different short-term outcomes. In comparison with children with RSV-only infections, multivariable models demonstrated that children infected with HRV alone or in combination with non-RSV viruses had a significantly shorter LOS. Children with RSV/HRV coinfections had a significantly longer LOS, even after adjusting for clinical and demographic factors associated with severity of illness. Therefore, on the basis of this large sample from across the United States, we submit that inpatient cohorting practices may not be as effective as once believed, that researchers consider testing for HRV more routinely in bronchiolitis studies, and that clinicians and researchers reconsider conventional wisdom that the infectious etiology of severe bronchiolitis does not affect short-term outcomes.

Although several studies have found that HRV lower respiratory tract infection in early childhood is associated with later wheezing^12,18,19 and asthma,^13,14,18 the short-term outcomes are less clear. Indeed, recent studies have shown that the clinical severity of HRV bronchiolitis is less than,⁵ no different than,⁴ and greater than²⁰ bronchiolitis due to RSV. Specifically, Marguet and colleagues⁵ performed a 4-center prospective study in France of 209 infants younger than 1 year with their first bronchiolitis hospitalization. In a multivariable analysis, they found that the 15 children with HRV-only infections had a reduced odds of staying in the hospital for 5 or more days (odds ratio, 0.13; 95% CI, 0.03-0.57) compared with children with RSV alone. However, a single-center study⁴ in Spain of 318 children younger than 2 years with severe bronchiolitis found no significant difference in LOS for the 24 children with HRV alone compared with RSV alone. A single-center prospective study²⁰ in Greece of 118 children younger than 18 months with severe bronchiolitis found that the presence of HRV (ie, alone or as coinfection) increased the odds of having a clinical severity score higher than the median (adjusted odds ratio, 4.9; 95% CI, 1.2-18.7). In the present multicenter study, we found that, on average, the 167 children with HRV-only infections had a shorter LOS than did children with RSV-only infections, even after controlling for 15 factors associated with severity of illness.^21,22

It may seem intuitive that children infected with more than 1 virus should have a more severe clinical course than children infected by only 1 virus, but the data on multiple pathogen infections are unclear.^23-26 Interestingly, we found that children with HRV in combination with viruses other than RSV had a shorter LOS. However, when HRV was paired with RSV, children with this specific coinfection had a longer LOS than did those with RSV alone. There are few data with which to compare our results, but Marguet and colleagues⁵ found that the 30 children with RSV/HRV coinfections had a reduced odds of staying in the hospital for 5 or more days (odds ratio, 0.26; 95% CI, 0.09-0.76) compared with children with RSV alone. A different and more acute measure of severity of illness is ICU admission or CPAP/intubation. Although Papadopoulos and colleagues²⁰ found that HRV increased an admission clinical severity score among 118 children with bronchiolitis, we did not find that the infectious etiology increased the odds of admission to the ICU or use of CPAP/intubation. Therefore, based on our data, children with RSV/HRV coinfections have a protracted severe illness but not necessarily a higher intensity of illness, as represented by the intensive care outcomes.

Although the pathophysiologic characteristics of the interactions between RSV and HRV are beyond the scope of this analysis, it is interesting that without RSV as a cofactor the clinical course of HRV parallels its more common, less severe, outpatient clinical course.^18,27 There are at least 2 plausible theories for the increased severity of illness of RSV/HRV coinfections. One possibility is that a diminished interferon γ response associated with RSV may allow for enhanced HRV replication^28,29; a similar pathogenesis occurs in airway epithelial cells from people with asthma.^30,31 Another possibility is that RSV-infected endothelial cells increase the cell surface expression of intercellular adhesion molecule 1,³² the major receptor for HRV,^33,34 setting the stage for a more severe HRV infection.³⁵

Of direct relevance to all hospitals that have 2 or more beds per room are the infection control issues raised by these data. The current point-of-care virology tests used to develop care plans for children with lower respiratory tract infections are influenza and RSV. If hospital coordinators group children with bronchiolitis together in a room with multiple beds, they usually do so by RSV status. However, given that 1 of 3 children with RSV and almost 1 of 4 without RSV will have a coinfection, the effectiveness of grouping children with RSV-positive or RSV-negative bronchiolitis together is questionable, especially given that some of the coinfecting pathogens require droplet precautions rather than just contact precautions. Some³⁶ have suggested routinely using PCR to test for multiple respiratory viruses in critically ill children with lower respiratory tract infections, but the expense of using molecular testing for all children with severe bronchiolitis may not outweigh the potential benefits for the family and clinicians. Although one possibility would be to limit the testing to RSV and HRV, the benefits of having a more complete picture of the infecting viruses—providing guidance about the potential severity of illness, possibly reducing antibiotic prescriptions,³⁷ and monitoring HRV-positive children closely for the development of asthma^12-14—most likely do not outweigh the expense of the molecular testing for HRV.

Testing for HRV in a clinical setting may not be practical currently, but we recommend that HRV testing become more common in bronchiolitis research. To date, no one has rigorously or effectively defined subgroups of children with severe bronchiolitis who may respond differently to medications and/or have different clinical outcomes. Our results suggest that categorization by infectious pathogen (ie, RSV and HRV) may be necessary to most accurately interpret the findings of randomized trials and other bronchiolitis research, especially when using LOS as an outcome.^38,39 Trials that combine all children with clinical bronchiolitis into one group or categorize children by RSV status alone may obfuscate real associations. Therefore, bronchiolitis investigators may be missing clinically meaningful results by not including HRV status in their analyses.¹⁹

The present study has potential limitations. Polymerase chain reaction detects low amounts of virus in children, and HRV in particular is detected in up to 24% of children younger than 1 year without fever or other respiratory symptoms.^40-43 Therefore, it is conceivable that the HRV we detected is a “bystander” virus⁴⁴ and that these HRV infections are asymptomatic.^40-43 Alternatively, we may be detecting a recent infection from which the children were recovering and not the causative agent related to the hospitalization. Although it remains possible that some of the children with HRV were asymptomatic or in recovery, on the whole, these data suggest that, in children with severe bronchiolitis, HRV plays a central role in the clinical course and is not asymptomatic. Another issue is that the study participants were hospitalized in academic medical centers. Consequently, these results are not necessarily generalizable to community medical centers or when the children are outpatients. Furthermore, bronchiolitis is a clinical diagnosis⁸ without a common international definition.^8,45 It is therefore possible that we included other respiratory disorders in this sample of children. However, when the data were restricted to resemble classic bronchiolitis, the results remained robust. Although the site teams enrolled 56% of the children and there were statistical differences in racial and ethnic groups, we do not believe that the level of enrollment or the statistical differences are clinically relevant or have provided biased results in this large multicenter study.

In summary, on the basis of these prospective, multicenter, multiyear data, we found that 1 in 3 children with severe bronchiolitis has a multiple virus infection, and we identified pathogen-based subgroups of children with different hospital LOS. Accordingly, we believe that these data raise questions about the effectiveness of RSV-based hospital cohorting practices. Moreover, severe bronchiolitis medication trials and other related research probably would benefit from inclusion of viral testing for both RSV and HRV so that lingering questions about differential effects by virus do not remain after the completion of otherwise rigorous trials. Most important, our data challenge current thinking that the infectious etiology of severe bronchiolitis does not affect short-term outcomes.

Correspondence: Jonathan M. Mansbach, MD, Department of Medicine, Children's Hospital Boston, 300 Longwood Ave, Main Clinical Bldg 9 S, Ste 9157, Boston, MA 02115 (jonathan.mansbach@childrens.harvard.edu).

Accepted for Publication: November 23, 2011.

Published Online: April 2, 2012. doi:10.1001/archpediatrics.2011.1669

Author Contributions: Drs Mansbach and Camargo had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Mansbach, Piedra, Teach, and Camargo. Acquisition of data: Mansbach, Piedra, Teach, Sullivan, Forgey, and Camargo. Analysis and interpretation of data: Mansbach, Piedra, Teach, Clark, Espinola, and Camargo. Drafting of the manuscript: Mansbach, Clark, and Camargo. Critical revision of the manuscript for important intellectual content: Mansbach, Piedra, Teach, Sullivan, Forgey, Clark, Espinola, and Camargo. Statistical analysis: Mansbach, Clark, Espinola, and Camargo. Obtained funding: Mansbach and Camargo. Administrative, technical, and material support: Sullivan, Forgey, and Camargo. Study supervision: Camargo.

Group Members: The MARC-30 site principal investigators were: Besh Barcega, MD, Loma Linda Medical Center, Loma Linda, California; John Cheng, MD, and Carlos Delgado, MD, Children's Healthcare of Atlanta, Atlanta, Georgia; Haitham Haddad, MD, Rainbow Babies & Children's Hospital, Cleveland, Ohio; Frank LoVecchio, MD, Maricopa Medical Center, Phoenix, Arizona; Charles G. Macias, MD, MPH, Texas Children's Hospital, Houston; Eugene Mowad, MD, Akron Children's Hospital, Akron, Ohio; Brian Pate, MD, Children's Mercy Hospital, Kansas City, Missouri; Mark Riederer, MD, and Paul Hain, MD, Children's Hospital at Vanderbilt, Nashville, Tennessee; M. Jason Sanders, MD, Children's Memorial Hermann Hospital, Houston; Alan Schroeder, MD, Santa Clara Valley Medical Center, Santa Clara, California; Nikhil Shah, MD, and Dorothy Damore, MD, New York Presbyterian Hospital–Cornell, New York, New York; Michelle Stevenson, MD, Kosair Children's Hospital, Louisville, Kentucky; Erin Stucky, MD, Rady Children's Hospital, San Diego, California; Stephen Teach, MD, MPH, Children's National Medical Center, Washington, DC; and Lisa Zaoutis, MD, Children's Hospital of Philadelphia, Philadelphia, Pennsylvania.

Funding/Support: This study was supported by grants U01 AI-67693, K23 AI-77801, and UL1 RR-031988 from the National Institutes of Health.

Disclaimer: The content of this manuscript is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Allergy and Infectious Diseases, National Center for Research Resources, or National Institutes of Health.