Chapter  69:  Management of Bronchiolitis in Infants and Children

Diagnosis

Specific literature regarding diagnosis of bronchiolitis was not found. The disease is clinically defined using well-accepted criteria. A large amount of data exists on the use of a variety of supportive laboratory tests such as specific respiratory syncytial virus (RSV) assays, complete blood counts (CBCs), and chest x-rays. However, only one study of 16 supported the clinical usefulness of such information. Thus, the existing data do not support the usefulness in testing to diagnose bronchiolitis.

The question of whether testing affects management and clinical outcome is more difficult to answer. Testing that can predict disease severity or worse clinical outcomes theoretically would be useful. One study suggests that testing may help identify patients likely to have more severe disease; however, five of the six predictors that emerged were based on history and physical examination (i.e., age, gestational age, general appearance, respiratory rate, and pulse oximetry).

Many clinicians are concerned that patients with more severe disease may have “bacterial superinfections.” This may result in the addition of antibiotics to a patient's treatment. Such concerns are typically based on illness severity, chest x-ray appearance, and an elevated white blood count. No data were found to support these assumptions.

Treatment

The authors reviewed the efficacy or effectiveness of several major classes of pharmaceutical agents that have been studied in multiple RCTs as interventions for bronchiolitis. These classes of agents included epinephrine, beta-2 agonist bronchodilators such as albuterol or salbutamol, ipratropium bromide, oral and inhaled corticosteroids, ribavirin, and antibiotics. In addition, they located several interventions for which limited, single-trial evidence existed, such as surfactant and nebulized furosemide. Treatments for bronchiolitis for which there was strong and convincing evidence of effectiveness were not identified. However, the investigators did find several interventions that they believe show some potential for being efficacious and should be subjected to rigorously designed, adequately sized trials. These include nebulized epinephrine, nebulized salbutamol plus ipratropium bromide, nebulized ipratropium bromide, oral or parenteral corticosteroids (preferably dexamethasone), and inhaled corticosteroids (preferably budesonide). Two interventions in this category are applicable only to the most severely ill children: inhaled helium-oxygen and surfactant for ventilated children. Given that there is no current best treatment for bronchiolitis, we would recommend that the above mentioned interventions should be studied in large, well-designed studies. In such studies, it is appropriate to use placebos in the comparison group when feasible; however, all subjects must be given standard supportive care.

This literature review also revealed several commonly used treatments for which the data are sufficient to reject, or at least doubt, their efficacy as treatments for bronchiolitis. These interventions are aerosolised ribavirin, antibiotics, nebulized furosemide, intravenous respiratory syncytial virus immunoglobulin (RSVIG IV) (as a treatment rather than as a prophylactic agent), inhaled alpha-interferon, and nebulized recombinant human deoxyribonuclease (rhDNase). Although the studies of these drugs were usually underpowered as well, because of lack of evidence of efficacy and a potential for increased harm with some, we recommend that clinicians not use these treatments routinely. These drugs should be considered for treatment only as part of rigorously designed, controlled trials.

This literature review found two treatments for which occurrence of adverse events in studies warrants caution in their use until such time as trials with adequate power to detect adverse events are conducted. These treatments are inhaled budesonide and alpha-2-interferon. This is particularly important in the case of inhaled budesonide because this agent also appeared to confer at least modest benefit for some outcomes in some studies of its use.

No evidence that any single agent can be recommended for treatment of bronchiolitis was identified. At present, evidence is insufficient to recommend any of the treatments studied over good supportive care of affected infants and children.

Prophylaxis

Although most children who have bronchiolitis do well and have an uncomplicated disease with a self-limited course, for some children it is a serious and sometimes life-threatening illness. For the most part, these severely affected infants and children have coexisting conditions that put them at increased risk of complications. One of the objectives of this review was to assess whether prophylactic therapy has a role for prevention of severe RSV bronchiolitis and in particular whether any subpopulations might realize greater benefit from prophylaxis. The largest group of at-risk children are those who are born prematurely, who often have concurrent chronic lung disease (CLD). Palivizumab or RSVIG IV given on a monthly basis is effective for prophylaxis in high-risk infants and children who have underlying CLD or have been born prematurely and are less than 6 months of age. Clinically, palivizumab has largely supplanted RSVIG IV because of the formerapos;s ease of administration, lower incidence of adverse events and increased efficacy.

None of the studies of immunization of at-risk infants with purified F protein (PFP) vaccines demonstrated benefit, although older children with cystic fibrosis in some studies did seem to obtain some benefit from a similar vaccine. However, these types of vaccines are at early stages of development and the studies were small. An effective vaccine would be a preferable strategy for prevention of RSV bronchiolitis in at-risk children compared to the passive immunity created by monthly injections of RSVIG. Because of the early nature of the research and the potential benefits, RSV vaccine research should be encouraged.

Costs of Prophylaxis

Findings from the published literature vary widely, depending on the cost of prophylactic therapy assumed, the hospitalization and other health care costs assumed, the baseline rate of hospitalization for children with RSV bronchiolitis, and reductions in hospitalization rates associated with the use of palivizumab. When all costs are adjusted to 2002 dollars, results from the previous studies suggest that prophylactic therapy for infants from 32 through 35 weeks of estimated gestational age (EGA) ranges from cost saving—meaning that the expected value of avoided health care utilization is greater than the costs of prophylactic therapy—to an upper bound of over $500,000. Given these variations, evidence is insufficient at the present time to calculate accurate expected incremental costs, or cost per hospitalization avoided, resulting from administration of a prophylaxis in infants who were born 32 through 35 weeks EGA or who are premature with comorbidities.

Priorities for Future Research

Diagnosis. Prospective trials of the utility of ancillary testing (chest x-rays, complete blood tests, respiratory syncytial virus [RSV] testing) are feasible and should be performed. Studies of diagnostic tools used in the management of bronchiolitis should measure clinical outcomes that are important to both parents and clinicians. An important intermediate outcome for studies of diagnosis in the management of bronchiolitis is the change in physician management.

Treatment. The following interventions should be studied with well-designed, rigorously conducted RCTs, preferably with placebo control: (a) nebulized epinephrine; (b) nebulized salbutamol plus ipratropium bromide; (c) nebulized ipratropium bromide; (d) oral corticosteroids, preferably dexamethasone; (e) inhaled budesonide; (f) inhaled helium-oxygen for severely ill children; (g) Chinese herbal therapy with Shuang Huang Lian (if its use can be practically accomplished in U.S. settings); and (h) surfactant for ventilated children. Studies of interventions should measure outcomes of primary interest to parents and clinicians, such as hospitalization, duration of hospitalization, need for more intensive care, and development of longer-term respiratory problems.

The treatment studies we reviewed were almost universally underpowered and as such do not give clinicians adequate guidance for management of bronchiolitis. There is substantial evidence that clinicians commonly use several interventions for which, currently, evidence is insufficient. These treatment interventions include inhaled bronchodilators, inhaled corticosteroids, and inhaled epinephrine. These drugs are all available as generic products and, therefore, relatively inexpensive; clinicians also consider them to be safe. We believe that clinicians will continue to use these types of treatments unless a large simple trial of these most common interventions is mounted. Such a trial would need to be large enough to examine each of the interventions not only in the overall population, but also in sub-populations of interest (e.g. infants with and without a history of atopy). This type of trial is unlikely to be funded by industry and would therefore require governmental support.

Prophylaxis. Use of prophylaxis in at-risk groups that were excluded from prior studies would need to be studied or reported before these agents can be recommended more broadly for other groups of infants and children at increased risk of more severe bronchiolitis. (At the time this report was written, findings from a study of prophylaxis with palivizumab including 1,287 children less than 2 years of age with congenital heart disease were expected to be reported at the American Academy of Pediatrics meeting on October 18, 2002. This study should give definitive evidence regarding prophylaxis for children with both cyanotic and acyanotic congenital heart disease.)

Studies of palivizumab prophylaxis should examine the effect on long-term outcomes such as the development of symptoms such as wheezing, development of bronchiolitis, hospitalization, and severe disease. The question of the relationship between bronchiolitis and asthma remains unanswered and is beyond the scope of this report. However, if the question is answered through a basic science study, and there is evidence of a causative relationship, this would have significant impacts on questions of prevention and the costs of prophylaxis.

RSV vaccine research should be encouraged as it would replace the need for prophylaxis.

Cost-effectiveness of Prophylaxis. Current cost-effectiveness analyses of palivizumab prophylaxis do not provide accurate incremental cost or cost-effectiveness ratios. Wide variations in available parameter estimates have resulted in wide ranges in reported incremental costs and costs per hospitalization avoided. Data on important parameters such as long-term health consequences, social costs, and the efficacy and safety of palivizumab on infants with comorbidities other than CLD were not available for previous analyses, but they may be available in the near future. The cost-effectiveness of palivizumab prophylaxis should be re-assessed as the new clinical trial data on palivizumab prophylaxis among infants in at-risk groups that were excluded from prior studies become available.

A new cost-effectiveness analysis should attempt to incorporate more social cost components and improved parameter values, and it should address as many subpopulations as possible by combining trial data on palivizumab safety and effectiveness from the IMpact-RSV and other new trials. Accurate social cost estimates for prophylaxis costs and hospitalization and outpatient utilization costs by cohort for each subgroup may influence cost-effectiveness ratios for each subpopulation. Prophylaxis cost estimates should reflect true costs to society, including identification of accurate palivizumab acquisition costs. As data become available, palivizumab's effects on long-term respiratory health should be addressed. Additional social costs would identify actual out-of-pocket expenses and productivity loss incurred by the family due to prophylaxis administration as well as RSV hospitalization and ambulatory care.

Accurate data on long-term consequences and family burden will help to integrate quality of life with costs in an economic evaluation. Current cost-effectiveness analyses report results in terms of incremental costs or cost per hospitalization avoided. Such measures do not fully quantify additional social burdens that RSV morbidity poses for infants and children and their families, and they do not provide guidance to policy-makers when faced with the decision of determining acceptable limits on cost-effectiveness.

General Guidelines for Future Research

Investigators should choose clinically relevant outcomes in future studies. Most of the outcomes studied in this literature are short-term and surrogate variables one measures such as oxygen saturation or respiratory rate at 15-minute intervals after treatment. Investigators should concentrate on measuring outcomes that are of interest to parents, clinicians, and health systems. Examples of these types of outcomes for intervention studies are rates of hospitalization, need for more intensive services in the hospital, costs of care, parental satisfaction with treatment and development of chronic asthma. An important intermediate outcome for studies of diagnosis in the management of bronchiolitis is the change in physician management.

Studies should be powered to detect meaningful differences in clinically relevant outcomes. Power calculations must include sufficient numbers to account for multiple comparisons if multiple outcomes are to be measured.

Few studies reported adverse events associated with treatments. This gap hampers any determination of whether the risks of particular treatments are sufficient to exclude their clinical use. Future investigations should carefully monitor and report adverse events associated with treatments.

Diagnosis

Diagnosis of bronchiolitis is based primarily on history and physical examination alone. Infants with fever, rhinitis, tachypnea and wheezing between November and May can be presumed to have bronchiolitis. Most bronchiolitis occurs in winter months. Because some types of parainfluenza virus are present in other months, bronchiolitis can be seen year round. Various laboratory studies can provide supportive data to the diagnosis, but none is highly sensitive or specific. Examples include chest x-ray and complete blood counts.

Specific testing can be done to determine the etiology of bronchiolitis (i.e., RSV vs. parainfluenza). Diagnostic methods include viral isolation, immunofluorescence, and enzyme-linked immunosorbent assays (ELISA) that detect antigen. Most clinicians use the RSV ELISA (e.g., a rapid test), which is performed on a specimen of nasal washing. These kits have sensitivities that range from 80 percent to 90 percent.⁶

The clinical utility of specific etiologic testing in cases of bronchiolitis is debatable. Such testing may be useful if other diagnoses are in the differential diagnosis (e.g., pneumonia or congestive heart failure) or if, in rare situations, treatment with ribavirin is being considered. In the vast majority of cases, however, determining that RSV is the cause of an individual case of bronchiolitis does little to change clinical course, management, or prognosis. In some institutions, evidence-based guidelines have been developed specifically to decrease the use of both RSV ELISA and supportive diagnostic testing.⁷

Treatment

Treatments for bronchiolitis can be categorized as specific and symptomatic. No specific therapy exists for parainfluenza virus. The only specific therapy for RSV is aerosolized ribavirin. Administration of ribavirin has been associated with improved oxygenation, improved clinical scores, and diminished levels of secretory mediators of inflammation associated with severe wheezing and disease. The use of ribavirin in certain infants at high risk of serious RSV disease was initially endorsed by the American Academy of Pediatrics (AAP) in 1993 based on initial carefully controlled clinical trials. However, the AAP modified the recommendation in 1996 from “should be used” to “may be considered” after several subsequent trials showed no significant effect on clinical outcomes. The use of ribavirin is further constrained by its high cost and possible risk to health care personnel who administer it.⁸

Among the popular symptomatic treatments are bronchodilators and corticosteroids. The widespread use of beta 2-agonist bronchodilators in bronchiolitis is likely explained by the similarity of symptoms and signs of bronchiolitis and asthma. However, the data to support their effectiveness in bronchiolitis are conflicting. Two systematic reviews have been published, the most recent one updated in 2001.^{9, 10} Kellner et al. examined 20 randomized controlled trials (RCTs) and found a statistically significant increase in the proportion of bronchodilator-treated infants demonstrating an improvement in their confidence interval [CI] 0.19 to 0.45).¹⁰ Bronchodilator recipients did not show improvement in measures of oxygenation with a difference favoring the control population (pooled difference 0.7; 95% CI 0.36 to 1.35). The rate of hospitalization was not significantly reduced in bronchodilator recipients compared with controls (odds ratio [OR] 0.7; 95% CI 0.36 to 1.35). Hospitalization duration was also not reduced in bronchodilator recipients (pooled difference 0.19 days; 95% CI -0.3 to 0.5).

Flores and Horwitz found no evidence that beta 2-agonists either improved oxygen by a clinically significant amount or reduced admission rates from outpatient and emergency department settings.⁹

Infants with bronchiolitis have been treated with corticosteroids because they are well-known anti-inflammatory agents acting at a multitude of cellular levels.¹¹ Clinicians have considered them for use in infants with acute bronchiolitis, partly because of the clear benefits of steroids in children with acute asthma. However, as with inhaled beta 2-agonists, data supporting the use of corticosteroids are conflicting. Clarification of potential benefit is of particular importance when the well-known adverse effects of corticosteroids are considered. Reported side effects from short-term administration include hypertension, hyperglycemia, hyponatremia, hypokalemic alkalosis, irritation and/or ulceration, and avascular necrosis in bones. However, serious side effects from short term administration over a few days such as might be used for bronchiolitis or an asthma exacerbation are rare.

Garrison et al. recently published a meta-analysis of six randomized trials performed with hospitalized infants.¹¹ Infants who received corticosteroids had a mean length of stay (LOS) or duration of symptoms (DOS) that was 0.43 days less than those who received the placebo treatment (95% CI: -0.81 to -0.05 days). The effect size for mean clinical score was -1.60 (95% CI: -1.92 to -1.28), favoring treatment. They concluded that the combined, published reports of the effect of systemic corticosteroids on the course of bronchiolitis suggest a statistically significant improvement in clinical symptoms, LOS, and DOS. Although the authors found a positive effect, they excluded several potentially relevant studies, and the clinical significance of an effect size of 1.6 is unclear. The 2000 Red Book states: “In previously healthy infants with RSV bronchiolitis, corticosteroids are not effective and are not indicated.”⁶

The AAP Committee on Infectious Diseases made recommendations about treatment for bronchiolitis in the 2000 Red Book.⁶ The group recommends supportive care as needed, including hydration, supplemental oxygen, and mechanical ventilation as the primary treatment modalities for bronchiolitis. Corticosteroids are judged to be ineffective and not indicated for previously healthy infants with RSV bronchiolitis. The committee states that antibiotics are rarely indicated as bacterial lung infection and bacteremia are uncommon in infants with bronchiolitis.

Prophylaxis for RSV infection with either RSVIG IV or palivizumab are recommended for infants and children younger than 2 years of age with chronic lung disease who have required treatment for chronic lung disease within 6 months prior to the anticipated RSV season. Palivizumab is the preferred agent for most children because of its ease of administration as an IM injection. Patients with more severe chronic lung disease may be considered for prophylaxis for two RSV seasons.

The recommendations also state that infants born at 32 weeks of gestation or earlier without chronic lung disease may benefit from prophylaxis with the primary considerations being gestational age and chronological age at the beginning of the RSV season. Infants born at 28 weeks of gestation or earlier may benefit from prophylaxis up to 12 months of chronological age while infants born at 29 to 32 weeks may benefit most up to 6 months of chronological age.

Until more data are available the AAP does not generally recommend these prophylactic agents for infants born between 32 and 35 weeks of gestation who do not have additional risk factors. Palivizumab and RSVIG IV are not currently licensed by the U.S. Food and Drug Administration (FDA) for patients with congenital heart disease. However, if the infant has chronic lung disease and/or was born prematurely and has asymptomatic acyanotic congenital heart disease then the Committee believes that such children may benefit from prophylaxis. The results of a large trial of prophylaxis in children with both cyanotic and acyanotic heart disease will be reported in mid-October 2002 and may change this recommendation. The AAP acknowledges that prophylaxis has not been evaluated in randomized trials in immunocompromised children, but notes that children with severe immunodeficiencies may benefit. In children who are receiving standard IGIV on a monthly basis for immunodeficiency, RSVIG IV can be substituted during RSV season.

Prophylactic Therapy

Respiratory syncytial virus immune globulin intravenous (RSVIG IV) was first licensed in 1996 for prevention of severe RSV disease in children. The AAP recommended use for younger than 24 months with chronic lung disease or a history of premature birth, given the higher burden of disease in this age group.¹² The AAP quickly endorsed its use. This therapy requires monthly intravenous infusions throughout the RSV season. In 1997 compelling data supporting an alternative therapy, palivizumab (an RSV monoclonal antibody administered intramuscularly) were published. The AAP issued new recommendations for palivizumab. The therapy is currently recommended for children younger than 24 months with chronic lung disease and infants born at 32 weeks gestation or earlier. It is not currently indicated in children with congenital heart disease, as evidence on its safety in this group of patients will only become available in late 2002. One systematic review of prophylactic immunoglobulin therapy concluded that it reduces admission to hospital and intensive care.

Cost of Prophylaxis

Although the effectiveness of prophylactic therapy is of critical importance in deciding whether it should be administered, cost is also an important factor.^{13, 14} The cost-effectiveness of RSV prophylaxis is very sensitive to the cost of the prophylaxis intervention and to the costs avoided as a result of the intervention. These costs are dominated by the acquisition cost of palivizumab and the cost of hospitalization, respectively.

Cost estimates used in published studies vary widely. Prophylaxis administration cost estimates used in previous analyses ranged from $2,754 to $4,957 per infant¹⁴ (updated to August 2002). Estimates can vary because of differences in acquisition and administration costs, the number and size of doses, and the amount of wasted palivizumab. Hospitalization costs average about $14,000 per infant but can vary widely, with studies reporting costs ranging from $11,336 to $118,336¹⁴ (updated to August 2002, adjusted to costs with cost/charge ratio of 0.6). Consequently, a summary of evidence from the literature on the cost-effectiveness of prophylactic therapy could prove valuable for deciding whether benefits are likely to outweigh costs.

Inclusion and Exclusion Criteria

Table 1. Management of Bronchiolitis in Infants and (more...)

Table 1. Management of Bronchiolitis in Infants and Children: Inclusion and Exclusion Criteria

Category	Criteria
Study population	Humans
	Infants and children
Study settings and geography	Inpatient, outpatient, home; all geographical locations subject to publication language and study design criteria
Time period	Systematic reviews, from 1966 through 2001
	Individual studies, published from 1980 through 2001
Publication languages	English only
Admissible evidence (study design and other criteria)	Original research studies that provide sufficient detail regarding methods and results to enable use and adjustment of the data and results
	For studies on diagnosis
	Randomized controlled trials (RCTs): double-blinded, single-blinded, and cross-over designs
	Non-RCTs: prospective cohort studies
	For studies on treatment and prophylaxis:
	RCTs: double-blinded, single-blinded, and cross-over designs
	For the cost-effectiveness component
	Studies that use an analytical method (e.g., cost, cost-effectiveness, cost-utility, or cost-benefit analysis)
	Patient populations must include infants and children
	Relevant outcomes must be able to be abstracted from data presented in the papers
	Sample sizes must be appropriate for the study question addressed in the paper; single case reports or small case series (fewer than 10 subjects) will be excluded

Based on consultation with the TEAG, the RTI-UNC team revised the specification of the patient population of interest for Key Questions 1, 2 and 3 from “infants and children ages 0–5” to “infants and children.” We made this revision because the age category 0 to 5 years did not reflect the fact that bronchiolitis is diagnosed primarily in children under 3 years of age. Also, the team wanted to be able to capture studies that looked at the long-term consequences of treatment of bronchiolitis in infancy or early childhood, even if those consequences were recorded at later ages. For Key Question 4 (cost-effectiveness of prophylaxis), the target populations for the cost-effectiveness question are (1) infants born 32 through 35 weeks' gestational age, and (2) infants born 32 through 40 weeks' gestational age with comorbid conditions.

The original geographic areas to which we intended to confine our literature searches and attention were North America, the United Kingdom, Australia, New Zealand, and Europe. Based on the recommendations of the TEAG, we removed this exclusion criterion for two reasons. First, some high-quality studies on this condition may well have been conducted elsewhere in the world, and we needed to be able to capture them. Second, including all areas may facilitate our examining information on different ethnicities and races in the report, as AHRQ and the professional societies had originally requested.

The criteria for study design were different for each key question based on the sufficiency and quality of evidence. Our diagnostic question (Key Question 1) was broad and required a lower admissibility standard. Therefore, we included both RCTs and prospective studies. The treatment and prophylaxis questions (Key Questions 2 and 3) were more specific and required greater strength of evidence; we thus elected to limit searches to RCTs. For the cost-effectiveness of prophylaxis (Key Question 4), we reviewed studies that employed economic analysis.

For all studies, key inclusion criteria included outcomes that were both clinically relevant and able to be abstracted. We set a minimum sample size of 10; small case series and single case reports were excluded. For Key Question 4 alone, we also excluded article abstracts that did not mention using an analytical method such as cost, cost-effectiveness, cost-utility, or cost-benefit analysis.

To ensure that we were reviewing therapies relevant to current clinical practice, we excluded individual studies before 1980. Our search was last updated on April 1, 2002, and contains all abstracts entered into the MEDLINE® and other databases until that date.

Search Terms

Table 2. Management of Bronchiolitis in Infants and (more...)

Table 2. Management of Bronchiolitis in Infants and Children: Search Terms

Topic	Search terms
Exploded terms for diagnosis	Bronchiolitis, diagnosis, differential diagnosis, thoracic radiography, laboratory techniques and procedures
Exploded terms for treatment	Steroidal anti-inflammatory agents, steroids, bronchodilator agents, antiviral agents, antimicrobial cationic peptides, antibiotics, antimicrobials, anti-infective agents
Exploded terms for prophylaxis	Primary prevention, immunoglobulins, bronchiolitis [prevention & control], isolation strategies, patient isolation
Exploded terms for cost-effectiveness	Costs and cost analysis
Study design for diagnosis	Prospective studies, longitudinal studies, cohort studies
Study design for treatment and prophylaxis	Randomized controlled trial, single-blind method, double-blind method, random allocation, meta-analysis
Outcomes for diagnosis	Fatal outcome, outcome and process assessment (health care), outcome assessment (health care), treatment outcome
Outcomes for treatment and prophylaxis	Morbidity, mortality, adverse effects or harms
Limiting terms for all	Human, year = 1980 through 2001, newborn infant (birth to 1 month) or infant (1 to 23 months) or preschool child (2 to 5 years)

Identification of Relevant Data Sources for Review

Table 3. Electronic Databases for Literature Searches (more...)

Table 3. Electronic Databases for Literature Searches

Database	Description
MEDLINE®	MEDLINE®, maintained by the National Library of Medicine (NLM), is the premier bibliographic database. Its 9.2 million records (with 31,000 new records added weekly) contain articles from more than 3,800 international biomedical journals (some chapters and articles from selected monographs are found in earlier years) covering the field of medicine, nursing dentistry, veterinary medicine, and the preclinical sciences. MEDLINE® contains citations for articles published in all languages from 1966 through the present; these citations are searchable, using NLM's controlled vocabulary, MeSH (Medical Subject Headings). For articles published in a foreign language, English abstracts are provided for 76 percent.
Cochrane Collaboration Resources	Among the Cochrane Collaboration resources are the Cochrane Controlled Trials Register, an electronic database providing reference information on RCTs and other controlled clinical trials in health care. Studies appearing in this database are part of systematic reviews conducted by various Cochrane Collaborative Review groups; all their studies have been reviewed for quality and additional information has been obtained from the original authors or through hard searches of the literature. The Cochrane Library houses the York Database of Abstracts of Reviews of Effectiveness (DARE), which provides access to structured abstracts of systematic reviews, American College of Physicians Journal Club abstracts, abstracts of reports prepared by the members of the International Network of Agencies for Health Technology Assessment, and other systematic reviews.
Health Economic Evaluations Database (HEED)	HEED provides structured summaries (reviews) of more than 20,000 articles appearing in the literature relevant to the economic assessment of health technologies. Each month electronic databases, leading journals and academic/government center publication lists are searched for relevant articles. An expert panel of academic reviewers complete the reviews of articles meeting specific inclusion criteria for cost analyses

Literature Assessment

For the clinical questions (Key Questions 1, 2 and 3), based on our initial search terms, we judged that 74 articles were possible inclusions, based upon full article review. The Scientific Directors independently evaluated each abstract for inclusion or exclusion, using the abstract review form (see Appendix B). When the Scientific Directors disagreed on an abstract, they reviewed it again together and came to a consensus. During the process of abstraction, we found that one article was a followup to a study not included by our search parameters. The original study was not classified under the MeSH term ‘bronchiolitis’. In order to capture any RCTs on bronchiolitic children that we may have missed, we conducted a systematic search titles and abstracts in MEDLINE® of the term ‘wheezing infants’ and identified 81 studies. After reviewing the abstracts, we included 10 articles for full review. Upon full review of the articles, we retained 4 articles in which the recruitment was conducted specifically during winter months and had children presenting with viral symptoms including wheezing. This suggests that the majority of the subjects had a clinical diagnosis of bronchiolitis rather than asthma.

For the cost-effectiveness of prophylaxis (Key Question 4), we identified 82 unique articles that mention the economic analysis of prophylaxis for the prevention of bronchiolitis in infants. Upon examination, we found that 21 article abstracts met our inclusion/exclusion criteria, and we obtained the full articles for review. We identified and ordered additional relevant articles based on a review of the reference lists from articles abstracted for any of the key questions. In all, we abstracted 41 articles for the cost-effectiveness questions.

In our review of the literature on prophylaxis and its costs for infants in the target populations, we identified published articles that describe two RCTs for RSVIG IV and one RCT for palivizumab; all met the inclusion criteria. We considered the possibility of pooling results from RCTs for RSVIG IV and palivizumab, but TEAG members discouraged this approach, citing that palivizumab exhibits higher efficacy, better safety, and ease of administration. TEAG members recommended that we consider only palivizumab in an evaluation of the cost-effectiveness of prophylaxis. However, because only one clinical trial has been conducted for palivizumab to date, data on outcomes for children in the intervention branch of the tree would necessarily be derived from this single study.¹⁰

Outcomes from the IMpact-RSV study are available for the following subpopulations of interest: (1) >32 weeks and = 35 weeks' gestational age and = 6 months at the time of randomization and (2) infants with a diagnosis of BPD and = 24 months upon randomization. The primary study outcome is rate of hospitalization. Although secondary endpoints included hospital length of stay, frequency and length of stay for intensive care unit (ICU), and mechanical ventilation, these results were not reported separately for the subpopulations of interest.

Our primary analysis focuses on the six articles that review the cost-effectiveness of palivizumab.

Abstractors and Trainers

The RTI-UNC EPC used both clinical and methods abstractors. All abstractors attended three training sessions. At the first session, we explained the process and goals of data abstraction, and then sent the abstractors home with an article to review. We then reconvened the group and, through a review of the test article, ensured that the abstractors understood what was expected of them. The reviewers abstracted an additional two test articles, reconvened, and reviewed their work. At this time the Scientific Directors determined that the abstractors were able to abstract the data as required, and we began the data abstraction process. The Research Coordinator monitored progress and routed the data abstractors' questions or issues to the Scientific Directors.

Data Abstraction Forms

For Key Questions 1, 2 and 3, the Study Director and the Scientific Directors created a single data abstraction form (Appendix C). This form was developed through multiple rounds of pretesting on different articles spanning the entire range of interventions to ensure that it would adequately capture all relevant issues. We solicited feedback from the data abstractors during training to refine further the data abstraction form.

For Key Question 4, we used a systematic approach to review and abstract economic data.¹⁵ We first developed and used a standardized abstraction form to identify information from each article about the study design, analytic perspective used, cost components included in the analysis, and value of the economic summary measure (e.g., cost, cost-effectiveness ratio, or cost-utility ratio). This form is an adaptation of the Economic Evaluation Abstraction Form (version 3.0) developed and used to evaluate economic studies for the Guide to Community Preventive Services. ¹⁶

We made adjustments to summary measures from the abstracted articles to facilitate comparisons across study findings. For example, to account for cost differences across studies attributable solely to price inflation, we used the medical care price index (MCPI) to adjust all estimated costs to constant 2001 dollars. The MCPI is a subset of the Consumer Price Index compiled by the Bureau of Labor Statistics; it includes medical care items such as prescription drugs and medical supplies, physicians' services, eyeglasses/eye care, and hospital services.

We also focused our comparisons on specific components of cost, such as treatment or hospital costs, rather than on total cost measures, because different studies may have included different resources in their total cost estimates. In some cases, we could not adjust study results because of differences in methods. For example, if costs were not presented separately for each component included in the study, we could not make adjustments to total cost estimates for comparability.

For articles that did not indicate the year for which costs were reported, we assumed that the costs were valued in constant dollars for the year prior to publication.

Development of Evidence Tables

After abstracting the included articles, we developed evidence tables to present the essential information to address Key Questions 2 and 3 relating to treatment and prophylaxis. These tables appear in at the end of this report and cover the following pieces of information:

• Setting of the intervention: country, patient setting;

• Followup: acute (48 hours after intervention), short-term (2–14 days after intervention) and long-term (14 days and more);

• Research design: randomized trials, including placebo-controlled, nonplacebo-controlled (both those comparing active treatment and control groups to nonplacebo), and crossover trials;

• Length of enrollment;

• Masking;

• Objective of the study;

• Inclusion/exclusion criteria;

• Number enrolled in and completed study;

• Sex;

• Mean age at enrollment and mean gestational age;

• Comorbidities;

• Interventions;

• Results and significance tests for primary and secondary outcomes and subgroup analysis;

• Adverse events;

• Quality; and

• Significant differences at baseline and other comments.

Given the wide range of reported outcomes, we assigned results in evidence tables as primary or secondary outcomes based on their clinical relevance to the key questions. In studies with multiple outcomes, we generally listed the more clinically important outcomes such as length of hospitalization or development of long-term sequellae as primary outcomes and the more physiologic measurements such as heart rate or respiratory rate as secondary outcomes. Applying this rule, however, depended on the nature of results presented in the study. When the authors presented pulmonary function tests as their primary outcomes and did not present data on length of hospitalization or development of long-term sequellae, the Scientific Directors may have chosen physiologic measurements as the more clinically relevant outcome from that study and placed them as the primary outcome for the purposes of the evidence table.

For primary outcomes, individual results for each study arm and P values were always recorded where possible. For secondary outcomes, P values were generally reported when results were positive.

Grading the Strength of Evidence

For Key Question 1 on diagnosis, we initially intended to assign quality scores to the diagnostic studies using standard criteria.¹⁷ However, several factors prevented this. First, no articles specifically assessed diagnostic tests or criteria for bronchiolitis. The literature, the TEAG, and our study team all agreed that bronchiolitis is a clinical diagnosis for which no true or “gold standard” test exists.

Second, the majority of diagnostic information extracted for review came from the 61 treatment studies. As such, the data had not been collected for the purposes of assessing their diagnostic utility. In most studies, viral studies, clinical scores, complete blood counts (CBCs), and chest x-rays were all used as baseline independent variables.

We did use selection criteria that ensured a minimal study validity. We took diagnostic data only from the RCTs in which all patients were tested (i.e., rather than at the discretion of the investigators or treating physicians). Of the non-RCT articles identified that included diagnostic data, all were prospective cohort studies.

For Key Questions 2 and 3, the Scientific Directors developed a quality assessment form for RCTs of treatment or prophylaxis (Appendix D). In prior work for AHRQ, the RTI-UNC EPC had developed an exhaustively peer-reviewed evidence report on systems to rate the strength of scientific evidence.¹⁸ We based our quality assessment tool on this work, with appropriate modifications for the literature on the management of bronchiolitis in infants and children. The quality assessment tool comprised four individual elements: randomization; masking, statistical analysis, and funding/sponsorship.

We rated each element as excellent, adequate, inadequate, or unable to determine. In addition to these four elements, we considered the appropriateness of the population studied, the clarity and relevance of outcome used, and the appropriateness of the statistical analysis used. Based on the composite of the assigned scores and their individual comments on each study, the Scientific Directors assigned an overall quality score on a four point scale, ranging from 1 (poor) to 4 (excellent). The subjectivity involved in this method of scoring the quality of evidence was reduced by the independent assessment of each article by both Scientific Directors. When they did not agree, they reviewed the article together and arrived at a consensus. Of the 61 articles that were scored for quality for Key Questions 2 and 3, the Scientific Directors had a 98 percent rate of agreement within one point.

For Key Question 4 on the cost-effectiveness of prophylaxis, we adopted the Guide to Community Preventive Services convention for the Centers for Disease Control and Prevention (CDC) of not scoring economic studies on quality. As Carande-Kulis et al. explain, differences in economic methods may be attributable to differences in study objectives; even when differences result from variations in quality, they may not have a large impact on study findings.¹⁵ Further, the number of economic studies available for review is quite limited, so we adopted the CDC approach of reviewing all available studies but adjusting results to account for differences in methods.

Key Question 1: Diagnosis

The TEAG agreed that we should retain this question because of its theoretical importance. TEAG members generally agreed that patients with bronchiolitis undergo many tests but that few influence clinical management or outcome; they do affect the costs of care. We identified 16 studies dealing with diagnosis. We also reviewed 61 clinical trials for additional data on diagnostic testing. The data available fell into several natural categories:

• Case definitions and inclusion criteria used in the clinical trials;

• Etiology of cases of bronchiolitis when all subjects were tested;

• Comparison of various virus isolation techniques;

• Predictors of disease severity or complications; and

• Studies in which standardized tests were performed on all patients as part of their evaluation (e.g., chest x-rays, complete blood counts).

Key Questions 2 and 3: Treatment and Prophylactic Therapy

Our assessment of the literature for Key Questions 2 and 3 suggested a range of interventions, with studies choosing to report a widely varying set of outcomes measured at different time intervals. Given the disparity in outcomes, we grouped studies by type of intervention rather than by outcomes for Chapter 3 (Results). This grouping resulted in 15 sets of interventions and evidence tables:

• Nebulized epinephrine versus nebulized saline placebo;

• Subcutaneous epinephrine versus saline placebo;

• Nebulized epinephrine versus nebulized bronchodilators (salbutamol or albuterol);

• Nebulized bronchodilators (salbutamol or albuterol) versus placebo or other treatments;

• Nebulized bronchodilators (salbutamol or albuterol) plus ipratropium bromide versus either nebulized salbutamol or nebulized albuterol alone and/or placebo;

• Oral corticosteroids versus placebo, with or without bronchodilators;

• Parenteral dexamethasone versus placebo;

• Nebulized corticosteroids versus placebo or usual care;

• Ribavirin versus placebo;

• Antibiotics versus no treatment or other antibiotics;

• RSVIG IV as treatment for bronchiolitis

• Other miscellaneous treatments for bronchiolitis;

• RSVIG IV versus placebo or standard care to prevent RSV bronchiolitis;

• Monoclonal antibody for prophylaxis of RSV bronchiolitis; and

• Vaccines to prevent RSV bronchiolitis.

Key Question 4: Cost-Effectiveness of Prophylactic Therapy

To determine whether we could assess the cost-effectiveness of prophylactic therapy using existing data, we used a decision analysis framework to describe the treatment options and the possible costs and outcomes that could result. Based on our initial findings from the literature and on input from the TEAG, we developed a decision tree to show treatment alternatives—administer prophylactic therapy versus no treatment intervention—and the possible outcomes associated with each alternative.

Figure 3. Cost Effectiveness Decision Tree

   Figure 3. Cost Effectiveness Decision Tree

Figure 3

Figure 3

Figure 3

In Chapter 3 we summarize results from existing economic analyses of prophylactic therapy for the prevention of bronchiolitis. However, as we discuss in some detail in Chapter 5 , the existing literature contains many gaps, and much of the data required for a cost-effectiveness analysis from the societal perspective are not available. Our discussion section in Chapter 4 summarizes these data gaps and offers recommendations about additional data needed to answer the question of whether prophylactic therapy is cost-effective when used in the target population.

Peer Review Process

We requested review of this report from several individual experts in the field and from relevant professional societies and public organizations. They are acknowledged at the beginning of this report. We revised the report in response to suggestions from these outside agents.

Case Definition and Inclusion Criteria

We reviewed the case definition and inclusion criteria from the clinical trials. Case definitions were quite similar: (a) 38 used tachypnea in either the case definition or inclusion criteria; (b) 39 used wheezing; (c) 30 used oxygen saturation; and (d) 28 used retractions. However, many studies simply stated that infants with signs and symptoms consistent with bronchiolitis were cases eligible for inclusion. Many authors referred to the historical definition of bronchiolitis published by Court.¹⁹

Eligibility criteria in the clinical trials varied to a greater extent, especially with respect to variables such as age, duration of symptoms, comorbidities (e.g., prematurity, chronic lung disease), history of previous wheezing, and severity of disease. This variation was determined by the specific objectives of the studies (e.g., numerous studies included only infants who were positive for RSV disease).

Most trials measured disease severity both as a baseline independent variable and as a dependent outcome (i.e., change in disease severity resulting from treatment). Disease severity was most commonly measured using clinical scales (43 of the 52 treatment trials). The variety of scales used made comparisons between studies difficult. Appendix A describes the numerous clinical scales used.

Some studies used clinical scales that had been validated in previous studies such as the Respiratory Distress Assessment Instrument (RDAI).^20–23 Others were created or modified by authors for their particular trial.^{24, 25} Despite this variation, the clinical scales all incorporated measures of respiratory rate, respiratory effort, severity of wheezing, and oxygenation.

Identification of Etiology of Bronchiolitis

Many, but not all, of the included studies attempted to identify the etiology of the enrolled cases. As mentioned above, a subset of the treatment trials enrolled only infants who were RSV positive.

Of the 52 treatment studies, 42 performed RSV testing on all subjects. In the studies that tested all and included all regardless of RSV status, the range in the prevalence of cases caused by RSV was 26 percent to 95 percent. Twelve studies tested patients for other viral etiologies (e.g., parainfluenza viruses) in addition to RSV. It is recognized that RSV testing of patients with bronchiolitis is justified in several situations. First, isolation of RSV as the etiology of fever in an infant under 3 months may support a clinician's decision to forego additional testing in the traditional “rule out sepsis” work-up. Second, RSV testing may be helpful in clinical situations where the diagnosis of bronchiolitis is not clear. Third, RSV testing will be essential in research settings where RSV-specific therapies are being evaluated for effectiveness. Finally, RSV testing is an important tool to epidemiologists and public health officials responsible for surveillance of lower respiratory tract infections in infants. However, most reported results as percentage positive for RSV versus “other viruses.”

Various techniques for identifying RSV as the causative agent of bronchiolitis were used, including viral cultures, rapid antigen detection tests (e.g., direct immunofluorescence assay [IFA], enzyme immuno-assays [EIA]), polymerase chain reaction (PCR), and measurements of acute and convalescent antibody titers. Rapid antigen detection tests for RSV were used most frequently. In many of these, viral cultures were performed on cases that were negative for RSV.

Comparison of Virologic Tests

Table 4. Studies Examining the Accuracy of Virologic (more...)

Table 4. Studies Examining the Accuracy of Virologic Tests

Author	“Gold Standard”a	Tests comparedb	Resultsc
Ahluwalia et al., 198726	Viral culture of NPA and NPS	EIA, IFA on both NPA and NPS specimens	EIA-NPA:
			Sn = 69, Sp =100
			EIA-NPS:
			Sn = 61, Sp =100
			IFA-NPA:
			Sn = 61, Sp = 89
			IFA-NS:
			Sn = 52, Sp = 78
Chattopadhya et al., 199227	Viral culture	IFA, EIA, EIA by blocking test	IFA:
			Sn = 89, Sp = 92
			EIA: Sn = 94, Sp = 74 EIA
			by blocking test :
			Sn = 94, Sp = 77
Eugene-Ruellan et al., 199828	Viral culture and/or IFA	PCR	97% “concordance”
Ong et al., 200129	IFA	PCR	IFA detected 27 cases
			PCR detected 28 cases
Waner et al., 199030	Viral culture and/or IFA	EIA	Sn = 86, Sp = 91

a

NPA, nasopharyngeal aspirate; NPS, nasopharyngeal suction; IFA, direct immunofluorescence assay.

b

EIA, enzyme immunoassays; PCR, polymerase chain reaction.

c

Sn, sensitivity; Sp, specificity.

Of interest from both the clinical and utilization points of view is the question of whether RSV testing is necessary in all patients with bronchiolitis. Although such testing is commonly used to document the etiology of bronchiolitis, the etiology rarely changes clinical management. Many institutions require testing all infants being admitted to the hospital; the rational involves assisting with identifying cohorts (i.e., to decrease nosocomial RSV infections). However, no good quality RCTs examine the effects of cohort segregation in preventing nosocomial transmission of bronchiolitis.³¹ As a result, many infection control policies recommend that all infants with acute lower respiratory infection (ALRI) be isolated, regardless of etiology. No study we reviewed addressed the issue of utility of RSV testing.

Predictors of Severe Disease or Complications

Table 5. Studies Measuring Predictors Of Disease Severity (more...)

Table 5. Studies Measuring Predictors Of Disease Severity

Author	Outcome predicteda	Indicators examinedb	Predictors
Cherian et al., 199734	Diagnosis of ALRI (based on abnormal findings on auscultation or CRX) using respiratory rate and subcostal retractions in undernourished infants and children with respiratory infections in India	• RR ≥60/min in infants < 2 months of age	The sensitivity and specificity of tachypnea, subcostal retractions or the presence of either sign in detecting ALRI did not differ among children in different nutritional categories.
		• RR ≥50/min in infants 2–12 months of age
		• RR ≥40 in children >12 months of age
		• Presence of subcostal retractions
Dawson et al., 199036	Clinical score (mild, moderate, severe or very severe)	CXR findings (i.e., hyperinflation, atelectasis and infiltrates)	There was no correlation between CXR findings and disease severity
Mulholland et al., 199033	Severity at the time of admission as assessed by oximetry and arterial blood gas results	• Demographics	Indicators of severity at time of admission:
		• Cyanosis	• young age
		• Crackles	• cyanosis
		• Chest wall indrawing	• crackles
	Oxygen requirements during admission	• RR >50/min
		• HR > 150/min	Predictors of oxygen requirement during admission:
		• Liver >2cm below costal margin	• young age
		• SaO2 <90%	• cyanosis
		• PaO2 <60 mm Hg	• crackles
		• PaCO2 >45 mm Hg	• high RR
		• RSV status	• chest wall indrawing
			• SaO2 <90%
			• PaCO2 >45 mm Hg
			• PaO2 <60 mm Hg
Shaw et al., 199132	“Mild disease” (defined as alert, active and able to take fluids throughout their disease, no O2 therapy, etc) vs. “Severe disease” (defined as all others without mild disease)	Historical information	Six independent clinical and laboratory findings were strongly associated with more severe disease using multiple-factor analysis
		• Cyanosis or apnea	• “III” or “toxic” appearance
		• Gestational age < 34 or 37 wks	• Oxygen saturation < 95%
		• Age < 3 mo	• Gestational age < 34 wks
		• Decreased po intake	• RR ≥ 70/min
		• Perinatal complications	• Age < 3 months
		• URI symptoms <3 days
		Physical examination and observations
		• “III” or “toxic” appearance
		• Yale Observation Scale ≥ 10
		• Accessory muscle use
		• Clinical Asthma Score > 2
		• RR > 60/min or > 70/min
		• Rales
		Laboratory
		• Pulse oximetry quiet
		• Pulse oximetry while sucking
		• CXR findings of atelectasis or hyperaeration
		• Isolation of RSV
Saijo et al., 199635	Finding of lobar pneumonia vs. bronchopneumonia vs. bronchiolitis in hospitalized infants with RSV ALRI	• WBC > 15,000/mm3	The percentages of all 4 indicators were higher in patients with RSV lobar pneumonia vs. bronchiolitis or bronchopneumonia
		• Neutrophil count > 10,000/mm3
		• ESR > 30 mm/hr
		• CRP > 3.0 mg/dL

a

ALRI, acute lower respiratory infection; CXR, chest radiograph.

b

RR, respiratory rate; min, minutes; HR, heart rate; WBC, white blood count; ESR, erythyrocyte sedimentation rate; CRP, C-reactive protein; Sa0₂, oxygen saturation; PaCO₂, arterial carbon dioxide pressure; RSV, respiratory syncytial virus; po,oral; URI, upper respiratory infection.

In contrast, Cherian et al. focused on determining the reliability of easily observed physical findings in diagnosing ALRI in developing countries, as used in current World Health Organization (WHO) algorithms.³⁴ The Saijo et al. study focused on using laboratory studies to predict three categories of RSV disease defined radiographically rather than clinically.³⁵ As such, these findings have limited usefulness to clinicians.

Most textbooks cite young age, history of prematurity or other comorbidities, toxic appearance at presentation, and rapid progression of symptoms as risk factors for severe disease. Two studies support these assertions.^{32, 33} Additional prospective studies of disease severity or clinical prediction models are lacking.

Utility of Chest Radiographs in Bronchiolitis

Table 6. Use of Chest X-Rays for Diagnosis of Bronchiolitis (more...)

Table 6. Use of Chest X-Rays for Diagnosis of Bronchiolitis

Author	Purpose of Study	Use of Chest X-ray	Results
Can et al., 199824	RCT of salbutamol vs. mist	Baseline assessment	CXR findings “consistent with bronchiolitis” were present in 88% 69% and 73% of the infants in the three study groups
Daugbjerg et al., 199372	RCT of nebulized tertbutaline, nebulized corticosteroid and systemic corticosteroid	Baseline assessment	Infiltrates in 27% overall, no differences between 4 study groups
Dawson et al., 199036	Cohort design specifically to look at the utility of routine CXR's in bronchiolitis by examining the relationship between clinical assessment (i.e., mild, moderate, severe or very severe) and CXR findings (i.e., hyperinflation, atelectasis and infiltrates)	Baseline assessment	No correlation between CXR findings and disease severity
Dobson et al., 199837	RCT of albuterol in hospitalized infants	Baseline assessment	CXR results not reported
Friis et al., 199038	Prospective cohort of children < 7 years old designed to correlate CXR findings with viral and bacterial studies.	Baseline assessment	The results for children with bronchiolitis not reported separately, so no conclusions can be drawn
Luchetti et al., 199839	RCT of porcine-derived surfactant in ventilated infants	Baseline assessment and to document clinical improvement	CXR results not reported
Nasr et al., 200140	RCT of rhDNase in hospitalized infants	Baseline assessment and at study end or time of hospital discharge	CXR improvement was a trial outcome (CXR scores improved in the treatment group but not in the control group) CXR findings were not used to assess disease severity or determine management
Rodriguez et al., 199725	RCT of RSVIG treatment of hospitalized young children at high risk for severe disease	Baseline assessment Repeated at time of hospital discharge	CXR results not reported
Rodriguez et al.,199741	RCT of RSVIG treatment of previously healthy hospitalized children	Baseline assessment Repeated at time of hospital discharge	CXR results not reported
Rodriguez et al., 198742	RCT of ribavirin in infants with RSV disease (included patients with bronchiolitis, pneumonia, and croup)	Baseline assessment	CXR results for infants with bronchiolitis not reported separately
Roosevelt et al., 199643	RCT of dexamethasone in acute bronchiolitis	Baseline assessment	• No data presented correlating CXR findings to disease severity
			• Infiltrates seen in 32% of treatment group and 20% of placebo group
			• 90% of infants with visible infiltrates were treated with antibiotics vs. 44% of those without these findings
Schuh et al., 199044	RCT of albuterol in ED	Baseline assessment	CXR results not reported
Shaw et a., 199132	Prospective cohort of 228 infants designed to “Mild disease” (defined as alert, active and able to take fluids throughout their disease, no O2 therapy, etc.) vs. “Severe disease” (defined as all others without mild disease)	Baseline assessment	Overall
			• 58% had hyperaeration
			• 9% had atelectasis
			Findings in patients with “Severe” vs. “Mild” disease
			• Atelectasis: 21% vs. 2% (RR = 2.7, CI 1.97–3.70)
			• Hyperaeration: 69% vs. 52% (RR = 1.58, CI 1.03–2.42)
Taber et al., 198345	RCT of ribavirin in hospitalized infants	Baseline assessment	• Hyperinflation: 24/26
			• Peribronchial thickening: 25/26

CXR, chest x-ray; RCT, randomized controlled trials; RSVIVIG, respiratory syncytial virus intravenous immunoglobulin, ED, emergency department; RR, relative risk; CI, confidence interval.

Two studies set out to examine the relationship between x-ray abnormalities and disease severity. Shaw et al.'s data show that the patients with atelectasis were 2.7 times more likely (95% CI: 1.97–3.70) to have severe disease than those without this x-ray finding.³² This association persisted when it was included in a multivariable analysis. In contrast, Dawson's data demonstrated no correlation between chest x-ray findings and baseline disease severity as measured by a clinical severity scoring system.³⁶

The Roosevelt et al. study showed that the presence of chest x-rays abnormalities was strongly correlated with the use of antibiotics.⁴³ The effectiveness of antibiotic treatment in these patients was not examined. The fact that bronchiolitis is usually a viral illness calls in to question this course of disease management.

These data suggest that in mild disease, chest x-rays offer no information that is likely to affect treatment and that, therefore, they should not be routinely performed. In fact, the Roosevelt et al. data suggest that such x-rays may lead to inappropriate use of antibiotics, although this was not the focus of their study.⁴³ Chest x-rays may be useful in predicting which patients are likely to have more severe disease in cases in which this assessment is not otherwise clear.

Utility of Complete Blood Counts in Bronchiolitis

Table 7. Studies Examining Complete Blood Counts Performed (more...)

Table 7. Studies Examining Complete Blood Counts Performed

Author	Purpose of study	Use of CBC in study	Results
Barry et al., 198646	RCT of ribavirin in acute bronchiolitis	Baseline assessment	CBC results not reported
		Completion of study
Can et al., 199824	RCT of salbutamol vs. mist	Baseline assessment	Mean WBC, neutrophils, eosinophils, Hgb and HCT similar in three study groups
Chipps et al., 199347	RCT of alpha-2A-interferon in hospitalized infants	Baseline assessment	CBC results not reported
		Day 5 of study
De Boeck et al., 199748	RCT of dexamethacone hospitalized infants	Baseline assessment	No difference in leukocyte count and eosinophilia between treatment groups
Friis et al., 198449	RCT of antibiotics in treatment of pneumonia and bronchiolitis	Baseline assessment	CBC results for bronchiolitis vs. pneumonia not compared
Kjolhede et al., 199550	RCT of vitamin A in ALRI	Baseline assessment	CBC results not reported
Kong et al.,199351	RCT of Chinese herbs in hospitalized infants	Baseline assessment	CBC results not reported
Rodriguez et al., 198742	RCT of ribavirin in infants with RSV disease (included patients with bronchiolitis, oneumonia, and croup)	Baseline assessment	No differences between treatment groups
Saijo et al, 199635	Finding of lobar pneumonia vs. bronchopneumonia vs. bronchiolitis in hospitalized infants with RSV ALRI	• WBC > 15,000/mm3	The percentages of all 4 indicators was higher in patients with RSV lobar pneumonia vs. bronchiolitis or bronchopneumonia
		• Neutrophil count > 10,000/mm3
		• ESR > 30 mm/hr
		• CRP >3.0 mg/dL
Taber et al., 198345	RCT of ribavirin in hospitalized infants	Baseline assessment	No differences between treatment groups
		Time of discharge	No differences from admission to discharge to followup
		Followup

CBC, complete blood count; RCT, randomized controlled trial; WBC, whole blood count; Hgb, hemoglobin; HCT, hematocrit; RSV, respiratory syncytial virus; ALRI, acute lower respiratory infection; ESR, erythyrocyte sedimentation rate; CRP, C-reactive protein.

Nebulized Epinephrine versus Nebulized Saline Placebo

Subcutaneous Epinephrine versus Saline Placebo

Nebulized Epinephrine versus Nebulized Bronchodilators (Salbutamol or Albuterol)

Nebulized Bronchodilators (Salbutamol or Albuterol) versus Oral Bronchodilators, Nebulized Ipratropium Bromide, or Saline Placebo or No Treatment

Nebulized Salbutamol or Albuterol plus Nebulized Ipratropium Bromide versus Bronchodilators or Ipratropium Bromide Alone and/or Saline Placebo

These studies included between 62 and 102 participants, but two studies divided subjects among four groups, resulting in small group sizes.^{63, 65} Chowdhury et al.'s study excluded significant numbers of children after randomization (13 of 102) primarily for subsequent findings of lung consolidation. All three studies included children up to 2 years of age.

Primary outcomes included duration of hospitalization, respiratory rate, and clinical score. Primary outcomes did not differ significantly for any of the treatment groups. Wang et al. demonstrated a statistically significant improvement in the mean change in oxygen saturation. A secondary outcome, considering salbutamol plus ipratropium bromide versus salbutamol alone and ipratropium bromide alone, showed no difference when compared to the placebo group.⁶⁵ Schuh et al. did not report any benefit of nebulized ipratropium bromide in addition to nebulized albuterol for vital signs, oxygen saturation, or clinical scores.⁶⁴

Chowdhury et al. did not report any adverse events;⁶³ Wang et al. noted that one infant in the salbutamol group was withdrawn for tremulousness.⁶⁵ As expected with use of these agents, Schuh et al. found a heart rate increase with use of albuterol.⁶⁴

Oral Corticosteriods versus Placebo, With or Without Bronchodilators

All studies compared oral corticosteroids (i.e., prednisolone, prednisone, or dexamethasone) to placebo. Except for Van Woensel et al., all employed bronchodilators as a cointervention in all arms of the study.^{23, 66, 69, 70} Val Woensel et al. allowed the use of bronchodilators as needed and reported no difference in use between study groups.^{67, 68} The studies by Goebel et al., Berger et al. and Schuh et al. used albuterol (Berger et al. allowed the use of either oral or nebulized albuterol); the Klassen et al. study used salbutamol as the bronchodilator of choice. The studies were small (51 and 72 subjects). Most of the studies enrolled children up to 2 years of age, although Berger et al. admitted infants up to 18 months and Klassen et al. included infants up to 15 months of age. Van Woensel et al. admitted infants with severe disease and comorbidities, including those on ventilators and with BPD. All studies used some type of symptom score as an outcome. Other primary outcomes included hospitalization, readmission, persistent symptoms, and need for other treatments. Adverse events, largely unreported, were limited to the expected side effects of bronchodilator use.

These research teams found few differences between study groups overall. Goebel et al. reported a statistically significant difference in clinical scores between days 0 and 2; the group that received both prednisolone and albuterol improved more than the placebo and albuterol groups.⁶⁶ Berger et al. demonstrated no difference in clinical scores, respiratory rate, or oxygen saturation between the prednisone and placebo groups.⁷⁰ They also were able to contact approximately three-quarters of the parents at 2 years after the initial study; for this group, they determined that infants who had received oral prednisone experienced more respiratory symptoms (35.7 percent in the prednisone group versus 28.6 percent in the control group, P-value not reported). Overall, about one-third of the followup population had persistent respiratory symptoms at 2 years. Schuh et al. found significantly lower rates of hospitalization (19 percent vs 44 percent), improved clinical scores at 240 minutes post-treatment, and less need for corticosteriods after discharge in the dexamethasone plus nebulized albuterol group compared with the placebo plus albuterol group.²³

The initial Van Woensel et al. study found a significantly greater mean decline in symptom score among the 39 nonventilated patients and a shorter duration of hospitalization among the 14 ventilated patients.⁶⁸ Five-year followup did not demonstrate any significant differences in long-term outcomes such as wheezing in the first year of life or persistent or late-onset wheezing.⁶⁷

Parenteral Dexamethasone versus Placebo

Nebulized Corticosteroids versus Placebo or Usual Care

Ribavirin versus Placebo

Antibiotics versus No Treatment or Other Antibiotics

RSVIG IV as Treatment for Bronchiolitis

Other Miscellaneous Treatments for Bronchiolitis

RSVIG IV versus Placebo or Standard Care to Prevent RSV Bronchiolitis

Groothuis and colleagues studied 249 children less than 48 months of age with BPD due to prematurity, congenital heart disease or cardiomyopathy, or a history of prematurity along with a chronological age of less than 6 months.⁸⁷ These children, who were all at high risk for RSV infections, were randomized to either high-dose (750 mg/kg IV every month) or low-dose (150 mg/kg IV every month) RSVIG or a standard care and control group. Primary outcomes included total and moderate-to-severe episodes of RSV and non-RSV respiratory illness. They found both significantly fewer total cases RSV-related lower respiratory infections and fewer severe cases in the higher-dose RSVIG IV group compared to the standard-care group. The low-dose group and the control group did not differ significantly on primary outcomes. Differences between the high- and low-dose groups were not reported. In secondary outcomes they also reported significantly fewer hospitalizations, hospital days, and ICU days for the high-dose group compared to the standard-care group. Eight-five percent of the 249 enrollees were followed into the subsequent RSV season and there was no suggestion of enhanced disease in either the high or low dose groups who were hospitalized for RSV infections. Enhanced disease had been a concern in early RSV vaccine trials such that these investigators were asked to specifically look for this adverse effect.

Groothuis et al. also published a subgroup analysis of the 162 premature infant from this study, excluding the children with congenital heart disease.⁸⁶ There were 102 preterm children with BPD and the remaining 60 had no evidence of lung disease. The analysis was further restricted to a comparison of the high-dose (n = 58) and control groups (n = 58) as the original analysis had not demonstrated efficacy of the low-dose therapy. Subjects were followed monthly during the 5 months of the intervention and then into the subsequent RSV season. Primary outcomes for this analysis included total incidence of RSV illnesses, incidence of severe RSV illness, hospitalizations for RSV infections, mean duration of ICU admission, and mean worst respiratory score. There were statistically significant differences favoring the high-dose group over the control group with the exception of the mean difference in duration of hospitalization which achieved borderline significance (P = 0.06). This study had potential problems with the masking of study personnel because an unblinded team was responsible for enrollment, examinations at the time of infusion, and well-infant examinations. A masked team was responsible for weekly followup and evaluation of all respiratory illnesses. Follow up of all of the preterm children into their second RSV season did not demonstrate any enhanced RSV illness upon infection with RSV.

Simoes et al. studied a group of 425 children under 48 months of age with congenital heart disease or cardiomyopathy; they randomized subjects to 750 mg/kg IV RSVIG every month during RSV season or to a control group that received no intervention.⁸⁸ As with the Groothuis et al. studies, the Simoes et al. team responsible for enrollment, treatment, and clinical assessment was not masked, whereas the team responsible for weekly surveillance and clinical evaluation of respiratory illnesses was masked. The primary outcomes were total acute respiratory illnesses, total upper and lower RSV-associated respiratory illnesses, and both RSV-associated and nonassociated lower respiratory tract illness hospitalizations.

The investigators reported significantly fewer acute respiratory illnesses (73 percent vs. 82 percent, P = 0.02) and total hospitalizations for lower respiratory tract illnesses (17 percent vs. 27 percent, P = 0.02) in the RSVIG group compared to the no-treatment group. In subgroup analysis they found fewer RSV hospitalizations in the treatment group under 6 months of age. They found no significant overall differences for RSV hospitalization by cardiac subgroup, but when they removed the group of children with biventricular heart disease with right-to-left shunt from the analysis, they detected a trend toward a decrease among infants with all other types of heart disease (biventricular without shunts, biventricular with left-to-right shunt, and single ventricle or hypoplastic left heart) included in the study (11 percent vs. 27 percent, P = 0.06.) A randomization imbalance resulted in more children with left-to-right cardiac shunt in the control group and more with right-to-left shunt in the treatment group. A significantly increased rate of serious adverse events related to cardiac surgery and increased rate of cyanotic spells was observed in children with cyanotic congenital heart disease receiving RSVIG IV and were thought due to receipt of the RSVIG IV treatment

The PREVENT Study Group conducted a multicenter trial involving 510 high-risk infants less than 2 years of age with BPD or who were premature (= 35 weeks) and under 6 months of age at the time of enrollment.⁸⁹ The intervention group received 750 mg/kg IV RSVIG monthly during RSV season; the control group received albumin placebo. Several significant positive differences between groups occurred, including fewer RSV-related hospitalizations (8 percent vs. 13.5 percent, P = 0.047), fewer total number of RSV-related hospital days (60 vs. 129, P = 0.045) and days in hospital requiring oxygen therapy per 100 children (34 vs. 85, P = 0.007). The RSVIG IV treatment group also experienced fewer hospital days with severe clinical scores per 100 children (49 vs. 106, P = 0.049), incidence of total respiratory hospitalizations (16 percent vs. 27 percent, P = 0.005) and total number of respiratory hospital days per 100 children (170 vs. 317, P = 0.005). In a set of subgroup analyses for prematurity, presence of BPD, age less than 6 months at trial entry, and weight under 4.3 kg, trends emerged toward fewer hospitalizations in all subgroups receiving RSVIG IV, but statistical testing was not performed for these exploratory secondary analyses. The paper does not mention statistical correction for multiple comparisons. When infusions were incomplete or prolonged because of an adverse event judged potentially related to the study drug, the problem occurred more often in the group receiving RSVIG IV (3.2 percent vs. 1 percent).

Monoclonal Antibody for Prophylaxis of RSV Bronchiolitis

The IMpact-RSV Study Group studied 1,502 high-risk infants who were premature (= 35 weeks) and under 6 months of age or were 24 months of age and younger with symptomatic BPD.⁹¹ Children were randomized in a two-to-one ratio to either palivizumab 15 mg/kg IM or placebo every 30 days for up to 5 month. The primary outcome was incidence of RSV hospitalizations. In the placebo group, 53 of 500 children (10.6 percent) were hospitalized for RSV infection, compared to 48 of 1,002 children (4.8 percent) in the palivizumab group (P < 0.001.) The majority of secondary outcomes showed statistically significant benefits of the treatment as well. Among these secondary outcomes were total numbers of hospitalizations and hospital days per 100 children (62.6 vs. 36.4 days, P < 0.001), total days of RSV hospitalizations requiring oxygen therapy per 100 children (50.6 vs. 30.3 days, P < 0.001), hospital days with a severe clinical score per 100 children (47.4 vs. 29.6 days, P < 0.001), and incidence of ICU care (3 percent vs. 1.3 percent, P = 0.026). The differences observed in secondary outcomes are attributable to decreased RSV incidence and severity in the palivizumab group as the incidence of respiratory hospitalization unrelated to RSV was similar between the groups (14 percent vs. 13 percent, P = 0.505). Subgroup analyses examined the incidence of RSV hospitalization by weight, prematurity without BPD and BPD alone. All of these subgroup analyses showed a significant benefit of palivizumab. Adverse events, including development of fever, nervousness/irritability, injection site reaction, and diarrhea were not significantly different between the treatment and control groups. The overall rate of reported adverse events judged to be related to the study drug was 10 percent in the placebo group and 11 percent in the palivizumab group.

Meissner and colleagues conducted a trial to evaluate the safety, pharmokinetics and immungenicity of SB 209763, a humanized monoclonal antibody against RSV fusion protein.⁹² The study population consisted of 43 infants with BPD or without BPD who had been born prematurely at less than or equal to 35 weeks of gestation. Infants were randomized to receive two doses of the antibody 8 weeks apart, at one of four dosage levels ranging from 0.25 to 10.0 mg/kg per dose at each administration. The so-called “placebo” group was actually a group of infants who received placebo at the first administration and then were crossed over to receive a dose of SB 209763 at the dosage level that had been assigned in their randomization scheme 8 weeks later. The 5.0 and 10.0 mg/kg doses of both SB 209763 and placebo were split into two syringes and administered one into each thigh. However, there was no attempt made to completely blind the administration of lower dose levels by giving two injections as well. There was a trend toward fewer episodes of proven RSV infection in the group that received the 10.0 mg/kg dose of SB 209763 vaccine compared to placebo (1 of 22 vs. 2 of 10, P = 0.20) this difference did not reach statistical significance. There was a lower rate of proven RSV infection at the three other dose levels as well, but the P-values ranged from 0.72 at the 0.25 mg/kg dose to 0.49 at the 5.0 mg/kg dose level. Four adverse events judged related to the study drug were identified and included three episodes of mild/moderate purpura and one episode of thrombocytosis. The authors suggested that the doses used might have been too low to confer adequate clinical immunity and that future trials test higher doses of monoclonal antibody.

Vaccines to Prevent RSV Bronchiolitis

Groothuis and colleagues randomized 21 infants under 12 months of age with BPD. All infants had a proven RSV infection in the previous RSV season. These infants had previously had influenza vaccination in the previous year and were then randomized to vaccination with PFP-2 vaccine or trivalent influenza vaccine in the subsequent year.⁹³ Their primary outcome was RSV infection in the subsequent season. One of 10 in the treatment group and six of 11 in the control group had subsequent season RSV infections. This result was borderline statistical significance with a P value of 0.06. Some of the immunological secondary outcomes, including such items as mean neutralizing antibody 1 and 6 months after vaccination, were found to be statistically higher in the group that received PFP-2 compared to the placebo group. This is obviously a small study lacking sufficient power to detect even large differences between groups.

Piedra and colleagues reported the results of two studies using PFP-2 vaccine in children at high risk from RSV infection because of underlying cystic fibrosis. The first study of 34 children randomized groups to PFP-2 or saline placebo.⁹⁴ There were baseline group imbalances with the PFP group being taller, older and with lower body fat composition. There were no differences demonstrated in the development of RSV or total days of RSV illness between groups. However, there were significantly more children with one or more than one acute lower respiratory tract infection (15 of 17 vs. 9 of 17, P = 0.024) and with more ill days per subject (67 vs.30.5, P < 0.001) in the control group compared with the vaccine group. The vaccine group had fewer antibiotic courses (4.5 vs.2.2, P < 0.001) and fewer acute lower respiratory tract infections per subject (2.1 vs.0.8, P = 0.005) than did the control group. There were no significant differences in adverse events between the groups, although the vaccine group did report more cases of tenderness at the vaccine site (P = 0.09).

A second study by Piedra was conducted to evaluate the effectiveness of sequential yearly administration of PFP-2 versus a single administration in children with underlying cystic fibrosis.⁹⁵ A group of 29 or the 34 children who had participated in the previous study of PFP-2 vaccine discussed above were recruited into this study of sequential annual administration of vaccine. They were enrolled in this open label study to PFP-2 vaccine and all enrollees received a 50 microgram dose of the vaccine in the second season. Thus there were two groups, one which received vaccine each autumn for two seasons or saline placebo in the first year followed by PFP-2 vaccine in the second season. The sequential vaccine group which received active vaccine in both seasons had fewer children with more than one acute lower respiratory tract infection during the second season (9 of 13 vs.15 of 15, P = 0.035.) The sequential vaccine group was also found to have fewer acute lower respiratory tract infections per subject (1.2 vs. 2.1, P = 0.004) and ill days per subject (36 vs. 64.8, P = 0.001) compared with the group that only received the active vaccine in the second season. There were no significant differences in total number of illnesses per subject or mean number of courses of antibiotics per subject. Although only a total of 11 children had confirmed RSV infections in the second season, the sequential vaccine group of RSV infected children did have significantly fewer episodes of acute lower respiratory tract infections, days of illness and courses of antibiotics per subject. There were baseline differences between the two groups with the control group being taller, older and more likely to attend day care. Given the nature of cystic fibrosis disease and day care exposures, these baseline differences could have accounted for the outcome differences seen between the two groups. Adverse events and their distribution was comparable to those which were seen in the first Piedra study.

Summary of Findings from the Literature on the Cost-effectiveness of Palivizumab Prophylaxis

As mentioned previously, six studies have assessed the cost or cost-effectiveness of palivizumab in preventing RSV bronchiolitis. For each of these studies, we provide a brief description, present key findings, and discuss limitations.

General Findings Across Economic Analyses

The CEAs summarized in the previous subsection varied greatly in the approaches used, estimates of key parameters, and findings. Although the Panel on Cost-Effectiveness in Health and Medicine has recommended that a societal perspective be used for economic evaluations of clinical interventions, only Joffe et al. attempted to incorporate a societal perspective; the other studies adopted a payers' or providers' perspective.¹³ Three factors had a large impact on cost-effectiveness results from all of the studies: hospitalization incidence, healthcare costs, and the costs of palvizumab therapy. In this subsection, we discuss differences identified in these factors across studies and how these differences are likely to affect the cost-effectiveness of palivizumab.

Incidence of Hospitalization

Table 9. RSV Hospitalization Rates

Table 9. RSV Hospitalization Rates

Source	Entry Criteria	Year	Baseline Hospitalization		Palivizumab Hospitalization
Very Premature
Sorrentino and Powers 2000101	<28 EGA	1998-99			15/445	3.4%
Cunningham et al., 1991102	=32 EGA	1985-86	19/130	14.6%
Impact RSV 199891	=32 EGA	1996-97	41/372	11.0%	43/739	5.8%
Paul et al., 2002103	=32 EGA	1999-00			0/175	0.0%
Joffe et al., 199913	23 to 32 EGA, >=28 O2, discharge Sept-Nov	1992-96		24.6%
Joffe et al., 199913	23 to 32 EGA, >=28 O2, discharge Dec-Aug	1992-96		10.7%
Joffe et al., 199913	23 to 32 EGA, <28 O2, discharge Sept-Nov	1992-96		8.0%
Joffe et al., 199913	23 to 32 EGA, <28 O2, discharge Dec-Aug	1992-96		3.1%
Sorrentino and Powers 2000101	28 to 33 EGA	1998-99			12/611	2.0%
Average RRR	69%			11.4%		3.6%
Less Premature
Groothuis et al., 199586	=35 EGA	1989-91	13/58	22.4%
The PREVENT Study Group, 199789	=35 EGA and/or BPD	1994-95	35/260	13.5%
Oelberg	=35 EGA	1994-96	53/378	14.0%
Impact RSV 199891	=35 EGA	1996-97	53/500	10.6%	48/1002	4.8%
Farina 2002100	=35 EGA or BPD	1998-99	10/42	23.8%
Impact RSV 199891	32 to 36 EGA	1996-97	12/128	9.4%	5/263	1.9%
Sorrentino and Powers 2000101	32 to 36 EGA	1998-99			8/548	1.5%
Joffe et al., 199913	33 to 36 EGA, >=28 O2, discharge Sept-Nov	1992-96		11.0%
Joffe et al., 199913	33 to 36 EGA, >=28 O2, discharge Dec-Aug	1992-96		4.4%
Joffe et al., 199913	33 to 36 EGA, <28 O2, discharge Sept-Nov	1992-96		3.2%
Joffe et al., 199913	33 to 36 EGA, <28 O2, discharge Dec-Aug	1992-96		1.2%
Schrand et al., 200198	28 to 37 EGA	1994-95	10/40	25.0%	1/61	1.6%
Sorrentino and Powers 2000101	>35 EGA	1998-99			0/26	0.0%
Average RRR	58%			7.7%		3.3%
Diagnosed with BPD
The PREVENT Study Group89	BPD	1994-95	26/149	17.4%
Groothuis et al., 198899	BPD	1985-86	11/30	36.7%
Impact RSV 199891	BPD	1996-97	34/266	12.8%	39/496	7.9%
Sorrentino and Powers 2000101	BPD	1998-99			42/1839	2.3%
Average RRR	78%			16.0%		3.5%
Average RRR for All Groups	66%			10.1%		3.4%

RSV, respiratory syncytial virus; EGA, estimated gestational age; BPD, bronchopulmonary dysplasia; RRR, relative risk reduction.

Estimates of the incidence of hospitalization for RSV bronchiolitis vary widely, and these differences can have a considerable impact on the estimated cost and cost-effectiveness of prophylactic therapy. Table 9 shows some of the RSV hospitalization rates found in the literature. Note that baseline hospitalization rates for infants from about 32 through 35 weeks EGA vary from 1.2 to 25 percent.

One possible reason for the limited evidence on hospitalization rates is because of the difficulty of obtaining consistent diagnoses of RSV bronchiolitis across hospital settings. Bronchiolitis is generally a clinical diagnosis, and therefore hospitalization incidence rates based on a diagnosis of bronchiolitis may under- or over-attribute RSV as the infectious agent. For studies that used universal antigen testing to determine the presence of RSV, variations in the epidemiology and prevalence of RSV by geographic or socioeconomic group as well as variations in virulence and subspecies, can greatly affect findings.

Cost of Health Care Resource Utilization

Table 10. Cost Estimates used in Analyses

Table 10. Cost Estimates used in Analyses

	Cost of Prophylaxis	Cost of Hospitalization	Value of Parents' Missed Work
			Hospitalization	Prophylaxis
Joffe et al. 199913	$3,648	$11,336	$466	$57
Lofland et al., 200014	$2,754 or $4,957	$11,551	-	-
Schrand et al., 200198	$3,968	$19,525	-	-
Marchetti et al., 199996 (charges)	-	$22,773	-	-
Average	$3,457 or $4,191	$14,019	$466	$57

Estimates are in August 2002 dollars.

Hospital and other medical resource costs can vary by severity of illness, geographical area, and institution. The source of cost information can also change the value of the estimate. Charges overstate costs to society, as most payers pay significantly less. Cost to charge ratios to convert charges can be calculated from Medicare data and indicate that costs are typically less than 60 percent as high as charges, but use of cost to charge ratios for non-Medicare hospitalizations introduces even more uncertainty into the actual costs. The values obtained from hospital cost accounting systems are likely to be the most accurate measures of cost available, although they best reflect medical costs for a particular geographical region and may not reflect any profit. In the economic analyses of palivizumab described in the previous subsection, hospitalization and other medical care cost estimates varied widely. These estimates, adjusted to 2001 dollars using the MCPI, are shown in Table 10. Diagnosis codes designating RSV hospitalization are now available, and may facilitate estimation of more accurate cost values.

Cost of Palivizumab Therapy

Table 11. 2001 Cost Per Hospitalization Avoided

Table 11. 2001 Cost Per Hospitalization Avoided

CEA	32–35 Estimated Gestational Age Population		Bronchopulmonary Dysplasia Population
Marchetti et al., 199996a	Savings (minimum bound)	$33,566 (maximum bound)	Savings (minimum bound)	$50,999 (maximum bound)
Lofland et al., 200014b	$26,439 (therapy-$2,500)	$59,487 (therapy=$4,500)	$43,614 (therapy=$2,500)	$87,805 (therapy=$4,500)
Schrand et al., 200198	Savings (internal ratesa)	$33,565 (Impact-RSV ratesc)	Savings (internal ratesa)	$61,138 (Impact-RSV ratesc)
Joffe et al., 1999104d	$117,265 (discharge Sept-Nov)	$328,343 (discharge Dec-Aug)	$29,707 (discharge Sept-Nov)	$85,995 (discharge Dec-Aug)
Computed using average hospitalization costs from CEAs, average incidence rates from Table 9	$54,500		$19,540

a

Converted to cost per hospital avoided with IMpact-RSV hospitalization rates for each subpopulation.

b

Cost per hospital avoided based on approximating similar rate reduction as seen in IMpact-RSV for each subpopulation.

c

IMpact-RSV rates calculated by substituting subpopulation rates from IMpact-RSV for the overall rates used by Schrand et al.

d

32–36 weeks EGA with less than 28 days oxygen approximating 32–35 week cohort, use of oxygen for = 28 days used as approximation of CLD cohort.

The single largest barrier to wide-scale use of palivizumab is its cost. Palivizumab cost estimates from the literature are shown in Table 11. Most of the analyses used the average wholesale price (AWP), or a catalog price, as an estimate of the cost of palivizumab. However, AWPs are not calculated from actual sales; they are essentially suggested wholesale prices and may not accurately reflect actual costs. Wholesalers sometimes use the AWP as a list price in catalogs and then negotiate discounts with customers. Physician practices and insurance companies, especially those that use group purchasing organizations and pharmacy benefits managers, may be able to obtain palivizumab at a much lower unit cost than the published AWP. Indeed, certain Federal agencies (Department of Defense [DoD], Veterans Affairs [VA], and Health and Human Services [HHS], and the Coast Guard) are able to purchase palivizumab for 48 percent less than the published AWP.

Another component of the cost of palivizumab that varies across economic analyses is the number of doses required for a successful prophylaxis program. Palivizumab is recommended to be taken monthly during the 5 months of RSV season, but infants born during RSV season may take less than the full five doses. Schrand et al. reported that all infants in their treatment group received all required doses, but that the average number of doses per infant was 3.28.⁹⁸ Analyses that used an estimate of five, or nearly five, doses may overstate the costs for full administration.

Cost of Prophylaxis

Six articles have considered the cost-effectiveness of prophylactic therapy for preterm or other high-risk infants.^{13, 14, 96–98, 100} Findings from these studies suggest that the cost per hospitalization avoided varies widely, depending on the cost of prophylactic therapy assumed, the hospitalization and other health care costs assumed, the baseline rate of hospitalization for children with RSV bronchiolitis, and reductions in hospitalization rates associated with the use of palivizumab. When all costs are adjusted to 2002 dollars, results from the previous studies suggest that prophylactic therapy for infants from 32 through 35 weeks of EGA ranges from cost saving (e.g., Marchetti et al.⁹⁶)—meaning that the expected costs associated with the outcomes along the treatment intervention branch of the decision tree are lower than the costs of no prophylactic therapy—to an upper bound of $328,000 (e.g., Joffe et al.).¹⁴ Typical results indicated costs per hospitalization avoided of about $40,000 to $50,000, but given the wide variation in results, current analyses do not provide a reliable estimate of the cost-effectiveness of RSV prophylaxis.

Previous analyses were limited in several respects. First, all but one of these studies used effectiveness data from the only large RCT to date for palivizumab—the IMpact-RSV study.⁹¹ This trial did not report statistically significant, secondary end-points for subpopulations. The IMpact trial did not include any comorbidities other than BPD. The results from a trial on infants with cardiac disease are not yet available. The study that used alternative data on the impact of prophylactic therapy on hospitalization rates for RSV bronchiolitis had only 40 infants in the control group and 61 infants in the treatment group; hence, the quality of their results is seriously limited by the small number of study observations.⁹⁸

Second, most economic analyses of palivizumab have focused on estimating costs from the payer or provider perspective, rather than from the societal perspective, which is the approach recommended by the Panel on Cost-Effectiveness in Health and Medicine.¹⁰⁸ Consequently, most of these studies have excluded important costs that may result from a child's infection with bronchiolitis, such as parents' lost time from work, the family's nonreimbursable travel or parking expenses, and the productivity losses associated with the premature death or chronic morbidity of the affected infant (if, for example, bronchiolitis has long-term negative outcomes, such as asthma). Although Joffe et al. included parents' productivity losses associated with a provider visit for obtaining palivizumab and a child's hospitalization for RSV bronchiolitis infection, estimated losses were based on ad hoc assumptions about the amount of parental time required for outpatient visits to obtain palivizumab therapy and parental time spent with a hospitalized infant.¹³

Third, the baseline (no prophylactic therapy) rate of hospitalization in infants with bronchiolitis is unknown and may vary depending on the characteristics of the patient population, region of residence and method of measurement. Estimates used in the literature on the cost-effectiveness of prophylactic therapy range from 1.2 percent to 25 percent for infants born prior to 36 weeks EGA. Such widely varying estimates of baseline hospitalization rates have the most significant impact on the results of cost-effective analyses.

Finally, the literature contained a broad range of estimated costs for palivizumab and hospitalization for RSV bronchiolitis. Differences in acquisition costs for palivizumab, administration costs, and the number of doses lead to differences across studies in the estimated cost of prophylaxis. The estimated costs of hospitalization found in the literature also varied widely. Some studies have used hospital charges rather than cost estimates, which explains part of the difference in hospitalization costs observed across studies. However, estimated hospital costs also vary because of differences in the course of treatment, the inclusion of expenses that are unrelated to a child's diagnosis with bronchiolitis (such as surgery), and differences in the allocation of hospital overhead expenses.

One possibility that we considered for assessing the cost-effectiveness of palivizumab was to conduct an analysis from the societal perspective that uses data from the literature and from secondary data sources. However, such an analysis would suffer from many of the same limitations identified in existing studies. In particular, estimates of the impact of palivizumab would need to be drawn from the IMpact-RSV trial; as mentioned already, the only outcomes readily available in the literature for the subgroups of interest are hospitalization rates. The impact of palivizumab on length of hospital stay and on incidence of ICU admission is provided only for the full study sample. Although a RCT of palivizumab that focuses on children with congenital heart disease is currently under way, study results will not be available until late fall 2002. Data on baseline hospitalization rates would also need to be drawn from the literature, and these vary widely because of differences in methods and differences in the underlying risk factors for RSV bronchiolitis infection in the population.

Estimating hospital charges for children with bronchiolitis is possible from national data sets, such as the Nationwide Inpatient Sample of the Healthcare Cost and Utilization Project at AHRQ. However, although diagnosis codes are now available indicating RSV, the presence of these codes may not accurately indicate the true burden of this disease as RSV-antigen testing is often not routinely done, and gestational age at birth is not indicated.

Another challenge to conducting an analysis of the cost-effectiveness of palivizumab from the societal perspective is that additional data would need to be collected to estimate the impact of a child's bronchiolitis on the family and to assess whether palivizumab therapy affects long-term outcomes, such as chronic asthma. Without these data, only analyses from the provider perspective are possible.

Given the gaps in the literature, high variation in parameters, and the wide ranges in results, the true cost-effectiveness of RSV prophylaxis among infants in the target population has not been demonstrated. Questions over cost-effectiveness and cost-benefit of palivizumab among infants 32 to 35 weeks EGA have been cited as reasons for reserving indication of prophylaxis, except for instances where there are specific risk-factors. Although infants born 32–35 weeks EGA may be expected to encounter lower RSV hospitalization rates than infants born less than 32 weeks EGA, the IMpact-RSV trial indicated that this difference may not be very significant, while also demonstrating that palivizumab had better efficacy in these more premature infants. Thus, although cost-effectiveness has not been quantified for this population, prophylaxis in this population cannot be assumed to be less cost-effective than among infants already indicated for palivizumab. Cost-effectiveness is one factor for use when deciding whether or not to use a health-care intervention. At this time, usable measures of cost-effectiveness of palivizumab prophylaxis for each of the target populations are not available.

Diagnosis

Prospective trials of the utility of ancillary testing (chest x-rays, complete blood tests, RSV testing) are feasible and should be performed. Studies of diagnostic tools used in the management of bronchiolitis should measure clinical outcomes that are meaningful to both parents and clinicians. An important intermediate outcome for studies of diagnosis in the management of bronchiolitis is the change in physician practices (i.e., whether results of diagnostic steps alter the ways that physicians elect to manage their patients with this condition).

Treatment

Our review revealed that for several interventions for bronchiolitis, data are simply insufficient to exclude them as possible effective treatments. Until these interventions are shown to be efficacious, our conclusion is that their clinical use ought to be limited to study situations. The following interventions, in particular, should be studied with well-designed, rigorously conducted RCT, preferably with placebo control: (a) nebulized epinephrine; (b) nebulized salbutamol plus ipratropium bromide; (c) nebulized ipratropium bromide; (d) oral corticosteroids, preferably dexamethasone; (e) inhaled budesonide; (f) inhaled helium-oxygen for severely ill children; (g) Chinese herbal therapy with Shuang Huang Lian (if its use can be practically accomplished in U.S. settings); and (h) surfactant for ventilated children.

The treatment studies we reviewed were almost universally underpowered and as such do not give clinicians adequate guidance for management of bronchiolitis. There is substantial evidence that clinicians commonly use several interventions such as inhaled bronchodilators, inhaled corticosteroids, and inhaled epinephrine for which, currently, evidence is insufficient. These drugs are all available as generic products and, therefore, relatively inexpensive; clinicians also consider them to be safe. We believe that clinicians will continue to use these types of treatments unless a large simple trial of these most common interventions is mounted. Such a trial would need to be large enough to examine each of the interventions not only in the overall population, but also in subpopulations of interest (e.g. infants with and without a history of atopy). This type of trial is unlikely to be funded by industry and would therefore require governmental support.

Prophylactic Therapy

Studies of the use of prophylaxis in at-risk groups that had been excluded from earlier studies will need to be released before this agent can be recommended more broadly for infants and children who are at increased risk of more severe bronchiolitis. Studies of prophylaxis should examine the effect on long-term outcomes such as the development of asthma.

Evidence is insufficient about the use of PFP-2 vaccine among high-risk infants with chronic lung disease or among children with cystic fibrosis. Our conclusion is that this vaccine ought not to be used except in the context of well-designed, properly powered RCTs to determine its effectiveness, safety, and cost-effectiveness.

Costs of Prophylaxis

Better estimates of the cost of palivizumab are needed to assess whether the drug is cost-effective. In particular, additional data are needed on the cost of administration. Key issues include typical dosage amount and number of doses, time required for parents and providers to administer it, and the actual cost of palivizumab to providers, which may be less than the average wholesale or catalog prices used in most previous analyses.

Estimates of baseline hospitalization rates for RSV bronchiolitis in the specific subgroups of interest (infants 32 through 35 weeks' EGA or with comorbidities) are needed to assess better whether prophylactic therapy is cost-effective for these populations. These analyses should also consider how hospitalization rates differ depending on socioeconomic characteristics of the population and region of residence.

To assess the cost-effectiveness of palivizumab from the societal perspective, data are needed on family costs. Family costs may be incurred for the receipt of prophylactic treatment (e.g., productivity losses and out-of-pocket expenditures) or for a child's infection with RSV bronchiolitis and subsequent treatment. Other data needed to estimate the societal costs of bronchiolitis are information on excess chronic morbidity for infants in the palivizumab treatment group (e.g., asthma) and premature death.

Data are needed to assess whether outpatient service utilization and costs and length of acute episodes differ between the prophylaxis and no-prophylaxis groups of infants for the populations of interest. Although the cost of outpatient services is largely dwarfed by hospitalization costs, if children who receive prophylactic therapy require much less ambulatory care and their families incur significantly less expenses and productivity loss, these differences may be significant.

Although many studies have attempted to measure the impact of EGA and the presence of comorbidities on RSV infection rates, the importance of other risk factors should also be considered. For example, the impact of day care attendance, multiple birth, exposure to secondhand smoke, room-sharing with siblings, socioeconomic status, and general hygiene should also be considered when assessing the impact of palivizumab on RSV infection and subsequent illness.

Outcomes

In the future, investigators should choose clinically relevant outcomes. Most of the outcomes studied in this literature are short term; often they are only surrogate outcomes such as oxygen saturation or respiratory rate at 15-minute intervals after treatment. Investigators should concentrate on measuring outcomes that matter to parents, clinicians, and health systems. Examples include rates of hospitalization or rehospitalization, duration of hospitalization, need for more intensive services in hospital, costs of care, parental satisfaction with treatment, and development of chronic asthma.

Design

Studies should be powered to detect meaningful differences in clinically relevant outcomes. Power calculations must include sufficient numbers to account for multiple comparisons if multiple outcomes are to be measured.

Reporting of Adverse Events

Few studies reported adverse events associated with treatments. Determining whether the risks of particular treatments are sufficient to exclude their clinical use is difficult. Future investigations should carefully monitor and report adverse events associated with treatments.