Respiratory syncytial virus (RSV) is the major cause of lower respiratory infection (LRI) and hospitalization among infants and toddlers in North America.¹ Nearly all children are infected by age 2 years and 1% to 2% of those infected require hospitalization.² Among those admitted to hospital with no apparent risk factors for severe disease, 4% to 15% are admitted to the intensive care unit (ICU), 1% to 5% require assisted ventilation, and <1% die.³ In contrast, among children with underlying heart/lung disease, prematurity (gestation 36 weeks) and young age (6 weeks) the corresponding figures for ICU, ventilation, and mortality range from 10% to 40%, 8% to 27%, and up to 10%, respectively.³

Ribavirin, first approved in the United States in 1986, is the only licensed antiviral therapy for RSV infection. In 1993 the American Academy of Pediatrics (AAP) recommended that ribavirin should be given to selected infants and children at risk for or already manifesting severe disease as well as for all ventilated patients.⁸ Although several randomized trials suggested that ribavirin was efficacious among nonventilated patients,⁹ the use of the drug remained controversial, due to the small numbers studied, and concern over the validity, generalizability, and clinical relevance of the outcome measures used. For ventilated patients, controversy existed as well, with one trial supporting a beneficial effect for ribavirin¹⁵ while another failed to show a difference.¹⁶ Large increases in the cost of ribavirin since 1993 raised further questions about its relative cost-benefit. In Canada over 90% of RSV-related hospital costs are due to daily bed charges.¹⁷ With the daily cost of ribavirin at approximately $1500 (Canadian funds), therapy would have to lead to a substantial reduction in the duration of hospital stay to justify the costs of ribavirin. Accordingly, the AAP recommendations for patients who should receive ribavirin were used to select cases enrolled in a prospective study of children hospitalized for RSV LRI and outcomes were examined to determine the relationship between ribavirin therapy and length of hospitalization.

METHODS

From the database, cases were selected if at least one of the AAP recommended indications for ribavirin use was present, including: any underlying CHD or CLD, PREM, INFANT, and EARLY HYPOXIA. Excluded from analysis were nosocomially acquired RSV LRI, defined as onset of symptoms 3 or more days after admission to hospital, and patients with known congenital or acquired abnormalities of the immune system.

The association between ribavirin use and hospital outcomes was examined separately in ventilated and nonventilated patients in each of the CHD, CLD, PREM, INFANT, and EARLY HYPOXIA subsets. The groups were not mutually exclusive and some children were included in two or more analyses. However, before analysis the subsets were separated according to the presence or absence of cardiopulmonary disease, such that the CHD/CLD group could include cases that fit the PREM, INFANT, or EARLY HYPOXIA criteria, but none of the latter three subsets included patients with CHD or CLD. The outcomes were duration-measured in days of: hypoxia, mechanical ventilation, management in the ICU, and RSV-attributable hospitalization. For children on oxygen supplementation before the onset of RSV infection, the duration of hypoxia was defined as the number of days until return to pre-illness baseline. RSV-attributable hospital days were defined as days required for specific or supportive therapy or observation relative to the RSV illness. Additional days spent in hospital for investigation of other problems, elective surgery, or delayed discharge due to social or transportation problems were not counted. Because the data were not normally distributed, the Wilcoxon test was used throughout to compare outcome duration. ² or Fisher's exact tests of proportion were used to compare distribution of demographic factors, coexisting risk factors, and indicators of disease severity among the ribavirin and no-ribavirin therapy groups.

Receipt of ribavirin was also added as an independent variable to a previously developed multiple regression model for predicting RSV-attributable length of hospitalization.¹⁸ Separate analyses were performed for the CHD/CLD subsets and the INFANT/PREM/EARLY HYPOXIA subsets. The other independent variables included in the model for both analyses were INFANT, PREM, EARLY HYPOXIA, aboriginal racial background, history of apnea or respiratory arrest before or at the time of admission to hospital, consolidation on the admitting chest radiograph, and hospital center. For analysis of children with prior heart or lung disease CHD and CLD were included as independent variables.

RESULTS

Study Groups

The cumulative proportion of ribavirin-treated patients whose therapy was initiated by hospital day 2 and 3 was 66% and 81%, respectively for children with cardiopulmonary disease, and 76% and 95%, respectively, for previously healthy children.

Specific Risk Group Analysis for Effect of Ribavirin

The variation in duration of hospitalization, hy- poxia, and ICU management stratified by receipt of ventilation and ribavirin are presented in Table 1. There were no significant differences in any of the four outcomes for ventilated cases regardless of risk group. For nonventilated patients ribavirin was associated with a prolongation of hypoxia and hospital days for each of the five risk subgroups and of ICU days for all but the CHD subgroup.

Table 1. Impact of Ribavirin on Total Hospital Days Attributable to Respiratory Syncytial Virus, as Well as Days of Hypoxia, Intensive Care Stay, and Mechanical Ventilation, Among Children With Community-acquired Respiratory Syncytial Virus Lower Respiratory Infection [View Table]

Table 2 shows the distribution of factors previously shown in a multiple regression model to predict the duration of RSV-attributable hospital days (CHD, CLD, PREM, INFANT) or to be evidence of early severe RSV disease (apnea as a presenting problem, infiltrate on an admitting chest radiograph, EARLY HYPOXIA) for patients who did or did not receive ribavirin within the ventilated and nonventilated strata of each risk subgroup. Significant differences (all by Fisher's exact test) between ribavirin treated versus those not given ribavirin, respectively, were: ventilated cases-a) CHD subgroup: coexisting CLD in 67% vs 0% (P = .002); b) CLD subgroup: coexisting CHD in 38% vs 0% and apnea as a presenting problem for 12% vs 50% (P < .05); and c) PREM subgroup: infiltrate on an admitting radiograph in 36% vs 78% (P < .05); nonventilated cases-a) CHD subgroup: infiltrate on an admitting radiograph in 77% vs 34% (P < .001) and EARLY HY- POXIA in 80% vs 53% (P = .03); b) PREM subgroup: EARLY HYPOXIA in 72% vs 43% (P = .03) and apnea as a presenting problem in 33% vs 11% (P = .02); c) EARLY HYPOXIA subgroup: INFANT in 54% vs 20% and PREM in 54% vs 18% (P < .001) and apnea as a presenting problem in 29% vs 8% (P = .002). There were no differences in the distribution of left to right shunt or pulmonary hypertension between treatment groups for the ventilated and nonventilated CHD subgroups. For the ventilated CLD group, a current or past requirement for home oxygen was less common in ribavirin recipients versus those not treated (38% vs 83%; P = .04). For nonventilated CLD patients 68% of each of the ribavirin-treated and untreated groups had a prior home oxygen requirement.

Table 2. Distribution of Factors Associated With Duration of RSV-Attributable Hospital Stay18 Among Specified Risk Groups According to Whether or Not They Received Ribavirin [View Table]

Effect of Ribavirin in a Multiple Regression Model for Duration of RSV-Attributable Hospitalization

Mortality

DISCUSSION

Given the absence of evidence that ribavirin therapy reduces mortality in the short-term, or improves pulmonary function in the long-term, to be cost-effective ribavirin therapy must reduce associated morbidity such as the need for intensive care management and/or mechanical ventilation, or shorten hospital stay. When outcomes in the CHD, CLD, INFANT, PREM, and EARLY HYPOXIA subgroups were examined using our database, the benefit of ribavirin could not be demonstrated. Recently, Moler et al²¹ showed that ribavirin therapy was not associated with reductions in duration of hospitalization, days in ICU, or days on mechanical ventilation for previously healthy term and premature infants who required ventilation during the course of community-acquired RSV LRI. Despite methodological differences between that study and the currently reported PICNIC study, the data were quite similar with respect to the median days spent in hospital, in ICU, and on a ventilator for the ribavirin and no-ribavirin treatment groups. The lack of significant difference observed for the ventilated subgroups in our cohort could reflect a lack of power to detect a difference given the small number of cases.

The major limitation of this analysis is that the administration of ribavirin was not randomized. Thus, selection of the sickest patients for ribavirin may have accounted for the greater morbidity. Yet when the ribavirin-treated and untreated subgroups were compared in terms of coexisting risk factors for prolonged RSV-attributable hospitalization, only a few significant differences were found. In general, as shown in Table 2, ribavirin was more likely to be given to patients with more than one of the criteria for ribavirin use as recommended by the AAP. However, this observation was not consistent for all risk groups. Furthermore, the association between ribavirin therapy and longer hospital stay was also shown using multivariate analysis, which controlled for many of the factors known to increase the risk of severe disease as well as some indicators of disease severity. Nevertheless, other factors not included in the model may have affected disease severity.

A second weakness of the analysis is that ribavirin therapy was not started at a uniform time during the hospital course, and in a few instances was delayed until the fourth to seventh hospital day. In such cases the increased number of days in hospital attributable to RSV may have been an artifact of the late introduction of therapy and subsequent requirement for extra days of hospitalization. This was unlikely to have a pronounced effect, however, because the majority of treated children were started on ribavirin within 2 to 3 days of hospital admission.

The only randomized placebo-controlled trial in which treatment was started within 72 hours of symptom onset, involving cases with significant cardiopulmonary disease, showed that none of 20 ribavirin-treated and none of 27 placebo-treated children required intensive care management.¹⁴ The only other randomized controlled trials that have used intensive care, ventilator, and total hospital days as outcome measures focused exclusively on ventilated patients.^15,16 When the control group received inhaled water, ribavirin-treated patients had significant reductions in the ventilator and total hospital days.¹⁵ In contrast, with inhaled normal saline as the placebo, these reductions were not observed.¹⁶ These two studies differed significantly in the type of patient studied in that the water placebo study focused on young infants of whom 75% had no known preexisting disease whereas the saline placebo study included older children, 54% of whom had known cardiopulmonary or other abnormalities. Neither study included cases ventilated because of apnea.

The prospective enrollment of cases at nine centers across Canada with specific recording of RSV-attributable days for total hospital stay increases the generalizability of the study. The data strengthen and extend similar observations made by Wheeler et al²² who found that the duration of hospitalization for RSV LRI was similar between a center that used ribavirin according to the AAP criteria and a center which used no ribavirin. It is possible that there are fewer restrictions to hospital admission in Canada, such that children are not as severely ill. Yet data summarizing the hospital course of RSV-infected children admitted to large centers in the United States suggest that the proportion of previously healthy children,³ as well as those with underlying cardiac,^6,19 pulmonary,^6,23 or immunodeficiency³ disease, admitted to the ICU or ventilated mechanically are similar to what we have observed among the centers participating in the PICNIC RSV studies.^7,18

Recently, the recommendations of the AAP regarding ribavirin use among high-risk infants and children were revised to read "may be considered" instead of "should be used."²⁴ For physicians who may be reluctant to be more restrictive in their approach to the use of ribavirin, these data should provide some assurance. Additional data from large randomized controlled trials are needed to determine if ribavirin is cost-effective in a clinically relevant sense.