Objective. To estimate the incidence of clinical deterioration leading to intensive care unit transfer in previously healthy infants with respiratory syncytial virus (RSV) infection hospitalized on a general pediatric unit and, to assess the hypothesis that history, physical examination, oximetry, and chest radiographic findings at time of presentation can accurately identify these infants.
Study Design. A virology database was used to identify and determine the disposition of all children ≤1 year of age admitted to the Children's Hospital at Strong (CHaS) with RSV infection during the 1985 to 1994 respiratory seasons. Index patients were all previously healthy, full-term infants admitted initially to the general inpatient services at CHaS or Rochester General Hospital, a second University of Rochester teaching hospital, whose clinical deterioration led to transfer to the pediatric intensive care unit (PICU). These infants were matched retrospectively (for year and date of infection, sex, chronologic age, and race) with two hospitalized controls who did not require PICU transfer. Chest radiographic findings, respiratory rate (RR), O2 saturation, and presence of wheezing at time of presentation to the emergency department (ED) were compared.
Results. During the study years, 542 previously healthy, full-term infants were admitted to the general pediatric unit at CHaS with proven RSV infection. Ten (1.8%; 95% confidence interval, 0.9%, 3.4%) were transferred subsequently to the PICU, primarily for close monitoring of progressive respiratory distress. Data for these patients and 7 patients transferred from Rochester General Hospital to the PICU at the CHaS were compared with those for control patients. The mean RR in the ED (63 vs 50), and O2 saturation in the ED (88% vs 93%) were modestly abnormal in cases compared with controls. Wheezing on examination at time of presentation and chest radiographic findings did not differ between the two groups. A RR >80 and an O2saturation <85% at time of presentation each had a specificity >97% for predicting subsequent deterioration. Each parameter, however, had a sensitivity ≤30%.
Conclusion. Clinical deterioration requiring PICU admission is an uncommon occurrence in previously healthy infants admitted to a general pediatric inpatient unit with RSV infection. Extreme tachypnea and hypoxemia were both associated with subsequent deterioration; however, only a small proportion of patients who clinically deteriorated presented in this way. The clinical usefulness of these parameters, therefore, is limited. respiratory syncytial virus, deterioration, healthy infants, prediction.
- RSV =
- respiratory syncytial virus •
- LRT =
- lower respiratory tract •
- PICU =
- pediatric intensive care unit •
- RR =
- respiratory rate •
- ED =
- emergency department
A rational approach to the treatment of previ ously healthy infants hospitalized for respi ratory syncytial virus (RSV) infection requires an understanding of the clinical course of this illness, as well as the ability to predict accurately the outcome in an individual infected child. Because most hospitalized RSV-infected infants receive only supportive care on a general inpatient unit for <5 days,1 it is possible that high quality and less costly care could be provided to selected patients outside of the usual hospital setting.
A major issue in the design of such strategies, and in their ultimate acceptance by pediatricians and families, is the concern that an infected infant will seriously deteriorate and require life-saving respiratory support that might be difficult to provide in a nonhospital setting. Despite a very low mortality rate (<1% of all hospitalized, infected children),1 RSV-infection can result in severe respiratory compromise. This is evident in the fact that up to 7% of previously healthy, infected children require mechanical ventilation at some point in their illness.2 The development of alternative care strategies for treating infants with RSV infection who present initially with milder symptoms, therefore, depends on the incidence and predictability of life-threatening, respiratory compromise after initial evaluation and disposition planning have been completed.
Considerable data are available on the clinical course of RSV infection leading up to the decision to hospitalize and also in predicting outcome variables such as death and mechanical ventilation. Most clinical markers associated with a high risk of these outcomes (ie, prematurity, underlying cardiopulmonary disease, bronchopulmonary dysplasia, and immunodeficiency) do not apply to previously healthy infants who comprise the majority of admitted children potentially eligible for alternate care strategies. Deterioration after admission might be predicted by different parameters. We limited our analysis, therefore, to previously healthy children already admitted to an inpatient pediatric unit for treatment of RSV infection.
We assessed the risk of serious clinical deterioration in these children and tested the ability of admission physical examination and radiographic studies to identify those infants at high risk for subsequent deterioration. This issue has not been well studied previously because of the rarity of serious deterioration in previously healthy children hospitalized with RSV. To obtain adequate numbers to address this question, therefore, we used data collected throughout 10 respiratory seasons. The extended study period also minimized confounding because of varying RSV subtype predominance in a single respiratory season.
Two study designs were used. First, a descriptive study was done to assess the incidence of clinical deterioration among previously healthy infants hospitalized with RSV infection. This analysis used a database of all children with RSV-lower respiratory tract (LRT) infection hospitalized at Children's Hospital at Strong. During the 1985 to 1994 respiratory seasons a nasopharyngeal swab for RSV enzyme-linked immunoabsorptive assay and viral culture were obtained from all inpatients <2 years old with respiratory symptoms. Historical and clinical data were collected prospectively for each child during each hospitalization. Approval from the institutional review board for human research studies was obtained for the collection of these data. Pertinent demographic information, past medical history, and disposition of study patients were extracted from this database.
The second study involved a matched case-control study to compare children who clinically deteriorated after admission to those who did not. All infants ≤1 year old with RSV-LRT infection and without a history of congenital heart disease, immunodeficiency, bronchopulmonary dysplasia, or prematurity who clinically deteriorated after admission, as indicated by transfer to the pediatric intensive care unit (PICU), were identified as cases. They were initially hospitalized on the general pediatric ward at one of two teaching hospitals affiliated with the University of Rochester Medical Center. Patients were admitted from the emergency department (ED) or pediatric clinic based on the clinical assessment of the pediatric attending. Standard of care on the two inpatient units did not vary substantially during the study period and included oxygen therapy, bronchodilator and/or antiinflammatory therapy, intravenous hydration, and ribavirin at the discretion of the admitting pediatrician. Transfer criteria to the PICU included: 1) potential need for intubation and mechanical ventilation as a result of either severe recurrent apnea with desaturation or clinical evidence of increasing respiratory distress; or 2) need for intensive nursing care/monitoring greater than the typical ratio provided on the general unit. Transfer was initiated by the covering physician after consultation and the agreement of the PICU attending. All patients requiring intensive care were treated at a single site. Patients admitted directly to the PICU from a referring physician or the ED were not included in this study.
We used the virology database mentioned above to identify cases and controls. Two controls were identified for each case. Cases were limited to infants born ≥35 weeks' gestation without underlying medical conditions. Cases and controls were matched to the extent possible for year and date of infection (within 60 days of admission date), sex, chronologic age, and race (in descending order of priority). The exclusion criteria and matching process were performed to limit confounding because underlying medical condition, virus strain, chronologic age, sex, and gestational age ≤34 weeks have been shown previously to be risk factors for more severe disease.3–9 Children with hospital-acquired infections were not included as cases or controls.
The medical records of cases and controls were reviewed. The parental report of number of days ill before presentation and the reason for seeking medical attention were obtained from the ED record. Respiratory rate (RR), presence of wheezing, and oxygen saturation in room air were determined from the initial ED examinations. The radiology report of a chest radiograph obtained within 12 hours of presentation was reviewed for presence of hyperinflation and/or focal infiltrate. For cases, the length of time before transfer, the reason for PICU transfer, and vital signs at time of transfer were coded based on the PICU attending's admission assessment. All medical record information was abstracted by a single reviewer.
SPSS-PC software (SPSS Inc, Chicago, IL) was used for statistical analysis. Paired t test analyses, with mathematical transformation when necessary, were done for continuous variables. χ2 or Fisher's exact methods were used for discrete variables.
Figures 1 and2 illustrate the disposition of infants ≤12 months old admitted for RSV infection during the study period. Seventy-seven percent of admissions in this age group were full-term, previously well infants. Only 1.8% of full-term, previously healthy infants subsequently deteriorated and were transferred to the PICU (95% confidence interval, 0.9%, 3.4%). Although the incidence of subsequent deterioration in infants born <35 weeks' gestation was nearly 7 times greater (11.8%), this still represented a relatively uncommon occurrence.
Information was obtained from the virology database on a total of 17 infants ≥35-weeks' gestational age that were initially admitted to the general pediatric unit and subsequently were transferred to the PICU. Ten patients were from the Children's Hospital at Strong pediatric inpatient service and 7 were from the inpatient service at Rochester General Hospital. One patient was excluded from analysis because of an unavailable medical record. Table 1 compares the demographic characteristics of case and control patients. They were similar with respect to sex, race, and chronologic age, verifying appropriate matching. The clinical characteristics of the index patients are shown in Table 2. Index patients were generally tachypneic with mild respiratory acidosis at time of transfer.
The association between each independent variable and subsequent PICU transfer was determined. Table 3 shows the distribution of case and control patients for each of the clinical variables. The number of days the patient was symptomatic before presentation did not differ between index and control groups. Infants who presented to the ED for evaluation of fever, rather than for a primary respiratory complaint, were less likely subsequently to deteriorate. Wheeze and chest radiograph findings were not statistically different between the case and control groups.
Figures 3A and 3B illustrate the distribution for RR and O2 saturation in the ED in case and control patients. Although extreme tachypnea and hypoxemia were only identified in children who subsequently deteriorated, values for RR and oximetry in index patients overlapped substantially with those of controls. The receiver-operator characteristic curves for RR and O2 saturation (Fig 4) illustrate the sensitivity and specificity of these parameters in predicting the need for subsequent PICU transfer. The sensitivity of these parameters is too low to be useful in identifying the majority of infants at risk.
These data demonstrate the low incidence of clinical deterioration in previously well, full-term infants with RSV-LRT disease who are deemed stable enough at admission to be hospitalized on a general pediatric unit. This patient population may differ in some ways from other groups of infants hospitalized with RSV infection. First, our infants were all previously healthy and full-term and did not include those with known risk factors such as bronchopulmonary dysplasia, congenital heart disease, or prematurity. Second, a previous study has shown that Monroe County has the lowest hospitalization rate for children with LRT illnesses in New York State,10suggesting that admission criteria in this county are stricter than most and admitted patients may be sicker than in other areas. The rate of deterioration reported in this community, therefore, may overestimate the actual risk in other areas.
Our study differed from many previous studies assessing the reliability of clinical parameters to predict outcome in previously healthy children with RSV bronchiolitis.5 ,6 ,11 ,12 The focus of this investigation was on serious deterioration after the decision to hospitalize had been made. Serious deterioration after hospitalization is a rare event making recruitment of sufficient study participants difficult. Large numbers of hospitalized children need to be followed to accurately predict the frequency of respiratory deterioration. We chose multiple respiratory seasons as a method for collecting an adequate study population. We limited the study period to 10 years to minimize the effect of changing standards of general and intensive care received by inpatients.
Our analyses showed that pathophysiologic parameters have limited clinical value for identifying infants at risk for clinical deterioration. Not unexpectedly, extreme hypoxemia was useful for predicting deterioration despite hospitalization and standard treatment. A minority of patients, however, presented with this finding. RR also had minimal usefulness for predicting clinical deterioration. Presentation with a chief complaint of fever was useful for identifying patients at low risk for subsequent deterioration. These infants often had mild respiratory symptoms and were admitted primarily for evaluation of potential bacteremia. No index patients presented in this manner. The usefulness of any individual clinical variable, therefore, is limited in a single patient by its low sensitivity for detecting risk of adverse outcome.
These findings are limited by the retrospective evaluation of clinical treatment and disposition criteria. The majority of infants admitted to the general unit received supportive care with oxygen and intravenous hydration. Bronchodilator and antiinflammatory agents also were used intermittently as indicated. A minority of previously healthy infants received ribavirin. This treatment did not change substantially during the study period.
During the entire study period, the primary reason for PICU transfer was the attending physician's assessment that the infant required closer nursing and medical observation than was available on the general inpatient unit. In this retrospective study summarizing the hospital course of index patients was difficult. An accurate assessment of the length of time a patient was hospitalized before deterioration was suspected or recognized by the medical and nursing staff was not possible. In most cases, however, there seemed to be a gradual worsening of respiratory status throughout a number of hours rather than minutes. Only 1 patient had apnea requiring urgent airway management.
These data have important implications for the design of alternative strategies of care for previously healthy, full-term infants with RSV infection. Because the risk of deterioration is so low, sites of care other than a standard hospital-based pediatric inpatient service are attractive. Nevertheless, there is a small risk of deterioration that is difficult to predict at time of disposition planning. In addition to providing supportive services, therefore, these alternative strategies need to recognize the small risk of serious deterioration and provide a level of supervision and resources for transfer to a higher level of care that occasionally will be required.
This work was supported in part by Grant HS09062 from the Agency for Health Care Policy and Research.
- Received October 27, 1998.
- Accepted March 24, 1999.
Reprint requests to (A.-M.B.) Box 667, 601 Elmwood Ave, Rochester, NY 14642. E-mail:
Presented in part at a poster session at the Society for Pediatric Research meetings; April 1997; Washington, DC.
Micah Aviram, MD, is currently at Soroka University Medical Center, Ben-Gurion University of the Negev, Beer-Sheva, Israel.
- Wang EEL,
- Law BJ,
- Stephens D,
- et al.
- Copyright © 1999 American Academy of Pediatrics