Patients and methods

All children less than 12 months of age admitted to the Geneva Children’s Hospital or the Sophia Children’s Hospital, Rotterdam with a virologically confirmed RSV infection in the winter seasons 1993–4 and 1994–5 were included in the study. The study was approved by institutional ethical committees from the two hospitals. The RSV infection was defined as a positive result in direct immune fluorescence assay performed on cells from nasopharyngeal washings using fluorescein isothiocyanate labelled RSV specific monoclonal antibodies (DAKO, Ely) or detection of the viral antigen by indirect ELISA17 and subsequent confirmation by viral culture. Duplicate samples collected in Geneva were frozen and sent to the department of virology of Erasmus University for confirmation analyses. Children referred by other hospitals and nosocomially infected children were excluded from the analyses to minimise the potential influence of different referral systems.

The Geneva Children’s Hospital is a university hospital providing primary, secondary, and tertiary care for a population of approximately 500 000 inhabitants with an annual cohort of 5800 births in 1994. As it is the only children’s hospital it admits all infants from this defined area, including all prematurely born infants. The Sophia Children’s Hospital, Rotterdam is a university hospital with a combined secondary-tertiary care function. As one of several hospitals in the area it admits only some of the children from the Rotterdam area requiring admission to hospital. Most patients seen in the emergency care outpatient clinic of the Sophia Children’s Hospital receive basic paediatric care (90%) and only 17% come from outside the Rotterdam area.18

Epidemiological and clinical variables were prospectively obtained on admission and discharge or at a control visit for the season 1994–5, and retrospectively from the patient charts for the season 1993–4. Demographic and clinical data were recorded on a standardised form with common definitions for all items. The demographic variables included gender, age, duration of pregnancy, existence of underlying disease (defined as congenital heart disease, bronchopulmonary dysplasia, or T cell immune deficiency), breast feeding, a positive family history of asthma or eczema, number of children in the household, day care attendance, and smoking in the household. The clinical data included the number of days with breathing problems before admission, feeding difficulties (defined as an increase of time required for feeding or a decrease in feeding volume), a positive history of apnoea (defined as either a history of respiratory arrest with cyanosis or an observation of respiratory arrest for a period of more than 20 seconds and/or bradycardia with accompanying cyanosis in the paediatric emergency room or during hospital admission), respiratory rate, the presence of wheezing (scored positive if wheezing could be heard without using a stethoscope) and retractions, fever (defined as a rectal temperature higher than 38.5°C), oxygen saturation (Sao ₂) in room air, carbon dioxide tension (Pco ₂), pH, and abnormalities on a radiograph (hyperinflation, consolidation, or atelectasis) as described by the radiologist. Sao ₂ was measured transcutaneously with the use of a pulse oximeter (Hewlett Packard Neonatal (Rotterdam), Nellcor N-180 (Geneva)). Intubation was indicated in both centres in the case of (a) respiratory insufficiency with hypercapnia (Pco₂ > 8 kPa and pH < 7.2), (b) hypoxia (Sao ₂ <85% with a fractional inspired oxygen >60%), (c) prolonged episodes of apnoea leading to severe bradycardia requiring stimulation or hand bag ventilation, or (d) sudden clinical deterioration. Discharge from either hospital required an adequate fluid intake for age, correction of tachypnea, and no oxygen requirement.

Data collected on the course of the disease and treatment included the occurrence of additional apnoea during the hospital stay, the maximum respiratory rate, the length of stay in hospital, admission to and length of stay in the intensive care unit, use of mechanical ventilation, administration of oxygen, bronchodilators, ribavirine and/or antibiotics, and number of deaths.

The clinical data from the Geneva and Rotterdam patients were compared in a χ² test, Fisher’s exact test, or Mann-Whitney U test when applicable. Differences in the clinical manifestations between Geneva and Rotterdam were tested again with multiple regression analysis, adjusting for possible confounders. Linear regression was used for continuous variables and logistic regression for dichotomous variables. The clinical parameters were entered in the regression model as dependent variables, and the confounders and location were entered as independent variables. Statistical significance was accepted at p < 0.05. To check for seasonal differences in clinical severity the analyses were also performed separately for the two winter seasons.

Results

RATES OF ADMISSION TO HOSPITAL IN GENEVA

We calculated the rates of admission to hospital for RSV infection in the Geneva Children’s Hospital for both term (n = 5800) and preterm (gestation <37 weeks, n = 312) infants younger than 12 months of age. The rate of admission to hospital was 5.3/1000 for term and 22/1000 for preterm infants for the 1994–5 winter season. The rate of admission to hospital in Rotterdam could not be reliably calculated as the size of the attachment population for the Sophia Children’s Hospital cannot be precisely defined.

DEMOGRAPHICS

A total of 208 children younger than 12 months of age (61 in Geneva, 147 in Rotterdam) were admitted to hospital with a diagnosis of RSV infection. A predominance of RSV subgroup A was found in nasopharyngeal specimens in the two centres (Rotterdam, 1993–4 60% subgroup A, 1994–5 100% subgroup A; Geneva, 1993-4, data not available, 1994–5 75% subgroup A). Six infants in Geneva (one secondary referral and five nosocomial infections) and 51 in Rotterdam (27 secondary referrals and 24 nosocomial infections) were excluded from the analysis.

Table 1 summarises the demographic data on the 151 children included in the study. Significant differences between centres were only noted for breast feeding (table 1) and age at admission (fig 1): 55% of children admitted in Geneva and 69% of those admitted in Rotterdam were younger than 4 months (χ² test; p = 0.03).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Age distribution of infants less than 12 months admitted to hospital for RSV disease in two European children’s hospitals.

DISEASE SEVERITY AT ADMISSION

Several disease severity parameters were reported differently by the two centres (table 2). Rotterdam infants were more often admitted with a history of apnoea and had a lower Sao ₂and a higher Pco ₂ at the time of admission than the subset of Geneva infants for whom Pco ₂values were available. Geneva infants, in contrast, had a higher mean respiratory rate and more often presented with wheezing and chest retractions. Feeding difficulties and chest retractions were the most common presenting symptoms in the two centres.

COURSE OF DISEASE AND TREATMENT

Treatment in the two centres included supplementary oxygen, empirical bronchodilator administration, and the use of antibiotics but no corticosteroids (table 3). Ribavirin was only used for a subset of infants in Rotterdam. Additional episodes of apnoea (two infants) were only reported in Rotterdam, as well as the death of a 4 week old infant with no known RSV risk factors who had been admitted with cardiorespiratory arrest after recurrent apnoea. The total length of stay in hospital was shorter in Geneva than in Rotterdam, where more children were admitted to the intensive care unit and required mechanical ventilation.

SEPARATE ANALYSIS OF SEASONS

To evaluate the influence of seasonal variability on the clinical severity of RSV, statistical analyses were performed independently for the prospectively studied 1994–5 winter season and the retrospective 1993–4 season. For the 1994–5 season, no difference in pre-existing risk factors was observed between Geneva (n = 31) and Rotterdam (n = 59) infants. Significant differences for Geneva infants were higher rates of breast feeding >1 month (74.2 v 30.6%, p < 0.001), higher respiratory rates (60.3 v 51.3, p = 0.019), a lower frequency of apnoea (3.2 v 23.7%, p = 0.013), lower Pco ₂ (5.65 v 6.91, p = 0.019), a shorter length of hospital stay (6.0 v 8.4 days, p = 0.01), and a lower rate of admission to the intensive care unit (6.5 v 32.2%, p < 0.001). In this small sample of 90 infants differences in the frequency of chest retractions, mean Sao ₂, and requirement for respiratory support did not reach statistical significance. The same trends were found for the 1993–4 season (24 infants in Geneva and 37 in Rotterdam). Significant differences for the Geneva infants were a higher rate of breast feeding >1 month (73.4 v 21.8%, p = 0.002), a higher mean respiratory rate (59.5 v 51.1, p = 0.03), a lower frequency of apnoea (0 v 24.3%, p = 0.009), a shorter length of hospital stay (7.3 v 9.9 days, p = 0.02), and a lower rate of admission to the intensive care unit (0v 21.6%, p = 0.02).

MULTIVARIATE ANALYSIS

Clinical parameters which differed significantly between Geneva and Rotterdam were again compared by multivariate analysis, adjusting for the identified epidemiological factors that differed between the two centres—namely, age and breast feeding. After correction for these potential confounders a higher percentage of apnoeas, a higher rate of admission to the intensive care unit, and a longer duration of hospital stay were still observed in Rotterdam, whereas a higher respiratory rate and higher percentage of wheezing on admission in Geneva remained significant.

Discussion

In this study we confirmed that the course of RSV infections is significantly more benign in Geneva than Rotterdam. Infants admitted to hospital in Geneva less often presented with apnoea or respiratory insufficiency and thus less often required admission to the intensive care unit or respiratory support than infants admitted in Rotterdam. Their more benign status was also reflected by a significantly shorter length of stay in hospital. This demonstration of a local variability of RSV disease severity is not restricted to the two centres studied. It is in accordance with at least two previous published observations. A striking difference in RSV morbidity was first reported in 1961 in two nearby nursery groups.19 More recently, a significant association between the hospital centre and parameters of clinical severity was reported in a multicentre study of RSV outcome in Canada.20 Furthermore, paediatricians from various European centres have subjectively recognised either of the two distinct clinical patterns of RSV disease presented here as representative of the situation prevailing in their area (personal communications to C A Siegrist at ESPID meeting, June 1996). Importantly, this study also shows (as 35 years ago19) that a detailed comparison of all the parameters previously reported as affecting disease severity does not identify the factors responsible for the observed differences in RSV disease severity.

The more benign course of RSV infection in Geneva does not appear to depend on a lower incidence of RSV infections in the first year of life. The rate of admission to hospital for an RSV infection during the first winter season at 5.3/1000 lies within the previously reported rates of admission to hospital of 1–20/1000 children.3 21 This rate of admission to hospital in Geneva is little affected by the variable severity of the RSV winter epidemic. Comparison of the two consecutive seasons of children admitted to hospital confirmed that the disease pattern and severity also remain constant. As rates of admission to hospital have been reported to be influenced by socioeconomic status, influencing the age at exposure and access to medical care,21-23 the potentially higher socioeconomic status of parents in Geneva would be expected to result in a reduction of rates of admission to hospital rather than of RSV disease severity. Thus the more benign course of RSV disease in Geneva than in Rotterdam essentially reflects a reduced severity of disease in the most severely sick infants who require admission to hospital.

Disease severity and the outcome of infants admitted to hospital is related to their pre-existing status such as prematurity, age less than 6 weeks, congenital heart disease, bronchopulmonary dysplasia, or immune deficiency.7 9 10 The lower severity of RSV infections in Geneva than Rotterdam is, however, not explained by a lower number of infants presenting with these underlying risk factors. Two factors found to differ between the two cohorts were a smaller percentage of children aged less than 4 months at admission and a higher percentage of breast feeding in Geneva. Interestingly, the percentage of breast fed infants in the Rotterdam cohort was also significantly lower than the overall rate of breast feeding in the Netherlands (65% at 1 month and 55% at 3 months of age24). As minimal or no breast feeding has been reported to increase the risk of admission to hospital for respiratory infections,21 mucosal protection could participate in the observed reduction of disease severity. Correcting for breast feeding and age in multivariate analyses did not correct the differences in disease severity, however.

Differences in subtype virulence have also been suggested to explain the yearly variation of disease severity,25 26 although no relation between clinical severity and RSV subtypes was found in a study in Rotterdam.27 In the present study a predominance of subtype A was observed in Geneva and in Rotterdam. Although virulence could still differ within strains of the same subtype, strain virulence differences are unlikely to result in a higher morbidity in the same centre over two consecutive winter seasons.

Other epidemiological factors that could explain the reported variation in disease severity (see under methods) were carefully compared and found to be similar in the two cohorts. Differences in referral systems were minimised in our study by only including primarily referred infants to either centre, but differences in hospital policies still affect rates of admission to the intensive care unit. In Rotterdam all RSV infected children less than 2 months of age or born prematurely are initially monitored in the intensive care unit, whereas admission to the Geneva intensive care unit depends exclusively on the clinical status. These hospital policies cannot, however, explain the differences in clinical parameters at admission or indication for mechanical ventilation.

Unexpectedly, we recognised two different disease patterns in our two cohorts: respiratory rate and frequency of wheezing and chest retractions were significantly higher in the Geneva infants. In contrast, respiratory insufficiency was more common in Rotterdam, although the duration of reported respiratory symptoms before admission was shorter. We postulate that the efficacy of compensatory hyperventilation in response to lung disease could be a critical factor distinguishing the two cohorts. This dissociation between an increased respiratory effort (previously described as a poor predictor of clinical severity28) and the clinical outcome suggests that environmental factors such as air quality may exert an influence on RSV morbidity by modulating the infant’s capacity to respond to pulmonary disease by compensatory hyperventilation. Parental smoking (similar in the two cohorts) and the use of wood burning stoves23 29 30 have been shown to increase the risk and severity of RSV infections. It is important to define the role of air humidity or temperature, either indoors or outdoors, or of industrial air pollution, which is responsible for an excess of cardiovascular deaths among adult or elderly patients31 and which could also affect the capacity of young infants to cope with respiratory infections.32 A relatively preserved air quality in medium sized cities such as Geneva compared with large industrialised urban agglomerations such as Rotterdam could well contribute to a lower morbidity of infant respiratory diseases. Air quality could thus contribute to the high RSV morbidity reported by large American or European centres, mostly located in dense urban environments.

In conclusion, parameters to be collected in multicentre studies assessing RSV disease severity have not yet all been identified. Whether air quality affects RSV disease in infants and elderly patients should be specifically addressed through prospective studies collecting air samples. Until these additional factors responsible for the geographical variations of RSV morbidity are identified, the many prophylactic or therapeutic strategies planned for the next decade should probably take into careful account the existence of different local disease patterns.