The detection of RSV in 23.1% of the samples in our study denoted a prevalence rate similar to those observed in other studies on the occurrence of lower respiratory tract infection, ranging from 23% to 61% [6, 20, 21]. The possible limitation of the study was the use of a less sensitive method (IF) as compared to RT-PCR that has shown improved sensitivity in the detection of RSV infection [14, 22]. In fact there is differential detection related to viral load. Infants usually have severe RSV disease associated with higher viral load hence a better chance of detection compared to the older counterparts leading to a differential misclassification (BIAS) . However, the large number of samples analyzed in this study minimizes the bias and did not alter the statistical significance of our results.
RSV has been referred in the literature as the main agent responsible for bronchiolitis and pneumonia during the first year of life. According to our study, 62.5% of children were younger than one year old, suggesting that illnesses caused by RSV can be severe in this age group, thus requiring hospitalization with prompt and effective medical intervention [5, 24].
Gender was not identified as a significant risk factor for RSV infection in our study in agreement with other published studies [21, 25]. Nevertheless, some findings in the literature have shown a male predominance, as reported by D’Elia et al. , particularly between 0 and 2 months.
Evaluation of CRP is very useful for clinicians because it may help differentiate between bacterial and viral etiologies. Shin et al.  used levels of CRP (≥1.87 mg/dL) as criteria to rule out serious bacterial infection in infants from self-limiting viral illness in febrile infants younger than three months. Diniz et al.  also found a statistically significant difference between nosocomial viral lower respiratory tract infection and levels of CRP less than or equal to 40 mg/L. In our study, the RSV-positive group showed a lower CRP mean level when compared to the RSV-negative group. Thus, low CRP levels may suggest a viral infection and, depending upon the clinical and radiological findings, enables the suspension of antibiotic therapy and helps considerably in the reduction of the hospital stay.
The radiological findings in our study have shown that 54.2% of RSV-positive patients developed interstitial infiltrate. Diniz et al.  found a significant correlation between nosocomial viral lower respiratory tract infection and interstitial infiltrate and it was observed that all patients with confirmed bacterial, fungal, or mixed infection presented alveolar infiltrate.
We found a predominance of RSV infection without bacterial co-infection in our study, in agreement with Duttweiler et al.  who found that concomitant bacterial sepsis was a rare event in 127 hospitalized RSV infected infants. However, when we compared the groups, positive culture was predominantly observed in the RSV-positive group than the RSV-negative group (10.0% vs 4.5%, p<0.028). Thorburn et al.  reported on pulmonary bacterial co-infection in children with severe RSV bronchiolitis showing that 40% of children with severe RSV infection were infected with bacteria in their lower airways. Unfortunately, we were unable to demonstrate if these infections would be either secondary or concurrent to viral infection.
It was possible to determine the RSV Group in 227 (93.4%) samples, and 6.6% were untypable probably due to problems in processing samples or RNA extraction. This study reports predominance of RSV Group B infection in children hospitalized with CAP, which is not unusual in other regions [9, 28]. Unfortunately, our data presents some limitation since we could collect samples for only one year and annual RSV Group epidemic may change. Suwanjutha et al.  identified in the first year of study the predominance of Group B, in contrast to the second year when Group A was more predominant.
Due to the high number of RSV Group B infection, comparison with.Group A (only 18 children) epidemiological data is not conclusive. However, in our findings, Group B was found among children with a mean age lower than that for Group A (11.0 vs 13.0; p<0.03), a finding similar to that of Papadopoulos et al.  who reported a predominance of RSV B infection in the youngest children, but the reasons are still unknown. Mlinaric-Galinovic et al.  and Oliveira et al.  did not find age differences between Groups, even though Mlinaric-Galinovic et al.  found that Group B infections occurred more frequently in males less than 12 months of age than in females. Oliveira et al.  on the other hand, found that the 57.8% of RSV A-infected children were male. Unfortunately, there is no plausible explanation for this variation.
We were able to determine that the CRP mean level in Group B was lower than Group A (13.0 vs 19.0; p<0.05) but the reasons for this remain unknown. Further research with an adequate number of samples needs to be conducted in an attempt to better understand RSV Groups in cases of CAP and their association with the epidemiological data of patients.
We have evaluated the signs and symptoms and approximately 98% of RSV-positive cases had cough at admission, but no statistically significant difference was observed in comparison with RSV-negative group (96.1%) (p>0.05). Rhinorrhea and nasal obstruction were predominantly observed in the RSV-positive group (p<0.05). Regarding the clinical signs and symptoms observed by Diniz et al.  in São Paulo city, in the preterm infants infected with RSV, wheezing, rhinorrhea, vomiting, and diarrhea were significantly more frequent while in our study vomiting and diarrhea were detected predominantly in RSV-negative group. In addition, the clinical symptoms do not predict the viral etiology because there are difficulties in establishing the general etiologic diagnoses of pneumonia by clinical profiles, which are quite varied in literature and depend on the infectious agent as well as the age and immune state of the host.
With regards to the seasonality, our study showed that RSV activity in hospitalized children during January to July 2007, with a peak during April to June, coincided with a heavy rainfall period in the region. Outbreaks occurring mainly during the months with low temperatures were reported in Uberlandia (Midwestern Brazil) by Costa et al. . This seasonal pattern was also observed in other Brazilian settings where RSV occurrence is not uniform. This has been noted in several Brazilian geographic regions and even among different states in the same region . In Campinas (São Paulo)  and in Vitória (Espírito Santo) , South-eastern region, the annual highest incidence of RSV-infections occurred between January and June, the same period of months observed in our study.
A greater understanding of the factors that determine RSV activity would make this timing even more precise. However, such studies are important as they are expected to delineate the clinical and epidemiological behavior of RSV in this age range and in this region. Even though our study presents a low prevalence of bacterial co-infection among the RSV-positive patients, monitoring RSV activity is necessary in order to restrict antibiotic use to the infants in real need and to provide better prophylactic therapies.