Search strategy

A systematic search was performed in indexed databases, including Chinese BioMedical Database (CBM), China National Knowledge Infrastructure (CNKI), Wanfang database and PubMed to identify available RSV studies in China. The following search terms were used: RSV or syncytial virus. Taking into account the quality of studies, high quality publications from the Chinese core journals (2014 edition) [11] were considered in the final analysis. The Library of Perking University evaluates all Chinese journals every four years and excludes lower quality journals using the quality measurement similar to the impact factors. To obtain recent data, the search strategy was limited to publications dated from January 2010 to Mar 2015. Details of the search strategy are presented in Appendix S1 in Online Supplementary Document(Online Supplementary Document) .

Inclusion and exclusion criteria

To be included, the following criteria had to be fulfilled: 1) studies in humans; 2) studies in patients with ARTIs; 3) studies that had at least one following outcome: etiology of acute respiratory infections; seasonality; gender; age group; serotypes; clinical features; 4) studies published in Chinese or English.

Publications were excluded if they were: 1) animal experiments or basic research (examples, studies focus on principles or mechanisms using cells and tissues); 2) case reports, systematic review or meta–analysis; 3) replicates (when the same population was studied in more than one publication, only the latest one or the one with the most complete data was considered for the meta–analysis).

ARTIs was defined as patients who were present of one or more respiratory symptoms, including watery eyes, rhinorrhea, nasal congestion or sinus congestion, otitis media, pharyngitis, cough, sore throat, sneezing, headache, and muscle pain. Meanwhile, patients had at least one symptom during acute infection, with high fever (body temperature ≥38°C) or chillness or normal/low leukocyte count or who were diagnosed with pneumonia by chest radiography previously. Chest radiography was conducted according to the clinical situation of the patient, and pneumonia was defined as an acute illness with radiographic pulmonary shadowing (at least segmental or in one lobe) by chest radiography.

Literature screening and data extraction

Literature reviewers were divided into two parallel groups. Using the set criteria of inclusion and exclusion, the reviewers independently screened the literature by title, keyword and abstract. Any disagreement was solved by a third reviewer. If they were not sure whether the study should be included, the decision was made based on further review of the full texts. NoteExpress 2 (Aegean Software Corporation, Shanghai, China) was used for the bibliography management.

Two parallel groups independently extracted the following data from eligible studies: general information, methodological quality and outcome data. Inconsistencies between two groups were checked after data extraction. Any disagreements were solved by the third reviewer.

Quality assessment

This meta–analysis included various types of studies with different outcomes. Therefore, no pre–existing scale is directly suitable for the assessment. The 5–item specific rating scale was developed to assess the quality of studies. These included 1) Did the study report patients’ information? 2) Did the study report diagnosis criteria of acute respiratory infection? 3) Did the study report specimen collection methods? 4) Did the study report pathogen detection methods? 5) Did the study report statistical methods? Each item was scored on three scales; 0 indicating low quality, 1 indicating medium quality, and 2 indicating high quality. The score for each item was then added to give a composite score for the study, with a highest total score of 10. If the total score was equal to or greater than 8, we regarded the study as “good” quality.

Statistical analysis

The MetaAnalyst (Beta 3.13; Tufts Evidence–based Practice Center, Boston, USA) was used to conduct meta–analyses for pooled proportions and odds ratios. Considering heterogeneity across all studies, we chose a random–effects model to carry out meta–analysis using Der–Simonian Laird method. The publication bias was determined via Stata 12.0 (StataCorp LP, Texas, USA) using an Egger’s test. Meta–analysis for combining the results of studies was weighted to provide the balanced results of all included studies.

Selection of studies

Of the total of 4852 studies identified through the databases, 135 studies were included in the analysis ( Figure 1 ). Of the 135 studies, 123 studies were in children less than 16 years old and the remaining studies were in both children and adults. Of the 135 studies, 19 studies were published in English. The detailed information about author, publication year, province, age, specimen type, detection methodology, number of specimen and study outcomes are listed in Appendix S2 in Online Supplementary Document(Online Supplementary Document) .

Figure 1.  Flow diagram of literature search and selection. *Including: 1) Did not discuss respiratory tract infections or respiratory syncytial virus; or 2) only as background to discuss; or 3) used the abbreviation “RSV”, but not referring to respiratory syncytial virus. †These studies reported/included charts or risk ratios and the numerators and denominators were not available.

For the 135 studies, the quality evaluation score ranged from 5 to 10 points, with a mean ± standard deviation of 7.7 ± 1.3. There were 80 studies with a score of ≥8. The summary of quality assessment is listed in Appendix S3 in Online Supplementary Document(Online Supplementary Document) .

Etiology

The overall positivity rate of RSV among patients with ARTIs was 18.7% (95% CI 17.1–20.5%), followed by HRV, human bocavirus (HBoV), IV, PIV, human metapneumovirus (HMPV), enterovirus, ADV and human coronavirus (HCoV). ( Table 1 ).

The prevalence of RSV was stratified into inpatients, outpatients, inpatients/outpatients and unknown category. RSV was found most frequently among inpatients 22.0% (95% CI 19.9–24.2%), followed by inpatients/outpatients, outpatients and unknown ( Table 2 ).

Seasonal characteristics

A total of 45 studies reported the seasonality of RSV infections. Of these 45 studies, 28 studies reported monthly isolation rates and the remaining 17 studies reported quarterly. The peak of RSV infections mainly occurred in winter and spring ( Table 3 ).

Gender characteristics

A total of 36 studies reported gender characteristics of RSV infections. Among 96 694 male patients, RSV was positive in 17 163 patients (20.4%, 95% CI 16.6–24.8%). Among 54 958 female patients, RSV was positive in 8364 (19.9%, 95% CI 16.0–24.4%).

RSV serotypes

There were 23 studies that reported RSV serotypes. Among 4172 RSV infected patients; type A, type B and mixed type A and B were detected in 2469 (63.1%, 95% CI 52.3–72.8%), 1611 (35.6%, 95% CI –6.0 to –46.6%) and 92 (1.2%, 95% CI 0.7–2.2%), respectively.

Clinical characteristics

Twenty–five studies reported the clinical manifestations of RSV infection. The most frequently reported clinical manifestations were cough, sputum production, wheezing and fever ( Table 4 ).

Sensitivity analysis and publication bias

The conclusions remained robust and the outcomes did not alter significantly when only ‘good’ quality studies were evaluated in the sensitivity analysis. The overall RSV positivity rates were 19.7% (95% CI 17.7–21.9%) in all patients (n = 79), 22.7% (95% CI 20.2–25.3%) in inpatients (n = 53), and 13.8% (95% CI 8.3–22.0%) in outpatients (n = 5).

Publication bias was tested using the Egger’s test. No publication bias was detected when verifying the 132 publications that reported RSV positive rates in all patients (–0.27, 95% CI –1.19 to –0.65, P = 0.563).