We obtained written informed consent from the mother, father and an adult primary caregiver for all children enrolled in the study. The Institutional Review Board at International Centre for Diarrhoeal Disease Research, Bangladesh, (icddr,b) approved the study and the Institutional Review Board at Centers for Disease Control and Prevention (CDC), USA deferred to icddrb's approval.

We conducted our study in ward three of Pallabi thana (“thana” is the lowest administrative unit in Bangladesh). This ward of low-income community is in the north-west part of Dhaka and covers an area of 3.4 sq. km. The estimated population of the ward is 69,960; half of its population is male. The mean literacy rate is 69% for males and 59% for females [18].

In January 2008, icddr,b initiated surveillance among a cohort of children in the study area to assess immunity to human amebiasis. All children who were born in the study area and lived within 1.5 kms of the study clinic, were enrolled at birth and followed prospectively. Our study was conducted on this existing birth cohort of children. In April 2009, when we started our longitudinal respiratory disease study we re-enrolled all children who were already participating in the amebiasis birth cohort and were aged 0–2 years. Subsequently we treated the population as an open cohort because we continued to enroll all children who were born in the study area after the April 2009 enrollment. We subsequently followed the cohort of children including children who were enrolled in the beginning of the study in April 2009 and children who were born after beginning of the study until March 2011. Children who were already part of the ongoing amebiasis birth cohort were not under any interventional study.

As part of on the ongoing amebiasis study, field assistants measured height and weight of each of the enrolled child within three days of birth. When we re-enrolled children into our study which began in April 2009, field workers trained in collecting surveillance data using hand held devices collected additional information from each child, including level of parental education and breast feeding practices. Prospectively, field workers visited each child at home twice weekly for two years of the study, to identify whether the child developed any major symptoms of respiratory illness, including subjective fever; rapid, laboured or noisy breathing; lethargy; cyanosis; inability to drink; or convulsion or minor symptomsincluding cough, rhinorrhoea, sore throat, muscle or joint pain, chills, headache, irritability, decreased activity and repeated vomiting within the last 72 hours of follow-up [19]. Field workers referred all children who developed at least one major symptom or two minor symptoms to the study clinic located within the study area.

In the clinic, study physicians collected a history of illness including signs and symptoms using a structured questionnaire, conducted a physical examination, and collected nasopharyngeal wash from each child every time a child presented with a new episode of reported or documented fever (T≥38°C) or respiratory symptoms including cough, nasal discharge or difficulty in breathing. An episode was considered to be new if the child had been symptom free in the preceding 7 days.

The study physician diagnosed a child with acute respiratory tract infection (ARI) if the child had cough and/or runny nose and diagnosed a child with pneumonia, if the child had cough or difficulty breathing with age-specific tachypnea using WHO classification [20] including the following respiratory signs: crepitation, ronchi and wheezing. In the study clinic all children received free clinical care using standard WHO recommendations [20] and also free referral services including hospitalization in tertiary hospitals when needed.

Study physicians collected a nasopharyngeal wash sample by attaching a butterfly catheter (needle removed), to a 10 ml syringe containing 5 ml normal saline, placing the child in a 30° semi-Fowler position with the head slightly angled forward, inserting the catheter between 2 and 3 cm into the nares, injecting the saline, and immediately applying suction while removing the catheter [19]. The wash sample was then placed in viral transport medium (containing Dulbecco Modified Eagle Medium), and stored at 4°C in the study clinic refrigerator. At the end of each day, within eight hours of sample collection, a field assistant transported all respiratory samples to icddr,b's virology laboratory using cold box. At the virology laboratory the samples were stored at or below −70°C temperature until analyzed. Virologists at icddr,b performed real-time reverse transcription polymerase chain reaction (rRT-PCR) assays to detect influenza A and B viruses, RSV, adenoviruses, HPIVs 1, 2 and 3, human rhinoviruses, and HMPV using primers and probes supplied by the Centers for Disease Control and Prevention (CDC), Atlanta [primer/probe sequences and assay protocols for non-influenza respiratory viruses available from CDC on request].

We observed each child from the start of enrollment of the child in the study and continued until the end of the follow-up period (March 2011), until the child left the study, or the family of the child moved out of the study area to reside in some other area. We calculated incidence as the number of new events divided by child-years at risk. For calculating virus-specific incidences of ARI and pneumonia we restricted our analyses to children who were aged <2 years during the follow-up period and divided the child-years in to three age groups (0–6months, 6–12 months and 12–24 months). We used Poisson estimation to calculate incidence and 95% confidence interval around incidences. Using the date of illness onset, and the proportions of viral pathogen detected in the respiratory samples, we plotted timing of viral pneumonia in a histogram.

Our study began in April 2009 and ended in March 2011. We followed 515 children for 730 child-years of observation (median time each child was observed was 1.7 years); 283 (55%) of the children were male. Fifty two (10%) of the 515 children did not complete the study because they migrated out of the study area. Among 515 children, 263(51%) were born after the study began in April 2009. The median age of the cohort at enrollment in to our study was 0.5 months (inter-quartile range [IQR] 0.1–8 months). The median birth weight of the cohort children was 2663 gms (a birth weight of 2500 gm is considered to be standard birth weight) [21] (Table 1).

Among the 515 children followed, 423 (82%) developed 1,322 episodes of respiratory infections during April 2009–March 2011. A total of 918 (69%) of the episodes were classified as ARI and 378 (29%) were diagnosed as pneumonia by the study physician. The remaining 26 (2%) episodes were diagnosed as otitis media or only cough by the study physician. Out of the 223 children who developed at least one episode of pneumonia, 47 (21%) developed two episodes and 22 (10%) developed three episodes. The main presenting symptoms among children diagnosed with pneumonia were cough (100%), tachypnea (100%), reported fever (98%) (53% of the pneumonia episodes also had documented fever), runny nose (97%), difficulty breathing (84%), crepitations (44%) and ronchi (33%). The study physician referred one child (0.3%) with pneumonia to a tertiary children hospital who was hospitalized due to aggravating symptoms even after treatment as per WHO guidelines. There was no other hospitalization associated with ARI or pneumonia and we did not observe any ARI or pneumonia related deaths during the study period.

We identified viral respiratory pathogens in 66% (610 of 918) of ARI episodes and in 77% (293 of 378) of pneumonia episodes (Table 2). The median age of children who developed an episode of ARI associated with detection of a respiratory virus was 9.5 months (IQR 6–16 months) and that of children who developed an episode of pneumonia associated with detection of a respiratory virus was eight months (IQR 5–14 months).

We also identified 41 cases of 2009 pandemic influenza A (2009 H1N1) virus infection among the cohort children. The first child with 2009 H1N1 infection was identified on 12 July, 2009, less than four weeks after the detection of the first 2009 H1N1 case on 19 June 2009 in Bangladesh [22]. Only six (15%) of the 41 children with 2009 H1N1 infection developed pneumonia. The median age at infection of children with 2009 H1N1 infection was 13 months (IQR 5.5–20 months); 67% were male.

The incidence of ARI was 125/100 child-years (95% CI 120–134) and pneumonia was 52/100 child-years (95% CI 47–57) in this cohort of children. The incidence of ARI and pneumonia associated with laboratory-confirmed respiratory virus infection was 80/100 child-years (95% CI 77–90) and 40/100 child-years (95% CI 36–45). Children aged 0–6 months were followed for 140 child-years at risk, children aged 6–12 months for 167 child-years and children aged 12–24 months for 297 child-years. The age-specific annual incidences of pneumonia/100 child-years associated with laboratory-confirmed respiratory virus infection were 32 (95% CI 21–39) for children aged 0–6 months, 51 (95% CI 41–63) for children aged 6–12 months and 44 (95% CI 37–52) for children aged 12–24 months. The annual incidence of pneumonia/100 child-years associated with specific respiratory viruses was 12.5 for RSV, 6 for rhinoviruses, 6 for HMPV, 4 for influenza viruses, 3 for HPIV 3 and 2 for adenoviruses (Table 3). Pneumonia associated with a respiratory virus infection was present throughout the year (Figure 1).