Use of data linkage to investigate the aetiology of acute lower respiratory infection hospitalisations in children

Moore, Hannah C; de Klerk, Nicholas; Keil, Anthony D; Smith, David W; Blyth, Christopher C; Richmond, Peter; Lehmann, Deborah

Aim: To document the aetiology of acute lower respiratory infection (ALRI) hospitalisations in Western Australian children by linking population‐based laboratory data with hospital morbidity data. Methods: Data from all ALRI hospitalisations and laboratory records related to respiratory pathogens between 2000 and 2005 were extracted and linked through a population‐based record linkage system. The proportion of specimens that were positive for each respiratory viral or bacterial pathogen was documented. Results: Eight thousand nine hundred and eighty (45.2%) ALRI hospitalisations were linked to a laboratory record. Admissions to a private hospital and admissions from non‐metropolitan areas were less likely to have a linked laboratory record. In 57.9% of linked hospitalisations, a respiratory virus and/or a bacterial pathogen was identified. Frequently identified viral pathogens included respiratory syncytial virus (RSV; n= 3226; 39.5% of those tested), influenza viruses (n= 664; 8.5%), parainfluenza virus type 3 (n= 348; 4.6%), picornaviruses (n= 292; 22.3%) and adenoviruses (n= 211; 2.7%). RSV was identified in 63.7% of bronchiolitis admissions in those aged under 6 months and 33.1% of pneumonia admissions in those aged under 12 months. Influenza viruses were identified in 81.6% of influenza‐coded admissions. When a test was requested, Bordetella pertussis was identified in 21.2% of ALRI hospitalisations (n= 354), including 86.8% of whooping cough‐coded admissions. Conclusions: This is the first report of population‐based data linkage between statewide laboratory data and hospitalisation records and demonstrates proof of principle. RSV continues to be an important pathogen in ALRI. As pathogens were identified across all diagnoses, relying on hospital diagnosis coding alone may not accurately estimate the burden of different categories of ALRI.