European Journal of Epidemiology

, Volume 18, Issue 5, pp 431–439

Can respiratory syncytial virus etiology be diagnosed clinically? A hospital-based case-control study in children under two years of age


  • Josef Alfons Isidor Weigl
    • Pediatric Infectious Diseases, Department of General PediatricsUniversity Children's Hospital Kiel
  • Wolfram Puppe
    • Pediatric Infectious Diseases, Department of General PediatricsUniversity Children's Hospital Kiel
  • Heinz-Josef Schmitt
    • Pediatric Infectious Diseases, Center for Preventive PediatricsUniversity Children's Hospital Mainz

DOI: 10.1023/A:1024213400297

Cite this article as:
Weigl, J.A.I., Puppe, W. & Schmitt, H. Eur J Epidemiol (2003) 18: 431. doi:10.1023/A:1024213400297


An unmatched, hospital-based case–control study was performed, to determine, whether respiratory syncytial virus (RSV) etiology in hospitalized young children can be predicted clinically. Children under 2 years of age admitted with a lower respiratory tract infection in three hospitals in northern Germany were included (one tertiary and two secondary centers). Cases were children tested positive for RSV by multiplex RT-PCR. One control group consisted of children tested negative for RSV in the multiplex-RT-PCR and a second control group consisted of patients in whom no PCR was done. A weighted backward stepwise logistic regression model was applied for multivariate analysis. RSV-etiology could be predicted with a sensitivity of 72.8% and a specificity of 73.2%. Young age, disease entity – pneumonia or bronchiolitis, center, intercostal retractions, absence of an underlying condition, low level of C-reactive protein, short duration of symptoms (all on admission), prematurity and epidemiologic year were predictive; anatomical infiltrates and wheezing were not. Pathogen specific diagnosis is necessary for individual therapy, allocation in observational studies or treatment trials and for surveillance of airway infections in children, since the positive predictive value is too low for an accurate diagnosis and decision making. Multivariate techniques are effective tools in complex clinical research for deconfounding.

Case–controlClinical predictionHospitalizationMultivariate analysisRespiratory syncytial virus

Copyright information

© Kluwer Academic Publishers 2003