Systematic review and meta-analysis of clinical features to exclude radiographic pneumonia in febrile neutropenic episodes in children and young people

Abstract

Aims Children presenting with febrile neutropenia no longer routinely have chest x-rays. Traditionally, the use of prolonged intravenous antibiotics may have made the clinical effect of undiagnosed pneumonia irrelevant. With shorter courses of antibiotics this could be a significant problem. This systematic review assessed if clinical features could exclude radiographic pneumonia in children with febrile neutropenic episodes. Methods The review followed “Systematic reviews: CRD9s guidance for undertaking reviews in healthcare” and was registered with the HTA Registry of Systematic Reviews, CRD32009100453. An electronic search strategy included multiple databases. Bibliographies of retrieved articles were examined. Published and unpublished reports were sought and no language restrictions applied. Two reviewers independently screened studies and undertook data extraction. Disagreements were resolved by consensus or with a third reviewer. Study validity was assessed using a modified Quality Assessment of Studies of Diagnostic Accuracy in Systematic Reviews questionnaire. Meta-analysis was used to synthesise sensitivity and specificity, with a univariate random-effects model to account for between-study variation. Results Electronic searches identified 2057 articles, with 3 further from bibliographies of systematic reviews and included studies. 89 were examined in detail, and 4 met the inclusion criteria. The four studies were undertaken in similar, appropriate populations. The prevalence of pneumonia ranged from 1.9% to 6.3%. Meta-analysis of the three higher quality studies estimated average sensitivity at 75% (95% CI 56.4% to 93.6%) and average specificity at 67.9% (95% CI 55.9% to 79.9%). With the prevalence of pneumonia at 5%, the negative predictive value was 98.1% (95% CI 96.1% to 99.6%) There was marked heterogeneity (sensitivity I2 =26%, specificity I2 =91%). Conclusion Signs and symptoms of lower respiratory infection have moderate sensitivity and specificity for pneumonia. Clinically, this means a child with absence of signs or symptoms of infection on clinical examination has probability of pneumonia of 1.9% (95% CI 0.4% to 3.9%).This justifies the routine withholding of chest radiographs in the general paediatric oncology population. However, one in 50 children (95% CI 26 to 240) will have an unsuspected pneumonia, and patients who have a predisposition to pneumonia, an unresolving fever, or who re-present after a short course of antibiotic therapy, may need a chest x-ray despite an absence of signs.