A Systematic Review of Clinical Prediction Rules to Predict Hospitalisation in Children with Lower Respiratory Infection in Primary Care and their Validation in a New Cohort

EClinicalMedicine

Master Chisale¹  Peter Harrington²  Richard J Drew³  Chris J Watson⁴  Joe Gallagher⁵  Mark T Ledwidge⁶  Dermot M Wildes⁶ 

[1] Corresponding author: Dr. Dermot Michael Wildes, gHealth Research Group, School of Medicine, University College Dublin, Dublin;Innovations, Mzuzu University, Malawi;;Biological Science Department, Faculty of Science, Technology &Irish Meningitis and Sepsis Reference Laboratory, Temple Street, Dublin 2;Wellcome-Wolfson Institute for Experimental Medicine, Queen's University Belfast, Northern Ireland;gHealth Research Group, UCD Conway Institute, School of Medicine, University College Dublin, Ireland;

关键词: Paediatrics;    Respiratory Medicine;    Infectious disease;    Global health;    Pneumonia;   

DOI  :  

来源: DOAJ

【 摘 要 】

Background: Our goal was to identify existing clinical prediction rules for predicting hospitalisation due to lower respiratory tract infection (LRTI) in children in primary care, guiding antibiotic therapy. A validation of these rules was then performed in a novel cohort of children presenting to primary care in Malawi with World Health Organisation clinically defined pneumonia.Methods: MEDLINE & EMBASE databases were searched for studies on the development, validation and clinical impact of clinical prediction models for hospitalisation in children with lower respiratory tract infection between January 1st1946-June 30th 2021. Two reviewers screened all abstracts and titles independently. The study was conducted in accordance with the Preferred Reporting Items for Systematic Reviews & Meta-Analyses guidelines.The BIOTOPE cohort (BIOmarkers TO diagnose PnEumonia) recruited children aged 2-59 months with WHO-defined pneumonia from two primary care facilities in Mzuzu, Malawi. Validation of identified rules was undertaken in this cohort.Findings: 1023 abstracts were identified. Following the removal of duplicates, a review of 989 abstracts was conducted leading to the identification of one eligible model. The CHARMS checklist for prediction modelling studies was utilized for evaluation. The area under the curve (AUC) of the STARWAVe rule for hospitalisation in BIOTOPE was found to be 0.80 (95% C.I of 0.75-0.85). The AUC of STARWAVe for a confirmed diagnosis of bacterial pneumonia was 0.39 (95% C.I 0.25-0.54).Interpretation: This review highlights the lack of clinical prediction rules in this area. The STARWAVe rule identified was useful in predicting hospitalisation from bacterial infection as defined. However, in the absence of a gold standard indicator for bacterial LRTI, this is a reasonable surrogate and could lead to reductions in antibiotic prescription rates, should clinical impact studies prove its utility. Further work to determine the clinical impact of STARWAVe and to identify diagnostic tests for bacterial LRTI in primary care is required.

【 授权许可】

Unknown