Community use of digital auscultation to improve diagnosis of childhood pneumonia in low resource setting

Abstract

Background: The World Health Organization9s Integrated Management of Childhood Illness (IMCI) algorithm for childhood pneumonia diagnosis has high sensitivity but low specificity. This study aims to evaluate whether the use of automated lung sound classification through digital auscultation may improve the accuracy of pneumonia diagnosis in first-level facilities. Methods: In a cross-sectional design, Community Health Workers (CHW) record lung sounds using a novel digital stethoscope (Smartscope) of 2426 under-5 children with possible pneumonia at first-level facilities in Bangladesh. A standardised paediatric listening panel is classifying the recorded sounds. A mobile app containing the Smartscope analysis system is also classifying the sounds and comparing with the reference paediatric panel’s classification. Results: As of 31 December 2019, 1957 children screened, 1070 eligible cases identified and 1029 enrolled (32.67% had IMCI pneumonia). The results of the data collected during the first six months will be presented. These results will describe CHWs ability to record quality lung sounds and agreement between human and machine interpretation. Conclusions: Auscultation and correct interpretation of lung sounds are often not feasible in the first-level facilities. Incorporation of the auto-classification of the Smartscope recorded lung sounds within the current IMCI pneumonia diagnostic algorithm may improve the accuracy of the diagnosis of childhood pneumonia at first-level facilities in LMICs. Method: This research was commissioned by the UK NIHR Global Health Research Unit on Respiratory Health (RESPIRE).