A study protocol for community use of digital auscultation to improve diagnosis of paediatric pneumonia in Bangladesh

IntroductionThe World Health Organisations Integrated Management of Childhood Illnesses (IMCI) algorithm relies on counting respiratory rate and observing respiratory distress to diagnose childhood pneumonia. IMCI performs with high sensitivity but low specificity, leading to over-diagnosis of child pneumonia and unnecessary antibiotic use. Including lung auscultation in IMCI could improve pneumonia diagnosis. Our objectives are: (i) assess lung sound recording quality by primary health care workers (HCWs) from under-five children with the Feelix Smart Stethoscope; and (ii) determine the reliability and performance of recorded lung sound interpretations by an automated algorithm compared to reference paediatrician interpretations. Methods and analysisIn a cross-sectional design, Community HCWs will record lung sounds of [~]1,000 under-five-year-old children with suspected pneumonia at first-level facilities in Zakiganj sub-district, Sylhet, Bangladesh. Enrolled children will be evaluated for pneumonia, including oxygen saturation, and have their lung sounds recorded by the Feelix Smart stethoscope at four sequential chest locations: two back and two front positions. A novel sound-filtering algorithm will be applied to recordings to address ambient noise and optimize recording quality. Recorded sounds will be assessed against a pre-defined quality threshold. A trained paediatric listening panel will classify recordings into one of the following categories: normal, crackle, wheeze, crackle and wheeze, or uninterpretable. All sound files will be classified into the same categories by the automated algorithm and compared with panel classifications. ConclusionsLung auscultation and reliable interpretation of lung sounds of children are usually not feasible in first-level facilities in Bangladesh and other low- and middle-income countries (LMICs). Incorporating automated lung sound classification within the current IMCI pneumonia diagnostic algorithm may improve childhood pneumonia diagnostic accuracy at LMIC first-level facilities. Ethics and disseminationEthical review has been obtained in Bangladesh (BMRC Registration Number: 09630012018) and in Edinburgh, Scotland, United Kingdom (REC Reference: 18-HV-051). Dissemination will be through conference presentations, peer-reviewed journals and stakeholder engagement meetings in Bangladesh. Trial registration numberNCT03959956 Article summayStrengths and limitations of this study O_LIEvaluating the quality of lung sound recordings in a first-level facility where auscultation is usually unavailable and challenging to obtain due to a typically crowded and noisy environment and providers may not get enough time to calm the child due to time pressure from a high-volume patient. C_LIO_LIThis study will assess the feasibility of recording lung sounds by front line community health workers who do not usually use conventional stethoscopes during clinical care. C_LIO_LITwo standardised paediatricians masked to the childs clinical status will independently classify the recorded lung sounds, and a third masked and independent paediatrician will arbitrate any discrepancies. C_LIO_LIA machine-learning algorithm developed by Johns Hopkins and Sonavi Labs will detect abnormal lung sounds and be compared with classifications by human listeners/paediatricians. C_LIO_LIThe study will not have chest radiography findings of enrolled children, which is considered by many a gold standard for pneumonia diagnosis, as chest radiography is not available at this level of the health system in Bangladesh. Instead, this study will measure the peripheral oxyhaemoglobin saturation and evaluate clinical examination findings, including respiratory danger signs data. C_LI