Improving Paediatric Pneumonia Diagnosis Using Digital Auscultation

Overview

This study aims to evaluate an automated interpretation algorithm of recorded lung sound by a digital stethoscope, name the Smartscope, among rural Bangladeshi children receiving community care in order to improve the diagnosis of childhood pneumonia at first level facility in low- and middle-income countries. A mixed-methods study will be conducted for a period of twelve months in rural Sylhet, Bangladesh. A total of 12 community health workers (CHWs) and 12 community healthcare providers (CHCPs) will be recruited and trained for this study. CHWs will conduct household surveillance to identify children with cough and difficult breathing and refer to nearby community clinic (CC). The CHCPs will screen the children at the CCs as per protocol and enroll the suspected cases with couth or difficult breathing. A total of 1003 children will be enrolled in this study. Enrolled children will be assessed for signs and symptoms of pneumonia including oxygen saturation. The children will have their lung sounds recorded by the Smartscope at four sequential locations. A listening panel comprises by pediatricians will generate one summary patient classification of normal, crackle, wheeze, crackle and wheeze, or uninterpretable. The Respiratory detector automated algorithm will be applied to the lung recording to generate an interpretation. The study hypothesis is more than 50% of patients will have quality lung sound recordings and the agreement between the automated computerized analysis by Respiratory Detector and an expert listening panel will be high (kappa \>0.5).

Background Childhood pneumonia is the second leading cause of under-five death accounting for about 0.921 million children globally.1 About 120 million new episodes of childhood pneumonia occur each year worldwide.2 The estimated incidence is 0.015 episodes per child-year in developed and 0.22 episodes per child-year in developing countries in this age group.3 In Bangladesh, the annual incidence of pneumonia is 36 per 100 child-years.4 Expanding pneumonia treatment beyond higher level facility and into first level facility settings with lay community health workers (CHWs) is an emerging strategy that can address gaps in care access. World Health Organization (WHO) recommended healthcare providers to use practical, standardised case management guidelines for childhood pneumonia care.5,6 However, these guidelines do not include lung auscultation in their pneumonia definition for frontline healthcare workers6 because of its high inter-observer variability and subjectivity, regardless of healthcare providers training level.7-11 Digital auscultation by electronic stethoscopes may overcome these limitations. An innovative, low-cost digital auscultation device was developed especially for children called Smartscope. The Smartscope improves lung signal strength by uniformly distributing highly sensitive electret microphone arrays across the stethoscope diaphragm so that the entire audible spectrum can be captured, a critical feature for identifying higher frequency pathologic lung sounds. It's rechargeable battery can power \>20 hours of use, important in rural communities with inconsistent electricity. The device mitigates movement artifact and tubular resonance by using an ergonomic design to better secure the device on the child's chest. It also eliminates the rubber stethoscope tubing, a source of ambient noise and friction contamination. It includes an integrated external microphone that removes unwanted ambient noises. The Smartscope also permits onboard data storage with a microSD card and is equipped with automated analysis software.12 Smartscope has been successfully validated in the laboratory against the Littmann 3200 electronic stethoscope and has demonstrated comparable results. This Smartscope has the potential to be a highly specific respiratory diagnostic tool that is feasible for use by community-based healthcare workers in LMICs. This project proposes to evaluate the potential impact of digital auscultation at first level facilities in rural Bangladesh. Objectives 1. To assess whether lung sounds recorded using Smartscope in children by Community Health Care Provider (CHCP) at CC meet pre-defined quality thresholds established by experts. 2. Determine the reliability and performance of the Smartscope Respiratory Detector automated analysis algorithm on lung sounds recorded by CHCP using the Smartscope, compared to reference interpretations by a paediatric listening panel. 3. Collect hypothesis-generating pilot data on the impact of digital auscultation on treatment decision-making by CHCPs, assuming lung recordings as the reference, as well as document equipment usability and durability during the project in order to understand whether additional device refinements are necessary. 4. Determine the acceptability of the Smartscope among CHCP and carers of children evaluated with the Smartscope. Methods A mixed-methods study will be conducted in rural Sylhet district of Bangladesh. Bangladesh has established about 13,000 CCs, one each for \~6,000 people. CCs are staffed by a CHCP with at least 12th-grade education and 3 months of pre-service training including Integrated Management of Childhood Illness (IMCI) guidelines. Twelve CCs will be selected purposively in Zakiganj and Kanaighat sub-districts of Sylhet district. Additionally, a total of 12 CHWs will be recruited and trained for this study. Each CHW will cover approximately \~13,000 population with \~1,400 under-five children. CHWs will visit each child aged 0-59 months in her catchment area every other month. The CHW will evaluate the child for signs/symptoms of pneumonia and question the mother or carer about any history of these symptoms. All suspected pneumonia cases (history/observed cough or difficult breathing) will be referred or accompanied to CCs with a referral slip for further evaluation and treatment. A study physician will be recruited for providing training and supervision of CHCPs and CHWs in clinical assessment, measurement of oxygen saturation and recording lung sound by the Smartscope. Screening: CHCPs in the CC will screen all under-five children as per protocol using a screening form. Approximately 7,200 children will be screened. If the child becomes eligible, the carer will be invited for consent to participate in the study. Enrolment: If the carer gives consent, the child will be enrolled in the study. A total of 1003 children with history/observation of cough or difficult breathing (possible pneumonia) will be enrolled. Lung auscultation: Lung sounds will be recorded using Smartscope. It is being used for its intended purpose only. 1. Sound collection: Enrolled children will have their lung sounds recorded at four sequential locations (two anterior and two posterior chest positions) for \~10 seconds/position. The recording procedure lasts approximately 60 seconds and captures 3-4 full breath cycles/position. Daily or alternate day Smartscope lung sounds will be uploaded to a server and securely stored. Later on, this lung sounds will be processed and used by the personnel involved in this study. 2. Sound processing: A novel sound-filtering algorithm will be applied to lung recordings to reduce ambient noise and optimize lung sound quality. 3. Listening panel: Three practicing pediatricians will form the panel. One will serve as the arbitrator and two as primary listeners. All panelists will be trained and standardized to the methodology developed and validated during PERCH. Panelists will generate one summary patient classification of normal, crackle, wheeze, crackle and wheeze, or uninterpretable. Two primary listeners will independently classify the recorded lung sounds, any discrepancies will be resolved through discussion, and if not resolved then the third listener will arbitrate. 4. Respiratory detector: The Respiratory Detector automated algorithm will be applied to the lung recordings to generate an interpretation. Pulse oximetry: Pulse oximetry will also be performed using a Masimo® Rad5 oximeter to all eligible consented children. If any child's oxygen saturation found \<90% will be referred to the sub-district health center or Sylhet Osmani Medical College Hospital. Collecting socioeconomic and confounders data: CHW will collect socio-economic and other related confounders' data including immunization information, weight, mid-upper arm circumference and length/height from all enrolled cases at the household level within seven days of enrolment. Focus group discussions: Four focus group discussions (FGDs) will be organized, two with carers (one with female and one with male) and one with CHCPs and one with CC management committee members and community leaders to share their opinions about the Smartscope. Each FGD will consist of 8-12 participants. A trained moderator will lead the discussion about perceptions of the Smartscope and the acceptability of Smartscope as a diagnostic for pneumonia.