Performance of automated respiratory rate counters for respiratory rate assessment to support pneumonia detection among children under five years of age in low- and middle-income settings: a systematic review of qualitative and quantitative evidence.

BACKGROUND

Accurate respiratory rate (RR) assessment is essential for pneumonia identification under World Health Organization Integrated Management of Childhood Illness (IMCI) guidelines; however, manual RR counting is often inaccurate and difficult in routine care. Automated RR counters have been developed to address these challenges, but evidence on their diagnostic performance and implementation in low- and middle-income countries (LMICs) remains limited. This systematic review evaluated the accuracy, usability, acceptability and time efficiency of automated RR counters for RR assessment to support identifying pneumonia in children aged 0-59 months in LMICs.

METHODS

We searched MEDLINE, EMBASE, Web of Science, and Scopus for studies published between 1 January 2014 and 15 July 2025. Studies conducted in LMICs assessing automated RR counters in children aged 0-59 months were eligible, including quantitative, qualitative, and mixed-methods designs. Study quality was assessed using the Joanna Briggs Institute appraisal tools. Quantitative findings were synthesised narratively, and qualitative findings were analysed thematically.

RESULTS

Fourteen studies from seven LMICs were included. Two automated RR counters - Masimo Rad-G and ChARM - were evaluated. Masimo Rad-G showed variable performance across studies (sensitivity = 75.9%-95.4%; specificity = 93.8%-98.3%; kappa = 0.55-0.85, n = 3), while ChARM demonstrated high diagnostic performance (sensitivity = 95.8%; specificity = 93.5%; kappa = 0.74-0.86, n = 2). Usability studies reported generally positive user experiences, though performance was affected by child movement, adherence to measurement procedures, and device design. Health workers and caregivers perceived automated tools as more reliable than manual counting, with visual indicators improving confidence in clinical decisions. Reported challenges included battery life, maintenance, workflow disruption, and the need for continuous training and supervision.

CONCLUSIONS

Although the available evidence remains limited, automated RR counters show promise for RR assessment in LMIC settings to support pneumonia identification. Future studies are needed to confirm their effectiveness and implementation feasibility across diverse routine-care settings. Successful adoption is likely to depend on adequate training, device optimisation, and integration into existing health-system workflows.

REGISTRATION: PROSPERO: CRD420251080564.