ABSTRACT
BACKGROUND: Acute fever is a common presenting symptom in low/middle-income countries (LMICs) and is strongly associated with sepsis. Hypoxaemia predicts disease severity in such patients but is poorly detected by clinical examination. Therefore, including pulse oximetry in the assessment of acutely febrile patients may improve clinical outcomes in LMIC settings. METHODS: We systematically reviewed studies of any design comparing one group where pulse oximetry was used and at least one group where it was not. The target population was patients of any age presenting with acute febrile illness or associated syndromes in LMICs. Studies were obtained from searching PubMed, EMBASE, CABI Global Health, Global Index Medicus, CINAHL, Cochrane CENTRAL, Web of Science and DARE. Further studies were identified through searches of non-governmental organisation websites, snowballing and input from a Technical Advisory Panel. Outcomes of interest were diagnosis, management and patient outcomes. Study quality was assessed using the Cochrane Risk of Bias 2 tool for Cluster Randomised Trials and Risk of Bias in Non-randomized Studies of Interventions tools, as appropriate. RESULTS: Ten of 4898 studies were eligible for inclusion. Their small number and heterogeneity prevented formal meta-analysis. All studies were in children, eight only recruited patients with pneumonia, and nine were conducted in Africa or Australasia. Six were at serious risk of bias. There was moderately strong evidence for the utility of pulse oximetry in diagnosing pneumonia and identifying severe disease requiring hospital referral. Pulse oximetry used as part of a quality-assured facility-wide package of interventions may reduce pneumonia mortality, but studies assessing this endpoint were at serious risk of bias. CONCLUSIONS: Very few studies addressed this important question. In LMICs, pulse oximetry may assist clinicians in diagnosing and managing paediatric pneumonia, but for the greatest impact on patient outcomes should be implemented as part of a health systems approach. The evidence for these conclusions is not widely generalisable and is of poor quality.