A Systematic Review of the Markers of Severity in Acute Respiratory Infections to Inform Primary Care Surveillance

Background: Primary care computerised medical records (CMR) are used to report the incidence of acute respiratory infections (ARI) for public health surveillance. These systems could increase their utility by also reporting population-level severity of ARI; however, this is rarely done. Objectives: To identify candidate markers of ARI severity suitable for use in primary care CMR-based surveillance. Methods: We undertook a systematic review of bibliographic databases and grey literature. Eligible studies reported characteristics for > 500 patients with ARI, severe ARI, influenza-like illness or suspected COVID-19. Studies had to report at least one potential marker of severity. A panel of clinical primary care informaticians reviewed candidate severity markers and assessed each for severity, specificity, relevance to primary care and whether it was likely to be recorded in a CMR. Results: We included 126 studies from 84 countries. Seventy-seven candidate severity markers were identified across 11 groups. These included four outcome groups (complications, hospital events, intensive care events and death) and seven predictor groups (symptoms, signs, scores, investigations, treatments, absenteeism and treatment-seeking behaviour). Thirty markers were considered most suitable for primary care CMR-based ARI surveillance: 7 outcomes (such as hospital admission, attendance and death) and 23 predictors (such as shortness of breath, oxygen levels, work absence and antibiotics). Predictors were generally considered more timely, as they are likely recorded during the consultation. Conclusions: This review provides a list of severity markers that could support the development of population-level severity indicators for ARI surveillance in primary care. This could improve real-time situational awareness during respiratory outbreaks.