ObjectiveTo assess the performance of currently available sepsis recognition tools in patients referred to a community-based acute ambulatory care unit.DesignService evaluation of consecutive patients over a 4-month period.SettingCommunity-based acute ambulatory care unit.Data collection and outcome measuresObservations, blood results and outcome data were analysed from patients with a suspected infection. Clinical features at first assessment were used to populate sepsis recognition tools including: systemic inflammatory response syndrome (SIRS) criteria, National Early Warning Score (NEWS), quick Sequential Organ Failure Assessment (qSOFA) and National Institute for Health and Care Excellence (NICE) criteria. Scores were assessed against the clinical need for escalated care (use of intravenous antibiotics, fluids, ongoing ambulatory care or hospital treatment) and poor clinical outcome (all-cause mortality and readmission at 30 days after index assessment).ResultsOf 533 patients (median age 81 years), 316 had suspected infection with 120 patients requiring care escalated beyond simple community care. SIRS had the highest positive predictive value (50.9%, 95% CI 41.6% to 60.3%) and negative predictive value (68.9%, 95% CI 62.6% to 75.3%) for the need for escalated care. Both NEWS and SIRS were better at predicting the need for escalated care than qSOFA and NICE criteria in patients with suspected infection (all P<0.001). While new-onset confusion predicted the need for escalated care for infection in patients ≥85 years old (n=114), 23.7% of patients ≥85 years had new-onset confusion without evidence for infection.ConclusionsAcute ambulatory care clinicians should use caution in applying the new NICE endorsed criteria for determining the need for intravenous therapy and hospital-based location of care. NICE criteria have poorer performance when compared against NEWS and SIRS and new-onset confusion was prevalent in patients aged ≥85 years without infection.