Impact of United Kingdom guidelines for stratification of risk in adult hospital inpatients identified with sepsis – an observational cohort study

United Kingdom National Institute for Health and Care Excellence (NICE) guidelines for evaluation of sepsis risk remain to be validated. This study aims to model the impact of the updated guidelines in patients identified as having a risk of sepsis based on the original guidelines.

A single centre cohort study, between 2019 and 2022. Adult inpatients at risk of sepsis defined by red / amber flags in the original guideline were included. NICE criteria, based upon aggregate National Early Warning Score (NEWS2) and additional risk-factors (NEWS2 single parameter score three, lactate ≥2mmol/l, skin changes, acute kidney injury or deteriorating trajectory) were applied to this cohort to derive a revised risk classification. The primary outcome was the net reclassification index (NRI). Secondary outcomes were the association with a composite of critical care admission or death in hospital and the association with ICD-10 sepsis / Sepsis-3 status.

1303 hospital inpatients were included. Between red flag sepsis and NICE High-risk classification, there was agreement in 57.6%. The NICE model decreased the risk-classification in 471 (42.4%), of whom 216 experienced the composite outcome. The event RI deteriorated (-0.321), whilst the non-event RI improved (0.355). NICE High-risk had a similar C-statistic (0.57(0.55,0.60) vs. 0.56(0.54,0.58)) but with lower sensitivity (59.1% vs. 91.2%), higher specificity (55.1% vs. 20.2%) and PPV (55.9% vs. 52.1%) as compared to red flags. NICE High-risk had a greater C-statistic for association with Sepsis-3 status (0.66 (0.63, 0.69) vs. 0.55 (0.54, 0.57)).

As implemented in this study, NICE criteria decreased the number of patients in a high-risk sepsis category, compared to a model using red flag criteria. These criteria decreased sensitivity and increased specificity for the outcome of critical care admission or death, whilst strengthening the association with Sepsis-3 criteria. Limited predictive performance advocates for clinically-led sepsis assessment.