Early warning scores to assess risk before emergency laparotomy

I read with great interest the excellent study from Portsmouth describing a new ‘evidence-based’ PRE-OP model to predict mortality before emergency bowel surgery [1]. The PRE-OP score showed good discrimination and was well calibrated. It requires external validation in other NELA centres and has several regulatory obstacles to negotiate before clinical implementation [2]. National Emergency Laparotomy Audits (NELA) report that many patients experience long delays in the Emergency Department before medical assessment and surgical referral [3]. The 8th NELA report quotes a median time of ≥15 h from arrival at the hospital to the time of surgery for patients with suspected sepsis at the time of arrival [3]. Initial triage and subsequent postoperative care in the UK are heavily impacted by one of the key contributory components of the novel PRE-OP model; namely the individual National Early Warning Score (NEWS2). Many acute surgical emergencies with NEWS2 ≤ 4 wait even longer for an operating theatre slot. Portsmouth previously reported excellent prediction of deterioration (ROC > 0.89) by NEWS in an earlier cohort of non-elective surgical patients. The original 2016 Portsmouth paper speculated that a lower clinical escalation NEWS threshold was required for surgical patients than their medical counterparts [4]. The dramatic reported reduction in NEWS2 performance (ROC < 0.71) in patients undergoing emergency bowel surgery is cause for concern [1]. It challenges the basic premise that identical NEWS2 escalation thresholds are not disadvantageous to surgical patients. This raises several questions. First, are the authors confident that the additive ‘consensus-based’ NEWS2 (ROC approximately 0.71) has clinical utility in acutely unwell surgical patients? Second, does the current NEWS ≥ 5 escalation threshold put surgical patients at increased risk of going under the clinical radar? Finally, does the widespread adoption of electronic observation charts signal the opportunity for a switch towards an ‘evidence-based’ logistic surgical model specifically calibrated for the patient group it aims to protect [5]?