Incremental predictive value of the Nurse Intuition Patient Deterioration Scale added to MEWS in an emergency observation unit: a retrospective validation and before-after analysis

While the Modified Early Warning Score (MEWS) is widely used for early detection of clinical deterioration, reliance on physiological thresholds alone may miss subtle changes in Emergency Observation Unit (EOU) patients. The Nurse Intuition Patient Deterioration Scale (NIPDS), which captures nurses’ clinical gestalt, has shown promise in ward settings but remains to be validated in the EOU. This study primarily aimed to evaluate the incremental predictive value of adding NIPDS to MEWS for 24-hour deterioration risk among EOU patients; a secondary, exploratory aim was to describe the association between the implementation period and event rates. This retrospective before-after study included 9,595 EOU patients from a tertiary hospital in two sequential periods: Period 1 (January–December 2024, MEWS alone, n = 4,881) and Period 2 (February 2025–January 2026, MEWS + NIPDS, n = 4,714). The primary analysis was a paired comparison within Period 2, where both MEWS and NIPDS were recorded for the same patients, to assess the incremental discrimination of the combined score. The secondary exploratory analysis compared event rates across periods using multivariable logistic regression. The primary outcome was a 24-hour composite of process/escalation events (unplanned transfer to the resuscitation room or ICU) and hard clinical events (cardiac arrest or death), which were also analysed separately because escalation decisions may themselves be influenced by the NIPDS assessment. Within Period 2 ( n = 4,714), the combined score demonstrated significantly better discrimination than MEWS alone (AUC 0.899, 95% CI 0.884–0.912 vs. 0.732, 95% CI 0.708–0.755; paired DeLong test Z = 15.64, P < 0.001), with higher sensitivity (0.855 vs. 0.732) and specificity (0.797 vs. 0.599) at exploratory Youden cut-offs. NIPDS alone achieved an AUC of 0.879 (95% CI 0.862–0.897), indicating that MEWS contributed additional predictive information beyond NIPDS. In the exploratory cross-period comparison, Period 2 was associated with lower adjusted odds of the composite outcome (aOR 0.83, 95% CI 0.73–0.94, P = 0.004); this association is susceptible to temporal confounding and incorporation bias and should not be interpreted as prevented deterioration. Adding NIPDS to MEWS improved within-period risk discrimination for 24-hour clinical deterioration in EOU patients. Because the composite outcome was dominated by escalation decisions that may themselves be influenced by the NIPDS assessment, whether this predictive advantage translates into improved patient outcomes requires prospective, preferably cluster-randomised, validation with blinded outcome adjudication.