Recognizing and Predicting the Syndrome of Imminent Death (≤14 Days): A Systematic Review.

Abstract

Recognizing and predicting the syndrome of imminent death (SID) is central to end-of-life (EOL) decisions. To systematically review the diagnostic/prognostic accuracy of bedside clinical approaches used to recognize or predict the SID (death ≤ 14 days). We searched MEDLINE, Web of Science, and Scopus to 2024. Eligible studies evaluated bedside approaches-individual signs/vitals, composite criteria/scores, clinician probability estimates, and device-based monitoring-against observed death within ≤14 days. Risk of bias was appraised; evidence was synthesized narratively. Certainty was assessed (GRADE-DTA). The protocol was registered preferred reporting items for systematic reviews and meta-analyses guidelines (PROSPERO). Thirteen studies (6447 patients; 465 clinicians) were included. At ≤48-hour, clustered signs-confusion, hypotension, low oxygen saturation, death rattle, reduced consciousness-showed positive predictive value (PPV) ≈ 95% and negative predictive value (NPV) ≈ 81% (strong rule-in). At ≤72-hour, single bedside signs had positive likelihood ratio (LR+) ∼9-16. A ≤ 3-day bedside diagnostic tree reported accuracy = 68.3%. In acute/triage cohorts (≤7-14 days), respiratory rate (RR) >28 per minute and heart rate (HR) ≥110 per minute were associated with higher short-term risk (odds ratio ≈ 12.7 and ≈ 4.9), amplified by uncontrolled disease/metastasis. Glasgow prognostic score (GPS) plus thrombocytopenia achieved specificity >95% and LR+ >5 for ≤3 days. The one day surprise question (SQ) showed sensitivity ∼82% and NPV ∼ 91% (≤24 hour); the three day SQ showed sensitivity ∼94% and PPV ∼54% (≤72 hour). Nurses outperformed physicians at ≤72 hour (concordance statistic 0.85 vs. 0.68). Continuous, nonwearable monitoring of RR and HR correlated with mortality across 24/48/72 h windows. Certainty ranged low to moderate. Short-horizon recognition of SID is best supported by clustered signs, simple vital-sign cut-offs, and disease context, supplemented by GPS-based composites and one-day/three-day SQ. Prospective external validation with standardized horizons and calibration reporting is needed.