Behavioral Telemetry in the ICU: Missing Orientation Assessment Predicts Mortality in Patients with Low Acute Physiologic Derangement

BackgroundBehavioral telemetry--the analysis of clinical actions NOT taken--may identify care process failures associated with adverse outcomes. While missed nursing care predicts outcomes in survey-based studies, objective EHR-derived measures are lacking. We hypothesized that missing routine cognitive assessment in ICU patients with low acute physiologic derangement would predict mortality independent of illness severity. MethodsRetrospective cohort study using MIMIC-IV (2008-2022, Beth Israel Deaconess Medical Center) with external assessment of documentation practices in eICU (208 US hospitals). We identified ICU admissions with SOFA 0-2 (low acute physiologic derangement), excluding neurological ICUs. Orientation documentation was classified within 24 hours. Primary outcome was in-hospital mortality. Multivariable logistic regression adjusted for age, sex, SOFA, and Charlson Index. ResultsAmong 46,004 ICU patients with SOFA 0-2, 4,737 (10.3%) had no orientation documentation within 24 hours. These patients had 24.68% mortality versus 7.57% early-assessed and 4.56% late-assessed. After adjustment, missing orientation was associated with 4.29-fold higher odds of death (95% CI 3.95-4.65; E-value 8.0). In SOFA=0 patients (N=23,670), the signal strengthened (OR 5.65, 95% CI 5.03-6.35; E-value 10.8). Late-assessed patients had the LOWEST mortality (OR 0.65), arguing against reverse causation. Patients without orientation had 22% MORE chart events (1,600 vs 1,309), arguing against neglect. External assessment revealed that among 166 eICU hospitals with [≥]100 eligible patients, only 5% documented orientation routinely--92% lack the infrastructure to detect this signal. ConclusionsIn ICU patients with low acute physiologic derangement, absence of orientation assessment is associated with 4-6 fold increased mortality. This association may identify care process failures not captured by severity scores, though prospective studies are needed to establish causality. What is Already Known on This TopicMissed nursing care--care omissions--predicts patient mortality in survey-based studies. Nurse staffing ratios are associated with mortality, but the mechanism is poorly understood. No objective, EHR-derived measure exists to detect care process omissions in real time. What This Study AddsFirst EHR-based operationalization of the missed nursing care construct, enabling objective, real-time detection. Missing orientation assessment associated with 4-6 fold increased mortality (OR 4.29 in SOFA 0-2; OR 5.65 in SOFA=0). Signal strengthens in SOFA=0 patients (E-value 10.8), suggesting finding is not driven by acute illness severity. Argues against reverse causation: late assessment has BETTER outcomes than early or no assessment. Argues against neglect: patients without assessment had MORE documentation, not less. Argues against immortal time bias: Never Documented patients had LONGER ICU stays (7.58 vs 3.09 days). Quantifies association: 10.3% of patients account for 27.2% of deaths. Reveals systemic gap: 92% of US ICUs lack the documentation infrastructure to detect this signal. Key PointsO_ST_ABSQuestionC_ST_ABSDoes absence of routine orientation assessment predict mortality in ICU patients with low acute physiologic derangement (SOFA 0-2), independent of illness severity? FindingsIn this cohort study of 46,004 ICU patients with SOFA 0-2, those without orientation documentation within 24 hours had 4.29-fold higher adjusted odds of death (95% CI 3.95-4.65). In SOFA=0 patients, the signal strengthened to OR 5.65 (E-value 10.8). Patients assessed late (6-24h) had the LOWEST mortality (OR 0.65), arguing against reverse causation. Among 166 eICU hospitals, only 5% document orientation routinely-- 92% lack the infrastructure to detect this signal. MeaningMissing routine cognitive assessment may identify care process failures associated with increased mortality. The finding that 92% of US ICUs lack the documentation infrastructure to detect this signal reveals a systemic gap in care process monitoring.