539 - Performance Evaluation of Clinician Concern and Disparity by Preferred Language

Publication Number: 1518.539

Sarah Kim, Childrens Hospital of Philadelphia, Philadelphia, PA, United States; Hannah R.. Stinson, Children's Hospital of Philadelphia, Philadelphia, PA, United States; Katie Smith, Childrens Hospital of Philadelphia, Philadelphia, PA, United States; Megan Bernstein, Children's Hospital of Philadelphia, Wynnewood, PA, United States; Jenn Jacob-Freese, Children's hospital of Philadelphia, blue bell, PA, United States; Kelly Papili, CHOP, Philadelphia, PA, United States; Christine Taylor, Childrens Hospital of Philadelphia, Lititz, PA, United States; Sanjiv D. Mehta, Children’s Hospital of Philadelphia and University of Pennsylvania, Philadelphia, PA, United States

Background: Early warning systems (EWS) are critical for timely recognition of clinically deteriorating hospitalized children. In 2024, our hospital deployed a Clinician Concern (CC) system to supplement the Clinical Deterioration Risk (CDR) score (automated modified bedside PEWS score). The CC system asks bedside nurses to document their subjective ("gut" instinct) assessment of risk of deterioration on a 4-point scale (none, slight, moderate, significant). Subjective data may improve detection but also risk introducing disparate care for patients with a preferred language other than English (LOE).

Objective: To evaluate the impact of CC on EWS sensitivity for unplanned ICU transfer (ICU-tx) and to evaluate differences in performance by primary language.

Design/Methods: We performed a single-center retrospective study of children ( < 21yo) admitted to medical-surgical wards at our hospital from July 2024 - September 2025. The outcome was ICU transfer. We calculated maximum CC and CDR scores for each 12-hour interval (excluding 2 hours prior to ICU transfers). Each 12-hour window was evaluated for ICU transfer during that period or following 12 hours. Alert was defined as CC > None and CDR ≥4. We assessed sensitivity, specificity, and positive predictive value (PPV). We classified patients as English preferred and LOE preferred. We compared CC vs. CDR using McNemar's Test and English vs. LOE using Fisher's Exact Test.

Results: Across 20,262 admissions for 14,360 patients, 424 (2.1%) had ICU transfer. Mean patient age was 7.9 (SD 6.3) years; 1349 patients (9.4%) had a preferred LOE (Table 1). Median CDR scores were 5 (IQR 4-7). There were 2591 (12.8%) Slight, 335 (1.7%) Moderate, 41 Significant (0.2%), and 17295 (85.3%) No Concern scores.

While CDR had higher sensitivity than CC (52.4% [95% CI: 48.8, 56.0] vs 38.5% [35.0, 42.1], p< 0.01), integrating both had significantly higher sensitivity than CDR alone (60.9% [57.3, 64.3], p< 0.01) (Table 2). Comparing English and LOE preferred, there was no significant difference in CC sensitivity (English: 38.1% [34.3, 42.0] vs LOE 40.7% [31.7, 50.1], p=0.60), CDR sensitivity (English 51.7% [47.8, 55.7] vs. LOE 55.9% [46.5, 65.1], p=0.40), or combined sensitivity (English 60.3% [56.4, 64.2] vs. LOE 63.6%, [54.2, 72.2], p=0.51).

Conclusion(s): Addition of a clinician's subjective risk assessment to a EWS improved sensitivity for PICU transfer and did not introduce disparities by preferred language status.

Table 1
Baseline Characteristics of Patients Admitted

Table 2
Performance of bedside clinician concern and elevated clinical deterioration scores for intensive care unit transfer within 2-24 hours