Validation of contemporary scoring algorithms for medical clearance of emergency department psychiatric patients.

BACKGROUND

In 2024, >6 million US ED visits for primary psychiatric complaints required medical clearance before psychiatric evaluation. However, clearance methods vary across institutions. Tools such as the SMART checklist, Shah's Score, and Triage Algorithm for Psychiatric Screening (TAPS) score aim to standardize this process. Our purpose was to compare the utility of these tools for the medical clearance of patients with psychiatric ED.

METHODS

We retrospectively reviewed adult ED patients' Electronic Medical Records (EMRs) with psychiatric chief complaints at a large US urban teaching hospital, excluding pregnancy, incarceration, or incomplete records. We applied the three scores retrospectively to assess the accuracy. Outcomes included disposition to the psychiatric unit or discharge vs. medical admission. The patients were followed up for 30 days to track revisits. The accuracy of Shah's, TAPS, and SMART checklists were calculated and analyzed using Pearson's χ2 test.

RESULTS

Of 1613 screened patients, 507 met the inclusion criteria. Their mean (SD) age was 39.1 (±14) years; most were male (66.7 %) and African American (49.7 %). Suicidality was the most common complaint (49.9 %); 55.0 % presented voluntarily, and 89.9 % underwent laboratory tests. Dispositions included ED discharge (61.9 %), psychiatric (35.5 %), or medical admission (2.6 %). The SMART checklist had the highest sensitivity (83.3 %, 95 % CI = 51.6 %-97.9 %). The TAPS and Shah's score sensitivities were 66.6 % (34.9-90.1 %) and 33.3 % (09.9 %-65.1 %), respectively. The specificities of Shah's score, TAPS score, and SMART were 76.9 %(73 %-80.6 %), 36.6 %(32.1-40.8 %), and 44.9 %(40.4 %-49.4 %), respectively. NPVs were 99.1 % (96.8 %-99.9 %), 97.8 % (94.5 %-99.4 %), and 97.9 % (96.0 %-99.1 %) for SMART, TAPS, and Shah's, respectively. The SMART Checklist missed two patients: one with thyroid cancer and one admitted for inability to perform activities of daily living (ADLs).

CONCLUSION

The SMART checklist showed higher sensitivity and NPV than the other tools for identifying psychiatric patients who did not require medical evaluation. However, further prospective validation is required to confirm these findings.