Measurement Properties of Delirium Tools in Pediatric Intensive Care: A Systematic Review

Abstract

CONTEXT: Delirium is common in pediatric intensive care units. Reliable tools are needed, but evidence on measurement properties remains fragmented. OBJECTIVES: To identify pediatric delirium tools and evaluate their measurement properties and certainty of evidence. DATA SOURCES: MEDLINE, EMBASE, PsycINFO, CINAHL, Cochrane Library, and Web of Science were searched without language or date restrictions. STUDY SELECTION: Original observational, cross-sectional, and validation studies evaluating at least 1 measurement property of a pediatric delirium tool in acute care were eligible. DATA EXTRACTION: Two reviewers independently screened studies, extracted data, and assessed risk of bias using the adapted COnsensus-based Standards for the selection of health Measurement INstruments (COSMIN) methods and Quality Assessment of Diagnostic Accuracy Studies-2 for diagnostic accuracy. Evidence was rated using COSMIN-adapted Grading of Recommendations Assessment, Development and Evaluation. RESULTS: From 8378 records, 41 studies were included, covering 15 language versions of 6 tools: Cornell Assessment of Pediatric Delirium (CAPD), preschool-Confusion Assessment Method for Intensive Care Unit, pediatric-Confusion Assessment Method for Intensive Care Unit, Sophia Observation withdrawal Symptoms scale - Pediatric Delirium (SOS-PD), PEdiatric Delirium Scale (PEDS), and Child Delirium Assessment Scale (CDAS). CAPD had the largest evidence base. High-certainty evidence supported criterion validity of CAPD and Confusion Assessment Method tools, with moderate-certainty evidence for reliability. SOS-PD showed moderate-certainty evidence for criterion validity and high-certainty evidence for convergent validity and reliability but was evaluated in fewer studies. Evidence for CDAS and PEDS was limited to single studies. LIMITATIONS: Evidence was uneven across tools and mainly addressed criterion validity and interrater reliability. Measurement error, cross-cultural validity, and subgroup performance were rarely assessed. Criterion validity is difficult because no perfect gold standard exists for diagnosis. CONCLUSIONS: SOS-PD showed the most favorable certainty profile across evaluated measurement properties but was evaluated in fewer studies than CAPD. Further studies should address measurement error, cross-cultural validity, subgroup performance, and implementation.