Reports of increasing incidence rates of delirium in critically ill children are reason for concern. We evaluated the measurement properties of the pediatric delirium component (PD-scale) of the Sophia Observation Withdrawal Symptoms scale Pediatric Delirium scale (SOS-PD scale).
In a multicenter prospective observational study in four Dutch pediatric ICUs (PICUs), patients aged ≥ 3 months and admitted for ≥ 48 h were assessed with the PD-scale thrice daily. Criterion validity was assessed: if the PD-scale score was ≥ 4, a child psychiatrist clinically assessed the presence or absence of PD according to the Diagnostic and statistical manual of mental disorders (DSM)-IV. In addition, the child psychiatrist assessed a randomly selected group to establish the false-negative rate. The construct validity was assessed by calculating the Pearson coefficient (rp) for correlation between the PD-scale and Cornell Assessment Pediatric Delirium (CAP-D) scores. Interrater reliability was determined by comparing paired nurse-researcher PD-scale assessments and calculating the intraclass correlation coefficient (ICC).
Four hundred eighty-five patients with a median age of 27.0 months (IQR 8–102) were included, of whom 48 patients were diagnosed with delirium by the child psychiatrist. The PD-scale had overall sensitivity of 92.3% and specificity of 96.5% compared to the psychiatrist diagnosis for a cutoff score ≥4 points. The rp between the PD-scale and the CAP-D was 0.89 (CI 95%, 0.82–0.93; p < 0.001). The ICC of 75 paired nurse-researcher observations was 0.99 (95% CI, 0.98–0.99).
The PD-scale has good reliability and validity for early screening of PD in critically ill children. It can be validly and reliably used by nurses to this aim.