Diagnostic Accuracy and Preliminary Psychometric Validation of the Spanish Stanford Proxy Test for Delirium in Non-intubated Mexican ICU and Intermediate Care Patients

AI Summary

S-PTD (cutoff ≥3) showed strong diagnostic accuracy: AUC 0.87, sensitivity 81%, specificity 89.9%, overall accuracy 88.3%.
High internal consistency (Cronbach's alpha 0.948) and exploratory factor analysis supported a predominantly unidimensional construct (first factor 64.2% variance).
Limitations include single-site design, exclusion of intubated or deeply sedated patients, limited delirium cases and no inter-rater reliability; multicentre validation required.

TL;DR (AI Generated)

OBJECTIVE: To evaluate the diagnostic performance, internal consistency, and construct validity of the Spanish Stanford Proxy Test for Delirium (S-PTD) in non-intubated critically ill Mexican patients admitted to intensive and intermediate care units.

DSM-5 delirium was identified in 21 patients (17.5%), whereas 27 patients (22.5%) screened positive using the S-PTD cutoff of ≥3.

Findings should be interpreted cautiously because of the single-site design, limited number of delirium cases, exclusion of mechanically ventilated and moderately/deeply sedated patients, absence of formal inter-rater reliability testing, and the small number of patients with documented cognitive impairment or dementia.

Basic summary

J Acad Consult Liaison Psychiatry. 2026 Sep 19:S2667-2960(26)00102-3. doi: 10.1016/j.jaclp.2026.09.004. Online ahead of print.

ABSTRACT

BASIC PROCEDURES: We conducted a single-site cross-sectional psychometric study in adult patients admitted to intensive care and intermediate care units within the first 48 hours of admission. Trained nursing staff administered the Spanish S-PTD at the end of the clinical shift, and results were compared with an independent delirium assessment performed by consultation-liaison psychiatrists using DSM-5 criteria. Patients with endotracheal intubation or Richmond Agitation-Sedation Scale (RASS) scores ≤ -3 were excluded by protocol. Diagnostic accuracy, receiver operating characteristic curve performance, internal consistency, and exploratory factor structure were examined.

MAIN FINDINGS: A total of 120 patients were analyzed. Mean age was 62.5 ± 18.1 years, 64 (53.3%) were men, and 87 (72.5%) were admitted for medical conditions. DSM-5 delirium was identified in 21 patients (17.5%), whereas 27 patients (22.5%) screened positive using the S-PTD cutoff of ≥3. At this cutoff, the S-PTD demonstrated an AUC of 0.87, accuracy of 88.3% (95% CI, 81.4%-92.9%), sensitivity of 81.0% (95% CI, 60.0%-92.3%), specificity of 89.9% (95% CI, 82.4%-94.4%), positive predictive value of 63.0% (95% CI, 44.2%-78.5%), and negative predictive value of 95.7% (95% CI, 89.5%-98.3%). Internal consistency was high (Cronbach’s α = 0.948). Exploratory factor analysis provided preliminary evidence supporting construct validity and a predominantly unidimensional structure (KMO = 0.84; Bartlett’s test p < .001). The first factor had an eigenvalue of 7.71 and explained 64.2% of the variance, whereas the second eigenvalue was 0.97.

PRINCIPAL CONCLUSIONS: In this non-intubated ICU/IMCU Mexican cohort, the Spanish S-PTD demonstrated promising diagnostic accuracy and psychometric properties when administered by trained nursing staff. Findings should be interpreted cautiously because of the single-site design, limited number of delirium cases, exclusion of mechanically ventilated and moderately/deeply sedated patients, absence of formal inter-rater reliability testing, and the small number of patients with documented cognitive impairment or dementia. Further multicenter studies, including mechanically ventilated patients, are needed before broad implementation in critically ill populations.

PMID:42763037 | DOI:10.1016/j.jaclp.2026.09.004

Document this CPD