Aims
To evaluate the diagnostic characteristics and accuracy of the 3D-CAM-TH used by medical students.
Methods
A cross-sectional study was conducted in 2024 at a single tertiary centre, including 100 patients aged ≥70 years from medicine wards. Delirium was assessed using the 3D-CAM-TH: 20 cases by a medical student and 80 cases by two trained nurses, and validated against the DSM-5 criteria as the reference standard by a geriatrician. Diagnostic performance was assessed through sensitivity, specificity, accuracy, positive predictive value (PPV), negative predictive value (NPV), positive likelihood ratio (PLR), and negative likelihood ratio (NLR).
Results
The mean participant age was 80.3 ± 6.94 years, with a majority being female. Delirium prevalence was 39%. When a medical student assessed the 3D-CAM-TH, it showed a sensitivity of 87.5%, specificity of 83.3%, and accuracy of 85%. PPV and NPV were 77.78% and 90.90%, respectively, with a PLR of 5.25 and an NLR of 0.15. When used by nurses, sensitivity was 88.9%, specificity 94.3%, and accuracy 92.5%. The inter-rater reliability (Cohen’s Kappa = 0.86) indicated excellent agreement between assessors.
Conclusions
3D-CAM-TH demonstrated strong diagnostic performance for delirium detection, even when used by medical students with limited clinical experience. Its high sensitivity supports its utility as a screening tool, while excellent inter-rater reliability indicates consistency across different healthcare providers. Although specificity among medical students was slightly lower than that of nurses, overall accuracy remained robust after brief training. These findings support the broader application of 3D-CAM-TH in clinical practice to enhance early delirium detection.