Abstract We recalibrated a risk score to identify difficult-to-treat resistant Pseudomonas aeruginosa among 197 ICU patients with bloodstream or respiratory infection. Prevalence was 6%. Models showed moderate discrimination (c-statistics 0.71–0.74), low sensitivity (18%), and high specificity (94%–96%), highlighting the need for local validation and consideration of performance tradeoffs.
Le et al. (Tue,) studied this question.