PROSPECTIVE VALIDATION OF AN ACUTE RESPIRATORY-DISTRESS SYNDROME PREDICTIVE SCORE

We derived an Acute Respiratory Distress Syndrome Score (ARDS Score) f rom previously described training set data. To validate its diagnostic accuracy for identifying a complicated course (early death or prolong ed intubation) in acute lung injury, 50 patients were prospectively sc ored using an ARDS Score decision threshold of greater than or equal t o 2.5 to discriminate between an uncomplicated (successful extubation after less than or equal to 14 d) and complicated course. Predictor fa ctors incorporated in the ARDS Score were collected on Day 4 and Day 7 of ARDS and included Pa-O2/PA(O2). ratio, required positive end-expir atory pressure (PEEP), and chest radiograph progression. The diagnosti c accuracy of the ARDS Score for determining a complicated course as w ell as overall survival was compared with three other available indice s. Using receiver operating characteristic (ROC) analysis, the ARDS Sc ore and Lung Injury Score (LIS) had the greatest diagnostic accuracy f or determining a complicated course, but the Simplified Acute Physiolo gy Score (SAPS Score) (score greater than or equal to 14) more accurat ely identified survival. The LIS components of static respiratory syst em compliance (C-rs) and chest radiograph description did not differ b etween patient groups. The interobserver concordance of the dynamic ch est radiograph interpretation included in the ARDS Score was significa nt (p < 0.05). We conclude that the previously derived ARDS Score has valid diagnostic accuracy for identifying patients with ARDS who will follow a complicated course.