Abstract
Background: We investigated the accuracy of predicting preload responsiveness by means of a passive leg raising test (PLR) using the perfusion index (PI) in critically ill patients showing signs of hypoperfusion in a resource-limited setting. Methods: We carried out a prospective observational single center study in patients admitted for sepsis or severe malaria with signs of hypoperfusion in Chattogram, Bangladesh. A PLR was performed at baseline, and at 6, 24, 48, and 72 h. Preload responsiveness assessed through PI was compared to preload responsiveness assessed through cardiac index (CI change ≥5%), as reference test. The primary endpoint was the accuracy of preload responsiveness prediction of PLR using PI at baseline; secondary endpoints were the accuracies at 6, 24, 48, and 72 h. Receiver operating characteristic (ROC) curves were constructed. Results: The study included 34 patients admitted for sepsis with signs of hypoperfusion and 10 patients admitted for severe malaria. Of 168 PLR tests performed, 143 had reliable PI measurements (85%). The best identified PI change cutoff to discriminate responders from non-responders was 9.7%. The accuracy of PLR using PI in discriminating a preload responsive patient at baseline was good (area under the ROC 0.87 95% CI 0.75-0.99). The test showed high sensitivity and negative predictive value, with comparably lower specificity and positive predictive value. Compared to baseline, the AUROC of PLR using PI was lower at 6, 24, 48, and 72 h. Restricting the analysis to sepsis patients did not change the findings. Conclusions: In patients with sepsis or severe malaria and signs of hypoperfusion, changes in PI after a PLR test detected preload responsiveness. The diagnostic accuracy was better when PI changes were measured at baseline.