Peripheral perfusion index combined with end-tidal carbon dioxide in predicting sustained return of spontaneous circulation in adults with in-hospital cardiac arrest: A prospective observational study

OBJECTIVE: To evaluate the associations of peripheral perfusion index (PI) and end-tidal carbon dioxide (PETCO2), individually and in combination, with sustained return of spontaneous circulation (ROSC) in adults with in-hospital cardiac arrest, and to assess their predictive performance. METHODS: This prospective observational study included 125 adults with in-hospital cardiac arrest, of whom 62 achieved sustained ROSC and 63 did not. Measurements obtained after ROSC or termination of resuscitation were excluded. Changes in PI and PETCO2 during resuscitation were analyzed using linear mixed-effects models. Event-average PI and PETCO2 were evaluated for their ability to predict sustained ROSC using receiver operating characteristic (ROC) analysis. The combined model was internally validated using 2,000 bootstrap resamples. RESULTS: The median time to ROSC was 11 min (IQR, 5-20). Significant group-by-time interactions were observed for both PI and PETCO2 (both P < 0.001). Event-average PI was higher in the ROSC group than in the non-ROSC group [0.62% (IQR, 0.42-0.90) vs. 0.37% (0.31-0.52)], as was event-average PETCO2 [15.5 mmHg (12.2-17.8) vs. 12.0 mmHg (10.4-13.7)] (both P < 0.001). The areas under the ROC curve (AUCs) were 0.756 (95% CI, 0.670-0.842) for PI, 0.753 (95% CI, 0.666-0.839) for PETCO2, and 0.808 (95% CI, 0.731-0.886) for their combination. The optimism-corrected AUC for the combined model was 0.800. Pairwise comparisons showed no significant differences among the AUCs. In the exploratory fully adjusted model, PETCO2 remained independently associated with sustained ROSC (adjusted OR, 1.626; 95% CI, 1.148-2.302; P = 0.006), whereas PI was not statistically significant (adjusted OR, 1.359 per 0.1% increase; 95% CI, 0.962-1.920; P = 0.082). CONCLUSION: Higher PI and PETCO2 during resuscitation were associated with sustained ROSC. Combining PI with PETCO2 yielded numerically better discrimination than either measure alone, although the differences in AUC were not statistically significant. External validation is needed before the combined model can be applied in clinical practice.