Yang Liu, MD
Intensivist
Nanjing Gaochun People's Hospital
Nanjing, China
Disclosure information not submitted.
xiaoshi li
Critical Care Intensivist
Nanjing Gaochun People's Hospital
Nanjing, Jiangsu, China (People's Republic)
Disclosure information not submitted.
xin zhang
Critical Care Intensivist
Nanjing Gaochun People's Hospital, United States
Disclosure information not submitted.
guoping zhou
Critical Care Intensivist
Nanjing Gaochun People's Hospital, United States
Disclosure information not submitted.
Title: False Positives of Pulse Pressure Variation in Spontaneously Breathing Patients
INTRODUCTION/HYPOTHESIS: Pulse pressure variation (PPVDI) obtained during deep inspiration maneuvers (DIM) has been shown to improve the prediction of fluid responsiveness in spontaneously breathing subjects. Given a positive influence of tidal volume (Vt) on PPV, we hypothesized that the DIM-induced increase in Vt would cause significant PPV even in patients not responding to volume expansion (False positive, FP). The aims of the study were to investigate the association between Vt and FP PPVDI results, and to test whether the indexation of PPVDI to Vt could avoid some of the FPs observed by using PPVDI alone.
Methods: Patients with sepsis or septic shock who were breathing spontaneously without mechanical ventilation and were considered for volume expansion (VE) were prospectively enrolled. VE was performed by infusing 500 mL saline over 20 minutes. Prior to VE, measurements including PPV and Vt were obtained during quiet spontaneous breathing (PPVTB and VtTB, respectively) and during the deep inspiration maneuver (PPVDI and VtDI, respectively). The absolute DIM-induced changes in PPV (ΔPPV = PPVDI - PPVTB) and Vt (ΔVt = VtDI - VtTB) were also calculated. Patients were classified as responders if stroke volume (SV) increased ≥ 15% after VE.
Results: Of the 33 patients included, VE led to an increased SV ≥ 15% in 16 (responders). The area under the receiver operating characteristic curve (AUC) of PPVDI was significantly greater than that of PPVTB (0.825 vs. 0.586, P = 0.05). PPVDI more than 12% yielded a sensitivity of 76.4% (95% confidence interval [CI], 50.1%-93.2%) and a specificity of 56.25% (30%-80.2%), respectively. ΔPPV was positively correlated with ΔVt (R2 =0.24, P = 0.004). VtDI (19.6 ± 1.8 vs 14.4 ± 4.6 ml/kg; P =0.01) was significantly higher in 7 nonresponders with a PPVDI > 12% (FP) than in 9 other nonresponders with a PPVDI ≤ 12% (true negative). Normalizing PPVDI by VtDI led to major improvement in predictive performance (AUC 0.96; 95% CI, 0.83-1.0; P=0.03 compared to PPVDI) by diminishing FP rate (43.75% [19.8%-70.1%] vs. 6.25% [0.2%-30.2%]; P=0.02).
Conclusions: In spontaneously breathing patients with sepsis, FP values of PPVDI may occur in the context of high Vt induced by DIM, and can be avoided in the majority of cases by adjusting PPVDI for Vt.