Dynamic preload markers to predict fluid responsiveness during and after major gastrointestinal surgery: an observational substudy of the OPTIMISE trial.

BACKGROUND Stroke volume variation (SVV) and pulse pressure variation (PPV), termed dynamic markers of preload responsiveness, may predict the response to i.v. fluid in critically ill patients. However, the predictive accuracy of these variables during gastrointestinal surgery remains uncertain. METHODS Observational study of patients aged ≥50 yr undergoing major gastrointestinal surgery, enrolled in the OPTIMISE trial. Patients received six 250 ml fluid challenges with i.v. colloid solution (three during and three after surgery), while SVV and PPV were measured using the LiDCOrapid monitor (LiDCO Ltd, UK). Fluid responsiveness was defined as a stroke volume increase ≥10%. Area under the receiver operating characteristic curve was calculated with 95% confidence intervals. Adjustment for covariates was performed by regression modelling and a clustering method was used to adjust for intra-patient correlation. RESULTS One hundred patients were recruited between August 2010 and October 2012. Five hundred and fifty-six fluid challenges were administered and 159 (28.6%) were associated with increased stroke volume. The predictive value of both variables was poor during surgery [SVV 0.69 (0.63-0.77); PPV 0.70 (0.62-0.77)], and also after surgery [SVV 0.69 (0.63-0.78); PPV 0.64 (0.56-0.73)]. The findings were similar when analysed according to whether patients were mechanically ventilated [SVV 0.68 (0.63-0.77); PPV 0.69 (0.61-0.77)] or breathing spontaneously [SVV 0.69 (0.61-0.77); PPV 0.63 (0.56-0.72)]. Predictive value improved slightly in a sensitivity analysis excluding outlier values of SVV and PPV. CONCLUSIONS In this study, the predictive accuracy of SVV and PPV for fluid responsiveness was insufficient to recommend for routine clinical use during or after major gastrointestinal surgery.