Paired Comparison of SVV and PVI Accuracy

Background and rationale. Goal-directed fluid therapy during laparoscopic major abdominal surgery remains variable because dynamic preload indices perform inconsistently under pneumoperitoneum and positional changes. Stroke Volume Variation (SVV, derived from the arterial pressure waveform) and the Pleth Variability Index (PVI, derived from the pulse oximeter) are both widely available and non-drug, non-investigational monitors. A head-to-head, paired accuracy study using standardized fluid challenges can clarify which index better identifies fluid responsiveness in this setting.

Objectives. Primary: Compare the diagnostic accuracy (area under the ROC curve, AUROC) of SVV vs PVI for predicting fluid responsiveness to a small crystalloid bolus during laparoscopic major abdominal surgery.

Key secondary objectives: (1) determine optimal decision thresholds and report sensitivity, specificity, predictive values, and likelihood ratios; (2) compare calibration and classification performance (Youden index, net reclassification); (3) assess effect modification by pneumoperitoneum status, body mass index, ventilation parameters, and vasoactive use; (4) describe peri-bolus hemodynamic changes.

Design and setting. Prospective, single-center, paired diagnostic-accuracy cohort. Adults undergoing elective laparoscopic major abdominal surgery under general anesthesia are enrolled. Each participant undergoes up to two standardized 250-mL crystalloid fluid challenges delivered over ~3 minutes at two time points when feasible: (A) before insufflation and (B) after establishment of pneumoperitoneum at a hemodynamic steady state. Because both indices are recorded around the same fluid challenges, each participant serves as his/her own control.

Index tests. SVV (from the arterial pressure monitor) and PVI (from the pulse-oximetry monitor) are captured immediately before ("pre-bolus") and at ~3 minutes after each bolus. Index values are recorded by trained research staff and kept masked from anesthesia clinicians to avoid influencing usual care.

Reference standard and definition of fluid responsiveness. The reference standard is change in stroke volume index (SVI) or cardiac index derived from arterial pulse contour analysis. Fluid responsiveness is defined a priori as a ≥10% increase in SVI after the 250-mL bolus. Sensitivity analyses will use a 15% threshold and will repeat analyses with cardiac index. If signal quality is inadequate or a clinical safety concern arises, the test is aborted and data flagged.

Anesthesia and peri-bolus standardization. Ventilation is volume-controlled with tidal volume ~6-8 mL/kg predicted body weight and PEEP according to institutional practice; settings and pneumoperitoneum pressure/position are recorded. Vasoactive drugs and additional fluids are allowed as clinically indicated but are time-stamped; fluid challenges for research are not given when active titration of vasoactives makes a stable baseline impossible.

Participants. Inclusion: adults (≥18 years), ASA I-III, elective laparoscopic major abdominal surgery, arterial line in place, informed consent. Exclusion: clinically significant arrhythmia (e.g., atrial fibrillation with irregular R-R), severe valvular disease, intracardiac shunts, pregnancy, severe right-heart failure, inability to obtain reliable arterial or plethysmographic waveforms, or any condition in which a fluid challenge is judged unsafe by the anesthesiologist.

Outcomes. Primary outcome: AUROC of SVV vs PVI for predicting fluid responsiveness. Secondary outcomes: sensitivity/specificity at prespecified thresholds (e.g., SVV 12-13%, PVI 13-15%), decision-curve analysis, subgroup AUROC by pneumoperitoneum status, and immediate hemodynamic responses (MAP, HR).

Sample size and power. The study plans to enroll ~300 adults, yielding up to ~600 evaluable fluid challenges. Assuming a fluid-responsive rate of ~40%, this sample provides >80% power to detect a paired AUROC difference of ~0.08-0.10 between SVV and PVI at α=0.05 using DeLong's test with clustering by subject.

Statistical analysis. ROC curves for SVV and PVI are constructed using clustered bootstrap resampling at the patient level (≥2000 replicates). AUROCs are compared with paired DeLong tests. Threshold performance metrics include exact 95% CIs with patient-clustered bootstrap. Mixed-effects logistic regression (random intercept for patient) will estimate adjusted odds of fluid responsiveness per unit change of each index; prespecified covariates include pneumoperitoneum status, BMI, tidal volume per kg, PEEP, and vasoactive use. Missing index or reference values are minimized; if >5% are missing, multiple imputation will be performed in sensitivity analyses. All analyses follow a prespecified statistical analysis plan.

Safety and risk/benefit. Risks relate to small fluid boluses (transient blood pressure or heart-rate changes, fluid accumulation in vulnerable patients). The bolus is withheld or stopped for any safety concern. No experimental drugs or investigational devices are used; all monitors are standard of care. Participants are unlikely to benefit directly, but the results may improve future intraoperative fluid management.

Confidentiality and data handling. Data are captured on a secure electronic case-report form with coded identifiers. Only the research team has access. Results will be presented in aggregate without personal identifiers and disseminated in peer-reviewed venues.

Regulatory status. This is not an FDA-regulated drug or device investigation and is conducted under institutional review board approval at a single academic hospital in the Republic of Korea.