Trending Ability and Accuracy of Minimally Invasive Pulse Wave Analysis Devices: A Systematic Review and Meta-Analysis.

Abstract

Minimally invasive pulse wave analysis devices are widely used for perioperative hemodynamic monitoring, but their accuracy and ability to track changes in cardiac output remain controversial. Previous reviews mainly addressed accuracy, whereas trending ability has been reported inconsistently. We conducted a systematic review and meta-analysis of studies comparing minimally invasive pulse wave analysis devices with reference techniques. Methodological quality was assessed through a predefined fitness appraisal (simultaneity, homoscedasticity, absence of systematic trend, independence). Random-effects models (metafor, R 4.3.2) were used with log and logit transformations as appropriate. We included 50 studies; 49 provided accuracy/precision data and all 50 provided trending ability data, yielding 69 and 70 lines of analysis, respectively. Subgroup analyses were performed by device, clinical setting, and conflict of interest. Across all devices and settings, pooled bias was 0.09 L/min (95% confidence interval [CI], -0.05 to 0.23) and pooled percentage error was 48.3% (unweighted mean 49.0%, 95% CI, 45.8-52.2). Global trending analyses showed pooled four-quadrant concordance of 72.6% (95% CI, 67.9-77.4) and polar plot ±30° agreement of 63.2% (95% CI, 54.9-71.5). Heterogeneity was extreme (I2 >95% throughout). Subgroup analyses showed lower percentage error in intensive care unit (ICU) (≈42%) than in noncardiac surgery (≈55%) and more favorable results in studies with declared conflict of interest (≈44% vs 51%, P < .05). MostCare and Argos achieved four-quadrant concordance >90% and polar agreement close to accepted thresholds, but these estimates were based on few studies. Sensitivity analyses restricted to high-quality datasets yielded results very similar to the overall estimates. Minimally invasive pulse wave analysis devices do not achieve sufficient accuracy or trending ability to be interchangeable with reference techniques. Extreme heterogeneity persists across devices, clinical settings, and methodological quality. Differences associated with conflicts of interest and between ICU and noncardiac surgery populations suggest that performance is context-dependent and influenced by study sponsorship. Taken together with recent evidence and new methodological standards for cardiac output validation, our findings underscore the need for more transparent algorithms, rigorous reporting of trending ability, and standardized validation frameworks before these technologies can be considered reliable for guiding hemodynamic management.