Advanced Hemodynamic Monitoring: Are We Asking the Right Questions?

“Does ‘advanced’ intra- or perioperative hemodynamic monitoring improve patient outcomes?”Image: J. P. Rathmell.Does “advanced” intra- or perioperative hemodynamic monitoring improve patient outcomes? In this issue of Anesthesiology, Ripollés-Melchor et al. report a multicenter randomized trial comparing new monitoring technology with “standard care.”1 In a non–industry-sponsored trial enrolling 917 patients undergoing moderate- to high-risk abdominal surgery, the authors compared a hemodynamic management bundle “triggered” by a proprietary machine learning algorithm (the Hypotension Prediction Index; HPI), a key component of an advanced hemodynamic monitoring interface (Edwards Lifesciences’ Acumen IQ system, USA), versus “usual” hemodynamic monitoring without the HPI/Acumen IQ system across 28 hospitals (primarily in Spain). The primary outcome was moderate-to-severe acute kidney injury, ascertained per international consensus guidelines. This trial did not demonstrate a significant decrease in acute kidney injury (22.2% in the HPI-triggered group compared to 25.6% in controls [odds ratio, 0.85; 95% CI, 0.65 to 1.11]) or secondary outcomes (overall complications, length of stay, renal replacement therapy, mortality) with HPI-triggered management Ensuring adequate organ perfusion during surgery is a fundamental role for anesthesiologists. Accordingly, biomedically oriented engineers (both physician and nonphysician) have developed numerous advanced hemodynamic monitors, beginning with invasive hemodynamic approaches, and more recently, a host of noninvasive approaches facilitated by advances in on-board computing and digital signal processing. While many advanced monitors were originally standalone devices, vendors have more recently moved to integrated displays in a dashboard format. Such displays appear to be using systematic approaches to detecting and treating hemodynamic disturbances. Edwards Lifesciences introduced the HPI into its monitoring suite several years ago along with indices of fluid responsiveness, arterial elastance, and contractility based on pulse contour waveform analysis measurements obtained via invasive (intra-arterial), and more recently, a noninvasive finger cuff sensor. Using this suite, some believe that the clinician may gain further insight into acute changes in cardiovascular function. However, questions have been raised regarding its efficacy, given some inherent limitations in physiologic assumptions made as well as by use of a proprietary “black box” approach in development and validation of the HPI.2 Clinical studies using this suite have been heterogeneous and have reached differing conclusions on its efficacy in reducing intraoperative hypotension. Some studies report robust decreases in the frequency of hypotension or time-weighted average below a certain preset threshold (usually mean arterial pressure less than 65 mmHg), whereas others have not.3,4 However, far fewer studies have explored the crucial next step in understanding whether approaches using the HPI provide clinical benefits.5 At its core, a seemingly simple question as to whether such monitors improve patient outcomes is challenged by a cascade of questions that must be addressed to precisely characterize conditions under which use of such monitors may provide added value. Such questions include (but are not limited to) the following: (1) In what population might an advanced hemodynamic monitor be beneficial? (2) What interventions should be triggered by certain values provided by the monitor? (3) To what extent should interventions be tempered by clinical conditions “unseen” by the monitor? (4) To what other monitors or processes of care should they be compared? and (5) What outcomes should be considered beneficial? Ripollés-Melchor et al. shed light on some of the key questions noted above, while herein we address crucial questions that remain in understanding potential clinical benefits of the HPI. After development and publication of initial validation data of the HPI (the controversies related to this have been extensively explored in the Journal and are not a focus of this editorial),2,6–9 understanding whether a given intervention such as an advanced hemodynamic monitoring strategy will result in improved outcomes requires estimating the degree to which outcomes are improved in the presence, compared to the absence, of the intervention. Randomized controlled trials represent the least biased method to estimate potential benefits, as assignment to the intervention is not influenced by characteristics that may also be associated with outcomes under study. However, randomized trials are expensive and typically take years to complete, and even then, results often do not directly generalize to all patients eligible to receive an intervention. Therefore, clinician researchers, in partnership with patients, clinicians, scientists, and policy makers, need to ensure that such trials are designed to answer the right questions to avoid research waste and effectively advance patient care. While many frameworks exist to describe the development and evaluation of interventions, a useful construct to conceptualize the evaluation of a technology like the HPI is the continuum between explanatory and pragmatic trials first described in 1967,10 and more recently operationalized as the PRECIS (PRagmatic Explanatory Continuum Indicator Summary)-2 tool.11 Specifically, trials are understood to exist in a space between a completely explanatory design and a completely pragmatic design. In an explanatory trial, all aspects of trial conduct, intervention delivery, control conditions, and data collection are tightly controlled and protocolized with groups differing only based on assignment to intervention versus control. The objective of an explanatory trial is to estimate changes in outcomes that are mechanistic, proximal, and specific to the controlled experimental condition. In contrast, a pragmatic trial only protocolizes assignment to the intervention versus control, with an objective to estimate changes in outcomes that are patient prioritized, often routinely collected, and generalizable to real-world settings. Where the current study sits on this continuum versus other hypothetical trials is shown in figure 1. Ultimately, most trials should not be fully explanatory or fully pragmatic. Instead, an optimally specified trial should reflect the balance between the need to understand postulated mechanistic pathways, as explored by explanatory components of trial design, versus a need to understand the effectiveness of an intervention to improve outcomes, when applied to settings consistent with real-world care, as explored by pragmatic components of trial design.Fig. 1.: Contexualization of the PRECIS-2 (PRagmatic Explanatory Continuum Indicator Summary-2) tool to the trials assessing the potential benefits of advanced hemodynamic monitoring devices on patient outcomes. HPI, Hypotension Prediction Index [Edwards Lifesciences, USA]; MAP, mean arterial pressure.For the HPI, the postulated causal pathway leading to improved patient outcomes is that earlier, accurate notifications of impending hypotension lead to earlier preventative actions by the anesthesiologist when compared to usual care. In turn, preventative actions will reduce the severity or duration of hypotension, and—to the extent that blood pressure is a causal mediator of end-organ perfusion—will mitigate complications related to organ hypoperfusion such as acute kidney injury.12,13 On the other hand, for the HPI, a practical healthcare delivery question must consider the nuanced and variable workflows of perioperative care. Can the HPI (or other alternatives) alter current clinical practice sufficiently to improve outcomes valued by patients and health systems? Considering the trial by Ripollés-Melchor et al., their report provides insights into steps along the proposed hypotension causal pathway, yields estimates relevant to everyday clinical care, and also highlights key gaps that remain in deriving a full understanding of whether a tool like the HPI should be incorporated into routine care. First, consider the pragmatic “usual care” comparator in this trial. Whereas trial enrollment was restricted to centers with clinicians experienced in using the Acumen IQ system (an explanatory component mitigating bias related to clinician familiarity with such devices), clinicians were allowed to manage control patients’ hemodynamics according to “local practice.” However, this pragmatic choice meant that hemodynamic management varied substantially within the control cohort, while also leading to important departures in care relative to the intervention group. For example, while use of the HPI and Acumen IQ sensor suite was strictly prohibited for control participants, 46% of “usual care” group patients still received other forms of advanced hemodynamic monitors (e.g., the Hemosphere system paired with a FloTrac sensor [both Edwards Lifesciences] or comparable monitors). Control participants’ clinicians could also respond to hemodynamic indicators as they saw fit, including the choice of whether to treat with fluid and/or the use of vasoactive medication(s). This relatively pragmatic approach could help results to generalize to centers like those participating in the trial (i.e., those experienced in routinely using advanced hemodynamic monitors). However, this approach concurrently limits our ability to untangle the mechanistic role of the HPI itself in improving outcomes, versus differences in hemodynamic treatment strategies, which, as described in the next paragraph, were materially different between the HPI-guided intervention arm and the usual care control group. In contrast to the pragmatic approach to monitoring and management of control group patients, the trial’s intervention group care was explanatory. Specifically, the when (if an HPI value greater than 80 was identified), how (fluid versus vasoactive medication based on an algorithmic approach informed by hemodynamic parameters provided by the Acumen IQ system [e.g., stroke volume variation, contractility, and systemic vascular resistance]), and what (vasoactive medications were limited to ephedrine bolus or norepinephrine infusion, with phenylephrine banned in the intervention bundle) used to manage potential hypotension were highly protocolized. Thus, the true intervention under study was a bundle of care combining the HPI value, other advanced hemodynamic parameters, and specific treatment strategies rather than solely the HPI. This explanatory choice meant that substantial variations in management emerged between treatment arms. For example, intervention patients were 12% more likely to receive vasoactive therapy than controls. Whereas intervention group patients were prohibited from receiving phenylephrine, 24% of control patients received phenylephrine as a vasoactive treatment. In this way, any attempt to draw inferences about HPI-guided versus non–HPI-guided care could be strongly influenced by the comparative effects of phenylephrine (standard care group) versus norepinephrine and ephedrine (HPI group), or treatment of hypotension with a more fluid-favoring (control group) versus vasoactive-favoring (HPI group) strategy. This approach to intervention design—which could be useful for an explanatory trial solely comparing the effects of specific blood pressure treatment strategies on outcomes, independent of hemodynamic monitoring strategies used—makes it difficult to separate the role of the monitor from the treatment protocol. A more explanatory approach where use of the HPI is directly compared to intra-arterial monitoring, with drugs and fluids standardized across arms, may be required to provide more direct causal insights into the HPI’s hypothesized benefits. Or conversely, if treatment strategies in both arms were to follow a nonprotocolized pragmatic approach, the effect of HPI interpretation could be isolated by providing anesthesiologists clinical scenarios (e.g., postinduction of general anesthesia, ongoing blood loss, versus an otherwise seemingly healthy patient) and studying their individually preferred treatment responses, based on HPI values accompanied by measures of stroke volume variation, contractility, and systemic vascular resistance. These potential strategies are illustrated in figure 1. Last, consider the trial’s primary outcome of moderate-to-severe acute kidney injury (AKI). Given the state of evidence regarding the HPI’s clinical efficacy, AKI is a clinically relevant outcome, is appropriately positioned between an explanatory (e.g., incidence or time with hypotension, which is a mechanistic outcome that has been previously studied) and pragmatic (e.g., patient-reported recovery or days alive at home, which may be too distal given the current state of evidence) measure, and causally should be sensitive to the quality of hemodynamic management. However, for those who remain unconvinced by available heterogeneous data that use of the HPI truly decreases hypotension in a clinically meaningful manner, a more proximal explanatory outcome that directly captures hemodynamic parameters may still be required. For the trial by Ripollés-Melchor et al., an unfortunate limitation was the lack of recording or reporting of intraoperative blood pressure data, which would have offered mechanistic insights into the effects of each treatment intervention along the postulated hypotension causal pathway. Such data would provide greater insights into whether the lack of significant difference in AKI occurred because of a lack of hemodynamic separation between groups, or despite such separation. Limitations notwithstanding, providing evidence-based perioperative care requires conduct and reporting of large, multicenter trials—an admirable and often herculean task. Ripollés-Melchor et al. should be congratulated for their work aiming to identify an improvement in a clinically meaningful outcome based on differing hemodynamic management strategies. As the trial’s results reflect a comparison between a hemodynamic management bundle that included the HPI with a heterogeneous control condition whose care may not be considered “usual” at all centers, along with a lack of quantitative hemodynamic data to understand mechanistic pathways, widely generalizable conclusions regarding the impact of an HPI “guided” protocol remain uncertain. Fortunately, anesthesiologist scientists are well positioned to answer such questions with finely tuned explanatory trials designed to first provide mechanistic insights into the HPI’s impact on intraoperative hemodynamics, followed by pragmatic designs targeted at improving patient- and system-centered outcomes in real-world care. Competing Interests Dr. McIsaac receives salary support from The Ottawa Hospital Anesthesia Alternate Funds Association (Ottawa, Ontario, Canada) and a Clinical Research Chair from the University of Ottawa Faculty of Medicine (Ottawa, Ontario, Canada). Dr. Mathis has received research grant support from the U.S. National Institutes of Health (National Heart, Lung, and Blood Institute and National Institute of Diabetes and Digestive and Kidney Diseases, Bethesda, Maryland) and industry-sponsored research (Chiesi USA, Cary, North Carolina) paid to his institution, unrelated to the present work. Dr. London is not supported by, nor maintains any financial interest in, any commercial activity that may be associated with the topic of this article. Dr. London is an Editor for Anesthesiology.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.