The association of intraoperative hypotension (IOH) with postoperative morbi-mortality (1–3) suggests that intraoperative hemodynamic stability is a significant determinant of the postoperative outcome and a potential target for hemodynamic management during general anesthesia (4). However, current management strategies for this instability are often reactive, with therapeutical interventions typically initiated only when significant derangements in cardiovascular physiology have already occurred, such as low blood pressure or cardiac output (CO) (5).

There is no consensus on how to define hemodynamic instability (HI) (6–8). However, this term is commonly used to describe an abnormal cardiovascular condition characterized by one or several out-of-range macrohemodynamic variables that, if not addressed, may lead to organ dysfunction and ultimately shock (6). Thus, HI can be a clinical manifestation of shock, but shock may be masked by physiologic compensatory mechanisms or external therapeutic support, such as vasopressors. Indeed, shock is not exclusively defined by the presence of hypotension, as it arises from an imbalance between oxygen supply and demand in the tissues. However, when clinical evidence of HI is apparent, prompt intervention is required to avoid the development or worsening of organ hypoperfusion.

HI often results from three potential and not mutually exclusive pathophysiological mechanisms: hypovolemia, myocardiac contractility impairment, or vasoplegia (6). Similarly, the etiology of IOH is multifactorial and based on patient-specific and procedure-related factors. However, whatever the mechanism causing IOH, they all eventually lead to organ hypoperfusion (8). Therefore, early and adequate hemodynamic support is crucial to prevent and avoid worsening dysfunction and postoperative complications. Besides, determining the root cause(s) of instability according to the underlying pathophysiological mechanism is a prerequisite for appropriate therapy selection, which can be a real challenge without proper hemodynamic monitoring.

The Hypotension Prediction Index (HPI, Edwards Lifesciences in Irvine, CA, USA), is a recently introduced tool that predicts the occurrence of arterial hypotension while the patient is still hemodynamically stable. The HPI algorithm is based on a supervised machine learning algorithm from the arterial pressure waveform trained to predict the likelihood of arterial hypotension, defined as mean arterial pressure (MAP) < 65 mmHg during at least 1 min. HPI values range from 0 to 100, with higher values indicating a greater likelihood of hypotension. An HPI value >80 indicates that the patient will experience a hypotensive episode in the near future with a high sensitivity and specificity, i.e., it warns that the patient is no longer hemodynamically stable (9).

In this study, we tested the hypothesis that intraoperative HI results from different pathophysiological mechanisms in a cohort of patients undergoing elective major abdominal surgery using well-defined hemodynamic patterns based on known parameters of preload-dependency (stroke volume variation, SVV), left ventricular contractility (arterial dP/dtmax) and arterial system (Eadyn, dynamic arterial elastance).

In this observational study, we retrospectively analyzed prospectively collected anonymized data from six Spanish hospitals in 2022. The study was approved by the Ethics Committee of the Hospital Universitario Infanta Leonor, Madrid, Spain (2022). The requirement for written informed consent was waived. All participating centers signed a data release agreement and received a training program in managing intraoperative arterial hypotension with the HPI. This study adheres to the applicable Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines (10). A data analysis and statistical plan was written after the data were accessed.

A total of 393 patients were included, 34% of whom were women. The mean age was 64.7 (13) and the body mass index was 26.3 (4.2) kg/m². The median monitoring time per patient was 4.78 h (3.07–6.13 h). A total of 2038hours of monitoring time was evaluated, and after the screening and filtering data, 1946hours met the inclusion for the analysis, representing 95.5% of the initial dataset. A total of 2,058 HI episodes were identified, with a total cumulative duration of 148.5 h (7.62% of the analyzed time). 360 (91.6%) patients experienced at least one episode of intraoperative HI. The mean number episodes per patients was 5.23 (5.58) and the mean duration of an HI episode was 4.33 (9.72) minutes (Table 1).

The most common type of HI was vasoplegia, which accounted for 50.1% (1,032 episodes) of all HI episodes. Absolute hypovolemia was the second most frequent type, contributing 30.7% (632) of HI episodes. The least common type was low contractility with vasoplegia (Supplementary Figure 1).

Table 1 shows the hemodynamic characteristics of each predefined type of HI. The deviation of SVV, Eadyn and dP/dtmax concerning the mean values for each type is illustrated in Figure 3 using spider graphs, while box plots show normalized hemodynamic variables for each intraoperative HI type (Figures 3–7).

In this study, we defined hemodynamic instability as an HPI value > 80 and identified various causes of HI based on a therapeutical protocol on the physiological mechanisms explaining arterial hypotension. These types of HI are readily identifiable by clinicians at the bedside, potentially allowing for immediate corrective treatment depending on the primary cause of HI. The most common cause of HI was vasoplegia, followed by absolute and relative hypovolemia.

Hemodynamic instability is not a specific diagnosis but a final common pathway resulting from different hemodynamic mechanisms that can coexist or act independently. Vasoplegia was the most prevalent type of HI in our cohort. It is a clinical condition characterized by decreased systemic vascular resistance in the presence of a normal or increased CO, which can lead to low blood pressure and poor tissue perfusion (11). Hypotension due to vasodilation caused by administering general or neuraxial anesthesia, even when transient, has been associated with adverse outcomes (12). Saugel et al. demonstrated that post-induction hypotension results from decreased arterial tone rather than venous vasodilation or diminished myocardial contractility (13). Although vasoplegia is a well-recognized phenomenon, it lacks a standard clinical definition. We characterized HI due to vasoplegia as an SVV value < 13%, indicating non-preload dependency, and a dP/dtmax value > 400 mmHg/s, suggesting no myocardial depression. Consequently, even if SVRI values in this group were near the lower normal range, our definition of vasoplegia was primarily pragmatic and therapeutically targeted.We categorized hypovolemia as either absolute or relative. Absolute hypovolemia refers to the loss of blood, plasma, or water from the vascular compartment, which can be defined as a decrease in intravascular volume relative to a normal-sized vascular compartment. In contrast, relative hypovolemia implies a normal or even increased blood volume that is inadequate to fill the vascular compartment due to increased capacity (14). We define both situations in a clinical and pragmatic manner, as these two situations typically require different treatments. HI due to absolute hypovolemia results from an elevated SVV with an Eadyn > 1, while relative hypovolemia is characterized by an SVV > 13% with an Eadyn < 1. These definitions illustrate two types of hypovolemia, their underlying pathophysiology, and distinct therapeutic approaches. The lack of tachycardia during episodes of absolute hypovolemia raises important questions regarding the sensitivity of autonomic compensatory mechanisms to varying degrees of volume loss. This finding emphasizes the need for a more nuanced understanding of the cardiovascular response to hypovolemia, suggesting potential variability in physiological thresholds for the activation of these autonomic compensatory mechanisms.

Both vasoplegia and hypovolemia are common causes of HI during major abdominal surgery. To address this, various hemodynamic algorithms have been developed, such as goal-directed hemodynamic therapy (15, 16). The IMPRESS study found that optimizing stroke volume (SV), in conjunction with an individualized management strategy aimed at achieving a systolic blood pressure > 10% of the reference value, improved postoperative outcomes (12). However, the use of SV optimization carries a risk of volume overload (17). To mitigate this risk, an algorithm based on HPI could be employed. According to our definitions, fluid administration would be limited to situations of HI (HPI > 80), regardless of SVV or the increase in SV after fluid administration. Moreover, fluid administration would be restricted to pressure-dependent patients, defined as an Eadyn value > 1, since only those patients will experience improved blood pressure, even if CO increases (18).

One potential cause of instability in the operative setting is the negative inotropic effects of anesthetic agents (19). Peripheral arterial dP/dtmax has been shown to correlate with left ventricular end-systolic elastance, a load independent index of ventricular systolic function (20). Since arterial pressure results from the combined interaction of the ventricular ejection and the arterial system properties, other potential factors such as afterload or preload-dependency could also contribute to the peripheral dP/dtmax (21). We therefore defined HI due to low contractility when SVV is below 13%, reflecting a non-preload dependency condition. Nevertheless, contractility changes are the most prominent determinant contributing to the arterial dP/dtmax (20). Defining myocardial depression as reduced SVI and CI, and normal pulse pressure variation (PPV), Kouz et al. found that this was the most frequent cause of IOH during major abdominal surgery (7). In contrast, based on pre-established definitions in our sample, myocardial depression was the least frequent cause of HI that could lead to IOH. This discrepancy can be attributed to various factors: our definitions for different types of HPI were a priori established with a pre-emptive decision in mind and based on a previous therapeutic protocol, while in the study of Kouz et al., endotypes of intraoperative hypotension were arbitrarily labeled after identifying six potential clusters from the studied hemodynamic parameters (SVI, heart rate, CI, SVRI, and PPV). Notably, this hemodynamic assessment did not include dP/dtmax nor Eadyn. Furthermore, the hemodynamic factors determining IOH are likely time-dependent, suggesting that the determinants defining hemodynamic instability, which may eventually lead to arterial hypotension, do not necessarily have to be the same as when hypotension is already evolving. As our identification of different patterns of HI is based on an HPI > 80 value, clinician can potentially intervene earlier to the primary cause leading to IOH.

Our study has developed a classification system with five distinct patterns of HI based on the clinical evaluation of hemodynamic parameters that underlie arterial hypotension. This classification system provides a more nuanced approach to the assessment and management of HI and offers the potential for more personalized care by tailoring interventions to the main cause of HI.