Background
According to international consensus definitions, acute kidney injury (AKI) concerns 30 to 40% of patients admitted to intensive care unit (ICU) [
1‐
3]. Renal replacement therapy (RRT) is required in 5 to 20% of patients [
1,
2] and is associated with high mortality [
2,
4]. Despite the improvement of intermittent hemodialysis management [
5,
6], intradialytic hemodynamic instability (IHI) remains a common issue [
5,
7‐
12] that could account for increased mortality and delayed renal recovery. Optimal timing to initiate RRT in critically ill patients is still uncertain [
4,
13,
14]. Apart from life-threatening complications, an international survey reported that indications for initiating RRT varied widely among intensivists [
15]. Recent interventional trials demonstrated that delaying RRT in the absence of life-threatening complications of AKI does not impair survival and allows some patients to avoid RRT [
4,
13]. Determination of risk factors for IHI could help clinicians to identify patients in whom initiation of RRT should be reconsidered in the absence of emergent criteria. However, factors associated with IHI remain poorly explored [
8‐
12]. The pathophysiology of IHI is complex with a reduction of venous return and also an alteration of endothelial function that limits vascular tone adaptation to intermittent hemodialysis-induced hypovolemia [
12,
16‐
18]. Pre-existing endothelial dysfunction, commonly observed in ICU patients [
19‐
23], might worsen the negative hemodynamic impact of intermittent hemodialysis.
The evaluation of endothelial dysfunction at the bedside is very challenging, but its consequences on tissue hypoperfusion could be assessed more easily [
24‐
26]. The prognosis value of lactate level and lactate clearance has been largely demonstrated [
27,
28] and a therapeutic strategy targeting lactate normalization is recommended during sepsis and septic shock [
29]. Capillary refill time (CRT) is an easy-to-use bedside [
30] marker of peripheral perfusion. Its relevance has been demonstrated in triage of patients in pre-hospital setting [
31] and in the emergency department [
32,
33]. CRT correlates with severity of organ failure and is strongly associated with prognosis during septic shock [
34]. Moreover, a recent multicenter study underlined the interest of a resuscitation strategy based on CRT monitoring in septic shock patients [
35].
We hypothesized that critically ill patients with altered hemodynamics and/or impaired tissue perfusion might be more prone to develop IHI. In this prospective study, we tested whether cardiovascular SOFA score, index CRT and lactate level could be predictive of IHI requiring therapeutic intervention. The analyses led to the construction of a bedside score predictive of IHI.
Discussion
In a prospective study including 211 hemodialysis sessions performed in a mixed ICU population, we found that IHI mostly occurred at the initiation of intermittent hemodialysis—during the first hour of the first session—despite the absence of fluid removal. A cardiovascular SOFA score ≥1, and two tissue hypoperfusion markers, index CRT ≥ 3 s and lactate level > 2 mmol/L, were associated with the occurrence of IHI. In addition, the risk of IHI increased with the number of abnormal parameters. A bedside score combining these three parameters, named SOCRATE score (cardiovascular SOFA score ≥ 1, index CRT ≥ 3 s and lactATE > 2 mmol/L), improved risk stratification with a good accuracy.
Intermittent hemodialysis is a key support therapy in ICU. Despite protocol-based optimization [
5], IHI remains a frequent issue in critically ill patients. IHI incidence varies from one study to another because of discrepancies in IHI definition and in preventive protocols [
5‐
11,
38,
39]. In our cohort, using a pragmatic definition of IHI, blood pressure drop requiring a medical intervention, we found an incidence of 23%. Using the same definition, Monnet et al. [
9] found higher incidence (33%), but all patients underwent fluid removal.
IHI mainly occurred during the first hour of treatment in the absence of ultrafiltration. During intermittent hemodialysis, due to partial redistribution of blood volume from the intravascular compartment to the extracorporeal circuit, mild hypovolemia is constantly induced even in the absence of ultrafiltration. However, blood pressure does not systematically decrease because of counter-regulatory mechanisms such as tachycardia, increased cardiac contractility, plasma refilling and peripheral vasoconstriction [
16,
40]. One could speculate that pre-existing alterations of macro- and microcirculation could interfere with these cardiovascular adaptation mechanisms and could promote IHI. Pre-existing microvascular endothelial dysfunction, commonly observed in ICU patients [
19‐
23], might alter such adaptive vasoconstriction [
20], promoting IHI. Hemodialysis, by itself, could also alter endothelial function. Meyer et al. [
17] showed an increase in “vasculotoxic” cell-free hemoglobin in the plasma of chronically dialyzed patients during intermittent hemodialysis. In addition, microbubbles generated in the circuit could damage endothelium glycocalyx, triggering activation of coagulation, platelets, neutrophils and promoting oxygen reactive species release [
41‐
43].
The evaluation of endothelial dysfunction at the bedside is very challenging [
24,
25], but its consequences in term of tissue hypoperfusion, could be assessed more easily. We speculated that tissue hypoperfusion parameters reflect microvascular endothelial dysfunction and we investigated two easy-to-use parameters, rapidly available at the bedside, the lactate level and the index CRT. Lactate level, widely used in ICU, is inversely correlated with sublingual microvascular perfusion [
44,
45] and is predictive of ICU mortality in septic shock patients [
28]. CRT measurement correlates with the pulsatility index, a surrogate ultrasound-derived parameter that reflects vascular tone of visceral organs [
46] and with objective parameters of tissue perfusion, such as tissue oxygen saturation in patients with septic shock [
47]. CRT is associated with hyperlactatemia and severity of critical illness addressed by SOFA score [
48] and predicts 14-day mortality in patients with septic shock independently of SOFA score [
34].
Here, we found for the first time that increased index CRT and hyperlactatemia were associated with IHI occurrence, as well as impaired global hemodynamics defined by a cardiovascular SOFA score ≥1. More interestingly, we found a cumulative predictive effect of these parameters similar to that we observed in the setting of severe pulmonary embolism [
49]. Combining them in a score which can be quickly and easily calculated at the bedside could be helpful to improve risk stratification. IHI was rare in the absence of macro- and micro-circulatory disorders (10% in first sessions, 3% in all the sessions) and increased progressively with the number of abnormal parameters reaching 80% when all three markers were present. In the absence of emergency criteria, intermittent hemodialysis initiation might be reconsidered in patients with higher SOCRATE score, indicating a high risk of IHI. Such a hypothesis is supported by the fact that optimal timing to initiate RRT in critically ill patients is still uncertain [
4,
13,
14]. Recent trials demonstrated that delaying RRT in the absence of life-threatening complications of AKI does not impair survival and allows some patients to avoid RRT [
4,
13]. These results plead in favor of a reasoned strategy balancing risks and benefits for initiating RRT in ICU patients.
Moreover, if intermittent hemodialysis indication is retained, SOCRATE score might be helpful for clinicians to identify patients who may benefit from a therapeutic intervention aiming at optimizing hemodynamic status before intermittent hemodialysis initiation in order to decrease the occurrence of IHI.
Our study has several limitations. It is a monocentric study, and the results need to be confirmed in a multicenter study including a larger population. Nevertheless, we analyzed a large number of intermittent hemodialysis sessions in a mixed non-selected medical ICU population. We did not include patients with dark skin because CRT was difficult measure. It would be interesting to test the prognosis value of a score combining lactate and cardiovascular SOFA alone, or in association with other validated clinical parameters of peripheral hypoperfusion such as central-to-toe temperature difference [
50]. As we could not include sessions performed in extreme emergency because data could not be recorded before the beginning of intermittent hemodialysis, IHI incidence was probably underestimated. Finally, our study focused on patients receiving intermittent hemodialysis. Future studies also including patients receiving continuous RRT are needed to test whether SOCRATE score could help to identify patients at risk of hemodynamic instability whatever the modality of RRT.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.