Anemia is a prevalent issue in critically ill patients (1, 2). Despite a decline in red blood cell transfusions (RBCT) usage over time (3), they remain commonly administered to intensive care unit (ICU) patients (4), even in the absence of active bleeding (5). It is important to note that RBCT may have either beneficial effects, such as increasing global oxygen delivery and improving tissue oxygenation, or harmful effects, including transfusion-related acute lung injury (TRALI), transfusion-associated circulatory overload (TACO), and immunosuppression with increased risk of infections (6–10). Consequently, deciding whether to transfuse an ICU patient is a significant challenge for clinicians and should be personalized based on specific triggers.

Current guidelines recommend a restrictive RBCT strategy with a hemoglobin (Hb) threshold of 7 g/dL for considering transfusions in non-bleeding ICU patients (11). However, a fixed Hb value does not adequately reflect tissue perfusion or oxygen requirements in this setting (12, 13). Additionally, hematocrit is lower in capillaries than in large vessels (arteries and veins), making systemic Hb an inadequate parameter for microvascular oxygen content and viscosity (14–16). Furthermore, systemic endpoints for estimating tissue perfusion and oxygenation to guide RBCT have yielded conflicting results (17–20). A recent meta-analysis demonstrated that RBCT significantly increased oxygen delivery and consumption, while reducing the oxygen extraction ratio, without affecting cardiac output, in critically ill patients (21). These effects were particularly evident in septic patients, with a wide range of RBCT effects among other subgroups (21). It is important to note that microvascular alterations and tissue hypoxia may persist in the presence of normal values for systemic variables, including blood pressure, cardiac output, lactate levels, or oxygen delivery (22–27). Consequently, RBCT administration should not solely rely on these systemic variables but should be based on tissue perfusion and/or oxygenation.

Some studies have indicated that RBCT have heterogeneous effects on sublingual microcirculation, with either improved or altered microvascular density, in critically ill patients (12, 13). Similar results were observed when measuring muscular oxygenation saturation and applying a vascular occlusion test (VOT), with those patients with impaired baseline microvascular reactivity showing an improvement while the others showed a deterioration of microvascular function after RBCT (28). A recent study suggested that low baseline brain oxygen pressure values could predict a significant improvement in brain oxygenation after RBCT in critically ill patients with acute brain injury (29). However, these methods have some limitations. Tissue oxygenation with VOT requires a dedicated probe and software that are no longer available, and repeated measurements could affect baseline results, limiting its routine applicability at the bedside (30–32). Sublingual microcirculation assessment using videomicroscopy can be challenging in awake and/or agitated patients. Brain oxygen pressure monitoring is expensive and invasive and is only feasible in brain-injured patients (33). In this context, Skin Laser Doppler (SLD) is a non-invasive and straightforward technique for directly assessing skin blood flow (SBF) and tissue perfusion (34, 35). This technique uses an optical fiber to direct light from a low-power laser source to the skin and to collect the back-scattered light; the shift in light wavelength is proportional to the red blood cell velocity in the studied skin area, providing a non-invasive measurement of SBF (35, 36). However, no studies have evaluated the effects of RBCT on tissue perfusion using this technology in non-bleeding ICU patients after resuscitation.

The aim of this study was therefore to evaluate the effects of RBCT on SBF and to identify predictors of skin blood flow response to RBCT in non-bleeding critically ill patients after initial resuscitation.

In this prospective study, we found that RBCT did not consistently affect skin tissue perfusion, as assessed using laser Doppler, in critically ill patients after resuscitation without acute bleeding. However, there was substantial interindividual variability in the response to RBCT; 44% of patients showed a significant increase in SBF_BT. Patients who responded to RBCT had lower SBF_BT values at baseline than non-responders, and the relative increase in tissue perfusion after RBCT was higher when baseline SBF_BT was lower. We found no relationship between skin tissue perfusion response to RBCT and storage time, the presence of sepsis, or systemic hemodynamics.

The ultimate goal of RBCT is to improve tissue oxygenation (12). However, RBCT is still often given based on a specific Hb trigger, despite Hb being a poor surrogate for estimating oxygen delivery (DO₂), tissue perfusion, oxygenation, or cellular metabolic needs (12, 40, 41). Thus, monitoring tissue perfusion, oxygenation, and microcirculation could be a useful tool for identifying patients who require interventions to improve microvascular flow or oxygenation, with RBCT being one possible intervention (15). Previous studies have reported heterogeneous effects of RBCT on microcirculatory parameters. In septic and trauma patients, RBCT did not affect sublingual microcirculation assessed by videomicroscopy, despite increases in Hb, MAP, and global DO₂ (12, 42). However, improved microcirculation was observed in patients with altered perfused capillary density at baseline, while the opposite was reported in those with normal microvascular flow. In septic patients, RBCT did not affect muscle oxygen saturation, oxygen consumption, or microvascular reactivity measured using near-infrared spectroscopy (NIRS) (43). However, changes in tissue oxygenation and oxygen consumption were negatively correlated with baseline values, suggesting a larger response to RBCT in patients with impaired tissue oxygenation at baseline (28). Other studies, including mixed ICU populations, trauma, hemorrhagic shock, or ECMO patients, reported that RBCT directly improved microcirculation assessed by videomicroscopy (13, 42, 44–46). While all studies reported that the effects of RBCT on microcirculation were independent of hemodynamic parameters, our study highlights how skin tissue perfusion can variably respond to RBCT and suggests that discrepancies among studies may be due to differences in patient selection and the techniques used.

Undoubtedly, RBCT cannot be considered the optimal intervention for consistently improving tissue oxygenation and perfusion. In critical illness, after initial resuscitation, some patients may have a normal microvascular function and should not receive RBCT based solely on hemoglobin values, as this could potentially be harmful, as demonstrated in our study and other studies. However, RBCT may provide beneficial effects on microvascular density, tissue perfusion, tissue oxygenation, and oxygen consumption when microvascular function is impaired, and systemic hemodynamics are restored within accepted therapeutic ranges. Our study found that 44% of patients had skin blood flow responders to RBCT. Therefore, it is essential to identify predictors of SBF response to RBCT to individualize transfusion policies in clinical practice. Specifically, lower baseline values of skin blood flow basal tone could predict skin tissue response to RBCT with moderate accuracy. These results are consistent with previous reports, which suggest low baseline functional capillary density and tissue oxygenation as predictors of RBCT response (12, 28). Although we did not concurrently assess sublingual microcirculation with videomicroscopy or tissue oxygenation with NIRS, assessing tissue perfusion using SBF has some potential advantages. First, the technique is easy to use, even in non-intubated and agitated patients, who are frequently difficult to assess with videomicroscopy (27, 34). Second, SBF requires a very short learning phase and provides accurate inter-operator reproducibility during pilot studies in our service (data not shown). Third, altered SBF has been associated with the severity of the underlying disease, has important prognostic value, and responds to other clinically relevant interventions in ICU patients (27, 47, 48).

The interindividual variability of tissue response after RBCT could be related to other factors. The presence of an underlying neurological disease was an independent predictor of non-response to RBCT. Microvascular skin flow is controlled by the autonomic nervous system (ANS) (49), and arterio-venous shunts provide a low-resistance pathway by which blood flow can bypass capillaries and thus be drained from the arterioles to venules. The shunts are maintained in a constricted state by a sympathetic tone (50). The neurological disease causes loss of this tone and then the shunts open diverting blood flow from the skin (51). The lack of significant changes in tissue perfusion in neurological patients might be related to ANS dysfunction. Higher age was also an independent predictor of skin tissue response to RBCT. Previous studies reported that age could affect peripheral tissue perfusion. One study showed that SBF measured on the dorsum of the hand was 30% lower in the elderly (i.e., around 70 years) than in younger (i.e., around 20 years) healthy volunteers (52). Another study reported that elderly volunteers (mean age of 74.6 ± 7.6 years) had significantly longer capillary refill time (CRT) than younger volunteers (mean age of 32.2 ± 7.2 years) (53)_. However, in the multivariate analysis, low SBF_BT remained independently predictive of a positive response, regardless of age. However, other factors did not significantly impact tissue response to RBCT. For example, storage time was not associated with an increase in SBF, despite alterations in RBCs with storage duration that might result in the release of vasodilators, a reduction in RBCs deformability, altered adhesiveness and aggregability, and the release of free Hb, which could alter microvascular flow (54). Our findings are consistent with previous reports (12, 28, 55). Also, sepsis was not independently associated with skin blood flow response to RBCT, although significant modifications in RBCs rheology were observed in septic patients (56). These findings may suggest that critical illness itself, rather than sepsis alone, could directly alter tissue perfusion and result in variable responses to different therapeutic interventions.

Another important finding in our study is that a significant reduction in the SBF was observed after RBCT in an important proportion of patients (i.e., “non-responder” group). Other clinical studies have also reported a decrease in the microvascular sublingual perfusion and tissue oxygenation when an RBCT was administered to patients who already have normal baseline tissue perfusion (12, 28). One could argue that the microcirculation is impaired by the specific rheological characteristics of the transfused RBCs. Friedlander et al. yielded that RBCT improved RBCs deformability in septic patients, probably by replacing rigidified, endogenous RBCs with more functional, or less dysfunctional exogenous RBCs (57). However, RBCT may be deleterious when performed in patients with stable microcirculatory conditions, preserved deformability, vasoreactivity, and/or tissue perfusion, probably because of an increased aggregability (13, 43). Additionally, because the flow in a capillary is proportional to the driving pressure, increments in CVP may affect the capillary perfusion pressure. Vellinga et al. reported an association between elevated CVP (>12 mmHg) and a reduction in microcirculatory blood flow in patients with sepsis (58). Other studies also reported impaired microvascular blood flow for mild increases in venous pressure in humans (59, 60). Although no significant changes in CVP were observed after RBCT in our population, we cannot exclude the possibility that increments in CVP in some of these patients could have decreased the driving pressure thereby reducing microcirculatory perfusion. However, it is also possible that this would reflect a normal physiological response of the microcirculation (i.e., vasoconstriction) in response to a supranormal oxygen delivery state.

Our study has several limitations. First, it was a single-center study, which limits its external validity. Second, during the intervention period, some interventions other than RBCT were ongoing (i.e., antibiotics, and intravenous fluids), and we cannot rule out the possibility that these changes may have influenced the final results, but they were a minority around 3.4%. Third, we did not collect other potential variables of interest, such as cardiac output, as it was not monitored in all our patients. This analysis would be informative to detect some incoherence between the microcirculatory and global systemic variables, as previously reported (12, 13, 45, 61–63). Fourth, although it would be of clinical interest to include measurements like the capillary refill time (CRT) or the mottling score, these data were not available. However, recent data confirmed the correlation between CRT and SBF measurements in critically ill patients (64). Fifth, the SLD technology itself has some limitations. SBF represents the average of the blood flow in arterioles, venules, and capillaries in the measured volume of tissue, and the relative contributions of arterial, venous, and capillary blood flow within the measured volume of tissue cannot be determined. Sixth, we cannot extrapolate skin measurements to other organs, and further studies are needed to clarify if the responses obtained at the skin microcirculatory level are the reflection of what happens in other microvascular networks.

In conclusion, the present study shows that RBCT results in variable tissue perfusion changes during critical illness. Altered skin tissue perfusion at baseline had a higher probability to improve after RBCT. Skin tissue perfusion monitoring might help to individualize transfusion strategies in this setting.

Due to ethical restrictions, the datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request. Requests to access the datasets should be directed to EC, elaine_meduece@yahoo.com.br.

The studies involving humans were approved by Erasme Hospital Ethical Committee. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

EC and FT conceived the study. EC, PB, and DO selected the population and collected the data. EC and FT conducted the statistical analysis and wrote the first draft of the article. PB, DO, J-LV, and JC revised the text for intellectual content. All authors contributed to the article and approved the submitted version.