Septic shock is a highly lethal condition associated with a mortality risk of 30 to 60%. Optimizing tissue perfusion and oxygenation is the aim to decrease mortality and morbidity in septic shock patients.

Persistent hyperlactatemia after initial resuscitation is particularly difficult to interpret, although optimizing systemic blood flow might reverse ongoing hypoperfusion. Nevertheless, if persistent hyperlactatemia is caused by non-hypoperfusion-related mechanisms, then sustained efforts aimed at increasing cardiac output (CO) could lead to detrimental effects of excessive fluids or inotropes. Another potential alternative resuscitation target is peripheral perfusion as assessed by capillary refill time (CRT), mottling score or central-to-toe temperature differences. Reversal of abnormal peripheral perfusion might represent improvement in tissue hypoperfusion with the advantage of a faster recovery than lactate.

Hypothesis: Peripheral perfusion guided resuscitation in septic shock is associated with lower mortality, less organ dysfunctions, less mechanical ventilation (MV), less vasopressor load, and less renal replacement therapies than a lactate-targeted resuscitation strategy.

Main Objective To test if peripheral perfusion targeted resuscitation in septic shock is associated with lower 28-day mortality than a lactate targeted resuscitation.

Design: Multicenter, Parallel Assignment randomized controlled study, conducted under supervision of an independent Data Safety Monitoring Board (DSMB).

Interventions:

1. Active Comparator- Peripheral Perfusion guided resuscitation
2. Active Comparator- Lactate guided resuscitation

Randomization: 1:1 the randomization using a block size of eight will be stratified according to participating centers.

Trial size: 400 randomized patients in 30 ICUs.

Early Goal Directed Therapy using a Physiological Holistic View. A multicenter study in Latin America: The ANDROMEDA-SHOCK Study

Hypothesis Peripheral perfusion guided resuscitation in septic shock is associated with lower mortality, less organ dysfunctions, less mechanical ventilation (MV), less vasopressor load, and less renal replacement therapies than a lactate-targeted resuscitation strategy.

Using a holistic view of optimizing tissue perfusion and oxygenation the investigators aim to decrease mortality and morbidity in septic shock patients.

A. Background Septic shock is a highly lethal condition associated with a mortality risk of 30 to 60%. It is currently the most frequent cause of death in the intensive care unit (ICU) as the investigators demonstrated in a recent Chilean prevalence study. Several pathogenic factors such as hypovolemia, myocardial depression, vasoplegia, and microcirculatory abnormalities can induce progressive tissue hypoperfusion in severe cases. In this context, persistent hyperlactatemia has been traditionally considered as the hallmark of ongoing tissue hypoxia during septic shock, and therefore lactate normalization is recommended as a resuscitation target by the Surviving Sepsis Campaign (SSC).

Pathophysiologic determinants of persistent hyperlactatemia

The physiologic basis of lactate generation or clearance during septic shock has been matter of active research. Hypovolemia-induced hypoperfusion is probably the predominant pathogenic mechanism during the early phase. Some patients resolve acute circulatory dysfunction and clear lactate after initial fluid resuscitation, while others evolve into a persistent circulatory dysfunction with hyperlactatemia. Several mechanisms have been associated to persistent hyperlactatemia besides hypoperfusion, but recent literature has highlighted the role of sustained hyperadrenergia with increased muscle aerobic glycolysis, a condition denominated stress hyperlactatemia, and also of impaired hepatic lactate clearance.

The investigators have explored the significance and potential determinants of hyperlactatemia in a series of clinical physiological studies performed over the last 15 years. These studies have addressed the three most relevant pathogenic factors involved in persistent hyperlactatemia: overt or occult hypoperfusion, hyperadrenergic state and impaired hepatic clearance. The complexity of this subject is also highlighted by a more recent study where the investigators demonstrated that lactate decrease during successful septic shock resuscitation exhibits a biphasic pattern, an early rapid decrease in parallel to normalization of more flow-sensitive variables (see below), followed by a slower recovery thereafter. The latter eventually related to non-flow dependent mechanisms such as hyperadrenergic state and/or delayed hepatic clearance.

Persistent hyperlactatemia after initial resuscitation is particularly difficult to interpret as suggested by the extensive research summarized above. Optimizing systemic blood flow might reverse ongoing hypoperfusion, a potential source of anaerobic lactate generation. Under this perspective, some of the pathogenic factors involved in hyperlactatemia are potentially flow-sensitive, and others are not. Distinction between the two scenarios could strongly impact further resuscitation. If persistent hyperlactatemia is caused by non-hypoperfusion-related mechanisms, then sustained efforts aimed at increasing CO could lead to detrimental effects of excessive fluids or inotropes, a fact now well demonstrated in the literature. The decision of when to consider that a patient has been fully resuscitated and as a consequence stop further interventions is a milestone, and appears as highly relevant since the results of a number of recent studies have increased awareness about the risk of fluid overload and/or of vasopressors and inodilators such as pulmonary edema, increased intraabdominal hypertension, acute kidney injury, delayed weaning, arrhythmias, hepatosplanchnic or myocardial ischemia, among other problems. By these means, over-resuscitation could eventually increase morbidity and/or mortality.

Is hyperlactatemia a valid resuscitation target in septic shock?

Not surprisingly, lactate clearance or normalization is used worldwide as resuscitation targets. Indeed, SSC the most ambitious and global collaboration in critical care has proposed to focus septic shock resuscitation on normalizing macrohemodynamic parameters and lactate. SSC guidelines are followed in many countries and adherence to recommended management bundles have been reported to be associated to improved survival, although the role of each individual component is not clear. Lactate clearance, defined by a change of lactate levels between two time-points, and expressed as a 10-20% hourly lactate reduction, or a decrease of at least 10% in 6h during early resuscitation has been related to survival, and tested as a goal in two important studies with conflicting results.

However, there are several unresolved aspects and concerns about the role of lactate as an appropriate resuscitation target. First, it is not clear if selecting lactate clearance versus lactate normalization as resuscitation goals is equivalent, but more importantly, if this decision leads to similar timely resolution of tissue hypoperfusion or hypoxia. Second, since non-hypoperfusion related causes of hyperlactatemia might predominate in an unknown number of patients, this could lead to over-resuscitation in at least some of them as stated above. Third, the dynamics of recovery of lactate might exhibit a biphasic pattern and therefore, the real-time response of lactate to fluid challenges could be not straightforward depending on the hypoperfusion context. Some survivors might even normalize lactate only after 24h of evolution. Therefore, to explore other potential resuscitation targets appears as mandatory.

Potential alternative resuscitation targets in septic shock

A foremost priority is to rule out ongoing hypoperfusion in septic septic shock patients under active resuscitation. The investigators recently proposed that a simultaneous analysis of central venous O2 saturation (ScvO2), central venous-arterial pCO2 gradient (P(cv-a)CO2), and peripheral perfusion as assessed by capillary refill time (CRT), mottling score or central-to-toe temperature differences, might be helpful in suggesting a hypoperfusion context for patients with or without hyperlactatemia. From a theoretical point of view, these three easily assessable perfusion-related variables offer an important advantage over lactate as potential resuscitation targets in septic shock patients: they are clearly flow-sensitive and exhibit a much faster dynamics of recovery after systemic blood flow optimization. In other words, these parameters might clear in minutes in fluid-responsive patients as compared to lactate, which sometimes takes hours to recover. The investigators demonstrated this by analyzing the dynamics of recovery of these parameters in a cohort of ultimately surviving septic shock patients. ScvO2, P(cv-a)CO2 and CRT where already normal in almost 70% of the patients after 2h of fluid resuscitation, as compared with only 15% in the case of lactate.

However, there are also a couple of drawbacks for some of these perfusion-related flow-sensitive parameters. ScvO2 is a complex physiological variable. It was widely used until recently as the resuscitation goal in critically ill patients, although several limitations may preclude a straightforward interpretation of its changes. For instance, normal or even supranormal ScvO2 values do not rule-out global or regional tissue hypoxia for several reasons that have been highlighted elsewhere, but that include severe microcirculatory derangements impairing tissue O2 extraction capabilities. Vallee et al found persistent abnormal P(cv-a)CO2 values in 50% of septic shock patients who had already achieved normal ScvO2 values after initial resuscitation. Nevertheless, in some hyperdynamic states a high efferent venous blood flow could be sufficient to wash out the global carbon dioxide (CO2) generation from hypoperfused tissues and thus, P(cv-a)CO2 could be normal despite the presence of tissue hypoxia. Another problem for these two variables is that they necessarily require a central venous catheterization to be assessed, a task that might be complex to perform in limited-resource settings or emergency departments (ED). Therefore, peripheral perfusion appears as the most promising alternative resuscitation target in septic shock patients.

Peripheral perfusion as a potential resuscitation target in septic shock patients

The skin territory lacks auto-regulatory flow control, and therefore sympathetic activation impairs skin perfusion during circulatory dysfunction, a process that could be evaluated by peripheral perfusion assessment. Indeed, peripheral perfusion can be easily evaluated in many ways at bedside, and therefore, it could be a valuable monitoring tool in any setting. The presence of a cold clammy skin, mottling or CRT are frequently described as indications to initiate fluid resuscitation in patients with sepsis-related acute circulatory dysfunction.

The concept of CRT, the most relevant parameter, is based on this assumption. It was proposed initially in trauma patients but some negative studies that found no correlation with systemic hemodynamics precluded further research on this variable. More recently however, Lima et al found that abnormal peripheral perfusion is associated with hyperlactatemia and organ dysfunctions in critically ill patients. Other authors confirmed this finding and built up a robust body of evidence supporting the strong prognostic value of abnormal peripheral perfusion in the intensive care unit (ICU) context.

The investigators observed that CRT was the first parameter to be normalized in a cohort of septic shock patients and this predicted lactate normalization at 24h and survival. Moreover, some recent clinical data suggest that targeting peripheral perfusion during septic shock resuscitation might improve outcome. van Genderen et al performed a randomized controlled trial comparing two resuscitation protocols; one targeted at normal peripheral perfusion and the other to standard management in 30 ICU patients. The study demonstrated that targeting peripheral perfusion is safe, and associated with less fluid administration and organ dysfunctions. Therefore, a parameter like CRT with a rapid-response time could be very useful to test the response to treatments with strong physiologic impact such as fluid loading, especially at the ED or in limited-resource settings. In a prospective non published study performed in a cohort of 100 patients just admitted to the ED, the investigators found that patients exhibiting a normal CRT after initial fluid loading had a hospital mortality of less than 10% as compared to 55% in patients with abnormal values.

How can fluid loading and resuscitation improve peripheral perfusion? There is an intricate relationship between macrohemodynamics and peripheral perfusion. Both are affected by hypovolemia and tend to improve in parallel in fluid-responsive patients. Their relative changes, though, are not well correlated. The beneficial effects of fluids and vasoactive drugs may be explained by an increase in CO or perfusion pressure, a decrease in the neurohumoral response to hypovolemia, and eventually by direct effects at the microcirculatory level. Whatever the mechanism, normalization of peripheral perfusion parameters appears to indicate a successful reversal of initial circulatory dysfunction.

There are some data that suggest that vasopressor adjustment and/or inodilators could induce favorable effects on peripheral perfusion or microcirculation under certain circumstances. Jhanji et al demonstrated that increasing mean arterial pressure (MAP) to 90 mmHg with norepinephrine (NE) doses up to 0.41 mcg/kg/min improved cutaneous tissue oxygen pressure (PtO2) and cutaneous microvascular red blood cell flux in a cohort of septic shock patients. The same group obtained similar results in another cohort of postoperative patients after major abdominal surgery but with an intervention consisting in stroke volume optimization with fluid challenges and an inodilator (dopexamine) in fixed dose. Dubin et al demonstrated that rising MAP to 85 mmHg with incremental doses of norepinephrine (NE) up to 0.74 mcg/kg/min improved sublingual microcirculatory flow in septic shock patients with the worst microcirculation at baseline. Dobutamine in fixed doses of 5 mcg/kg/min improved sublingual microcirculatory flow in another cohort of septic shock patients. On the other hand active vasodilation with nitroglycerine induced a clear improvement of peripheral perfusion parameters in a group of shock patients, despite a mean fall in MAP of 14 mmHg. Based on these findings and other data, it was proposed that permisive hypotension could eventually improve microcirculatory driving-pressure in patients with acute circulatory failure. In summary, it appears that pharmacological therapies aimed at improving peripheral perfusion might be individually tailored but could imply increasing or lowering vasopressors and MAP, inodilators or pure vasodilators according to the clinical context.

More recently Bakker et al, added another important piece of information after performing a pilot study in 30 septic shock patients subjected to early resuscitation. In this study, CRT and skin mottling were correlated with the pulsatility index, a sonographic surrogate of vascular tone, of visceral organs. This means that improvement in peripheral perfusion might move in parallel with improvement in hepatosplanchnic perfusion, eventually explaining the good prognosis associated with recovery of CRT and other related parameters.

Using peripheral perfusion to target resuscitation in septic shock has also several potential drawbacks. First, there is some degree of subjectivity and inter-observer variability in some of the parameters used to assess it such as CRT and mottling. Second, it cannot be evaluated in some settings such as dark skin patients. Third, and more importantly, the corpus of evidence that supports that improvement of peripheral perfusion is associated with resolution of profound tissue or microcirculatory hypoperfusion, or hypoxia is still scanty.

However, the excellent prognosis associated with CRT recovery, the rapid-response time to fluid loading, the simplicity of its assessment, its availability in limited resource settings, and recent data suggesting that it might change in parallel to perfusion of physiologically more relevant territories such as the hepatosplanchnic region, constitute a strong background to promote studies evaluating its usefulness to guide resuscitation in septic shock patients.

Why to compare peripheral perfusion with lactate as targets for septic shock resuscitation?

Summarizing the theoretical background stated above, it is plausible that normalization of peripheral perfusion as compared to normalization or a rapid decrease (>20%/2h) of lactate might be associated with less fluid resuscitation and secondarily less positive 24h fluid balances. Eventually, less positive fluid balances might be associated with less organ dysfunctions, especially respiratory (oxygenation, mechanical ventilation days), renal (less increase in creatinine and renal replacement therapy), and gastrointestinal (less increase in intra-abdominal pressure). In addition, peripheral perfusion targeted-resuscitation might be also associated with less vasopressor load and inodilator use thus preventing other set of potential complications such as hepatosplanchnic hypoperfusion, arrhythmias or myocardial ischemia. At the end, this could result in less mortality for a combination of the previous reasons.

Hypothesis Peripheral perfusion guided resuscitation in septic shock is associated with lower mortality, less organ dysfunctions, less mechanical ventilation (MV), less vasopressor load, and less renal replacement therapies than a lactate-targeted resuscitation strategy.

Design Multicenter, open-label randomized controlled study, conducted under supervision of an independent Data Safety Monitoring Board (DSMB).

Main Objective To test if peripheral perfusion targeted resuscitation in septic shock is associated with lower 28-day mortality than a lactate targeted resuscitation.

Primary Outcome All-cause 28-day mortality

Secondary outcomes Need of mechanical ventilation Need of renal replacement therapies (RRT) Days free of MV, vasopressors and RRT in 28-days Sequential Organ failure Assessment (SOFA) at 8, 24, 48 and 72h Acute kidney injury (AKI) Intra-abdominal hypertension Fluid balances at 8, 24, 48 and 72h All-cause hospital and 90-day mortality Intensive care unit (ICU) and hospital length of stay

I. Patients Inclusion Criteria

Adult patients (≥18 years) will be screened for the following inclusion criteria:

Septic shock diagnosed at ICU admission according to the Sepsis-3 Consensus Conference, (basically septic patients with hypotension requiring norepinephrine (NE) to maintain a mean arterial pressure (MAP) of ≥ 65 mmHg, and serum lactate levels > 2 mmol/l after initial fluid resuscitation with at least 20/ml kg in one hour.

Exclusion Criteria

1. Pregnancy
2. Anticipated surgery or dialysis procedure during the first 8h after septic shock diagnosis
3. Do-not-resuscitate status
4. Child B or C liver cirrhosis
5. Active bleeding
6. Acute hematological malignancy
7. Severe concomitant acute respiratory distress syndrome (ARDS)
8. More than 4h after officially meeting septic shock criteria

II. Randomization Recruited patients will be randomized to a peripheral perfusion-targeted resuscitation (group A) with a goal of normalizing capillary refill time (CRT), or a lactate-targeted resuscitation (group B) with a goal of either normalizing lactate or achieving a >20% decrease per hour during the 8h study period.

The randomization sequence will be generated by an external statistician of the DSMB with the use of a computer program and an allocation of 1:1. The randomization using a block size of eight will be stratified according to participating centers.

Allocation concealment will be maintained by means of a web-based central, automated randomization system, available 24 hours a day (RedCap Cloud). The group to which the patient is allocated will only be disclosed after the information is recorded in the electronic system. Such measure prevents the investigator and the medical team from predicting to which treatment group the patient will be allocated. To include a patient in the study, investigators must simply access the study website and fill in a short medical record form.

Statisticians and the researcher responsible for event assignment will be blinded to the group allocation.

Treatment assignment will not be recorded in the medical chart or electronic patient data monitoring system and clinicians on general wards, who care for the patients after ICU discharge, will not be aware of the treatment assignment.

III. Assessments Baseline Demographics, comorbidities, acute physiology and chronic health evaluation (APACHE) II, sepsis source and treatment

pre-ICU resuscitation and fluid balance

Biomarkers: procalcitonin (PCT) or c-reactive protein (CRP), and adrenomedullin (MR-ProADM) sampling

SOFA + AKI criteria

Hemodynamics: heart rate (HR), systolic blood pressure (SBP), diastolic blood pressure (DBP), MAP, central venous pressure (CVP), dynamic predictors (DP) of fluid responsiveness (FR), intraabdominal pressure (IAP), NE levels, diuresis

Perfusion: lactate, ScvO2, P(cv-a)CO2, hemoglobin (Hb), central venous and arterial blood gases, CRT, mottling score, perfusion index (where available)

Evolution SOFA and AKI criteria at 8, 24, 48 and 72h Hemodynamics hourly up to 8h Fluid administration and balance at 8, 24, 48 y 72h Complete perfusion assessment when the targeted parameter is normalized and then at 8, 24, 48 and 72h Register of vasoactive drugs and dobutamine/milrinone use Register of MV and RRT Source control re-analysis at 4h Rescue therapies: high-volume hemofiltration (HVHF), vasopressin, epinephrine, steroids, others Echocardiography at least once during the study period Follow-up till 28 days for use of MV, RRT and vasopressors All cause mortality at hospital discharge, 28 and 90 days Cause of death

IV. Principles of general management Sepsis source identification and treatment should be pursued as a priority of first line treatment. A central venous catheter and an arterial line are inserted in all, and the use of a pulmonary artery catheter or a PiCCO device is recommended for patients with a past medical history of heart failure or with concomitant ARDS.

Echocardiography will be performed routinely as soon as possible after admission to evaluate basal cardiac function, and repeated as necessary to aid in assessing preload status through inferior vena cava distensibility when necessary.

NE will be the vasopressor of choice and adjusted to a MAP ≥ 65 mmHg in all patients.

Hemoglobin concentrations will be maintained at 8 g/dl or higher to optimize arterial O2 content. Mechanical ventilation settings are adjusted according to current recommendation. Rescue therapies such as epinephrine, vasopressin analogues, steroids or different blood purification techniques like high-volume hemofiltration will be decided following usual practice of the involved centers in patients evolving with refractory septic shock.

C. Study protocol

A sequential approach to resuscitation will be followed in both groups. Time 0 is the starting point when after randomization, a central venous catheter (CVC) and an arterial line are in place, and the basal measurements are performed including hemodynamics and blood sampling.

The study period will be of 8 hours and after this, attending intensivists can continue treatment according to their usual practice or department protocol.

I. Tests and Procedures during the study period Capillary refill time assessment CRT will be measured by applying firm pressure to the ventral surface of the right index finger distal phalanx with a glass microscope slide. The pressure will be increased until the skin is blank and then maintained for 10 seconds. The time for return of the normal skin color will be registered with a chronometer, and > 3 seconds is defined as abnormal.

Lactate measurements A normal lactate value is defined as less then 2 mmol/l. Lactate will be assessed with the technique more easily available for each center, including arterial serum levels (with common gas analyzers at the central lab), or capillary levels with lactate scout strips.

Fluid responsiveness (FR) This is the first step. FR will be assessed with a structured approach. Basically, dynamic predictors (DP) will be evaluated depending on the patient background status.

In sedated adapted mechanically ventilated patients without arrhythmias, pulse pressure variation (PPV) or stroke volume variation (SVV) will be used as first choice. A FR+ status is established with values ≥ 13% and 10%, respectively. If negative, PPV and SVV will be reassessed after transiently increasing tidal volume (VT) to 8 ml/kg (one minute). An increase >3.5% and 2.5% in PPV or SVV, respectively will be considered as FR+.

In patients with arrhythmia, the preferred tests will be the end expiratory occlusion test with a 15 sec pause (> pulse pressure >5% considered as positive), or echocardiography assessing inferior vena cava (IVC) distensibility index (>15% considered as +).

In spontaneous breathing patients or non-sedated patients under MV, a passive leg rising (PLR) maneuver will performed with an early (<1min) increase in pulse pressure > 10% considered as FR+. If this is not obtained and to rule out a false negative response, the maneuver will be repeated assessing aortic velocity time integral (VTI) with echocardiography before and after PLR with a >15% increase in VTI accepted as FR+.

Fluid Challenge In FR+ patients, the first resuscitation step is to administer a fluid bolus (FB) of 500 ml of crystalloids every 30 min until CRT is normalized in group A, or DP becomes negative in group B. DP and CVP will be measured before and after each bolus in both groups.

Safety measures during fluid challenges Central venous pressure (CVP) and FR will be reevaluated after any fluid challenge. If CVP increases <5 mmHg and FR is still +, another FB is administered and so on while the goal is not reached.

If CVP increases ≥ 5 mmHg or FR is -, fluids will be stopped and the patient will be moved to the next step.

Vasopressor test In FR- patients with persistent abnormal CRT or with a still abnormal lactate that decreased <20%/2h, a vasopressor test will be performed.

In previously hypertensive patients, MAP will be increased to the range of 80-85 mmHg by transiently rising NE doses. CRT and lactate rechecked (CRT at one h and lactate at 2h). If CRT is normal in the group A, or lactate normalizes or decreases >20% in group B, resuscitation will be stopped and NE dose maintained. If not, NE will be reduced to the pre-test doses, and the protocol moves to the next step.

In all the other patients, MAP will be reduced to the range of 60-65 mmHg by transient decreases in NE doses, with the same objectives and principles as stated above.

Use of inodilators Dobutamine 5 mcg/kg/min or Milrinone 0.25 mcg/kg/min in fixed doses will be started, and CRT or lactate rechecked (CRT at one h and lactate at 2h). If, the goals are not reached, drugs will be discontinued and no further action will be taken during the study period, except recheck FR every hour and restart fluid challenges if patients get FR+ again.

In responders (same criteria as with vasopressor test), the inodilator will be continued along the study period.

As a safety measure, inodilators will be stopped if HR increases >15%, or arrhythmias, ischemia or hypotension develop.

Group A. Management of peripheral perfusion-targeted resuscitation.

In this group, the goal is to normalize CRT by following the next steps in the given order:

1. Assessment of FR
2. Fluid challenges until CRT is normal, the patient is fluid unresponsive or a safety measure is met
3. Vasopressor test
4. Inodilator test

As a safety measure, resuscitation will be stopped even with normal CRT only in the presence of a stable macrohemodynamics as demonstrated by heart rate (HR) <120 beats per minute, and stable MAP with no increase in vasopressors during the last hour.

After CRT normalization at any step, CRT will be reassessed hourly during the study period. If at any point, it turns abnormal again the resuscitation sequence will be restarted.

Group B. Management of lactate-targeted resuscitation.

In this group the goal is to normalize lactate levels or get a decrease rate of at least 20%/2h, by following the next steps in the given order, always reevaluating lactate at 2h intervals.

1. Assessment of FR
2. Fluid challenges until FR- or safety CVP limit is reached during the bi-hourly intervals between lactate assessments
3. Vasopressor test
4. Inodilators

Lactate will be assessed every two hours during the 8h study period. If after obtaining the lactate goal, lactate gets abnormal again or the decrease rate turns <20%/2h at any of the following bi-hourly controls during the study period, the resuscitation sequence will be restarted.

D. Sample size Mortality in patients with increased lactate levels in circulatory dysfunction has been shown to exceed 40%. In addition, several studies have shown that abnormal peripheral perfusion is associated with a mortality exceeding 40%.

The investigators should enroll 420 patients. With these sample size the study will have 90% power to detect a reduction in 28-day mortality from 45% to 30%, at a significance level of 5%, considering time-to-event analysis. The investigators consider a decrease of 15% in mortality to have direct clinical implementation effect. Similar effects on mortality have been shown in early resuscitation studies. In addition limiting fluid administration in patients with septic shock and normal peripheral perfusion has been shown to decrease organ failure, which is the leading cause of death in these patients.

When aiming for a smaller decrease in mortality (like 10%), this sample size would only have 57% power to detect benefit. Therefore the investigators will use an adaptive approach that will allow for a sample-size re-estimation at the interim analysis when 75% of the sample has been recruited. The sample-size re-estimation will be conducted by the DSMB if the effect size observed in the interim analysis is between 10% and <15% absolute reduction in mortality. A smaller effect size would turn achieving adequate sample size unfeasible, whereas a larger effect size will provide the study with more than 90% power to detect the effect.

E. Statistical analysis plan

A detailed statistical analysis plan will be prepared before proceeding to patient enrolment. The essential characteristics of this statistical analysis plan are described below.

All analyses will be based on the intention-to-treat principle. The investigators will assess the effect of peripheral perfusion-targeted resuscitation (group A) compared to lactate-targeted resuscitation (group B) on the primary outcome through hazard ratio with 95% of confidence interval (CI) and Kaplan-Meier curve comparison (using log-rank test). Binary secondary outcomes will be compared through relative risks, 95%CIs, and chi-square tests. Results for continuous outcomes with normal distribution will be expressed as mean difference, 95%CI, and P-value calculated by the t-test. Continuous outcomes with asymmetrical distribution will be assessed by the Wilcoxon test.

The investigators will analyze the effects of the study fluids on the primary outcome in the following subgroups:

1. Patients with lactate > 4.0 mmol/l as set by SSC
2. Patients without a confirmed source of infection (as this could increase the translation of the study to other critically ill).
3. Patients with low APACHE II / SOFA scores
4. Patients with a more than 10% difference in lactate level between the very first one measured and the baseline when starting the study.

Inclusion Criteria:

• Septic shock diagnosed at ICU admission according to the Sepsis-3 Consensus Conference, (basically septic patients with hypotension requiring norepinephrine (NE) to maintain a mean arterial pressure (MAP) of ≥ 65 mmHg, and serum lactate levels > 2 mmol/l after initial fluid resuscitation with at least 20/ml kg in one hour.

Exclusion Criteria:

• Pregnancy
• Anticipated surgery or dialysis procedure during the first 8h after septic shock diagnosis
• Do-not-resuscitate status
• Child B or C liver cirrhosis
• Active bleeding
• Acute hematological malignancy
• Severe concomitant acute respiratory distress syndrome (ARDS)
• More than 4h after officially meeting septic shock criteria