Severe metabolic acidemia in the critically ill (pH equal or less than 7.20; PaCO2 equal or less than 45mmHg and bicarbonate concentration equal or less than of 20 mmol/l) is associated with a 50% rate of day 28 mortality. Moderate to severe acute kidney injury is a frequent cause of metabolic acidemia in the critically ill. When both severe metabolic acidemia and moderate to severe acute kidney injury are observed, day 28 mortality is approximatively 55-60%. Severe acidemia has been shown to be a biomarker of severity but may also contribute by itself to outcome. Investigators recently performed a multiple center randomised clinical trial (BICARICU-1) that suggests that sodium bicarbonate infusion titrated to maintain the pH equal or more than 7.30 is associated with a higher survival rate (secondary endpoint) in patients presenting both severe metabolic acidemia and moderate to severe acute kidney injury patients. Whether sodium bicarbonate infusion may improve long term survival (Day 90, primary outcome) in these severe acute kidney injury patients is currently unknown.

Acute acidemia is frequently observed during critical illness, its reported incidence varying from 14%to 42%. Persistent acidemia has been associated with poor prognosis, with a mortality rate as high as 57% when the pH stays below 7⋅20 more than 24h. Along with case-specific treatment, improvement of tissue perfusion and supportive measures such as mechanical ventilation and renal-replacement therapy are the cornerstones of severe metabolic acidemia management in critically ill patients. Given that an acidotic cellular environment can cause cellular dysfunction, intravenous bicarbonate administration to increase the pH may also be beneficial. In a survey performed in North America, more than two thirds of the program directors in nephrology or intensive care units (ICU) declared that they used bicarbonate for the treatment of acidemia with hyperlactatemia.Despite the frequency of its use in ICUs across the world, the effect of bicarbonate infusion for the treatment of metabolic acidemia remains controversial. Small physiological studies,along with retrospective and/or observational studies, have failed to draw clear conclusions. Reluctance to use bicarbonate may be related to the absence of cardiovascular effects in two physiological studies and to the potential side effects, principally intracellular acidification due to the accumulation of carbon dioxide and the risk of hypocalcemia. However, bicarbonate could compensate the deleterious effects of acidotic cells on cardiovascular and oxygen delivery and might delay or avoid unnecessary early renal-replacement therapy. Among the organ failure that are associated with acidemia, acute kidney injury (AKI) is often observed. Moderate to severe AKI (KDIGO 2-3) occurs in 35-40% of the critically ill patients . The mainstay of AKI management is to identify and correct causative factors while providing supportive care and treating acute complications. Clinical manifestations include encephalopathy and pericardial effusion. Other potentially dangerous complications include metabolic abnormalities of hyperkalaemia and acidosis, and fluid overload. Despite correcting causative and providing supportive care AKI-associated mortality remains high. In intensive care settings, in-hospital mortality rates of moderate to severe AKI with severe acidemia has been reported to be over 50% . Medical treatment of acidemia with AKI may involve administration of intravenous sodium bicarbonate. Interestingly, a recent literature review performed by the Cochrane group showed that no prospective randomized study has ever evaluated the impact intravenous sodium bicarbonate as a medical treatment to correct severe acidemia during AKI .

Investigators conducted a prospective multicenter, randomized controlled trial to evaluate whether bicarbonate infusion would improve outcome in critically ill patients with severe metabolic acidemia (defined as an arterial pH equal or less than 7.20; PaCO2 equal or less than 45mmHg and bicarbonate concentration equal or less than of 20 mmol/l). Specifically, investigators hypothesized that, compared with no bicarbonate, early bicarbonate infusion would result in an improvement in the primary outcome (ie, composite criteria of organ failure at day 7 and any cause of death at day 28).

The findings of the BICARICU-1 trial suggest that in the overall non-selected patients, sodium bicarbonate infusion is not associated with clinical outcome (no difference in the primary outcome and the Kaplan-Meier method estimate of the probability of survival at day 28 between the control group and bicarbonate group: (46% [95% CI 40-54] vs 55% [49-63]; p=0⋅09)). In the overall non-selected patients, the absolute risk reduction of the composite outcome was 5.5%, with the possibility of being as large as 19.4% (the lower limit of the confidence interval), and concerning the hard endpoint mortality at 28 days, the absolute risk reduction was 9% (NNT=12), with the possibility of being as large as 19.4% (NNT=5) (p=0.07). Moreover, in multivariate analysis, after adjusting for important clinical covariates, the effect of sodium bicarbonate on mortality at 28 days became statistically significant (HR=0.727, 95% CI 0.54-0.98, p=0.035).

In the a-priori defined clinical stratum of patients with moderate to severe acute kidney injury (Acute Kidney Injury Network scores of 2 or 3 at enrolment), sodium bicarbonate infusion was associated with an improvement in the primary outcome (ie, composite criteria of organ failure at day 7 and any cause of death at day 28) and a reduced rate of mortality from enrolment to day 28 between the control group and bicarbonate group : 63% [95% CI 52-72] vs 46% [35-55]; p=0⋅0283. Additionally, the number of days alive and free from renal-replacement therapy was higher in the bicarbonate group than in the control group both in the overall study population and in the a-priori defined stratum of patients with moderate to severe acute kidney injury.

Knowledge gap and research hypothesis

Although investigators previously reported that sodium bicarbonate infusion might be associated with less renal replacement therapy in critically ill patients with severe acidemia and that sodium bicarbonate infusion might be associated with a better day 28 survival in patients showing both severe acidemia and moderate to severe AKI, there is currently no study that has ever evaluated the room for sodium bicarbonate infusion in critically ill patients with both severe acidemia and moderate to severe AKI.

The hypothesis is that sodium bicarbonate infusion will be associated with a better long term (Day 90) survival compared to no sodium bicarbonate infusion. Unpublished post hoc analysis of the BICARICU-1 trial suggests indeed a 10% drop of day 90 mortality in patients treated with sodium bicarbonate infusion during their ICU stay.

In a previous trial (BICARICU-1 trial,), Investigators have shown that sodium bicarbonate infusion is efficient and safe to increase the arterial pH. Titrating the infusion to target a pH equal or above 7.30 is feasible and the two groups (sodium bicarbonate infusion vs no sodium bicarbonate infusion) were different among time for the pH value and the plasma bicarbonate concentration making possible to impute the difference in outcome to the treatment arm. Physicians in charge will (in the intervention group) use a 4.2% sodium bicarbonate solution and will administer from 125ml to 250ml in 30min to 240min. Although a strict calculation of bicarbonate deficit would be of interest, they believe that it would not reflect the daily routine practice in most of the centers. Moreover, bicarbonate deficit is calculated with a controverted formula and our trial is designed as a pragmatic trial.

Originality and innovative aspects of the study This will be the first randomised clinical trial investigating whether sodium bicarbonate infusion is associated with day 90 mortality in critically ill patients with both severe acidemia and moderate to severe AKI. If sodium bicarbonate, a medication worldwide available for almost no additional cost is associated with a better outcome, it would change the way of treating these patients around the globe.

Inclusion Criteria:

• Aged from 18 years old
• Admitted in the ICU where the BICARICU-2 trial takes place
• Within 6h before enrolment, the patient MUST present on the same arterial blood gas (the last available before enrollment) the 3 following criteria:

pH ≤ 7.20 ; Bicarbonatemia ≤ 20 mmol/l ; AND PaCO2 ≤ 45mmHg ;

• Moderate to severe acute kidney injury ("Kidney Disease Improving Global Outcome", KDIGO group of 2 or 3)
• Within 48h of ICU admission, a total SOFA ≥ 4 OR an arterial lactate concentration ≥ 2 mmol/l
• Signed informed consent form. According to the French law, considering the severity of the illness, the fact that most of these patients would be unable to consent (sedation or potential delirium) and that their proxies might not be contactable at the time of inclusion, a deferred consent process for emergency situations will be enabled. When deferred consent will be used, written permission to pursue the research will be obtained from the patient or proxy as soon as possible. If this consent is not obtained, the patient's data will not be used and they will be withdrawn from the trial.
• Subjects must be covered by public health insurance

Exclusion Criteria:

• Pure respiratory acidosis (defined by pH 7.20, PaCO2 >50 mmHg, bicarbonatemia equal or greater than (PaCO2-40)/10 + 24), digestive or urinary tract proven loss of fluid (equal or greater than 1500ml/24h) with concomitant loss of sodium bicarbonate, chronic kidney insufficiency (creatinine clearance ≤ 10ml/min), proven tubular acidosis, ketoacidosis, exogenous acids poisoning (aspirin, methanol, ), PaCO2 > 45 mmHg and spontaneous breathing, sodium bicarbonate infusion or renal replacement therapy within 24h prior to screening prior to screening or imminent in the next 6h.
• Pregnant or breast feeding patient
• Patient who is in a dependency or employment with the sponsor or the investigator
• Patient who was enrolled in another study and who is in the exclusion period for any enrolment in the present study
• Life expectancy less than 48h
• Consent refusal from the patient or his/her next of kin and the impossibility to enrol using the emergency procedure
• Patients protected by law (Art.L 1121-5, 1121-6, 1121-8 du Code de la santé publique)
• Absence of a French Health Care Insurance coverage

De-identified data used in the published manuscript will be shared in accordance with the applicable informed consent under which the data was collected

12 months after the main publication

Data are provided to qualified investigators free of charge. Required documents to request data include a summary of the research plan, request form, and IRB review. Dataset will be shared after careful examination by the study board of investigators.