During diabetic ketoacidosis, what route of insulin delivery should be preferred?
R3.1—Insulin should probably be administered intravenously rather than subcutaneously in patients with diabetic ketoacidosis (GRADE 2+, STRONG AGREEMENT).
Rationale Two literature reviews have considered the optimal route for administration of insulin in diabetic ketoacidosis [
82,
83]. Four controlled and randomized trials compared subcutaneous (SC) insulin with intravenous (IV) insulin in management of diabetic ketoacidosis in adults [
84‐
87]. All evaluated the rate of correction of acidosis or the normalization of blood glucose. Three evaluated the length of hospital stay [
85‐
87]. A lack of precision in the reported results meant that one of these studies [
84] could not be included in the meta-analysis of the rate of correction of acidosis or normalization of blood glucose. This trial described a correction of ketosis and a significantly greater decrease in blood glucose at 2 h in the IV group, but non-significant results 4, 6, and 8 h after the start of therapeutic management. A meta-analysis of two trials comparing similar insulins [
85,
87] found no significant difference in the rate of correction of acidosis or the normalization of blood glucose (difference = 0.2 h; 95% confidence interval [− 1.7–2.1];
p = 0.81). The last trial [
86] reported similar results (
d = − 1 h [− 3.2–1.2];
p = 0.36). The meta-analyses found no significant difference in the effect of the route of administration on the length of hospital stay. The literature data do not show that IV insulin therapy is preferable to SC insulin therapy, in terms of the rate of correction of acidosis, the normalization of blood glucose, or the length of hospital stay. However, few patients were included and they presented uncomplicated ketoacidosis. In addition, SC injections of insulin were performed regularly and the frequency of injections could be a source of discomfort or even pain.
As a venous route was often necessary, the continuous IV route seems preferable so as to facilitate restoration of water–electrolyte balance, avoid repeated SC injections, and reduce the risk of hypoglycemia, while ensuring better control of the insulin dose administered.
During diabetic ketoacidosis, should an insulin bolus be administered before starting continuous intravenous insulin therapy?
R3.2—An insulin bolus should probably not be administered before starting continuous intravenous insulin therapy in patients with diabetic ketoacidosis (GRADE 2−, STRONG AGREEMENT).
Rationale A literature review of the use of an initial insulin bolus before initiation of continuous intravenous insulin therapy identified just one randomized controlled trial [
88] and one observational study [
89]. In the latter, the normalization of blood glucose and the length of hospital stay did not differ significantly between the bolus and non-bolus groups (change in blood glucose 60.1 ± 38.2 vs 56.0 ± 45.4 mg/dL/h, respectively;
p = 0.54; length of hospital stay 5.6 ± 5.3 vs 5.9 ± 6.9 days;
p = 0.81). The authors noted more cases of hypoglycemia in the bolus group, but the difference was not statistically significant (6 vs 1%;
p = 0.12). The randomized controlled trial compared three groups: a low-dose insulin bolus then a low insulin dose (0.07 IU/kg, then 0.07 IU/kg/h), a low insulin dose without an initial bolus (0.07 IU/kg/h), and double-dose insulin (0.14 IU/kg/h) without an initial bolus. The rate of correction of acidosis, the normalization of blood glucose, and the length of hospital stay did not differ between the three groups. It is important to note that this study did not evaluate the insulin dose commonly used in continuous intravenous administration, i.e., 0.1 IU/kg/h.
During diabetic ketoacidosis, should high or low continuous intravenous insulin doses be administered?
R3.3—Low continuous intravenous insulin doses should probably be administered in the treatment of diabetic ketoacidosis (GRADE 2+, STRONG AGREEMENT).
R3.4—The experts suggest using an initial dosage of 0.1 IU/kg/h without exceeding 10 IU/h, and to increase it in the absence of hypokalemia, if the targets for correction of blood ketones (0.5 mmol/L/h), bicarbonate (3 mmol/L/h), and capillary blood glucose (3 mmol/L/h) are not reached after the first hours of treatment (EXPERT OPINION).
Rationale The literature data, essentially from the 1970s, indicate that low continuous intravenous insulin doses are as effective as higher doses [
90,
91]. A literature review found two trials (with no control group) reporting a decrease in blood glucose that was similar for low and high insulin doses. The risk of hypokalemia, hypoglycemia, or cerebral edema possibly associated with high doses and efficacy of low doses have justified their use in practice for several decades. However, if the targets for correction of blood ketones (0.5 mmol/L/h) or failing that of bicarbonate (3 mmol/L/h) and blood glucose (3 mmol/L/h) are not reached, it is possible to envisage increased doses, provided there is no hypokalemia.
Should sodium bicarbonate infusion be used in severe metabolic acidosis and, if so, in what situations?
R3.5—The experts suggest administering sodium bicarbonate to compensate for gastrointestinal or renal base loss in case of poor clinical tolerance (EXPERT OPINION).
Rationale The administration of sodium bicarbonate could limit the deleterious cardiovascular, respiratory, and cellular energy effects of loss of bicarbonate [
2]. Sodium bicarbonate should be administered carefully as it is associated with a risk of hypokalemia, hypernatremia, hypocalcemia, rebound alkalemia, and water–sodium overload [
2].
R3.6—Sodium bicarbonate should probably be administered to intensive care patients with severe metabolic acidemia (pH ≤ 7.20, PaCO2 < 45 mmHg) and moderate to severe acute renal insufficiency (GRADE 2+, STRONG AGREEMENT).
Rationale Metabolic acidosis accompanying states of shock is often multifactorial, with hyperlactatemia and renal insufficiency being involved first and foremost, plus potentially associated loss of bicarbonate. Several retrospective, observational, single-center [
92‐
94] or prospective, multicenter studies [
95] were insufficient to draw conclusions regarding the role of sodium bicarbonate. Two randomized, prospective, crossover, single-center physiological studies in 10 [
96] and 14 patients [
97] concluded that administration of sodium bicarbonate did not have a more favorable effect than saline solution on hemodynamic parameters measured by pulmonary arterial catheter in patients with metabolic lactic acidosis (blood bicarbonate ≤ 22 or 17 mmol/L and arterial blood lactate > 2.5 mmol/L).
A controlled, randomized, prospective multicenter study in 400 patients (pH ≤ 7.20, blood bicarbonate ≤ 20 mmol/L and PaCO2 ≤ 45 mmHg and blood lactate > 2 mmol/L or SOFA score > 4) compared the effect of sodium bicarbonate administration (4.2% q.s. pH ≥ 7.30) with the absence of such administration on a principal composite endpoint (day-28 mortality and/or presence of at least one organ failure at day 7, according to the SOFA score). The authors reported no effect of alkalinization (71% of patients in the control arm and 66% of patients in the bicarbonate arm reached the composite endpoint; the estimated absolute difference was − 5.5% ([95% CI − 15.2% to 4.2%], p = 0.24). The probability of day 28 survival was 46% [95% CI − 40% to 54%] in the control group and 55% [95% CI 49% to 63%]; p = 0.09 in the bicarbonate group.
In the a priori defined stratum “acute renal insufficiency—AKIN 2–3,” 74 (82%) of the 90 patients of the control group and 64 (70%) of the 92 patients of the bicarbonate group reached the composite endpoint (estimated absolute difference: − 12.3%, 95% CI − 26.0% to − 0.1%; p = 0.0462). The probability of survival at day 28 was 46% [95% CI 35% to 55%] in the control group and 63% [95% CI 52% to 72%] in the bicarbonate group (p = 0.0283).
These results were confirmed in multivariate analysis. In the general population and the “acute renal insufficiency” stratum, the patients randomized to the control arm received renal replacement therapy (RRT) more often and for longer than the patients of the bicarbonate arm (52% need for RRT in the control arm vs 35% in the bicarbonate arm,
p < 0.001) [
98].
R3.7—Sodium bicarbonate should not be administered routinely in the therapeutic management of circulatory arrest, apart from pre-existing hyperkalemia or poisoning by membrane stabilizers (GRADE 1−, STRONG AGREEMENT).
Rationale Since the 1999 French consensus conference, the role of sodium bicarbonate alkalinization in the therapeutic management of the cardiac arrest has been evaluated in 5 retrospective studies [
99‐
103] and a prospective, randomized, double-blind, controlled multicenter study [
104]. Four retrospective studies showed an increase in the frequency of resumption of spontaneous circulatory activity in patients treated with sodium bicarbonate [
99,
101‐
103] and one reported decreased hospital survival in patients treated with sodium bicarbonate [
100]. The randomized clinical trial (792 patients) found no difference in survival between the patients treated with sodium bicarbonate (7.4%) and those receiving a placebo (6.7%,
p = 0.88). The use of sodium bicarbonate in patients could be reserved for pre-existing hyperkalemia or poisoning by membrane stabilizers [
105].
R3.8—Sodium bicarbonate should probably not be administered to patients with diabetic ketoacidosis (GRADE 2−, STRONG AGREEMENT).
Rationale The administration of sodium bicarbonate transiently increases pH and may limit the deleterious cardiovascular and cellular energy effects of acidemia. However, the administration of sodium bicarbonate is associated with a risk of hypokalemia, hypernatremia, hypocalcemia, rebound alkalemia, and water–sodium overload [
2]. A pathophysiological study in 39 patients has recently shown altered microvascular endothelial reactivity at the acute phase of diabetic ketoacidosis. This endothelial dysfunction was more marked when the arterial pH was below 7.20 and the vascular reactivity improved after 24 h of treatment. However, the administration of sodium bicarbonate was not tested in this observational study [
106].
Since the 1999 French consensus conference, the role of sodium bicarbonate alkalinization in therapeutic management of ketoacidosis was reassessed in a retrospective single-center study [
107] comparing 44 patients treated with bicarbonate and 42 untreated patients. The authors found no effect of sodium bicarbonate on the rate of correction of acidemia, as in previous studies, all of which were conducted in small populations [
108].
R3.9—The experts suggest administering sodium bicarbonate in the therapeutic management of salicylate poisoning, whatever the pH value (EXPERT OPINION).
Rationale Salicylate poisoning is rare and potentially fatal. Toxicological expertise is needed to ensure optimal therapeutic management. The aim of bicarbonate administration is twofold: induce alkalemia to limit the passage of salicylate into the central nervous system and alkalinization of urine to promote renal excretion of salicylate [
109,
110]. An old observational study in a small number of patients suggested that simple alkalinization leads to renal excretion of salicylate equal to or even greater than that of forced diuresis, alkaline diuresis, or not [
111]. The administration of sodium bicarbonate should be subject to close monitoring as it is associated with a risk of hypokalemia, hypernatremia, hypocalcemia, alveolar hypoventilation, and fluid overload [
2,
109]. In the case of severe poisoning, the experts suggest renal replacement therapy (cf. R3.13) and continued alkalinization between renal replacement therapy sessions until salicylate is completely eliminated.
Should renal replacement therapy be used in severe metabolic acidosis, and if so in what situations?
R3.10 In case of shock and/or acute renal insufficiency, the experts suggest initiation of renal replacement therapy if the pH is below or equal to 7.15 in the absence of severe respiratory acidosis and despite appropriate treatment (EXPERT OPINION).
Rationale There are no randomized controlled studies with mortality as the main endpoint that compare the initiation or not of renal replacement therapy in severe metabolic acidosis. The recommendations presented here come mostly from retrospective observational studies and case reports.
According to a questionnaire administered by the European Society of Intensive Care Medicine, 74% of intensivists consider metabolic acidosis (without indication of severity) to be a criterion for initiation of renal replacement therapy [
112].
The plasma bicarbonate or pH cut-off authorizing renal replacement therapy could be deduced from the results of randomized studies comparing the effect on mortality of early or delayed initiation of renal replacement therapy in acute renal insufficiency. In 101 surgery patients, Wald et al. [
113] found no difference in mortality as a function of the timing of renal replacement therapy, and plasma bicarbonate at its initiation was similar in the two groups: 20.7 ± 4.3 vs 20.1 ± 4.4 mmol/L. In 231 surgery patients with KDIGO stage-2 acute renal insufficiency, Zarbock et al. [
114] found at initiation of renal replacement therapy similar plasma bicarbonate levels in the early and late arms: 20.9 ± 3.6 mmol/L vs 20.7 ± 3.7 mmol/L. Mortality was significantly lower in the early initiation group.
In the AKIKI study [
115] in 619 patients with KDIGO stage-3 acute renal insufficiency, intention-to-treat analysis showed that pH and plasma bicarbonate were significantly lower in the late renal replacement therapy group (hard criteria for renal replacement therapy, including pH ≤ 7.15, rate of renal replacement therapy: 50%) than in the early renal replacement therapy group (6 h after inclusion, rate of renal replacement therapy: 100%): bicarbonate 16.6 ± 5.6 vs 18.9 ± 4.9 mmol/L (
p < 0.001) and pH 7.25 ± 0.15 vs 7.30 ± 0.12 (
p < 0.001). There was no difference in mortality between groups.
The IDEAL ICU study [
116] included 488 septic shock patients with RIFLE stage F acute renal insufficiency randomized to 2 arms (renal replacement therapy started within 12 h following inclusion, rate of renal replacement therapy 97% versus renal replacement therapy started 48 h after inclusion in the absence of resolution of acute renal insufficiency, rate of renal replacement therapy: 62%). There was no difference in mortality (58 vs 54%) and the study was stopped as medical care was deemed futile. A pH ≤ 7.15 was a criterion for initiation of renal replacement therapy. Of the 41 patients in the late arm, 20 had a pH of 7.10.
The BICAR-ICU study [
98] compared intravenous administration of 4.2% sodium bicarbonate (q.s. pH > 7.30) with a control arm without infusion of bicarbonate in patients with severe metabolic acidosis (pH ≤ 7.20, bicarbonate < 20 mmol/L and PaCO
2 ≤ 45 mmHg) and a SOFA score ≥ 4 or arterial blood lactate ≥ 2 mmol/L. This randomized, controlled, intention-to-treat study was stratified according to age, AKIN stage 2 or 3 acute renal insufficiency, and septic shock. Renal replacement therapy was used if 2 of the 3 following criteria applied: pH < 7.20 after 24 h, hyperkalemia, or urine output < 0.3 mL/kg/h over 24 h. In the acute renal insufficiency sub-group of 182 patients, the probability of survival at day 28 was 46% [95% CI 35% to 55%] in the control group and 63% [95% CI 52% to 72%] in the bicarbonate group (
p = 0.0283).
R3.11—In case of lactic acidosis suggestive of metformin poisoning, the experts suggest early initiation of renal replacement therapy when there is an organ dysfunction or in the absence of improvement in the first hours of therapeutic management (EXPERT OPINION).
Rationale Metformin-associated lactic acidosis is defined by arterial lactate above 5 mmol/L and pH below 7.35 during metformin treatment. Its incidence is low: from 10 to 12/100,000 [
117,
118]. A 2015 literature review identified 175 publications (no randomized trial) reporting high mortality (30 to 50%) [
119].
Yeh H-C et al. [
117] collated case reports and studies from 1977 to 2014 (3 studies, 142 case reports) in 253 patients and found a mortality of 16.2%. Factors associated with mortality were mechanical ventilation and lactate level (17 vs 22 mmol/L,
p < 0.01), but not pH, plasma bicarbonate, or level of metformin. A lactate level above 20 mmol/L was significantly associated with mortality.
A retrospective study conducted in Northern Italy from 2010 to 2015 collated 117 cases and reported 78.3% survival [
118]. On average, at initiation of renal replacement therapy, the pH was below 7.04 and blood lactate above 12 mmol/L.
As the metformin dose is not always available and its prognostic value is subject to discussion [
119], renal replacement therapy should be initiated without delay when there is an organ dysfunction or when there is no improvement in the first hours of therapeutic management. Renal replacement therapy is intended to correct water–electrolyte and acid–base disturbances and to ensure metformin clearance [
119].
R3.12—In case of methanol or ethylene glycol poisoning, the experts suggest initiation of renal replacement therapy if the anion gap is above 20 mEq/L or if there is renal insufficiency or visual impairment (EXPERT OPINION).
Rationale Methanol poisoning and ethylene glycol poisoning are rare and potentially fatal. Expertise is needed to ensure optimal therapeutic management including, if necessary, specific intensive care procedures.
In alcohol poisoning (methanol and ethylene glycol), the pH at admission is correlated with the prognosis [
120,
121]. A pH below 7.0 is predictive of death [
122], whereas a pH above 7.22 is associated with survival [
123]. The plasma anion gap (> 24 mEq/L or > 20 mEq/L in the case of hemodynamic instability) is correlated with the level of formate and with the prognosis [
124].
Circulating methanol is removed by the kidney with a clearance of about 5 to 6 mL/min, which represents approximately 25 to 50% of its systemic elimination before its conversion to formic acid (responsible for the toxicity). This conversion is inhibited by intravenous administration of ethanol or fomepizole. The clearance of methanol achieved by intermittent hemodialysis ranges between 77 and 400 mL/min, and between 17 and 48 mL/min if renal replacement therapy is continuous [
125,
126].
R3.13—In metabolic acidosis associated with salicylic acid poisoning, the experts suggest initiation of renal replacement therapy when there is neurological involvement and/or if the salicylic acid concentration is above 6.5 mmol/L (90 mg/dL) and/or if the pH is less than or equal to 7.20 (EXPERT OPINION).
Rationale Salicylate poisoning is rare and potentially fatal. Expertise is needed to ensure optimal therapeutic management comprising, if necessary, specific intensive care procedures.
A 2015 literature review by a group of experts [
110] found 84 publications, 80 of which related to case reports or patient cohorts and to a randomized controlled trial, and collated 143 patients with salicylate poisoning. The authors concluded that salicylic acid is highly dialyzable and that intermittent hemodialysis is the preferred modality. They also concluded that development of acidemia should be considered as a warning sign because it indicates the onset of an organ dysfunction (lactic acidosis, ketoacidosis, renal, and/or respiratory insufficiency). In addition, the presence of acidemia increases the entry of salicylate into the central nervous system and the risk of cerebral edema.
A more recent retrospective study [
127] in 56 mechanically ventilated patients with blood salicylate above 50 mg/dL reported 76% mortality. Failure to use renal replacement therapy was associated with increased mortality and survival was zero when blood salicylate was above 5.8 mmol/L, i.e., 80 mg/dL. However, no data were available on potential poisoning with other compounds or on the causes of death.
Given the limited volume and quality of the data, it is difficult to determine a toxic threshold accurately. However, it appears that above 6.5 mmol/L (90 mg/dL) the risk of death is high, even in the absence of clinical signs.
Should minute ventilation be increased in mechanically ventilated patients with metabolic acidosis?
R3.14—The experts suggest compensating for acidemia by increasing respiratory frequency without inducing intrinsic positive end-expiratory pressure, with a maximum of 35 cycles/min and/or a tidal volume up to 8 mL/kg of body mass, and by monitoring plateau pressure. The aim of ventilation is not to normalize pH. A target pH greater than or equal to 7.15 seems reasonable. Medical treatment of metabolic acidosis and of its cause should be envisaged concomitantly, as ventilatory compensation can only be symptomatic and temporary (EXPERT OPINION).
Rationale The control of breathing brings into play three types of interconnected structures: the control center commonly called the “respiratory centers” in the central nervous system at the level of the brainstem; the motor components of the respiratory system comprising the muscles of the upper airways, the thoracic cage, and the abdomen; and the receptors (chemoreceptors, muscle proprioceptors, airway and lung receptors) that transmit setpoints constantly to the respiratory centers (PCO
2, pH, lung distension, respiratory muscle load…). Thus, the respiratory centers receive sensory and humoral information that enables homeostasis, while optimizing the energy cost of each respiratory cycle. The central chemoreceptors on the ventral side of the brainstem respond rapidly and strongly to minimal variations in the pH and PCO
2 of the cerebrospinal fluid and blood. The hydrogen ion seems to be a determinant stimulus [
128].
In metabolic acidosis, the physiological response is an increase in alveolar ventilation [
129] that is constant, whatever the cause and severity of acidosis [
130]. The stimulation of chemoreceptors in metabolic acidosis is responsible for an increase in tidal volume rather than tachypnea [
130,
131]. Its efficacy depends not only on alveolar ventilation, but also on the hemodynamic state and integrity of the respiratory system [
129,
132].
As yet there are no specific data concerning ventilatory management of intubated-ventilated patients with metabolic acidosis. Though acidosis is conventionally associated with a poor prognosis [
133], it has potentially protective effects. Apart from the severity of acidosis, its mechanism and how it arises seem to be prognostic factors that should be taken into account.
The correction of metabolic acidosis by increasing respiratory frequency and/or tidal volume is questionable. Current data on protective ventilation are abundant and recommend keeping a tidal volume of about 6 mL/kg of body mass. Given the hemodynamic effects of metabolic acidosis, it seems reasonable to adapt the respiratory frequency to achieve a pH greater than or equal to 7.15 [
134‐
136], without exceeding 35 cycles/min, as data in animal models suggest that a high minute ventilation has deleterious effects [
137,
138], which are more marked when there is lung involvement.