The use of bicarbonate is a source of eternal disagreement.  Bicarbonate has a shameful history of being abused in situations where it’s unhelpful (e.g. cardiac arrest).  This has impugned its reputation, giving it an aura of ignorance and failure.  Consequently, bicarbonate is underutilized in some situations where it might actually help.

Warming up:  Some context for BICAR-ICU

Which metabolic acidoses are most treatable with bicarbonate? (1)

Not all metabolic acidoses are created equal.  My pre-study bias is as follows:

• NAGMA (Non-Anion-Gap Metabolic Acidosis, a.k.a. hyperchloremic metabolic acidosis) essentially represents a bicarbonate deficiency (2).  Giving bicarbonate makes physiologic sense in treating this.  Patients with healthy kidneys will eventually regenerate bicarbonate on their own, but this can take days;  exogenous bicarbonate hastens recovery.
• Uremic acidosis might be treated with bicarbonate (especially in the context of hyperkalemia, more on this below).  Bicarbonate has long been used in uremia to stave off acidosis, in attempts to avoid dialysis.
• Ketoacidosis (especially diabetic ketoacidosis) probably doesn’t benefit from bicarbonate therapy.  The most effective approach is to treat ketoacidosis itself.  When you're tempted to give bicarbonate for diabetic ketoacidosis, give some extra IV insulin instead.
• Lactic acidosis doesn’t appear to benefit from bicarbonate therapy either.  Again, treatment is ideally directed at the underlying cause of lactic acidosis.

A brief word on bicarbonate for the hyperkalemic patient with metabolic acidosis

Three physiologic mechanisms seem to explain the effect of bicarbonate solutions on potassium: (3)

1. In the context of metabolic acidosis, increasing the pH shifts potassium into cells and thereby improves hyperkalemia.
2. Some bicarbonate solutions are strongly hypertonic.  Administration of hypertonic fluids pulls water out of cells, which pulls potassium out along with it (a phenomenon known as solute drag).  This will tend to increase the serum potassium level.
3. Large volumes of potassium-free fluid can decrease the potassium level simply via dilution.  This effect comes into play when giving substantial volumes of isotonic bicarbonate (e.g., over a liter)(4).

Ampules of concentrated 8.4% bicarbonate have a neutral effect on potassium level (Blumberg 1988, Blumberg 1992, Kim 1996, Kim 1997).  In this situation, pH effect (#1) seems to cancel out the solute drag effect (#2).  Little volume is administered, so mechanism #3 is irrelevant.

On the flip side, isotonic 1.3% sodium bicarbonate appears to reduce potassium levels in the context of acidosis (Gutierrez 1991, Blumberg 1992)(5).  In this situation pH shifts and direct dilution both serve to decrease the potassium, whereas there is no solute drag effect.  The BICAR-ICU study uses 4.2% sodium bicarbonate, which might be expected to cause a mild decrease in potassium (based on its intermediate tonicity).

A brief word on bicarbonate therapy & sodium concentration

Ampules of 8.4% bicarbonate are quite hypertonic.  Giving several ampules to a patient with a normal baseline sodium level will rapidly cause problems regarding hypernatremia.

Isotonic 1.3% bicarbonate solves the hypernatremia problem but creates a problem regarding volume overload.  For example, a 70-kg person with a bicarbonate of 12 mEq/L would require about two liters of isotonic bicarbonate to normalize their bicarbonate level.  For patients with euvolemia or hypervolemia, this can rapidly become problematic.

BICAR-ICU utilizes 4.2% bicarbonate.  This is a clever compromise between 8.4% bicarbonate (which would cause lots of hypernatremia) and 1.2% bicarbonate (which would cause a large volume load).

pH-guided fluid resuscitation

Whenever you are giving fluid to a patient, that is an opportunity to affect their pH.  A poor choice of fluid will exacerbate abnormalities, whereas a clever choice may help.  pH-guided fluid resuscitation is summarized below (more detail here).

pH adjustment should ideally be achieved during resuscitation, because significant changes in pH will require giving a large volume of fluid.  If fluid resuscitation is performed without attention to pH, it will be too late to adjust the pH later on (for example, after the patient has reached euvolemia then it would be dangerous to then give several additional liters of isotonic bicarbonate).

Use of bicarbonate to avoid dialysis

Bicarbonate would be expected to avoid dialysis for various reasons:

1. Acidosis is a common dialysis indication.  Administration of bicarbonate to a patient with uremic acidosis will improve the bicarbonate level and the pH.
2. Hyperkalemia is a common dialysis indication.  Administration of bicarbonate that isn't excessively hypertonic may reduce the potassium level.
3. Administration of bicarbonate to a patient with hyperchloremic metabolic acidosis might improve renal function due to resolution of intra-renal vasoconstriction (more on this here).

Bicarbonate has been used for decades to stave off uremic acidosis and avoid dialysis.  However, it has never been shown that this strategy truly improves clinical outcomes.  For example, some might argue that early administration of dialysis is beneficial (in which case, giving bicarbonate to forestall dialysis could actually be dangerous).

BICAR-ICU trial

This is an open-label multicenter RCT involving 389 patients within 48 hours of ICU admission.  Inclusion required the presence of metabolic acidosis (pH < 7.20, PaCO2 < 45 mM) plus either SOFA score >3 or lactate >2 mM (6).  Exclusion criteria included ongoing loss of bicarbonate from the GI tract or kidneys (defined as volume loss >1500 ml/day or renal tubular acidosis), stage IV chronic kidney disease, ketoacidosis, or exogenous acid load (e.g. salicylate or methanol poisoning).  Patients in the bicarbonate arm were treated with 4.2% bicarbonate as shown here:

Patients were well matched at baseline.  57% were medical and 43% surgical.  Most patients had a primary diagnosis of sepsis (53%), hemorrhagic shock (22%), or cardiac arrest (9%).  Patients were mostly intubated (83%) and on vasopressors (80%).  Baseline serum bicarbonate levels were identical, with a median of 13 mEq/L and interquartile range of 10-15 mEq/L.  There were no differences in the volume of crystalloid administered, so outcomes don’t simply reflect the consequences of giving more fluid.

Neutral primary endpoint

The primary outcome was a composite of all-cause mortality at 28 days or the presence of at least one organ failure at day 7.  This is problematic for several reasons:

• It’s nearly impossible for an ICU study to demonstrate a significant reduction in all-cause mortality (discussed further here).
• Ideally, composite outcomes should combine similar items to retain some sort of meaning (e.g. adverse cardiovascular outcomes).  It’s strange to combine survival a month with the presence of single-organ failure after one week.  What is this melange supposed to mean?
• The study was powered to detect a 15% difference in the primary endpoint, based on a retrospective study that correlated acidosis with poor outcomes.  Correlational studies generally over-estimate causation.  Over-estimation of the effect size led to an under-estimate of the number of patients needed for the trial, causing the trial to be under-powered.
• 24% of patients in the control group received sodium bicarbonate, violating the study protocol.  This further degrades the power of the study, increasing the likelihood of a false-negative result.
• 109 patients were excluded from the study because they had already received bicarbonate.  Selective removal of patients whom the clinicians judged to definitely need bicarbonate skewed the trial population towards patients who might benefit less from bicarbonate – again increasing the likelihood of a false-negative result.

As might be expected, the primary endpoint was neutral (with a non-significant trend towards bicarbonate reducing mortality/morbidity).  Within a pre-specified subgroup of patients with more severe kidney injury at randomization (AKIN score of II-III), bicarbonate did improve mortality (46% vs. 63%, p=0.017).

Positive major secondary endpoint:  Dialysis avoidance

Like most studies involving mortality as a primary endpoint, the primary endpoint is doomed to fail so the secondary endpoints are actually more useful.  Rather than throwing out the study, it may be reasonable to designate one of the secondary endpoints as the major secondary endpoint.  For this study, the rate of requiring dialysis would be a natural choice as major secondary endpoint (7).  As discussed previously on this blog, the primary rationale for resuscitation with bicarbonate is avoidance of dialysis.

Bicarbonate reduced the need for dialysis from 52% to 35% (p=0.0009).  This correlates with a number needed to treat (NNT) of six patients to prevent one patient from requiring dialysis.  Among patients who received dialysis, bicarbonate delayed the initiation of dialysis by about twelve hours (p<0.0001), providing further evidence that bicarbonate tends to avert the need for dialysis.  As discussed above, these results come as no surprise.  Nonetheless, it’s nice to see this born out in a prospective multi-center RCT.

Specific indications for dialysis are shown above.  Bicarbonate therapy caused reductions in hyperkalemia, acidemia, and oliguria.  This suggests that bicarbonate may be acting via several mechanisms (discussed earlier above), rather than solely tweaking the pH.

What is the clinical significance of this result? (8)

• Emergent dialysis is invasive (requiring catheter placement), uncomfortable, and expensive. Avoiding dialysis is a meaningful patient-centered outcome on its own right.
• Dialysis can cause numerous complications (e.g. line infection, thrombosis, cytokine release as leukocytes pass through the filter, risks associated with anticoagulation for the filter, delayed physical therapy, hemodynamic instability due to fluid shifts).  The fact that bicarbonate reduced mortality among patients with more severe kidney injury hints that avoiding dialysis might offer the patient additional downstream benefits.  At the very least, it shows that using bicarbonate to avert dialysis isn’t harmful.

Effect on potassium level (minor secondary endpoint)

As discussed above, prior evidence suggests that 4.2% bicarbonate might cause a moderate reduction in serum potassium level.  Patients in the bicarbonate group did indeed experience a lower potassium level than patients in the control group, as shown above.  Bicarbonate reduced the incidence of hyperkalemia (32% vs. 49%; p= 0.0006).  This supports the use of less concentrated bicarbonate solutions for treatment of hyperkalemia in patients with metabolic acidosis.

Side-effects & weaknesses of the study

Perhaps the most important weakness of the study is that it doesn’t differentiate between NAGMA, lactic acidosis, and uremic acidosis.  It’s possible that bicarbonate could be beneficial in some conditions, but not others.  However, pre-specified subgroup analysis of patients with kidney injury does allow the study to suggest that bicarbonate is beneficial for uremic acidosis.

Patients treated with bicarbonate did experience an increased rate of metabolic alkalosis.  This likely represents “rebound alkalosis” occurring when lactic acidosis is treated with bicarbonate (after lactate is cleared, residual exogenous bicarbonate causes a metabolic alkalosis).  Other side effects of bicarbonate were increased rates of hypernatremia and hypocalcemia.

Another potential limitation is that the study investigated the use of hypertonic bicarbonate (4.2%)(9). The clinical effects of bicarbonate are probably mostly due to its effect on acid-base status.  However, the effects of 4.2% bicarbonate may not be completely generalizable to other concentrations of bicarbonate (10).

Caveats to applying this study in clinical practice

Evidence-based medicine requires integration of clinical evidence from numerous sources (ranging from RCTs to basic science studies).  No RCT is ever perfectly applicable to all of our patients.  BICAR-ICU provides some general support for the use of bicarbonate to treat metabolic acidosis, but it doesn’t prove whether to use it for specific different indications.  My opinion on these situations is as follows:

• NAGMA: Bicarbonate is a rational therapy for this.  Bicarbonate is already standard therapy in some forms of NAGMA (e.g. renal tubular acidosis).  With the evidence from BICAR-ICU, this therapy is further supported and is reasonably well justified.
• Uremic acidosis: The subset of patients in BICAR-ICU with renal failure seemed to derive the greatest benefit (in terms of mortality reduction and dialysis avoidance).  This supports the use of bicarbonate for uremic acidosis, which is already fairly common practice.
• Pure lactic acidosis:  There is currently little evidence to support the use of bicarbonate here.  BICAR-ICU doesn’t provide sufficient evidence to support the use of bicarbonate for a patient with an isolated lactic acidosis (11).

It should also be kept in mind that BICAR-ICU was performed on sick patients within 48 hours of ICU admission.  It doesn’t necessarily apply to a stable patient with chronic acidosis.  Thus, BICAR-ICU is not a justification to start blindly blanketing every acidotic patient with bicarbonate.

Implications for choosing saline vs. balanced crystalloid

Large volume normal saline resuscitation will cause NAGMA.  In turn, NAGMA may be treated with bicarbonate to return the patient to a normal pH status.  These are the facts and they are undisputed (12).

What is disputed is whether NAGMA is actually harmful, or whether it just represents a meaningless numerical foible that we needn’t fret about.  Accumulating evidence suggests that preventing NAGMA with the use of balanced crystalloid is beneficial (most notably demonstrated in the recent SMART and SALT-ED trials).

The fact that bicarbonate appears to be clinically beneficial further bolsters the concept that NAGMA is detrimental.  Essentially, the BICAR-ICU trial is the flip side of the SMART and SALT-ED trials.

• Treatment of metabolic acidosis with bicarbonate in the ICU didn’t cause a measurable effect on mortality in this 389-patient study (nor should anyone expect it to).
• Bicarbonate therapy reduced the need for dialysis (35% vs 52%, p=0.0009).  The number needed to treat (NNT) is only six patients to avoid placing one patient on dialysis.
• The significance of these results is unclear because the study combined patients with different types of metabolic acidosis.  However, preexisting data suggests that bicarbonate is more effective for NAGMA and uremic acidosis, compared to lactic acidosis.  Pre-specified subgroup analysis shows that bicarbonate improves mortality and avoids dialysis in patients with more severe acute kidney injury, supporting the use of bicarbonate in this context.
• This study indirectly argues against the use of normal saline as a resuscitative fluid (since normal saline causes NAGMA, exerting a physiologically opposite effect on pH compared to bicarbonate).
• Secondary analyses suggested that 4.2% bicarbonate decreases potassium levels (please note, however, that highly concentrated 8.4% bicarbonate doesn’t work for hyperkalemia).

Related

• The Bottom Line Review of BICAR-ICU by Steve Mathieu
• pH guided resuscitation.
• Hyperchloremia & balanced solutions
  • Hyperchloremic metabolic acidosis may injure the kidney.
  • Get SMART: 9 reasons to stop using saline.
• Modern treatment of hyperkalemia

Notes

1. Use of bicarbonate as compensation for respiratory acidosis is a different topic for another day. It’s on the docket, we will get there eventually.  Also, please note that I’m not necessarily claiming this approach is perfect or the only possible approach – but it helps to have some starting ground before we jump into BICAR-ICU.
2. Or, if you prefer the Stewart Acid-Base scheme, a chloride excess… it’s all the same damn thing, however.
3. A fourth mechanism which may come into play is that excretion of bicarbonate by the kidneys tends to favor excretion of potassium as well. Thus, pushing the serum bicarbonate level to a mildly elevated range could theoretically promote renal excretion of potassium.  I haven’t listed that mechanism here in the interest of space and also because it is a delayed effect which also may not work in patients who are oligo/anuric.
4. Yeah, this is a bit like cheating, but it still works. The extracellular fluid volume of an average person is about 14 liters.  If you give two liters of isotonic bicarbonate, this will expand the extracellular fluid volume by ~15%.  If your baseline serum potassium is in the range of 6-9 mEq/L, this will drop the potassium level by ~1 mEq/L solely due to dilutional effects (ignoring any additional physiologic effects of the bicarbonate on pH, potassium shifting etc.).
5. Isotonic bicarbonate is generated by adding 150 mEq of sodium bicarbonate (e.g. three 50-mEq ampules) to a liter of 5% dextrose (D5W).
6. The study also technically required that the serum bicarbonate level be <20 mEq/L. However, this inclusion criteria is redundant with other inclusion criteria. In order to have a pH <7.20 with a pCO2 below 45mm, the bicarbonate must be <17 mEq/L based on the Henderson-Hasselbach equation.  The bicarb <20 mEq/L inclusion criteria may actually be a bit misleading because it implies that some patients in the study had a bicarbonate of 19 mEq/L – which doesn’t appear to be possible.
7. A prior blog outlined a series of criteria for a secondary endpoint in order for it to be considered a “Major Secondary Endpoint.” Dialysis rate meets all of these criteria:  it is pre-specified, it could have conceivably been the primary endpoint, the entire study population was included in this analysis, it is no more distal than the primary endpoint, the original study is large and of high quality, and it is a clinically relevant patient-centered endpoint.
8. Some will doubt the significance of this result, because it is a secondary endpoint.However, please consider the number of protocol violations plusselective removal of patients already treated with bicarbonate.  Both of these factors will tend to reduce the apparent effect of bicarbonate.  The fact that any positive results were found in this study at all is actually a bit surprising.
9. The authors cannot be faulted for selecting 4.2% bicarbonate, because use of an isotonic bicarbonate solution would have caused logistic problems regarding volume overload. In bedside clinical practice, however, it might be ideal to select the bicarbonate concentration based on the patient’s volume status, acid/base status, and sodium level.   For example, a patient with hyponatremia and NAGMA will be well treated by hypertonic bicarbonate (which will improve both processes).
10. If you wish to mimic the precise treatment used in this study (4.2% bicarbonate), this may be achieved as follows. Give the patient a 1:1 ratio of 8.4% bicarbonate (the concentration used in most ampules in the United States, 1 mM/ml) and 5% dextrose in water (D5W). After these fluids equilibrate with the patient’s blood in circulation, this will be equivalent to administration of 4.2% bicarb (with a wee bit of extra glucose).
11. Bicarbonate could be considered in a patient with a mixed picture involving a combination of anion-gap and non-gap acidosis (with the intention of treating the non-gap component).
12. Please see: A Few Good Men. It’s official folks, the PulmCrit blog is now only one degree of separation from Kevin Bacon.

• Author
• Recent Posts

Social Me

Josh Farkas

Josh is the creator of PulmCrit.org. He is an associate professor of Pulmonary and Critical Care Medicine at the University of Vermont.

Social Me

Latest posts by Josh Farkas (see all)

• PulmCrit Wee – Loading infusion auto-titration (LIAT) for infused medications with intermediate half-lives - March 23, 2025
• PulmCrit Wee: Michelin Chest Syndrome - March 15, 2025
• PulmCrit: ADAPT and SCREEN trials are full of sound and fury, signifying little - December 13, 2024