Introduction
The Δ anion gap (ΔAG) and Δ bicarbonate (ΔHCO3) ratio (ΔAG/ΔHCO3) is used to detect coexisting acid-base disorders in patients with high AG metabolic acidosis. In general, the ΔHCO3 accompanies an equivalent change in the ΔAG, and this apparent 1:1 stoichiometry has been used to identify concurrent acid-base disorders, such as metabolic alkalosis or normal AG metabolic acidosis; a ΔAG/ΔHCO3 below one suggests a coexisting normal AG metabolic acidosis, whereas a ΔAG/ΔHCO3 more than one to two suggests a coexisting metabolic alkalosis (1).
In lactic acidosis, traditional belief holds that lactate anions tend to remain in the extracellular (EC) fluid compartment, whereas protons that accompany the lactate are buffered outside of the EC fluid, in cells and bone. Additionally, lactate excretion by the kidney is usually decreased because of lactate absorption by sodium-lactate transporters, hypoperfusion, and acute renal dysfunction. Regardless of the explanation, the net result is a ΔAG/ΔHCO3 of more than one in lactic acidosis, usually approximately 1.6–1.8.
Notably, the duration of the lactic acidosis is thought to affect the ΔAG/ΔHCO3. The ΔAG/ΔHCO3 of 1.6–1.8:1 is thought to occur after the acidosis persists for several hours (1). Within the first 60 minutes of onset of lactic acidosis, classic teaching holds that the ΔAG/ΔHCO3 is initially 1:1, increasing with time over several hours (1). However, this classic 1:1 stoichiometry described early in the development of lactic acidosis is derived primarily from animal models (2,3), and only limited human data has investigated the ΔAG/ΔHCO3 in early lactic acidosis (4,5). The objective of this study was to examine the ΔAG/ΔHCO3 within the first hours of the development of lactic acidosis. A secondary objective was to examine potential pathophysiologic explanations for the observed ΔAG/ΔHCO3.
Materials and Methods
The study was a reanalysis of data by Rudkin et al. (6), which examined the correlation between arterial and peripheral venous pH and base excess in patients who experienced trauma. They concluded that venous blood gas (VBG) pH and base excess cannot be used interchangeably with the corresponding arterial blood gas (ABG) measurements. The study was a prospective study that enrolled a convenience sample of adult (age >18 years) trauma-designated patients from a single level-1 trauma center. The study enrolled 385 patients. When an ABG was obtained for clinical purposes, a peripheral VBG was drawn as soon as possible. Venous samples, including a chemistry panel and serum lactate, were drawn before initiation of intravenous fluid resuscitation. Data collected included collection times of the ABG and VBG, physiologic data including BP, and indicators of patient severity (Glasgow Coma Scale score, Trauma and Injury Severity scores).
Statistical Analyses
The association between duration of lactic acidosis and the ΔAG/ΔHCO3, Δlactate and ΔAG, arterial pH and the ΔAG/ΔHCO3, and serum chloride and the ΔAG/ΔHCO3 were examined using Pearson correlation, and linear regression models were constructed. Least-squares regression lines were calculated and plotted. Additionally, the association between ΔHCO3 and ΔAG was examined using Pearson correlation, a linear regression model was constructed, and 95% prediction intervals were computed.
Results
The reanalysis included 108 patients. A total of 63 patients had normal serum lactate levels (≤2.1 mmol/L), and 45 patients had elevated serum lactate levels (>2.1 mmol/L). The final sample of the original study included 346 patients. In the reanalysis, 148 patients were excluded because they were missing serum lactate or serum HCO3 values, and five patients were excluded because of clerical errors, leaving 193 patients remaining. To determine the group with the elevated serum lactate measurements (>2.1 mmol/L), 67 patients with an AG <7.1 mEq/L were excluded, 68 patients with a serum HCO3 >24 mEq/L were excluded, and 13 patients with a serum lactate <2.1 mmol/L were excluded, resulting in 45 patients in the elevated serum lactate group. To determine the group with normal serum lactate measurements (<2.1 mmol/L), of the 193 patients remaining, 63 patients were found to have serum lactate measurements of <2.1 mmol/L.
The patients with normal serum lactate levels had a mean AG of 7.1 mEq/L and a mean lactate level of 1.5 mmol/L; these values were used to calculate subsequent ΔAG and Δlactate values in the patients with elevated serum lactate levels. The mean lactate for the elevated lactate group was 4.89 mmol/L, with an SD of 2.36 mmol/L and range between 2.2 and 11.1 mmol/L. ΔAG/ΔHCO3 and Δlactate/ΔHCO3 were then calculated for the 45 patients who had elevated serum lactate levels (>2.1 mmol/L). In the group with the normal serum lactate levels, the mean serum potassium was 3.82 mEq/L with an SD of 0.42 mEq/L and a range of 2.9–5.0 mEq/L; the mean serum creatinine was 0.95 mg/dl, with an SD of 0.32 mg/dl and a range of 0.3–2.3 mg/dl. The elevated serum lactate group had a mean serum potassium of 3.6 mEq/L, with an SD of 0.48 mEq/L and a range of 2.5–4.7 mEq/L; the mean serum creatinine was 1.04 mg/dl, with an SD of 0.37 and a range of 0.4–2.8 mg/dl. Table 1 shows the patient characteristics. For the patients with elevated lactate levels, the average (±SD) patient age was 38.6±18.8 years old, with a preponderance of males (71%). The mechanism of injury was predominantly blunt trauma (71%).
Table 1. -
Patient characteristics
| Characteristics |
Normal Lactate (≤2.1 mmol/L) |
Elevated Lactate (>2.1 mmol/l) |
|
N
|
63 |
45 |
| Age (yr), mean±SD |
43.6±21.0 |
38.6±18.8 |
| Male, n/N (%) |
44/63 (70) |
32/45 (71) |
| Intubated, n/N (%) |
6/63 (10) |
10/45 (22) |
| Hypotensive (systolic BP ≤90 mm Hg), n/N (%) |
3/63 (5) |
8/45 (18) |
|
Mechanism of injury, n/N (%)
|
|
|
| Blunt trauma |
49/63 (78) |
32/45 (71) |
| Penetrating trauma |
6/63 (10) |
8/45 (18) |
| Burns |
5/63 (8) |
3/45 (7) |
The mean ΔAG/ΔHCO3 was 1.86, with an SD of 1.40; the mean Δlactate/ΔHCO3 was 1.21, with an SD of 1.06. The mean Δlactate/ΔHCO3 was also calculated using the reported upper range of normal for serum lactate (2.1 mmol/L), yielding a value of 0.95 with an SD of 0.91.
Figure 1 shows the linear regression model examining duration of lactic acidosis and ΔAG/ΔHCO3. The r=−0.14, with a P value of 0.37. The duration of lactic acidosis was estimated by determining the period of time that elapsed between the time from activation of emergency medical services to the draw time of venous blood in the emergency department. This time ranged from 25.0 to 166.0 minutes, with a mean time of 81.5 minutes (SD, 31.4).
;)
Figure 1.: Pearson correlation between duration of lactic acidosis and the ratio of Δ anion gap to Δ bicarbonate ( Δ AG/ Δ HCO 3 ). There is no association between ΔAG/ΔHCO3 and duration of lactic acidosis.r=−0.14, P=0.37.
Figure 2 shows the linear regression model examining Δlactate and ΔAG. r=0.50 with P=0.001. The R2 is 0.25. Figure 3 shows the linear regression model examining arterial pH and ΔAG/ΔHCO3. The r=0.10 with P=0.52. Figure 4 shows the linear regression model between serum chloride and ΔAG/ΔHCO3. The r=−0.15 with P=0.33. Supplemental Figure 1 shows the linear regression model between ΔHCO3 and ΔAG. The r=0.70, with P<0.001. Dashed lines represent the 95% prediction interval.
Figure 2.: Pearson correlation between Δ lactate and Δ anion gap ( Δ AG). The Δlactate explains 25% of the observed variance in the ΔAG. r=0.50, P=0.001.
Figure 3.: Pearson correlation between arterial pH and Δ AG/ Δ HCO 3 . There is no association between arterial pH and ΔAG/ΔHCO3. r=0.10, P 0.52.
Figure 4.: Pearson correlation between serum chloride and Δ AG/ Δ HCO 3 . There is no association between serum chloride and ΔAG/ΔHCO3.r=−0.15, P=0.33.
Discussion
The relationship between the ΔAG, which reflects changes in the concentration of unmeasured anions, and ΔHCO3 has been used to evaluate for complex acid-base disorders in patients with underlying high AG metabolic acidosis (1). The accumulation of a nonchloride-containing acid, such as lactic acid, in the blood results in a reduction in serum HCO3. The accompanying anion, such as lactate, is retained to maintain electroneutrality, resulting in a rise in the serum AG. Theoretically, the reduction in serum HCO3 corresponds to an equivalent increase in the AG, resulting in a ΔAG/ΔHCO3 of one. Hence, any deviation from this 1:1 stoichiometry may reflect a coexisting acid-base disorder in addition to the AG metabolic acidosis.
The existing literature reveals there is variable stoichiometry of ΔAG/ΔHCO3, depending on the specific type of organic acidosis. In lactic acidosis, the traditional belief is that, in the first few hours, the ΔAG/ΔHCO3 is approximately 1:1 (1). However, as the lactic acidosis persists beyond a few hours, the mean ΔAG/ΔHCO3 ratio is approximately 1.6–1.8 (1); the increased ΔAG/ΔHCO3 has most commonly been ascribed to the theory that, after a few hours, hydrogen ion buffering in cells and bone reach completion, while only a small fraction of the lactate remains in the intracellular fluid space, preferentially residing within the EC fluid compartment.
Only limited human data have investigated the ΔAG/ΔHCO3 in early lactic acidosis. The main objective of this study was to specifically examine the ΔAG/ΔHCO3 within the first hours of the development of lactic acidosis. Patients in this study had undergone trauma and, therefore, the lactic acidosis was a result of hypovolemic shock. The patient characteristics, including the relatively young mean age (38.6 years) and male predominance (71%), are consistent with the demographics typically seen in patients who have experienced trauma. The mean ΔAG/ΔHCO3 was 1.86 within the first hours of the development of lactic acidosis due to hypovolemic shock, and was not associated with duration of lactic acidosis (Figure 1). This contradicts the traditional belief that, in lactic acidosis, the ΔAG/ΔHCO3 is 1:1 within the first few hours, subsequently increasing to approximately 1.6–1.8 as hydrogen ion buffering in cells and bone reach completion (while the lactate preferentially resides in the EC fluid compartment) (1). Notably, of the 45 patients that had elevated serum lactate levels (>2.1 mmol/L), 19 patients had a pH >7.4 and respiratory alkalosis, which, in the setting of trauma, was likely related to factors such as pain, anxiety, and head injury. Of these 19 patients, seven also had a concurrent metabolic alkalosis, likely related to factors such as vomiting.
Although reviews of the literature have concluded that both animal and human studies indicate that the ΔAG/ΔHCO3 is 1:1 early in the course of lactic acidosis (1), a more detailed re-evaluation of the literature reveals divergent results in animal and human studies. Oster et al. (2) and Madias et al. (3) both conducted animal studies in dogs and rats, respectively. The mean ΔAG/ΔHCO3 in the first hour of lactic acidosis in these studies ranged from 1.00 to 1.25. Notably, in these animal models, lactic acidosis was produced via a lactic-acid infusion. Therefore, the lactic acid directly enters the EC space, presumably facilitating EC buffering of protons by HCO3. Concurrently, the lactate remains in the same compartment, resulting in a 1:1 stoichiometry and the ΔAG/ΔHCO3 of approximately one observed in these animal studies.
In contrast, two small human studies have yielded more conflicting results (4,5). Orringer et al. (4) found that, in early lactic acidosis, the mean ΔAG/ΔHCO3 ranged from 1.5 to 1.86 within the first 30 minutes after grand-mal seizures. A second study by Brivet et al. (5) also examined patients after grand-mal seizures and determined that the mean ΔAG/ΔHCO3 was 1.28, with 33% of patients exhibiting a ratio of <0.8. The mean ΔAG/ΔHCO3 of 1.86 in early lactic acidosis in our study serves to confirm the findings by Orringer et al. It is worth highlighting that both Orringer and Brivet examined lactic acidosis occurring after grand-mal seizures, with pathophysiology resulting from enhanced metabolic rate and accelerated aerobic glycolysis. In contrast, to our knowledge, this is the first study to evaluate early lactic acidosis from shock in humans, with pathophysiology resulting from hypoperfusion and decreased oxygen delivery.
As discussed above, in animal models of early lactic acidosis, the mean ΔAG/ΔHCO3 was approximately one in the first hour of lactic acidosis, attributable to the lactic acid infusion directly entering the EC space and facilitating EC buffering of protons by HCO3. In contrast, our results indicate that, in humans, endogenous lactic acidosis manifests a mean ΔAG/ΔHCO3 of approximately 1.8 within the first few hours, a ratio which does not appear to change as the acidosis persists beyond a few hours (78–9). In distinction to animal models of lactic acidosis, which are initiated in the EC space, early endogenous human lactic acidosis originates intracellularly. This has been postulated by some to result in intracellular buffering of protons while lactate is predominantly distributed in the EC fluid, resulting in a mean ΔAG/ΔHCO3 approaching 1.8 as the acidosis persists beyond a few hours. The ΔAG/ΔHCO3 of one reported early in the course of lactic acidosis has been theorized to occur because the hydrogen ion buffering in cells and bone takes a few hours to reach completion. Our findings that the mean ΔAG/ΔHCO3 early in the course of lactic acidosis is 1.86 suggest that, if this elevated ratio is indeed related to the differing distribution spaces of lactate and hydrogen ions, the hydrogen buffering in cells and bone occurs much more rapidly than previously described.
In addition to establishing that the mean ΔAG/ΔHCO3 in lactic acidosis during the first few hours is approximately 1.8, our study helps elucidate the reasons for the high ΔAG/ΔHCO3. Four possible explanations for the deviation in 1:1 stoichiometry have been proposed. First, it has been suggested that only a small fraction of lactate generated by cellular metabolism remains in the intracellular space, which, combined with decreased urinary excretion of lactate anion because of reduced renal function, results in lactate retention in the EC fluid compartment; in contrast, a significant proportion of hydrogen ions that accompany the lactate are buffered in cells and bone. The disparity between space of distribution of hydrogen ions compared with lactate has been proposed to result in a ΔAG/ΔHCO3 of more than one, usually approximating 1.6–1.8. The assumption underlying this theory is that the increase in AG relative to the decrement in ΔHCO3 reflects an increase in EC lactate. Our data demonstrates a mean ΔAG/ΔHCO3 of 1.86, consistent with prior studies of lactic acidosis in humans (4,78–9). By contrast, the mean Δlactate/ΔHCO3 ranged between 0.95 and 1.21, depending on the baseline lactate value. Additionally, Figure 2 demonstrates that the Δlactate can only explain 25% of the observed variance in the ΔAG. Taken together, these data suggest the high ΔAG that results in an increased ΔAG/ΔHCO3 does not appear to be primarily a result of increased EC lactate, as this proposed model would suggest. Second, in theory, organic or inorganic anions or cations may exhibit a pH-dependent contribution to the AG as the pH decreases. For example, the albumin concentration, which is the main contributor to the AG, increases with a rise in pH, resulting in an elevated AG in metabolic alkalosis. Conversely, a drop in pH may change the degree to which certain anions and cations contribute to the AG, resulting in a high AG and an elevated ΔAG/ΔHCO3. However, Figure 3 demonstrates there is no statistically significant association between arterial pH and ΔAG/ΔHCO3, with a P value of 0.52. Therefore, the pH-dependent contribution of anions or cations does not explain the increased ΔAG/ΔHCO3 observed in lactic acidosis. Third, Madias et al. (3) suggested that, on the basis of an animal model, hypochloremia may account for 30%–50% of the increment in AG seen in lactic acidosis, explaining the deviation from 1:1 stoichiometry and elevated ΔAG/ΔHCO3. The decrement in serum chloride results from extrusion of cellular cations and resultant expansion of the EC compartment during the buffering process in lactic acidosis. However, there is not a significant correlation between serum chloride and ΔAG/ΔHCO3 (Figure 4), arguing that hypochloremia does not play a significant role in the elevated ΔAG/ΔHCO3 observed in lactic acidosis in humans. Fourth, it is likely that the increase in AG that results in an elevated ΔAG/ΔHCO3 is caused by unknown organic anions (or, less likely, due to a decrease in unmeasured cations). Our data, consistent with previous literature, show that, in lactic acidosis, up to 75% of the observed variance in the AG is not explained by blood lactate and, therefore, lactic acid does not entirely account for the AG metabolic acidosis (1011–12). Although attempts to identify specific unknown organic anions in lactic acidosis have not been uniformly successful, some studies have identified increased concentrations of Krebs cycle intermediates, including citrate, isocitrate, α-ketoglutarate, succinate, malate, and d-lactate (13,14). Importantly, these unmeasured anions may better predict clinical outcomes than serum lactate levels (15,16). Given that our data argue against other possible explanations of the high ΔAG/ΔHCO3 seen in lactic acidosis, these unmeasured anions are the most likely cause and further work to identify them needs to be carried out.
Although ongoing research attempts to identify the unmeasured anions in lactic acidosis and better explain its pathophysiology, the ΔAG/ΔHCO3 remains a widely used tool to detect coexisting acid-base disorders in patients with lactic acidosis and other high AG metabolic acidosis. The wide 95% prediction interval suggests that ΔAG/ΔHCO3 should be used cautiously in the diagnosis of mixed acid-base disorders (Supplemental Figure 1). For example, although the mean ΔAG/ΔHCO3 was 1.86, which is consistent with prior studies, the SD was 1.40 and 15 of 45 patients had a ΔAG/ΔHCO3 of less than one. Additionally, it should be recognized that the AG is an insensitive screening tool for elevated blood lactate (10,17).
To our knowledge, our study was the first to evaluate the ΔAG/ΔHCO3 in early lactic acidosis from shock in humans. However, the study does have some limitations. This study was a reanalysis of data from a prior study, which prospectively enrolled a convenience sample of adult, trauma-designated patients. Although the patient selection did not involve formal random sampling, this is unlikely to have resulted in systematic bias, because our demographics are similar to a typical population of patients with trauma seen in the emergency department, and the ranges of laboratory values (including serum lactate, serum HCO3, pH, and AG) span the clinically important range. Because most forms of type-A lactic acidosis are due to marked tissue hypoperfusion, it is likely that the results of this study will apply not just to hypovolemia, but also to other pathophysiologic states characterized by tissue hypoperfusion, including sepsis, cardiac failure, or cardiopulmonary arrest.
Secondly, this study used mean normal values for serum AG and plasma HCO3. In the past, the normal range for the AG has been 12±4 mEq/L, with variations depending on the specific blood gas analyzer used. More recently, new technology and use of ion-selective electrodes has resulted in reporting of higher serum chloride concentrations and, consequently, lower AGs (18). In fact, several studies have reported the normal range for AG to be 6±3 mEq/L (19,20). Given that the AG will vary among different laboratories, the mean AG (7.1 mEq/L) from the patients with normal serum lactate levels was used as an approximation of the true normal values for the specific study patient population in the study center. There is clearly a wide interindividual variability in AG, and use of the actual normal baseline values of individual patients would ideally be used for calculation of ΔAG/ΔHCO3, although this was not feasible in our population of patients experiencing acute trauma and in most study settings. In addition to our study, virtually all prior clinical studies examining ΔAG/ΔHCO3 used mean normal values for AG and HCO3. Regardless, it is important to note that the practice of using mean normal values likely has an important effect on the calculation of the ratio and subsequent conclusions about the underlying pathophysiology. For example, in this study, patients with an AG ≤7.1 mEq/L were excluded because the mean AG of 7.1 mEq/L was the value used to calculate subsequent ∆AG values. It is important to note that this mirrors what occurs in clinical practice when using a normal value for AG; some patients invariably have an AG that falls below the normal value used, precluding calculation of the ∆AG. To avoid this, the actual normal baseline values of individual patients, if known, should be used.
Thirdly, the dataset did not include some pertinent parameters. It did not include serum albumin, so it was not possible to correct the AG for albumin level. The dataset also did not include serum phosphorus, although, due to its low concentration in the EC fluid, the buffering by inorganic phosphate in the EC fluid is likely negligible compared with that of HCO3. The dataset also lacked information about comorbidities.
Lastly, this reanalysis had a relatively small sample size. Only 45 patients had elevated serum lactate levels. However, the prior human studies of lactic acidosis after grand-mal seizures by Orringer et al. (4) and Brivet et al. (5) only included eight and 35 patients, respectively. Therefore, to date, this is the largest human study investigating the ΔAG/ΔHCO3 in early lactic acidosis.
The mean ΔAG/ΔHCO3 was 1.86:1 within the first hours of the development of lactic acidosis due to hypovolemic shock, not 1:1 as previously thought. The traditional belief is that this deviation in 1:1 stoichiometry results from intracellular buffering of protons while lactate is predominantly distributed in the EC fluid, although our study implicates unmeasured anions as the cause.
Disclosures
Original data collection was supported by National Institutes of Health/National Center for Research Resources grant M01 RR00827 (via the Emergency Medicine Foundation, Emergency Medicine Basic Research Skills, and the General Clinical Research Center, University of California, Irvine).
Funding
This work was supported by National Institutes of Health/National Center for Advancing Translational Science grant UL1TR000124.
Supplemental Material
This article contains the following supplemental material online at http://kidney360.asnjournals.org/lookup/suppl/doi:10.34067/KID.0000842019/-/DCSupplemental.
Supplemental Figure 1. Pearson correlation between ΔHCO3 and ΔAG. r=0.689, P<0.001. Dashed lines are the 95% prediction interval.
Acknowledgments
We thank the UCLA CTSI for their invaluable assistance with statistical analyses.
Portions of this manuscript were presented in poster form (SA-PO898, FR-PO271) at the American Society of Nephrology Kidney Week 2015 and 2018.
Author Contributions
T.R. Grogan and S.E. Rudkin were responsible for project administration; T.R. Grogan, S.E. Rudkin, and R.M. Treger were responsible for data curation; T.R. Grogan and R.M. Treger were responsible for formal analysis; S.E. Rudkin and T.R. Grogan were responsible for funding acquisition; and R.M. Treger conceptualized the study, was responsible for investigation and methodology, provided supervision, and wrote the original draft.
References
1. Kraut JA, Madias NE: Serum anion gap: Its uses and limitations in clinical medicine. Clin J Am Soc Nephrol 2: 162–174, 2007
2. Oster JR, Perez GO, Vaamonde CA: Relationship between blood pH and potassium and phosphorus during acute metabolic acidosis. Am J Physiol 235: F345–F351, 1978
3. Madias NE, Homer SM, Johns CA, Cohen JJ: Hypochloremia as a consequence of anion gap metabolic acidosis. J Lab Clin Med 104: 15–23, 1984
4. Orringer CE, Eustace JC, Wunsch CD, Gardner LB: Natural history of
lactic acidosis after grand-mal seizures. A model for the study of an anion-gap acidosis not associated with hyperkalemia. N Engl J Med 297: 796–799, 1977
5. Brivet F, Bernardin M, Cherin P, Chalas J, Galanaud P, Dormont J: Hyperchloremic acidosis during grand mal seizure
lactic acidosis. Intensive Care Med 20: 27–31, 1994
6. Rudkin SE, Kahn CA, Oman JA, Dolich MO, Lotfipour S, Lush S, Gain M, Firme C, Anderson CL, Langdorf MI: Prospective correlation of arterial vs venous
blood gas measurements in trauma patients. Am J Emerg Med 30: 1371–1377, 2012
7. Kim HY, Han JS, Jeon US, Joo KW, Earm JH, Ahn C, Kim S, Lee JS, Kim GH: Clinical significance of the fractional excretion of anions in metabolic acidosis. Clin Nephrol 55: 448–452, 2001
8. Waters WC 3rd, Hall JD, Schwartz WB: Spontaneous
lactic acidosis. The nature of the acid-base disturbance and considerations in diagnosis and management. Am J Med 35: 781–793, 1963
9. Oh MS, Carroll HJ, Goldstein DA, Fein IA: Hyperchloremic acidosis during the recovery phase of diabetic ketosis. Ann Intern Med 89: 925–927, 1978
10. Levraut J, Bounatirou T, Ichai C, Ciais JF, Jambou P, Hechema R, Grimaud D: Reliability of anion gap as an indicator of blood lactate in critically ill patients. Intensive Care Med 23: 417–422, 1997
11. Mikulaschek A, Henry SM, Donovan R, Scalea TM: Serum lactate is not predicted by anion gap or base excess after trauma resuscitation. J Trauma 40: 218–222, 1996
12. Mecher C, Rackow EC, Astiz ME, Weil MH: Unaccounted for anion in metabolic acidosis during severe sepsis in humans. Crit Care Med 19: 705–711, 1991
13. Forni LG, McKinnon W, Lord GA, Treacher DF, Peron J-MR, Hilton PJ: Circulating anions usually associated with the Krebs cycle in patients with metabolic acidosis. Crit Care 9: R591–R595, 2005
14. Forni LG, McKinnon W, Hilton PJ: Unmeasured anions in metabolic acidosis: Unravelling the mystery. Crit Care 10: 220–224, 2006
15. Dondorp AM, Chau TTH, Phu NH, Mai NTH, Loc PP, Chuong LV, Sinh DX, Taylor A, Hien TT, White NJ, Day NPJ: Unidentified acids of strong prognostic significance in severe malaria. Crit Care Med 32: 1683–1688, 2004
16. Kaplan LJ, Kellum JA: Initial pH, base deficit, lactate, anion gap, strong ion difference, and strong ion gap predict outcome from major vascular injury. Crit Care Med 32: 1120–1124, 2004
17. Iberti TJ, Leibowitz AB, Papadakos PJ, Fischer EP: Low sensitivity of the anion gap as a screen to detect hyperlactatemia in critically ill patients. Crit Care Med 18: 275–277, 1990
18. Kraut JA, Nagami GT: The serum anion gap in the evaluation of acid-base disorders: What are its limitations and can its effectiveness be improved? Clin J Am Soc Nephrol 8: 2018–2024, 2013
19. Jurado RL, del Rio C, Nassar G, Navarette J, Pimentel JL Jr: Low anion gap. South Med J 91: 624–629, 1998
20. Winter SD, Pearson JR, Gabow PA, Schultz AL, Lepoff RB: The fall of the serum anion gap. Arch Intern Med 150: 311–313, 1990
