Metabolic acidosis frequently is present in postsurgical critically ill patients1,2; however, the clinical significance of postoperative metabolic acidosis is unclear. Although some studies have found an independent association between low pH or base excess and mortality rate,3–5 others have not.6
In a 2006 retrospective study, Gunnerson et al7 observed that critically ill patients with lactic acidosis had a greater rate of mortality than patients with hyperchloremic acidosis; however, mortality did not differ between patients with hyperchloremic acidosis and those without acidosis.
Recent data also have suggested the adverse effects of hyperchloremia.8 A trial in surgical patients in 2001 randomized intravenous fluids in either a 9% saline solution or a balanced solution (Hartmann’s solution and 6% hetastarch in balanced electrolyte and glucose), and found greater rates of hyperchloremic acidosis and reduced gastric mucosal perfusion in those who received 0.9% saline solution.9 Other studies10,11 also have reported greater rates of renal dysfunction with hyperchloremic solutions.
In contrast, in a double-blind randomized comparison of 9% saline solution with Ringer’s lactate in patients undergoing aortic reconstruction surgery, researchers found greater rates of acidosis in the 9% saline solution group but found no difference in mortality or complications.12
To clarify the potential role of metabolic acidosis in high-risk postoperative patients, we assessed the incidence of metabolic acidosis in high-risk surgical patients. We hypothesized that different types of acidosis would have different relationships with postoperative complications and 30-day mortality.
This observational multicenter study was approved by the Research Ethics Committee of the School of Medicine, University of São Paulo, the Ethics Committee for the Analysis of Research Projects of the Hospital das Clinicas, the Hospital do Servidor Público Estadual, and the Hospital do Câncer de Barretos. It was registered in the National System of Information on Ethics in Research Involving Humans (Opinion No. CAAE-0084.1.338.000-09). The participating hospitals have a total of 90 intensive care unit (ICU) beds, and the functional organization charts of all ICUs were similar in the referred hospitals.
Patients who had undergone high-risk surgeries and been admitted to the ICU between September 3, 2012, and December 3, 2013, were enrolled in this study. We modified the previously published criteria13 to identify the high-risk surgeries and included patients with a requested ICU postoperative stay and at least one of the following clinical conditions: severe cardiorespiratory illness (coronary artery disease or some problem with the aorta, chronic obstructive pulmonary disease, or ischemic stroke), extensive surgery for carcinoma (esophagectomy, total gastrectomy, liver resection, pancreatectomy, colectomy, rectal resection, or cystectomy), a duration of surgery greater than 6 hours, or severe multiple trauma. Intraoperative criteria for high risk included acute massive blood loss (hematocrit <20%), circulatory shock (mean arterial blood pressure <60 mm Hg) requiring the use of vasopressors, or previously severe nutritional disorder. In addition, those aged >65 years with limited physiological reserve in at least 1 vital organ (liver, bilirubin above normal range; kidney, creatinine above normal range; heart, New York Heart Association II to IV; and lungs, dyspnea after walking on level ground).
We obtained signed informed consent from all patients. If the patient was unable to provide consent preoperatively, a responsible adult (legal representative) provided informed consent for the patient.
Patients were excluded from the study if their ICU length of stay was <24 hours, or if they had a short life expectancy (such as those with cancer without treatment perspective or treatment limitations, such as do not intubate and resuscitate). In addition, we excluded patients with liver failure (Child–Pugh class B or C), patients with renal insufficiency (50 mL/min creatinine clearance as assessed by the Cockcroft–Gault formula or prior hemodialysis), or those with a previous diagnosis of diabetes or 2 measured glucose levels of >126 mg/dL after 8 hours of fasting during the perioperative period.14 We excluded patients in these last 2 groups because they may have developed metabolic acidosis from the underlying disease and not the surgical process. Moreover, greater rates of hyperlactatemia could still occur during liver failure because of the low lactate clearance.15 Exclusion criteria were evaluated in all patients until 12 hours from ICU admission, except for patients with an ICU stay <24 hours, who were automatically excluded.
All patients underwent 2 blood sample collections, 1 at the time of admission to the ICU (D1) and the other 12 hours after admission (D2), to determine the levels of arterial blood gases, lactate, serum sodium, albumin, and chlorine. Upon admission to the ICU, pH and base difference results were used to divide patients into 2 groups—those with metabolic acidosis and those without. At the same time, patients with metabolic acidosis were classified further according to their lactate and albumin-corrected anion gap (AG) values into 3 groups: hyperlactatemic metabolic acidosis, metabolic acidosis with increased AG, and metabolic acidosis with normal AG (hyperchloremic).
According to the prespecified standardized protocol, patients with metabolic acidosis were defined as those with a base excess of arterial blood gas <−4.0 mmol/L.1 Hyperlactatemia was diagnosed when serum lactate values exceeded 2.0 mmol/L.16 The measurement of AG was corrected by albumin and was considered as elevated when values were >12 mmol/L.17 To avoid the overlap among acidosis groups, the lactate values in the nonhyperlactatemia groups were ≤2.0 mmol/L. By definition, if a patient presented with increased AG and hyperlactatemia, then that patient was not classified as having hyperchloremic acidosis. If the patient had an increased AG and normal lactate values (≤2.0 mmol/L), then they were not classified as hyperlactatemic or hyperchloremic.
We used the AG to estimate the presence of unmeasured anions and considered a value of 12 mmol/L to be normal. We thus corrected the AG for abnormal albumin levels using the following equation18 (taking 4.5 as the normal concentration of albumin in g/dL):
All patients were followed until hospital discharge or for 30 days after surgery, regardless of duration of hospital stay. The primary outcome of the study was 30-day postoperative mortality rate. Secondary outcomes included in-hospital mortality, and the incidence of postoperative ICU complications defined as follows. (1) Cardiovascular dysfunction—the need for vasoactive drugs for more than 1 hour despite a central venous pressure of 8.0 mm Hg and pulse pressure variation <13% and titrated to achieve a mean arterial blood pressure >60 mm Hg.19,20 In any instance in which the monitoring of central venous pressure or pulse pressure variation was not used, fluid challenge with crystalloids was applied up to 1000 mL before any vasopressor drug was administered. (2) Respiratory dysfunction—Pao2/fraction of inspired oxygen ratio <200 in patients with no previous cardiac disease or need for reintubation or difficulty removing the endotracheal tube during the postoperative period (defined by >1 failed spontaneous breathing trial). Trials were conducted once patients were judged as being ready for extubation. A spontaneous breathing test was performed with 7 cm H2O pressure support and a positive end-expiratory pressure of 5 cm H2O for at least 30 minutes. Success was defined as tidal volumes at least 6–8 mL/kg of the ideal body weight and a respiratory frequency below 25 breaths per minute. (3) Renal insufficiency—creatinine increase of 30% or urine output of less than 400 mL over 24 hours or the need for renal replacement therapy or renal Sepsis-Related Organ Failure Assessment (SOFA) score of 2 or greater.21 Renal dysfunction was determined by a renal SOFA score >2 during the first 8 postoperative days. (4) Neurologic disorder—assessed by changes in the Richmond Agitation-Sedation Scale (RASS) score22: acute and fluctuating change in RASS over 24 hours, <>0 and psychomotor agitation was determined if the patient received a RASS score of 2+ or greater. (5) Coagulation dysfunction—30% drop in platelets with respect to preoperative values, and platelet count value below 100,000 µL−1.23 To minimize possible errors and differentiate chronic and acute postsurgical complications, a specially trained physician and nurse team evaluated all patients before and after surgery.
In addition, we assessed each patient for the presence of infection during his or her ICU stay, as well as the length of ICU and hospital stays. Infections were classified according to the locations of the infectious foci, etiologic agents, and severity. The characterization of foci and infectious agents were based on the criteria of the Centers for Disease Control and Prevention.24
All outcomes were analyzed by group classification during period D2, which determined the persistence of patients in each group after a 12-hour ICU stay, thus allowing the assessment of behavior associated with metabolic acidosis and of outcomes between groups.
In addition to acid–base tests, we also calculated the Simplified Acute Physiology Score (SAPS 3)25 at the time of admission, and the clinical and laboratory variables during the perioperative period were used to calculate the Physiological and Operative Severity Score for the Enumeration of Mortality and Morbidity (POSSUM).26 The assessment of organ dysfunction was determined by use of the SOFA score.21 The American Society of Anesthesiology physical status classification27 was used to identify the severity of the patient’s illness.
An initial pilot study28 was used to estimate the sample size. In this pilot study, the mortality for the patients with acidosis was 15%, compared with 2% for the patients without acidosis. These data are in accordance with the mortality rate for surgical patients in the general population (∼5%)29 and in the high-risk population (15%).30–33 Using a 1:1 ratio in accordance with the data from the pilot study and the minimal clinically significant difference between the groups, we calculated the required sample size to detect a 15% mortality rate in patients with acidosis and a 5% mortality rate in patients without acidosis. With a 95% confidence interval and 80% statistical power, it was determined that a sample size of at least 310 patients would be required (155 patients per group) to compare the mortality rates for 30 days (MedCalc 11.5.1; sampling: comparison of proportions).
All data were entered into an electronic database (Microsoft Excel; Microsoft, Redmond, WA) and subsequently analyzed with the statistical programs SPSS Version 20.0 (IBM Corp., Armonk, NY) and MedCalc version 13.2.0 (MedCalc Software, Ostend, Belgium). Initially, we described the demographic, clinical, and physiological characteristics of the patients included in the study. Frequencies and percentages were calculated for the description of categorical variables. Quantitative variables were described by the use of measures of central and dispersion tendencies.
The characteristics of patients with metabolic acidosis were compared with those of patients without acidosis. Assumptions that variables were normally distributed were tested with the analysis of histograms by skewedness and kurtosis value and by the Shapiro–Wilk test. Continuous variables with asymmetric distributions were assessed using the Kruskal–Wallis test when comparing more than 2 variables and when comparing 2 variables Mann–Whitney U test was applied. When analysis of variance was used, the assumption of normality was assessed plotting standard residuals for each group. The χ2 test was used to measure associations in categorical variables. The χ2 test for linear trends was used for categorical variables ordered in some sense. This procedure assesses whether there was a trend in the proportions with the characteristic over the categories of the second factor.
For multiple comparisons of categorical variables, we used Log-linear analysis in a saturated model to verify the main effects and interactions from groups compared with centrals outcomes. This process allowed comparing the outcome in the highest order effect among acidosis groups. The goodness-of-fit test was used to assess model fit; this statistic was nonsignificant. All statistical tests were 2-tailed, and for multiples comparison, the nonmetabolic acidosis group was the reference. In addition, the level of significance was performed among groups of comparison and in pairwise combinations of independent group proportions. The Bonferroni correction set the significance cutoff P value; it was used to detect the time points at which the differences were significant in multiple comparisons.
Homogeneity between the groups with regard to age, SAPS 3, and POSSUM scores was assessed with analysis of variance. Statistical differences between the acidosis groups, nonacidosis group, and with respect to 30-day mortality rates postoperatively were assessed via the Log-linear analysis, as well as with respect to renal dysfunction until 8 days postoperatively.
We performed a Cox progressive and conditional regression model, applying stepwise selection with backward elimination to determine the risk of death, comparing patients who remained or did not remain in the metabolic acidosis groups after a 12-hour ICU stay.
Variables considered for Cox regression analysis were used if they exhibited a statistically significant association (P < .05) in univariate analysis or if they presented pairwise interaction possibilities (sex, age, and prognosis score SAPS 3). Afterward, variables with interactions were tested in the main regression model.
Centers were included in the regression model to verify any possible confounding effects of variability in the clinical practices of the 3 hospitals. P value and confidence intervals (CIs) of the groups in comparison with 30-day mortality were obtained from this regression model. Subsequently, estimates of survival after 30 days were calculated for the groups using the Kaplan–Meier curve adjusted by Cox regression, with comparisons made using Mantel Cox test for linear trends.
From September 2012 to December 2013, 2638 high-risk surgical patients were admitted to the ICU. Of these patients, 1839 were excluded from the study, with the main reason for exclusion being diabetes (Figure 1).
On admission to the ICU, 59.1% of the patients were diagnosed with metabolic acidosis, and 40.9% of the patients were classified as not having acidosis. Approximately 23.9% of the patients with acidosis were hyperchloremic, 21.3% were hyperlactatemic, and 13.9% had increased AG acidosis (Figure 2).
No significant differences were observed in the severity scores (SAPS 3, POSSUM, and American Society of Anesthesiologists), age, sex, and length of surgery between the groups (Table 1). The median (interquartile range) base difference in the patients with acidosis was −6.5 (−9.3 to −4.3) mmol/L vs −0.6 (−1.5 to 0.6) mmol/L in the patients without acidosis (Mann–Whitney U test, P < .001).
Patients with normal AG acidosis (hyperchloremic) received more 0.9% saline solution during the intraoperative period and exhibited greater chloride levels during the postoperative period. The amount of other types of fluids received did not differ between groups (Table 2).
More patients with persistent acidosis (67.8%) exhibited postoperative complications than did patients without acidosis (59.3%; χ2 test, P = .03). The types of complications also differed between the different acidosis groups. Patients who had hyperlactatemia that persisted in the postoperative period had worse overall outcomes in terms of cardiovascular and renal dysfunction (Table 3). In Log-linear analysis, patients with hyperlactatemic acidosis were more likely to develop renal dysfunction in the first 7 days (higher-order effects, P < .001; Figure 3).
ICU and hospital mortality were both greater in individuals whose acidosis persisted 12 hours postsurgery. The ICU mortality rate in the hyperlactatemic group was 22.6% vs 15.7% in the increased AG group and 14.5% in the hyperchloremic group. Mortality in the nonacidosis group was 10.3% (χ2 for trend test P < .001). Hospital mortality was also greater (χ2 for trend test P < .001) in the individuals with persistent hyperlactatemia (30.1%) compared with other groups—increased AG (22.8%), hyperchloremia (17.1%), and nonacidosis (10.3%) during the postoperative period (Figure 4).
By the 30-day follow-up, 110 patients (17.8%) had died. The hyperlactatemic metabolic acidosis group had a mortality rate of 30.1%, the increased AG group had 24.3%, and the hyperchloremic acidosis group had 18.4%. The nonacidosis patients had a 10.3% rate of mortality. Log-linear analysis—interaction acidosis groups × 30 days’ mortality rate likelihood ratio χ2 = 15.59, P = .001.
In patients who were still acidotic after 12 hours, multivariate Cox regression, adjusted for confounding factors, such as sex, age, SAPS 3 score, gastrointestinal and neurological surgeries, and the centers involved, identified an independent mortality effect in the group with hyperlactatemic metabolic acidosis (P = .029, hazard ratio [HR] 1.7, 95% CI 1.02–2.97) compared with the group without acidosis after the 30-day follow-up period; however, results obtained in patients with hyperchloremic metabolic acidosis (P = .380, HR 1.47, 95% CI 0.75–2.90) and in those with an increased AG (P = .157, HR 1.68, 95% CI 0.85–3.32) were not statistically significant in Cox regression model. In addition, no substantive interaction with SAPS 3 was observed, and the proportional hazards assumption was satisfied (Table 4).
We also used an adjusted survival curve by Cox regression to assess the trend in the 30-day follow-up among the groups; it showed a worse 30-day survival in hyperlactatemic group compared with the other groups (P = .03; Figure 5).
This study showed that surgical patients exhibiting different types of postoperative acidosis had different morbidity and 30-day mortality rates and that patients with persistent hyperlactatemic metabolic acidosis had worse outcomes than those with other forms of acidosis.
Previous studies that assessed the relationship between metabolic acidosis and outcomes in critically ill patients have focused on specific types of acidosis, for example, lactate acidosis,34,35 or the severity of the acidosis.3,36,37 Some studies38, however, have raised the question as to whether acidosis is a marker or cause of postoperative complications.
Evidence suggests that acidosis may interfere with hemodynamics39 and innate immunity.40 Research also suggests that different types of acidosis are associated with different clinical consequences.41 Although observational studies cannot show a causal relationship, our study indicates that different types of metabolic acidosis in high-risk surgical patients are associated with different mortality and complication rates.
If the severity of metabolic acidosis (characterized by base deficit) is the primary determinant of perioperative outcome, the type of metabolic acidosis should not play a significant role. However, we found that the type of acidosis has an independent effect on the outcomes of high-risk surgical patients.
In comparison with the other etiologies of metabolic acidosis, we found that hyperlactatemic acidosis resulted in the greatest 30-day and hospital mortality rates, a finding that is not surprising as lactic acidosis has been associated with high mortality rates in critically ill patients.16,32,34,42 The relevance of high perioperative lactate levels, however, may differ between surgical patients and other types of critically ill patients.43 During surgery, cellular oxygen consumption may be lower,44 and liver clearance may be reduced.43 Thus, in the immediate postoperative period, serum lactate levels may not represent the actual clinical conditions of surgical patients. Our study demonstrates that patients with persistent hyperlactatemic acidosis after surgery also have poorer outcomes when compared with patients with other types of acidosis.
Although recent evidence suggests that hyperchloremic acidosis also may adversely affect outcomes,45,46 our study found no effect of hyperchloremic acidosis on mortality. This finding is consistent with that of others47 and suggests that the clinical relevance of hyperchloremic acidosis requires further study.
Our study has some limitations. Because of the observational design, we were unable to establish a causal relationship between acidosis and outcomes. Another limitation with our study was the exclusion of diabetic patients, because they are frequently subjected to high-risk surgeries. These patients were excluded from the study because lactate levels in diabetics correlate with glucose levels.48 This relationship raises the possibility that lactic acidosis in diabetic patients may be because of either altered glucose metabolism or hypoperfusion. The ability of epinephrine to stimulate muscle Na+/K+ pumps may play an enhanced role in diabetic patients, because stress and insulin deficiency enhance the production of counter-regulatory adrenergic hormones, including epinephrine.49 We chose to exclude diabetic patients from the study because of this alternative mechanism for lactic acidosis beyond tissue hypoperfusion. We also excluded patients with short ICU stays and low life expectancies (such as those with cancer without treatment perspective) because their outcomes could not be measured properly. Moreover, we excluded those with Child–Pugh class B or C cirrhosis and those with creatinine clearance <50 mL/min because they may have developed metabolic acidosis from the underlying disease and not because of the surgery. Because of these exclusions, our data may not be generalizable to all patient groups.
In addition, we did not classify acidosis using stronger ion definitions; however, several studies7,50,51 have reported good correlations between the model of AG adjusted for albumin and other models for the characterization of metabolic acidosis, apart from the fact that the model adjusted for albumin is easier to use. Finally, procedures conducted during the postoperative period and the possible influences on patient outcomes were not assessed. This effect was minimized, however, because outcome assessments were adjusted for confounders and were based on the states of the patients 12 hours after ICU admission. Although some issues were discussed, few prospective multicenter studies on surgical patients in the literature have assessed the effect of different causes of metabolic acidosis on the prognosis of this population.
High-risk surgical patients frequently exhibit metabolic acidosis during the postoperative period. We found that patients with postoperative hyperlactatemic acidosis had worse outcomes than patients with other types of acidosis or without acidosis. Different types of acidosis, however, are associated with divergent outcomes, with hyperlactatemic and elevated anion gap acidosis having the worst outcomes compared with patients without acidosis. In our study, persistent hyperlactatemia represented an independent effect on mortality in the high-risk surgical population. More studies are needed to determine the mechanisms and clinical significance of perioperative metabolic acidosis.
We thank Henrique Katayama, MD, for helping in preparation of the research project and statisticians Rogério Ruscitto do Prado, PhD, and Cleyton Zanardo de Oliveira, MSc, for assistance in statistical review.
Name: João Manoel Silva Jr, MD, PhD.
Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript.
Name: Amanda Maria Ribas Rosa de Oliveira, MD.
Contribution: This author helped design the study, conduct the study, and write the manuscript.
Name: Fernando Augusto Mendes Nogueira, MD.
Contribution: This author helped conduct the study.
Name: Pedro M. M. Vianna, MD.
Contribution: This author helped conduct the study.
Name: Cristina Prata Amendola, MD.
Contribution: This author helped conduct the study and write the manuscript.
Name: Maria José Carvalho Carmona, MD, PhD.
Contribution: This author helped design the study and write the manuscript.
Name: Luiz M. Sá Malbouisson, MD, PhD.
Contribution: This author helped design the study, analyze the data, and write the manuscript.
This manuscript was handled by: Avery Tung, MD, FCCM.
1. Gauthier PM, Szerlip HM. Metabolic acidosis in the intensive care unit. Crit Care Clin. 2002;18:289308, vi.
2. Kraut JA, Madias NE. Metabolic acidosis: pathophysiology, diagnosis and management. Nat Rev Nephrol. 2010;6:274285.
3. Davis JW, Parks SN, Kaups KL, Gladen HE, O’Donnell-Nicol S. Admission base deficit predicts transfusion requirements and risk of complications. J Trauma. 1996;41:769774.
4. Dunham CM, Siegel JH, Weireter L, et al. Oxygen debt and metabolic acidemia as quantitative predictors of mortality and the severity of the ischemic insult in hemorrhagic shock. Crit Care Med. 1991;19:231243.
5. Smith I, Kumar P, Molloy S, et al. Base excess and lactate as prognostic indicators for patients admitted to intensive care. Intensive Care Med. 2001;27:7483.
6. Balasubramanyan N, Havens PL, Hoffman GM. Unmeasured anions identified by the Fencl-Stewart method predict mortality better than base excess, anion gap, and lactate in patients in the pediatric intensive care unit. Crit Care Med. 1999;27:15771581.
7. Gunnerson KJ, Saul M, He S, Kellum JA. Lactate versus non-lactate metabolic acidosis: a retrospective outcome evaluation of critically ill patients. Crit Care. 2006;10:R22.
8. Chowdhury AH, Cox EF, Francis ST, Lobo DN. A randomized, controlled, double-blind crossover study on the effects of 2-L infusions of 0.9% saline and plasma-lyte® 148 on renal blood flow velocity and renal cortical tissue perfusion in healthy volunteers. Ann Surg. 2012;256:1824.
9. Wilkes NJ, Woolf R, Mutch M, et al. The effects of balanced versus saline-based hetastarch and crystalloid solutions on acid-base and electrolyte status and gastric mucosal perfusion in elderly surgical patients. Anesth Analg. 2001;93:811816.
10. Shaw AD, Bagshaw SM, Goldstein SL, et al. Major complications, mortality, and resource utilization after open abdominal surgery: 0.9% saline compared to Plasma-Lyte. Ann Surg. 2012;255:821829.
11. Yunos NM, Bellomo R, Hegarty C, Story D, Ho L, Bailey M. Association between a chloride-liberal vs chloride-restrictive intravenous fluid administration strategy and kidney injury in critically ill adults. JAMA. 2012;308:15661572.
12. Waters JH, Gottlieb A, Schoenwald P, Popovich MJ, Sprung J, Nelson DR. Normal saline versus lactated Ringer’s solution for intraoperative fluid management in patients undergoing abdominal aortic aneurysm repair: an outcome study. Anesth Analg. 2001;93:817822.
13. Gurgel ST, do Nascimento P Jr. Maintaining tissue perfusion in high-risk surgical patients: a systematic review of randomized clinical trials. Anesth Analg. 2011;112:13841391.
14. American Diabetes Association. Standards of medical care in diabetes—2013. Diabetes Care
. 2013;36(suppl 1):S11S66.
15. Luft FC. Lactic acidosis update for critical care clinicians. J Am Soc Nephrol. 2001;12(suppl 17):S15S19.
16. Rishu AH, Khan R, Al-Dorzi HM, et al. Even mild hyperlactatemia is associated with increased mortality in critically ill patients. Crit Care. 2013;17:R197.
17. Fencl V, Jabor A, Kazda A, Figge J. Diagnosis of metabolic acid-base disturbances in critically ill patients. Am J Respir Crit Care Med. 2000;162:22462251.
18. Figge J, Jabor A, Kazda A, Fencl V. Anion gap and hypoalbuminemia. Crit Care Med. 1998;26:18071810.
19. Leone M, Asfar P, Radermacher P, Vincent JL, Martin C. Optimizing mean arterial pressure in septic shock: a critical reappraisal of the literature. Crit Care. 2015;19:101.
20. Donati A, Loggi S, Preiser JC, et al. Goal-directed intraoperative therapy reduces morbidity and length of hospital stay in high-risk surgical patients. Chest. 2007;132:18171824.
21. Vincent JL, de Mendonça A, Cantraine F, et al. Use of the SOFA score to assess the incidence of organ dysfunction/failure in intensive care units: results of a multicenter, prospective study. Working group on “sepsis-related problems” of the European Society of Intensive Care Medicine. Crit Care Med. 1998;26:17931800.
22. Sessler CN, Gosnell MS, Grap MJ, et al. The Richmond Agitation-Sedation Scale: validity and reliability in adult intensive care unit patients. Am J Respir Crit Care Med. 2002;166:13381344.
23. Stéphan F, Hollande J, Richard O, Cheffi A, Maier-Redelsperger M, Flahault A. Thrombocytopenia in a surgical ICU. Chest. 1999;115:13631370.
24. Horan TC, Andrus M, Dudeck MA. CDC/NHSN surveillance definition of health care-associated infection and criteria for specific types of infections in the acute care setting. Am J Infect Control. 2008;36:309332.
25. Silva Junior JM, Malbouisson LM, Nuevo HL, et al. Applicability of the simplified acute physiology score (SAPS 3) in Brazilian hospitals. Rev Bras Anestesiol. 2010;60:2031.
26. Bennett-Guerrero E, Hyam JA, Shaefi S, et al. Comparison of P-POSSUM risk-adjusted mortality rates after surgery between patients in the USA and the UK. Br J Surg. 2003;90:15931598.
27. Soliani P. New classification of physical status. Anesthesiology. 1963;24:111.
28. Silva JMJ, Oliveira AMRR, Marti YN, et al. Outcome of surgical patients who present acidosis postoperatively. Critical Care. 2011;15:P64.
29. Pearse RM, Moreno RP, Bauer P, et al.; European Surgical Outcomes Study group for the Trials groups of the European Society of Intensive Care Medicine, the European Society of Anaesthesiology. Mortality after surgery in Europe: a 7 day cohort study. Lancet. 2012;380:10591065.
30. Irish Critical Care Trials Group. Intensive care for the adult population in Ireland: a multicentre study of intensive care population demographics. Crit Care. 2008;12:R121.
31. Lobo SM, Rezende E, Knibel MF, et al. Epidemiology and outcomes of non-cardiac surgical patients in Brazilian intensive care units. Rev Bras Ter Intensiva. 2008;20:376384.
32. Lobo SM, Rezende E, Knibel MF, et al. Early determinants of death due to multiple organ failure after noncardiac surgery in high-risk patients. Anesth Analg. 2011;112:877883.
33. Lobo SM, Salgado PF, Castillo VG, et al. Effects of maximizing oxygen delivery on morbidity and mortality in high-risk surgical patients. Crit Care Med. 2000;28:33963404.
34. Broder G, Weil MH. Excess lactate: an index of reversibility of shock in human patients. Science. 1964;143:14571459.
35. Vincent JL, Dufaye P, Berré J, Leeman M, Degaute JP, Kahn RJ. Serial lactate determinations during circulatory shock. Crit Care Med. 1983;11:449451.
36. Rutherford EJ, Morris JA Jr, Reed GW, Hall KS. Base deficit stratifies mortality and determines therapy. J Trauma. 1992;33:417423.
37. Siegel JH, Rivkind AI, Dalal S, Goodarzi S. Early physiologic predictors of injury severity and death in blunt multiple trauma. Arch Surg. 1990;125:498508.
38. Rocktaeschel J, Morimatsu H, Uchino S, Bellomo R. Unmeasured anions in critically ill patients: can they predict mortality? Crit Care Med. 2003;31:21312136.
39. Kellum JA, Song M, Venkataraman R. Effects of hyperchloremic acidosis on arterial pressure and circulating inflammatory molecules in experimental sepsis. Chest.2004;125:243248.
40. Kellum JA, Song M, Li J. Science review: extracellular acidosis and the immune response: clinical and physiologic implications. Crit Care. 2004;8:331336.
41. 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. 2004;32:11201124.
42. Bakker J, de Lima AP. Increased blood lactate levels: an important warning signal in surgical practice. Crit Care. 2004;8:9698.
43. Silva Junior JM, Oliveira AMRR, Silveira BR, et al. Intraoperative lactate measurements are not predictive of death in high risk surgical patients. Rev Bras Ter Intensiva. 2010;22:229235.
44. Rivers EP, Ander DS, Powell D. Central venous oxygen saturation monitoring in the critically ill patient. Curr Opin Crit Care. 2001;7:204211.
45. Silva Junior JM, Neves EF, Santana TC, Ferreira UP, Marti YN, Silva JM. The importance of intraoperative hyperchloremia. Rev Bras Anestesiol. 2009;59:304313.
46. McCluskey SA, Karkouti K, Wijeysundera D, Minkovich L, Tait G, Beattie WS. Hyperchloremia after noncardiac surgery is independently associated with increased morbidity and mortality: a propensity-matched cohort study. Anesth Analg. 2013;117:412421.
47. Burdett E, Dushianthan A, Bennett-Guerrero E, et al. Perioperative buffered versus non-buffered fluid administration for surgery in adults. Cochrane Database Syst Rev. 2012;12:CD004089.
48. Cox K, Cocchi MN, Salciccioli JD, Carney E, Howell M, Donnino MW. Prevalence and significance of lactic acidosis in diabetic ketoacidosis. J Crit Care. 2012;27:132137.
49. Bolli G, Cartechini MG, Compagnucci P, et al. Adrenergic activity and glycometabolic compensation in patients with diabetes mellitus. Minerva Med. 1979;70:37833795.
50. Mallat J, Michel D, Salaun P, Thevenin D, Tronchon L. Defining metabolic acidosis in patients with septic shock using Stewart approach. Am J Emerg Med. 2012;30:391398.
Copyright © 2016 International Anesthesia Research Society
51. Zampieri FG, Park M, Ranzani OT, et al. Anion gap corrected for albumin, phosphate and lactate is a good predictor of strong ion gap in critically ill patients: a nested cohort study. Rev Bras Ter Intensiva. 2013;25:205211.