Clinical and Experimental Nephrology

https://doi.org/10.1007/s10157-020-01867-y

ORIGINAL ARTICLE

Acid–base effects of continuous infusion furosemide in clinically

stable surgical ICU patients: an analysis based on the Stewart model
Kathryn A. Connor1 · Kelly Conn1 · David C. Kaufman2 · Curtis E. Haas3

Received: 27 January 2020 / Accepted: 25 February 2020

© Japanese Society of Nephrology 2020

Abstract
Objectives  We sought to test the strength of correlation between predicted and observed systemic acid–base status based
on the Stewart model equations during continuous infusion (CI) furosemide therapy.
Design, setting and participants  This was a prospective, single-center, observational study conducted in the Surgical ICU
of a large academic medical center. Ten critically ill patients who received CI furosemide were included.
Main outcomes and measures  The primary purpose was to characterize the relationship between changes in serum elec-
trolyte and acid–base status and the excretion of electrolytes in the urine during infusion of CI furosemide in critically ill
patients. As a secondary endpoint, we sought to evaluate the predictive application of the Stewart model. Over 72-h, intake
and output volumes, electrolyte content of fluids administered, plasma and urine electrolytes, urine pH, and venous blood
gases were collected. Predicted and observed changes in acid-based status were compared for each day of diuretic therapy
using Spearman’s correlation coefficient.
Results  The mean (SD) strong ion difference (SID) increased from 45.2 (3.2) at baseline to 49.6 (4.0) after 72 h of continuous
infusion furosemide. At Day 1, the mean SID (observed) (SD) was 47.5 (3.5) and the predicted SID was 49.5 (5.8). Day 1
observed plasma SID was positively correlated with the predicted SID (rs = 0.80, p = 0.01). By Days 2 and 3, the correlations
of observed and predicted SID were no longer statistically significant.
Conclusions and relevance  Using the Stewart model, increases in SID as an indicator of metabolic alkalosis due to the
chloruretic effects of furosemide were observed. Predicted and observed SID correlated well over the first 24 h of treatment.

Keywords  Acid–base balance · Critical care · Electrolytes · Nephrology · Surgery

Introduction

The analysis and comprehension of acid–base disturbances are

* Kathryn A. Connor essential for appropriately diagnosing and treating complicated
[email protected]
and vulnerable critically ill patients, but remain challenging
Kelly Conn for many clinicians, partly due to the lack of standardized
[email protected]
methodology for interpretation. Acid–base problems have tra-
David C. Kaufman ditionally been analyzed using a bicarbonate-based approach
[email protected]
that has been criticized for some of its limitations, such as the
Curtis E. Haas inability to completely diagnose complex patients [1–3]. A
[email protected]
more contemporary, quantitative approach to acid–base analy-
1
Department of Pharmacy Practice and Administration, St. sis, the Stewart method, is based on a physiochemical model
John Fisher College, Wegmans School of Pharmacy, 3690 that recognizes there are three independent factors which affect
East Ave., Rochester, NY 14618, USA acid–base status in mammals: changes in strong ion difference
2
Department of Surgery, The University of Rochester Medical (SID), partial pressure of ­CO2 ­(PCO2), and total weak acids and
Center, 601 Elmwood Ave., Rochester, NY 14642, USA their conjugate bases (Atot). Stewart’s principles provide a prac-
3
Department of Pharmacy, The University of Rochester tical and clinically relevant approach to interpreting acid–base
Medical Center, 601 Elmwood Ave., Rochester, NY 14642, disturbances, and are described in detail elsewhere [3–14].
USA

13
Vol.:(0123456789)
 Clinical and Experimental Nephrology

Surgical critical care patients often require aggressive vol- February to August 2012. This study was approved by and
ume resuscitation in the early phase of their Intensive Care conducted in compliance with requirements of the Univer-
Unit (ICU) care to maintain organ perfusion and decrease sity of Rochester Institutional Review Board, and informed
the risk of multiple organ failure [15]. After the capillary consent was obtained from patients or their surrogate, as
leak associated with the Systemic Inflammatory Response appropriate. Patients were included if they were clinically
Syndrome (SIRS) has resolved and the patient is hemody- stable and an order was written to start a CI loop diuretic
namically and clinically stable (e.g., afebrile, not septic, off (furosemide) during their admission to the SICU service.
vasopressor support and not requiring volume resuscitation), Patients were excluded if they were pregnant at the time
relatively aggressive intravenous (IV) loop diuretic therapy is of study enrollment, had open or dehisced abdominal fas-
appropriate to initiate. This diuresis is often crucial in revers- cia, diarrhea, vomiting or nasogastric drainage > 1L/day or
ing fluid overload and shortening the time to extubation and had recent transfusion needs [at least two units packed red
transfer out of the unit, thereby potentially avoiding adverse blood cells (PRBCs) in 48 h]. Demographic information
sequelae associated with prolonged mechanical ventilation was collected, including age, sex, height, weight, and race.
and extended ICU stays. Local practice at our institution usu- Clinical data were collected over a 72-h period, including
ally involves administering a furosemide continuous infu- intake and output volumes, electrolyte content of IV fluids
sion (CI) to efficiently achieve aggressive diuresis. However, administered, plasma electrolytes, urine electrolytes, urine
IV loop diuretic administration commonly causes untoward pH, and venous blood gases (VBGs). Urine and serum elec-
metabolic alkalosis with elevation of bicarbonate, without trolytes and VBGs were measured simultaneously as close to
other clinical evidence of intravascular volume depletion, the start of the diuretic infusion as possible, and then every
thereby limiting the achievement of overall net fluid loss. 12–24 h throughout the 72-h study period. If the infusion
This has been historically explained by describing the effect was stopped prior to 72 h, a final sample was obtained at
of diuretics on renal bicarbonate handling, however, this is the time of the termination of the infusion. Table 1 lists all
not consistent with Stewart’s physiochemical acid–base theo- specific laboratory parameters collected and the timing of
ries. A net loss of chloride results in a positive shift in the these collections, from time 0 h through 72 h. The amount of
SID, and a metabolic alkalosis [1]. diuretic received per day and any acetazolamide administra-
To our knowledge, there is currently no published clinical tion were also recorded.
data characterizing the effects of CI furosemide using the The following equation was used to calculate SID every
Stewart model. Thus, we sought to better understand the 12 h:
ability to predict the development of this therapy-limiting
SID = Na+ + K+ + Ca2+ + Mg2+ − (Cl− + lactate− ),
( )
adverse event using changes in SID, which may reveal man-
agement options to effectively prevent or delay the develop-
where ­Na+, ­K+, ­Ca2+, ­Mg2+, and ­Cl− represent the serum
ment of metabolic alkalosis. The purpose of this study was
concentrations of sodium, potassium, calcium, magnesium
to primarily characterize the relationship between changes in
and chloride, and lactate is the serum concentration of lac-
serum electrolyte and acid–base status and the excretion of
tate. All concentrations were reported as mEq/L. The mean
electrolytes in the urine during a CI loop diuretic in critically
SIDs (observed) were then calculated for Days 1–3. Sin-
ill patients. As a secondary endpoint, we sought to evaluate
gle point determinations were extrapolated over the time
the predictive application of the Stewart model.
between measurements.
The following equation was developed by the authors
based on principles of acid–base analysis, and was used to
Materials and methods
compare the mean observed vs predicted SIDs on Days 1–3:
This was a prospective, observational study in the Surgi-
[( )
Predicted SID = SIDIn × VolumeIn
cal ICU (SICU) at the University of Rochester Medical ( ) ( )]
− SIDOut × VolumeOut + SIDBL × ECFBL ∕
Center Strong Memorial Hospital, which is a large academic ( )
teaching hospital in Rochester New York. Ten critically ill ECFBL + Net Volume ,
adult patients were included over a 6-month period, from

Table 1  Laboratory variables Laboratory value Laboratory value

collected throughout 72-h study
period Serum chemistry, collected q12 h: Sodium, potassium, chloride, ­CO2, calcium
Urine chemistry, collected q12 h pH, sodium, potassium, chloride, ­CO2
Venous blood gas, collected q 24 h pH, ­pCO2, bicarbonate

13
Clinical and Experimental Nephrology

where ­SIDIn is the SID of fluids administered (mEq/L), Table 2  Baseline and demographic characteristics (n = 10)
­VolumeIn is the volume of fluids administered (L), ­SIDOut Characteristic Result
is the SID of the urine (mEq/L), ­VolumeOut is the volume
of ­urine, ­SIDBL is the baseline SID (mEq/L), ­ECFBL is the Age, years (mean, SD) 77.8 (6.4)
predicted extracellular fluid volume at baseline (L), and Net Sex, N (%)
Volume is the Net Ins and Outs (L).  Male 4 (40.0)
Predicted values were based upon Stewart’s principles  Female 6 (60.0)
as noted above. We estimated patients’ extracellular fluid Race, N (%)
volume (ECF) at the start of the diuretic infusion (baseline)  White 9 (90.0)
using standard equations (total body weight on hospital  African American 1 (10.0)
admission in kg × 0.6 in men or 0.5 in women) and an esti-  Furosemide received Day 1, mg (mean, SD) 85 (41.2)
mate of additional liters positive in patients with fluid over-  Furosemide received Day 2 86.4 (51)
load. The SID used for IV fluids was the physiologic SID  Furosemide received Day 3 112.5 (104.4)
(not the chemical SID of the fluid in the bag).  SID baseline, mEq/L (mean, SD) 45.2 (3.2)
Data were collected utilizing a case report form (CRF) SID strong ion difference
and stored in a secure database developed by the investi-
gators. Manual data collection from the electronic medical
record was mainly conducted by K. Connor, Surgical ICU
Table 3  Predicted and observed SIDs (n = 10); descriptive and com-
Clinical Pharmacy Specialist. parative statistics
All statistical analyses were completed using SPSS Statistics
(v24, SPSS Inc.). Descriptive statistics were used to summarize Day Mean SID Predicted Correlation
mEq/L coefficient, (rs),
baseline demographics and baseline clinical data. Predicted and (observed) p-value
observed systemic acid-based status were assessed for normal-
ity and compared using Spearman’s correlation coefficient. 1 median (IQR) 47.8 (44.7– 51.0 (43.7– rs = 0.8, p = 0.01
50.6) 55.1)
Serum acid-based laboratory values included pH, P ­ CO2 (par-
2 median (IQR) 49.4 (46.7– 47.3 (43.8– rs = 0.6,
tial pressure of carbon dioxide in mmHg), ­HCO3− (bicarbonate 52.1) 51.5) p = 0.088
concentration in mmol/L), BE (base excess in mmol/L), SID 3 median (IQR) 50.6 (44.9– 45.9 (43.7– rs = 0.476,
(strong ion difference in mEq/L), N ­ a+ (sodium concentration in 53.1) 51.6) p = 0.233
+
mmol/L), ­K (potassium concentration in mmol/L), C ­ l− (chlo-
ride concentration in mmol/L). Data, reported as median values Descriptive data (medians (IQR)) are shown for each time point and
then each of these continuous variables (not normally distributed)
and interquartile ranges, were also compared across measured were compared with a nonparametric test (Spearman’s correlation
time points within the 72-h study period using the Friedman coefficient)
Test. This study was based on a convenience sample of criti- SID strong ion difference
cally ill patients receiving CI loop diuretic. As all analyses were
descriptive and exploratory, no sample size or power analyses
were conducted. Data analyses were mainly conducted by K. received enteral nutrition, with one of these patients hav-
Conn, biostatistician and epidemiologist. ing received both enteral nutrition and TPN.
Table 3 provides the observed and predicted SIDs for
Days 1–3, as well as the results of the correlation analy-
ses between observed and predicted; data were not nor-
Results
mally distributed. On Day 1, the mean SID (observed) was
47.5 (3.5) and the predicted SID value was 49.5 (5.8); the
In our study sample, patients were primarily elderly, white,
observed plasma SID was positively correlated with the
and 60% were women. Patients underwent general, vascu-
predicted SID value (rs = 0.80, p = 0.01). The predicted
lar, or liver transplant surgery. Baseline and demographic
SID was most closely correlated with the observed SID
characteristics of the 10 study patients are presented in
on Day 1, and the correlation was not statistically signifi-
Table 2. Patients received the highest amount of diuretic
cant on Days 2 or 3. One patient received acetazolamide
on Day 3 at a mean daily dose of 112.5 mg and had a mean
500 mg IV × 1 on Days 1 and 2 during the study period.
baseline SID of 45.2  mEq/L. No patients in our study
Table  4 describes acid–base equilibrium at baseline
received continuous IV fluids concomitantly with their
and during furosemide infusion. Based on a review of the
diuresis, other than two patients receiving total parenteral
descriptive laboratory values, the serum chloride measure-
nutrition (TPN) and three other patient receiving low-flow
ments decreased, and the observed SID and plasma pH
rate IV fluids (e.g., ‘keep vein open’ fluids). Six patients
increased steadily over the 72 h infusion, reaching a median

13
 Clinical and Experimental Nephrology

SID of 50.8 on Day 3. Patients became progressively more metabolic alkalosis is also referred to as chloride-depletion
alkalotic on aggressive IV diuresis. When comparing acid- metabolic alkalosis (CDA), commonly caused by the loss
based laboratory values across each of the time points of chloride ions in high volumes of gastric secretions or the
measured, there was a statistically significant difference administration of chloruretic diuretics, such as furosemide.
for vPH (p = 0.032), SID (p = 0.001), ­K+ (p = 0.004) and Other contributing factors include the delayed or incomplete
­Cl− (p = 0.013). recovery from hypercapnic respiratory acidosis, large vol-
ume blood transfusions, alkali administration, and liver fail-
ure [17–19]. IV acetazolamide, a carbonic anhydrase inhibi-
tor, is effective in reducing hyperbicarbonatemia in critically
Discussion ill patients with mild-to-moderate metabolic alkalosis by
decreasing the serum SID [20–24]. At our institution, aceta-
This is the first study evaluating the application of Stewart’s zolamide is frequently used as a first-line treatment for loop
principles of acid–base analysis for critically ill surgical diuretic-induced CDA by administering up to 2 doses of IV
patients undergoing forced IV diuresis over several days. acetazolamide (e.g., 500 mg daily × 2 days) during ongoing
Our study findings are supported by the Stewart model diuresis if metabolic alkalosis occurs. In our study, only one
of acid–base physiology, and may contribute to the better patient received acetazolamide, but this did not appear to
understanding and management of metabolic alkalosis. appreciably affect the serum SID, which remained close to
The appropriate assessment of acid–base disorders is 50. When acetazolamide is ineffective or only transiently
vital to the optimal management of complex critically ill effective, consideration is given to administering a dilute
patients. The Stewart approach of analyzing acid–base dis- HCl infusion to treat persistent moderate to severe metabolic
turbances recognizes that there are only three independent alkalosis [15].
determinants of the H ­ + concentration: P
­ CO2, SID, and A
­ TOT. We believe that Stewart’s physiochemical model provides
+
Thus, neither H ­ nor H ­ CO3−concentrations as dependent a rational explanation of acid–base physiology and mecha-
variables will change unless at least one of the three inde- nistic explanations for common acid–base disorders and
pendent variables change, and while it is accurate to ana- their treatments. This work helps to emphasize the impor-
lyze alterations in acid–base status using observed changes tance of distinguishing between independent and dependent
in ­H+ and H­ CO3− concentrations, this does not explain the variables to better understand the primary derangement and
cause of the changes in acid–base status. [7]. Therefore, the provide appropriate management. Stewart’s theories are sup-
positive change in SID observed following administration ported by our analyses, and we suggested a predictive equa-
of chloride-wasting loop diuretics caused the increase in pH tion to assist with bedside analysis of acid–base disturbances
and ­HCO3−concentration which are dependent variables in [3]. At our institution, Stewart’s principles of acid–base anal-
acid-based physiology, consistent with what was expected ysis are utilized at the bedside and taught in our local medi-
based on the Stewart model.[16]. cal and pharmacy school curriculums, in conjunction with
Metabolic alkalosis is a common finding in critically ill the traditional, bicarbonate-based approach. Furthermore,
patients, and is classically described as being either “chlo- the relevance of Stewart’s principles has been increasingly
ride responsive” or “chloride resistant,”, and hypochloremic, appreciated and recognized. For example, there is a website
hyperbicarbonatemic alkalosis (“chloride responsive”) is the (www.acid-base.org) and a free mobile application to assist
predominant type in the ICU [1, 15]. Chloride-responsive

Table 4  Plasma Acid–Base equilibrium at baseline and during furosemide infusion (n = 10)

Laboratory result Baseline median (IQR) Day 1 median (IQR) Day 2 median (IQR) Day 3 median (IQR) p value

pH 7.41 (7.35–7.43) 7.44 (7.35–7.47) 7.44 (7.4–7.48) – 0.032

PCO2 (mmHg) 50 (41.75–52) 47 (41–53) 45 (42.5–51.3) – 0.670
HCO3-(mmol/L) 28 (26.3–33.3) 29 (25.5–33.8) 30.5 (27–34.8) – 0.180
BE (mmol/L) 1.5 (1–8.3) 4.5 (-1–9.3) 6.5 (2–9) – 0.250
SID (mEq/L) 44.8 (42.3–47.3) 48.9 (45.6–49.8) 49.1 (45.6–52.5) 50.8 (44.5–53.3) 0.001
Na+ (mmol/L) 141 (136–143) 141 (134.5–143.5) 141 (136.5–145) 138 (135.5–144.5) 0.371
K+ (mmol/L) 4.3 (4–4.8) 4.2 (3.7–4.6) 4.1 (3.6–4.4) 3.9 (3.5–4.2) 0.004
Cl− (mmol/L) 105 (101.5–111.5) 101 (97–109) 101 (97.5–111) 99 (96–109.5) 0.013

PCO2 partial pressure of carbon dioxide, H ­ CO3− bicarbonate concentration, BE base excess, SID strong ion difference, N ­ a+ sodium concentra-
tion, ­K+ potassium concentration, C
­ l− chloride concentration. Data are reported as median values and interquartile ranges

13
Clinical and Experimental Nephrology

the clinician with analysis of acid–base disturbances using may have had a small impact on acid–base status in this
Stewart’s theories [9, 10, 25, 26]. population.
Although utilization of the Stewart approach may give
a better understanding of the underlying mechanisms of
acid–base disorders, its clinical application has been criti-
cized for being complex, time-consuming, impractical, and Conclusions
misleading [16, 27–29]. In our study, we did not find the
necessary calculations to be cumbersome, and many of the This study described the impact of CI furosemide on sys-
variables were readily available with routine ICU clinical temic acid–base status, with observed changes consistent
care. with the Stewart physiochemical model. A predictive equa-
Compared to a study conducted by Zazzeron and col- tion based on the Stewart method was proposed, which cor-
leagues, our study was unique in that it prospectively related with observed values early in therapy but was not
evaluated patients receiving CI furosemide over a longer, reliable over the entire course of treatment.
72-h period of time [30]. These investigators evaluated
quasi-continuous changes in urinary electrolytes following Acknowledgements  We gratefully acknowledge Erin Bodekor, NP,
Jacinta Gardner, NP, Matthew Grunert, MD, Kathryn Maloney, NP,
a single low-dose intravenous administration of furosem- Jennifer Mercandetti, NP, Tedd Vineyard, NP, and the bedside nurses at
ide, and reported a detailed analysis of urinary electrolytes, the University of Rochester Medical Center Strong Memorial Hospital
but did not examine any predictive models. Similar to the for their assistance with informed consent and data collection. Pre-
Zazzeron study, our study showed that patients receiving sented as a poster in October 2017 at the American College of Clinical
Pharmacy (ACCP) Annual Meeting in Phoenix, AZ.
furosemide had increased BE, plasma pH, and plasma SID
and decreased plasma Cl− over the 72-h study period. Fur- Funding  This study was supported in part by a St. John Fisher Col-
thermore, our study demonstrated the expected effects of lege Faculty Development Grant. This sponsor had no role in the
furosemide on electrolytes and the underlying mechanism of study design; in the collection, analysis and interpretation of data; in
loop diuretic-induced metabolic alkalosis, which is mainly the writing of the report; and in the decision to submit the article for
publication.
chloride depletion and increased plasma SID.
Limitations of this study are those inherent to it being
Compliance with ethical standards 
small, single-center, and descriptive, which limited our abil-
ity to evaluate the relationship between changes in electro- Conflict of interest  The authors have declared that no conflict of inter-
lytes and acid–base status, to determine the reliability of the est exists.
proposed SID prediction equation, and to evaluate the practi-
cal application of the Stewart model. The ability to predict Ethical approval  All procedures performed in studies involving human
participants were in accordance with the ethical standards of the insti-
changes in SID over the 72 h observation period was likely tutional and/or national research committee at which the studies were
limited by the infrequent point determination of electrolyte conducted (IRB approval number RSRB00032569) and with the 1964
concentrations in plasma and urine that were extrapolated Helsinki Declaration and its later amendments or comparable ethical
over the time between measurements. The further into the standards.
duration of the furosemide infusion, the amount of predic- Informed consent  Informed consent was obtained from all individual
tion dependent on extrapolated values increases, leading to participants included in the study.
an understandable loss of predictability. Testing at a very
frequent interval to improve the predictability of the model
was not considered justifiable in this critically ill sample
of patients. The inability of the equation to predict SID is
References
evidenced by the fact that observed SID increased day to day
as expected, yet the predicted SID decreased. It is possible 1. BL Erstad (2016) Critical care pharmacotherapy. In: Lenexa KS,
that the timing of the urine collections misrepresented the The American College of Clinical Pharmacy
overall average value for the day or that there was an element 2. Kellum JA, Kramer DJ, Pinsky MR. Strong ion gap: a methodol-
ogy for exploring unexplained anions. J Crit Care. 1995;10:51–5.
missing from the equation (e.g., effect of the metabolism of
3. Rastegar A. Clinical utility of stewart’s method in diagnosis and
amino acids in TPN or enteral nutrition). In addition, other management of acid-base disorders. Clin J Am Soc Nephrol.
markers of renal function (e.g., serum creatinine) were not 2009;4:1267–74.
measured, which could confound the evaluation of acid–base 4. Fencl V. Leith DE: Stewart’s quantitative acid-base chem-
istry: applications in biology and medicine. Respir Physio.
status and the efficacy of diuresis. Last, Atot was not able to
1993;91:1–16.
be shown in this study due to missing the data, although 5. Gomez H, Kellum JA. Understanding acid base disorders. Crit
expected post-surgical changes in albumin and phosphorus Care Clin. 2015;31:849–60.

13
 Clinical and Experimental Nephrology

6. Hickish T, Farmery AD. Acid–base physiology: new concepts. 22. Marik PE, Kussman BD, Lipman J, et al. Acetazolamide in the
Anaesth Intensive Care Med. 2018;19:239–44. treatment of metabolic alkalosis in critically ill patients. Heart
7. Kellum JA. Disorders of acid-base balance. Crit Care Med. Lung. 1991;20:455–9.
2007;35:2630–6. 23. Mazur JE, Devlin JW, Peters MJ, et al. Single versus multiple
8. Kimura S, Shabsigh M, Morimatsu H. Traditional approach versus doses of acetazolamide for metabolic alkalosis in critically ill
Stewart approach for acid-base disorders: Inconsistent evidence. medical patients: a randomized, double-blind trial. Crit Care Med.
SAGE Open Med. 2018;6:1–9. 1999;27:1257–61.
9. Magder S, Emami A. Practical approach to physical-chemical 24. Moviat M, Pickkers P, van der Voort PH, et al. Acetazolamide
acid-base management. Stewart at the bedside. Ann Am Thorac mediated decrease in strong ion difference accounts for the cor-
Soc. 2015;12:111–7. rection of metabolic alkalosis in critically ill patients. Crit Care.
10. Seifter JL. Disorders of fluids and electrolytes: integration of acid- 2006;10:R14.
base and electrolyte disorders. N Engl J Med. 2014;371:1821–31. 25. Langer T, Scotti E, Carlesso E, et al. Electrolyte shifts across the
11. Sirker AA, Rhodes A, Grounds RM, et al. Acid-base physiol- artificial lung in patients on extracorporeal membrane oxygena-
ogy: the “traditional” and the “modern” approaches. Anaesthesia. tion: interdependence between partial pressure of carbon dioxide
2002;57:348–56. and strongion difference. J Crit Care. 2015;30:2–6.
12. Stewart PA. Independent and dependent variables of acid-base 26. Elbers PW, Regenmortel NV, Gatz R. Over ten thousand cases
control. Resp Physiol. 1978;33:9–26. and counting: acidbase.org is serving the critical care community.
13. Stewart PA. How to understand acid-base: a quantitative acid-base Anaesthesiol Intensive Ther. 2015;47:441–8.
primer for biology and medicine. New York: Elsevier; 1981. 27. Adrogué HJ, Madias NE. Assessing acid-base status: physi-
14. Todorović J, Nešovic-Ostojić J, Milovanović A, et al. The assess- ologic versus physicochemical approach. Am J Kidney Dis.
ment of acid-base analysis: comparison of the “traditional” and 2016;68:793–802.
the “modern” approaches. Med Glas. 2015;12:7–18. 28. Moviat M, van Haren F, van der Hoeven H. Conventional or physi-
15. Guffey JD, Haas CE, Crowley A, et al. Hydrochloric acid infusion ochemical approach in intensive care unit patients with metabolic
for the treatment of metabolic alkalosis in surgical intensive care acidosis. Crit Care. 2003;7:R41–R4545.
unit patients. Ann Pharmacother. 2018;52:522–6. 29. Siggaard-Andersen O, Fogh-Andersen N. Base excess or
16. Kurtz I, Kraut J, Ornekian V, et al. Acid-base analysis: a critique buffer base (strong ion difference) as measure of a non-respir-
of the Stewart and bicarbonate-centered approaches. Am J Physiol atory acid-base disturbance. Acta Anaesthesiol Scand Suppl.
Renal Physiol. 2008;294:F1009–F10311031. 1995;107:123–8.
17. Adrogue HJ, Madias NE. Management of life-threatening acid 30. Zazzeron L, Ottolina D, Scotti E, et al. Real-time urinary elec-
base disorders: second of two parts. N Engl J Med. 1998;338:107– trolyte monitoring after furosemide administration in surgical
11 (comment). ICU patients with normal renal function. Ann Intensive Care.
18. Khanna A, Kurtzman NA. Metabolic alkalosis. Respir Care. 2016;6:72.
2001;46(354–65):26.
19. Webster NR, Kulkarni V. Metabolic alkalosis in the critically ill. Publisher’s Note Springer Nature remains neutral with regard to
Crit Rev Clin Lab Sci. 1999;36:497–510. jurisdictional claims in published maps and institutional affiliations.
20. Berthelsen P. Cardiovascular performance and oxyhemoglobin
dissociation after acetazolamide in metabolic alkalosis. Intensive
Care Med. 1982;8:269–74.
21. Dickinson GE, Myers ML, Goldbach M, et al. Acetazolamide in
the treatment of ventilatory failure complicating acute metabolic
alkalosis. AnesthAnalg. 1981;60:608–10.

13