8a 28 1!2 1# # !1 ' ! 18# c8 b ! 128 181 818 # 8# 8b 1 a 8 ... · chloremia. (e) lactatemia....

8A 28 1 2 1 11 8 C8 B 12 8

1 8 1 81 8 8 8B 1 A 8( - (0

hbe ph vmpsc R SE yau

w i

(0I r Jl n UE If r T

d) ( t oP E

419M H P R P

n iU P R P %nP R P I J• 3

n li e e• a D ge

5C < 0 ;B : 1 B 1: ; %%. % - 8 + 7 42- %,, + ,

n ; + ax ub )

n ; ( lstvysaT: 1 , (broh• 7=4m a f chob Gl r g h

o st tulnm o eih

< RM 5 CO KPK E >K 9S K GP DK KUK K GHSPOI SU O KJ GIOJK OG SR OR KR O KIGUK RO GJ O OSR/ ORIOJKRIK TUSMRS O GRJ GJ ORO UG OSR SL H KU NKUGT / GTUS TKI O K P OTPKIKR KU J UO GUK ( 0 +/ C()- B ;7/ ( ..+-,.

l stvysT: 1 , ( : A) 1 (( 58 � +

BG A( 2 + : A) - (

l stvysT: 1 , ( BG A( 3 +

: A) 2 BG A( (

l stvys l r dhpl

6.8

6.9

7.0

7.1

7.2

7.3

7.4

7.5

pH

SurvivorsNon survivors

# * *

##

#

!

4

8

12

16

20

24

28

32 SurvivorsNon Survivors

##

* *

#

"

20

30

40

50

60

70

80

PaC

O2

(mm

Hg)


# #

# #

#

90

94

98

102

106

110

114

118

122

Chl

orem

ia (m

mol

/l)


$

0

4

8

12

16

Lact

atem

ia (m

mol

/l)

SurvivorsNon Survivors

* #

#

#

*

%

10

14

18

22

26

30

34

38

Corre

cted

Ani

on G

ap (m

mol

/l)


&

*#

#

#*

*

$AY�� $AY�� $AY�� $AY�� $AY�� $AY��

$AY�� $AY�� $AY�� $AY�� $AY�� $AY��

$AY�� $AY�� $AY�� $AY�� $AY�� $AY��

Bic

arb

on

atem

ia (m

mo

l/l)

.3.3

.3.3

.3 .3

Figure 2 General features of acid-base metabolism evaluated during the first three days. (A) pH. (B) Bicarbonatemia. (C) PaCO2. (D)Chloremia. (E) Lactatemia. (F) Corrected anion gap. *P < 0.05, Tukey’s post hoc analysis between survivors (n = 66) and nonsurvivors (n = 89). #P< 0.05, Tukey’s post hoc analysis vs day 0. Bars represent standard deviations. NS = not significant. PaCO2 = partial pressure of carbon dioxide.

Jung et al. Critical Care 2011, 15:R238http://ccforum.com/content/15/5/R238

Page 6 of 9%% 17 I 45 17 C0 % 3 %% 8

R ,U T % 2U H

365 % / 1.

– 7 I

n e i K M KH KD– 7 /an N D J I

n xuUo 0,6 hl parE U D nK – a

e i r – 7a P t2?7A 1 7; ,;: . 9? ; C 30

:ARHIE 5 ;ETIEPE > 3I APD ER A 6 ECR N BICAPBNMARE RHEPAO NM HELND MALIC AMD RI ENV EMARINM IM OARIEMR IRH ACRIC ACIDN I - A OPN OECRITE CNMRPN ED C IMICA R D 4PIR 4APE :ED,, . ,- =:85- ,

n >4 CPN NTEP DE I Mn mor op 74< /

1 =24 M0• 4PE 1 L D9g• v =A4< / LL7

5<2 NP 5<3k hdle M0n : ;A NP : L9• w L9 EC• eb i f ky fnu s

t i x f kyca

C ,- 3 2

4J =MJG 00 0AM M CD < <M GM SML 1HB AML D GD NS HL GD D DL MEPDN HB PGMBI, PDBMLC JMMI 7L D L 4 D :DC - , <:73,

n y i y ha

n mu nW

n so =DNPHP Wb k• ; l L/ W• y W 0<0264 77a , tsr p L/ W

n ; M9 I 71 G• N6 . a dc ha

n eg f ==25 d _ l c

the control group (p = 0.45). Although there was no sig-

nificant difference in the pH values between the two groupsat the onset of sepsis, there was a more rapid normalization

of pH (pH [ 7.35) on day 3 in those who received bicar-

bonate infusion (p \ 0.001). There were no adverse eventsreported from bicarbonate infusion. Five patients in the

bicarbonate group dropped their ionized calcium levels

below 0.9 mmol/L and required calcium supplementcompared to two patients in the control group (p = 0.43).

However, hypokalemia was more frequently observed inthe bicarbonate arm (16 cases vs. 6 cases in the control

arm; p = 0.02).

Reversal of shock

The proportion of patients who underwent reversal ofshock were similar among patients who received bicar-

bonate therapy (78%; 95% CI 61–90%) and those who did

not (72%; 95% CI 55–86%; p = 0.78) (Fig. 1). The med-ian time until reversal of shock was shorter in the bicar-

bonate group (44.5 h [95% CI 34–54]) than in the control

group (55.0 h [95% CI 39–60]) but the difference did notattain statistical significance (p = 0.09). Similarly, the time

to reversal of shock in those who were on bicarbonate

infusion was comparable between those with positive andthose with negative blood cultures (p = 0.72).

Secondary outcomes

The secondary endpoint analyses showed that the time to

weaning off mechanical ventilation was significantlyshorter in the bicarbonate group versus the control group

(p = 0.02) (Fig. 2). The median time to liberation of

mechanical ventilation was 10 days (95% CI 5.0–13.0) inthe bicarbonate group and 14 days (95% CI 9.0–19.0) in

the control group (p = 0.02). Three patients required res-

cue bi-level positive airway pressure post extubation, twowere in the bicarbonate group and the other was a control

subject. None of these patients had to be reintubated for the

period ending at 28 days post sepsis.The length of ICU stay was significantly shorter in the

surviving patients who received bicarbonate compared to

controls [median 11.5 days (95% CI 6.0–16.0) vs.16.0 days (95% CI 13.5–19.0), respectively; p = 0.01].

However, there was no difference in 28-day mortalitybetween the two study groups. Overall, there were 10 death

in the bicarbonate group (28%; 95% CI 14–45%) and 12 in

the control group (33%; 95% CI 19–51%, p = 0.79).

Discussion

The result of this study show that bicarbonate infusion in

patients with septic shock and increased blood lactate doesnot reduce the time to reversal of shock. However, bicar-

bonate replacement is associated with a shorter duration of

mechanical ventilation and reduced length of ICU stay.Over the last two decades, expert opinion on the treat-

ment of acidosis has swung from supporting the infusion of

bicarbonate in septic patients with hyperlactatemia to dis-couraging its use. Acidosis has been implicated in a

decreased cardiac contractility, increase in the time to peak

left ventricular pressure, and retarding ventricular relaxa-tion [22] while alkalinization was thought to improve

cardiovascular performance [23, 24]. In 1990s, this concept

was challenged after two clinical trials had suggested thatthe infusion of bicarbonate in critically ill patients with

metabolic acidosis and increased arterial plasma lactate

levels failed to improve hemodynamic variables or ame-liorate tissue oxygenation [8, 9]. The lack of potential

benefit was attributed to the observed decrease in plasmaFig. 1 Kaplan–Meier curves for time to reversal of shock (log-ranktest: v2 = 2.71; p = 0.09)

Fig. 2 Kaplan–Meier curves for time to weaning from mechanicalventilation (log-rank test: v2 = 5.11; p = 0.02)

Intern Emerg Med (2010) 5:341–347 345

123

the control group (p = 0.45). Although there was no sig-

nificant difference in the pH values between the two groupsat the onset of sepsis, there was a more rapid normalization

of pH (pH [ 7.35) on day 3 in those who received bicar-

bonate infusion (p \ 0.001). There were no adverse eventsreported from bicarbonate infusion. Five patients in the

bicarbonate group dropped their ionized calcium levels

below 0.9 mmol/L and required calcium supplementcompared to two patients in the control group (p = 0.43).

However, hypokalemia was more frequently observed inthe bicarbonate arm (16 cases vs. 6 cases in the control

arm; p = 0.02).

Reversal of shock

The proportion of patients who underwent reversal ofshock were similar among patients who received bicar-

bonate therapy (78%; 95% CI 61–90%) and those who did

not (72%; 95% CI 55–86%; p = 0.78) (Fig. 1). The med-ian time until reversal of shock was shorter in the bicar-

bonate group (44.5 h [95% CI 34–54]) than in the control

group (55.0 h [95% CI 39–60]) but the difference did notattain statistical significance (p = 0.09). Similarly, the time

to reversal of shock in those who were on bicarbonate

infusion was comparable between those with positive andthose with negative blood cultures (p = 0.72).

Secondary outcomes

The secondary endpoint analyses showed that the time to

weaning off mechanical ventilation was significantlyshorter in the bicarbonate group versus the control group

(p = 0.02) (Fig. 2). The median time to liberation of

mechanical ventilation was 10 days (95% CI 5.0–13.0) inthe bicarbonate group and 14 days (95% CI 9.0–19.0) in

the control group (p = 0.02). Three patients required res-

cue bi-level positive airway pressure post extubation, twowere in the bicarbonate group and the other was a control

subject. None of these patients had to be reintubated for the

period ending at 28 days post sepsis.The length of ICU stay was significantly shorter in the

surviving patients who received bicarbonate compared to

controls [median 11.5 days (95% CI 6.0–16.0) vs.16.0 days (95% CI 13.5–19.0), respectively; p = 0.01].

However, there was no difference in 28-day mortalitybetween the two study groups. Overall, there were 10 death

in the bicarbonate group (28%; 95% CI 14–45%) and 12 in

the control group (33%; 95% CI 19–51%, p = 0.79).

Discussion

The result of this study show that bicarbonate infusion in

patients with septic shock and increased blood lactate doesnot reduce the time to reversal of shock. However, bicar-

bonate replacement is associated with a shorter duration of

mechanical ventilation and reduced length of ICU stay.Over the last two decades, expert opinion on the treat-

ment of acidosis has swung from supporting the infusion of

bicarbonate in septic patients with hyperlactatemia to dis-couraging its use. Acidosis has been implicated in a

decreased cardiac contractility, increase in the time to peak

left ventricular pressure, and retarding ventricular relaxa-tion [22] while alkalinization was thought to improve

cardiovascular performance [23, 24]. In 1990s, this concept

was challenged after two clinical trials had suggested thatthe infusion of bicarbonate in critically ill patients with

metabolic acidosis and increased arterial plasma lactate

levels failed to improve hemodynamic variables or ame-liorate tissue oxygenation [8, 9]. The lack of potential

benefit was attributed to the observed decrease in plasmaFig. 1 Kaplan–Meier curves for time to reversal of shock (log-ranktest: v2 = 2.71; p = 0.09)

Fig. 2 Kaplan–Meier curves for time to weaning from mechanicalventilation (log-rank test: v2 = 5.11; p = 0.02)

Intern Emerg Med (2010) 5:341–347 345

123

[ nv t pr76 4, 01 B 576 4, 01 , B , 5

. %

t IN % ] Csh rl 3 , a

t ] P26 8 4, 01 B 5

0 : = 9 8 4, 01 ,B 5. ,

% oP d26 % 0 : = 9 . *,

6FH 35 JI >6 /I = 1DD AN JD MJ FOH FA L JI N HFIFMNL NFJI JI HJLN FN FI K NF INMFNE ANFA AF JMFM L NLJMK ANFP I MFM ;7J :I , ;840PaCO2. Infused sodium bicarbonate combines with hydrogen ions

to form H2CO3. H2CO3 dehydrates to H2O and CO2. Therefore,sodium bicarbonate administration physiologically increases CO2.If adequate ventilation is prohibited, sodium bicarbonate admin-istration worsens the acidemia caused by the combination ofrespiratory acidosis and lactic acidosis. Furthermore, even withadequate ventilation, increased CO2 due to sodium bicarbonateworsens intracellular acidosis in severely ill patients with circula-tory failure, although sodium bicarbonate increases the arterial pHin the short term [7,17]. In this study, sodium bicarbonate was notadministered for two patients with PaCO2.50 mmHg, althoughthey had an initial pH ,7.15. However, one of two patients withsepsis maintained pH .7.15 just with mechanical ventilation bymaintaining PaCO2,20 mmHg and finally recovered from lacticacidosis without sodium bicarbonate administration. One patientrecovered from heart failure and lactic acidosis, but developedalkalemia in the absence of sodium bicarbonate administration.Alkalemia was induced by the respiratory alkalosis to mechanicalventilation with a high minute volume and increased bicarbonateconversion from lactic acid. Further investigations are necessary toevaluate whether lower PaCO2 using mechanical ventilation canovercome acidemia in patients with lactic acidosis.

The causes of lactic acidosis can be classified as type A, which isassociated with marked tissue hypoperfusion, such as shock, ortype B, which is not associated with tissue hypoxia. The mostcommon cause of severe lactic acidosis is sepsis [2]. Sepsis was themost common cause of lactic acidosis, and the most commoncause of mortality in this study. We only had two days to improvelactic acidosis because the median survival time was just two daysin this and other studies [2]. Therefore, intensive management isnecessary to increase the survival rate in patients with sepsis andlactic acidosis.Jansen and Bakker et al. showed that the SOFA score is closely

associated with the lactic acid level in patients in the ICU [18].The SOFA score is composed of six components. Two compo-nents are related to tissue hypoperfusion, and another two areassociated with kidney and liver function. Therefore, the SOFAscore accurately reflects lactate production due to hypoxia andlactate clearance. The SOFA score, which can be easily calculated,was independent factor for mortality, and the APACHE II score,which has fourteen components, was correlated with mortality inthis study. Further studies are necessary to confirm the role of theSOFA score in predicting mortality in patients with lactic acidosis.This study has some limitations because the sample size is small,

and the study design is retrospective. In addition, non-calibrationand potential inaccuracy of APACHE II score to predict deathand a false severity correction make a limited interpretation.Despite these limitations, we found that sodium bicarbonateadministration may be harmful in patients with lactic acidosis.Further prospective studies are necessary to confirm the effect ofsodium bicarbonate on mortality in patients with lactic acidosis.In summary, we have found that lactic acidosis is a common

cause of high anion gap metabolic acidosis and has a very highmortality rate especially in patients with sepsis. Therefore, thelactic acid level should be checked to evaluate the cause ofmetabolic acidosis when patients present with high anion gapmetabolic acidosis especially those with sepsis. The SOFA score isan independent factor for predicting mortality and the initial andfollow-up lactic acid levels are associated with mortality in patientswith lactic acidosis. In addition, sodium bicarbonate should beprescribed with caution in cases of lactic acidosis because sodiumbicarbonate administration may affect mortality.

Author Contributions

Conceived and designed the experiments: WSA. Analyzed the data: HJKYKS WSA. Wrote the paper: HJK WSA.

References

1. Jordan JC, John TH, Jerome PK, Nicolas EM (1986) Lactic acidosis. KidneyInternational 29: 752–774.

2. Stacpoole PW, Wright EC, Baumgartner TG, Bersin RM, Buchalter S, et al.(1994) Natural history and course of acquired lactic acidosis in adults. DCA-Lactic Acidosis Study Group. Am J Med 97: 47–54.

3. Trzeciak S, Dellinger RP, Chansky ME, Arnold RC, Schorr C, et al. (2007)Serum lactate as a predictor of mortality in patients with infection. IntensiveCare Med 33: 970–977.

4. Luft FC (2001) Lactic acidosis update for critical care clinicians. J Am SocNephrol 12: S15–19.

5. Mitchell JH, Wildenthal K, Johnson RL Jr (1972) The effects of acid-basedisturbances on cardiovascular and pulmonary function. Kidney Int 1: 375–389.

6. Wildenthal K, Mierzwiak DS, Myers RW, Mitchell JH (1968) Effects of acutelactic acidosis on left ventricular performance. Am J Physiol 214: 1352–1359.

7. Adrogue HJ, Rashad MN, Gorin AB, Yacoub J, Madias NE (1989) Assessingacid-base status in circulatory failure: Differences between arterial and centralvenous blood. N Engl J Med 320: 1312–1316.

8. Mathieu D, Neviere R, Billard V, Fleyfel M, Wattel F (1991) Effects ofbicarbonate therapy on hemodynamics and tissue oxygenation in patients withlactic-acidosis: a prospective, controlled clinical study. Crit Care Med 19: 1352–1356.

9. Cooper DJ, Walley KR, Wiggs BR, Russell JA (1990) Bicarbonate does notimprove hemodynamics in critically ill patients who have lactic acidosis. Aprospective, controlled clinical study. Ann Intern Med 112: 492–498.

10. Forsythe SM, Schmidt GA (2000) Sodium bicarbonate for the treatment of lacticacidosis. Chest 117: 260–267.

11. Rachoin JS, Weisberg LS, McFadden CB (2010) Treatment of lactic acidosis:appropriate confusion. J Hosp Med 5: E1–E7.

12. Knaus WA, Draper EA, Wagner DP, Zimmerman JE (1985) APACHE II: aseverity of disease classification system. Crit Care Med 13: 818–829.

13. Ferreira FL, Bota DP, Bross A, Melot C, Vincent JL (2001) Serial evaluation ofthe SOFA score to predict outcome in critically ill patients, JAMA 286: 1754–1758.

14. Valenza F, Pizzocri M, Salice V, Chevallard G, Fossali T, et al. (2012) Sodiumbicarbonate treatment during transient or sustained lactic acidemia in normoxicand normotensive rats, PLoS One 7: e46035.

15. Susantitaphong P, Sewaralthahab K, Balk EM, Eiam-ong S, Madias NE, et al.(2012) Short-term and long-term effects of alkali therapy in chronic kidneydisease: a systemic review. Am J Nephrol. 35(6): 540–7.

16. Kraut JA, Kurtz I (2006) Use of base in the treatment of acute severe organicacidosis by nephrologists and critical care physicians: results of an online survey.Clin Exp Nephrol 10: 111–117.

Table 4. Logistic regression analysis for factors that affectmortality.

Univariate Multivariate

Variable Exp(B) (95% CI)p value Exp(B) (95% CI) p value

Sex, male 2.95 (1.08–8.08) 0.035 1.55 (1.00–24.68) 0.756

Age 0.95 (0.90–1.00) 0.040 0.71 (0.47–1.08) 0.110

Albumin 0.35 (0.13–0.92) 0.032 112.6 (0.1–109347.3) 0.178

SOFA 1.50 (1.21–1.86) 0.000 3.02 (1.23–7.42) 0.016

APACHE II 1.10 (1.01–1.21) 0.035 0.69 (0.46–1.04) 0.078

Initial lactate 1.03 (1.00–1.05) 0.016 1.01 (0.97–1.05) 0.705

Follow-up lactate 1.03 (1.00–1.05) 0.020 1.02 (1.00–1.06) 0.222

Bicarbonate use 4.16 (1.50–11.52) 0.006 15.83 (1.00–251.47) 0.050

Ventilator use 3.55 (1.18–10.68) 0.024 0.91 (0.10–8.56) 0.934

SOFA, sepsis related organ failure assessment; APACHE II, acute physiologic andchronic health evaluation; Exp(B), exponentiation of the B coefficient, which isan odds ratio; CI, confidence interval.doi:10.1371/journal.pone.0065283.t004

Sodium Bicarbonate on Mortality in Lactic Acidosis

PLOS ONE | www.plosone.org 6 June 2013 | Volume 8 | Issue 6 | e65283

sn h fm

t vo dSbw

cS a

7 A . HHJ 7

I iWu

:2/a yYp rlWu ea

4H FDLC 27 3HJGD 84 5D LHG 7 0D J HG LB LCBJ IO GDGLJ B M J D H D 1 DG D 8HG -- /-(. -( , 9 2. -)- -,

594 D. J. A. Goldsmith et al.

Ethics committee approval was obtained as appropriate.

Measurement of pHi and buffering capacity

Leucocytes were loaded for 30min at 37°C with 5 mmol/l BCECF-AM [2’,7‘-bis-(2-~arboxyethyl)- 5(6)-carboltyfluorescein acetylmethoxy ester] (Mole- cular Probes, Eugene, OR, U.S.A.) as described pre- viously [14]. pHi was monitored with a Perkin-Elmer luminescence spectrometer (Model LS50). Calibra- tion was performed using established methods [15]. Least-squares linear regression was employed to lin- earize the fluorescence/pH relationship (r always > 0.995). After equilibration with BCECF-AM the cells were washed and resuspended in buffer. The effect on pHi of several buffering systems was studied (described below). pHi measurements were performed after 20 min of equilibration and intracellular buffering capacity was calculated from the change in pHi after the addition of 3 mmol/l ammonium chloride to an aliquot of the cell suspension [16]. Extracellular pH was measured with a Whatman pH meter with micro-probe attachment. Samples of buffer were analysed for pH, Pcoz and standard bicarbonate using a Corning 168 pH and blood gas analyser. The effect of addition of sodium bicarbonate sufficient to increase the extracellular bicarbonate concentration by 12 mmol/l was then studied (subsequently termed a large bolus).

Buffering solutions

Leucocytes were exposed to four buffering systems: normal bicarbonate buffer (24 mmol/l), low bicarbonate buffers (6 mmol/l and 12 mmol/l) and Hepes buffer (10 mmol/l). In each case, the effect of the addition of sodium bicarbonate was studied. Earle’s medium was used as the standard buffer, containing physiological concentrations of sodium (140 mmol/l), potassium (6 mmol/l), calcium (1.8 mmol/l), magnesium (0.8 mmol/l), phosphate (0.8 mmol/l), glucose (5 mmol/l) and sodium bicarbonate (24mmol/l). After equilibration with 5% C02 the pH was 7.40. The low bicarbonate buffers were pre- pared in an identical manner except that the bicarbonate concentration was varied to produce final concentrations of 6 mmol/l and 12 mmol/l, the concentration of chloride being adjusted in a recip- rocal fashion. Equilibration of these buffers with 5-10% COz resulted in an extracellular p H ranging from 6.90 to 7.40. Hepes buffer was also used where the sodium bicarbonate was replaced by 10 mmol/l Hepes. The pH of this buffer was titrated to pH 7.40 with sodium hydroxide. Choline bicarbonate (6-24 mmol/l, equilibrated with either 5 or 10% COZ) and choline-Hepes buffers were used in order to pro- vide sodium-free conditions where minimizing the effect of Na+/H+ exchange between the leucocytes and the extracellular buffer was required. Acidifica-

tion of the leucocytes was achieved by addition of lactic acid (10 mmol/l) and propionic acid (5 mmol/l) to the extracellular fluid. All chemicals were supplied by Sigma Chemical Company (Poole, Dor- set, U.K.) and used without further purification.

RESULTS

Hepes buffer

After equilibration of leucocytes in Hepes buffer the pHi was 7.32k0.04 and buffering power was calculated as 16.7k2.3 mmol l-’ pH unit-’ (n = 16, where n = number of experimental data points, each being performed in duplicate). On addition of a large bolus of sodium bicarbonate a fall in pHi of 0.22k0.15 was observed over 20 s (Fig. 1). This was associated with an increase in buffer Pcoz from ~ 0 . 1 kPa to 3.5k0.3 kPa. Over the next 120 s an increase in pHi of 0.05 k 0.02 was observed. When choline bicarbonate was substituted for sodium bicarbonate this later rise in pHi did not occur, implying that it was due to Na+/H+ exchange.

Normal bicarbonate buffer (24 mmol/l, Pcol 5.1-5.3 kPa, pH 7.4)

pHi averaged 7.24 _+ 0.02 and buffering power averaged 33.6k5.2 mmol 1-’ pH unit-’ (n = 12). Both were significantly different from the values obtained with the Hepes buffer (PxO.05). Addition

7.417 7.35

1 ’ 7.15 I 1 I I I I I I

0 10 20 30 40 50 60 70 Time (seconds)

Fig. I . Effect of the addition of 12 mmol/l (‘large bolus’) sodium bicarbonate to a suspension of human leucocytes in Hepes buffer s I5 ui P yl

o 1; m

n 5B LCO NH MGLBBJSsc n

hb p tva

n r : ce d• 51;( c FFH 8

a

5 N E

n 5 gnO rKM J Ni M

n E dP ePO lsgnKL O Ia

• o PL - Put ocKM

2A DC 1, 4 : 5 3 9C 9 9 : 5 1 4;: 740.

a -

n

H 3n 3O

n N-• l C•

() C G

5 < <1 <

H 7

0 220 0 0 1 - ., - , 00 , 0 0 , . 0 .0

, 0 0 0 0 , 00 1 .0 6,. . ,. 0 , .

0, 0. 0 , 0 , 0 ,6 1 0 0 .0

. 1 .n N 7 Hn r p dt G a I 5H I 5

n o H I 57C cS

i IHn nu 5

��

Articles

www.thelancet.com Published online June 14, 2018 http://dx.doi.org/10.1016/S0140-6736(18)31080-8 1

Sodium bicarbonate therapy for patients with severe metabolic acidaemia in the intensive care unit (BICAR-ICU): a multicentre, open-label, randomised controlled, phase 3 trialSamir Jaber, Catherine Paugam, Emmanuel Futier, Jean-Yves Lefrant, Sigismond Lasocki, Thomas Lescot, Julien Pottecher, Alexandre Demoule, Martine Ferrandière, Karim Asehnoune, Jean Dellamonica, Lionel Velly, Paër-Sélim Abback, Audrey de Jong, Vincent Brunot, Fouad Belafia, Antoine Roquilly, Gérald Chanques, Laurent Muller, Jean-Michel Constantin, Helena Bertet, Kada Klouche, Nicolas Molinari, Boris Jung, for the BICAR-ICU Study Group*

SummaryBackground Acute acidaemia is frequently observed during critical illness. Sodium bicarbonate infusion for the treatment of severe metabolic acidaemia is a possible treatment option but remains controversial, as no studies to date have examined its effect on clinical outcomes. Therefore, we aimed to evaluate whether sodium bicarbonate infusion would improve these outcomes in critically ill patients.

Methods We did a multicentre, open-label, randomised controlled, phase 3 trial. Local investigators screened eligible patients from 26 intensive care units (ICUs) in France. We included adult patients (aged ≥18 years) who were admitted within 48 h to the ICU with severe acidaemia (pH ≤7⋅20, PaCO2 ≤45 mm Hg, and sodium bicarbonate concentration ≤20 mmol/L) and with a total Sequential Organ Failure Assessment score of 4 or more or an arterial lactate concentration of 2 mmol/L or more. We randomly assigned patients (1:1), by stratified randomisation with minimisation via a restricted web platform, to receive either no sodium bicarbonate (control group) or 4⋅2% of intravenous sodium bicarbonate infusion (bicarbonate group) to maintain the arterial pH above 7⋅30. Our protocol recommended that the volume of each infusion should be within the range of 125–250 mL in 30 min, with a maximum of 1000 mL within 24 h after inclusion. Randomisation criteria were stratified among three prespecified strata: age, sepsis status, and the Acute Kidney Injury Network (AKIN) score. The primary outcome was a composite of death from any cause by day 28 and the presence of at least one organ failure at day 7. All analyses were done on data from the intention-to-treat population, which included all patients who underwent randomisation. This study is registered with ClinicalTrials.gov, number NCT02476253.

Findings Between May 5, 2015, and May 7, 2017, we enrolled 389 patients into the intention-to-treat analysis in the overall population (194 in the control group and 195 in the bicarbonate group). The primary outcome occurred in 138 (71%) of 194 patients in the control group and 128 (66%) of 195 in the bicarbonate group (absolute difference estimate –5⋅5%, 95% CI –15⋅2 to 4⋅2; p=0⋅24). The Kaplan-Meier method estimate of the probability of survival at day 28 between the control group and bicarbonate group was not significant (46% [95% CI 40–54] vs 55% [49–63]; p=0⋅09. In the prespecified AKIN stratum of patients with a score of 2 or 3, the Kaplan-Meier method estimate of survival by day 28 between the control group and bicarbonate group was significant (63% [95% CI 52–72] vs 46% [35–55]; p=0⋅0283). Metabolic alkalosis, hypernatraemia, and hypocalcaemia were observed more frequently in the bicarbonate group than in the control group, with no life-threatening complications reported.

Interpretation In patients with severe metabolic acidaemia, sodium bicarbonate had no effect on the primary composite outcome. However, sodium bicarbonate decreased the primary composite outcome and day 28 mortality in the a-priori defined stratum of patients with acute kidney injury.

Funding French Ministry of Health and the Société Française d’Anesthésie Réanimation.

Copyright © 2018 Elsevier Ltd. All rights reserved.

Published Online June 14, 2018 http://dx.doi.org/10.1016/ S0140-6736(18)31080-8

See Online/Comment http://dx.doi.org/10.1016/S0140-6736(18)31305-9

*The BICAR-ICU study investigators are listed in the appendix

Saint Eloi ICU (Prof S Jaber MD, A de Jong MD, F Belafia MD, Prof G Chanques MD) and Département de Médecine Intensive et Réanimation (V Brunot MD, Prof K Klouche MD, Prof B Jung MD), Montpellier University Hospital, PhyMedExp, INSERM, CNRS, Montpellier, France; AP-HP, Département Anesthésie et Réanimation, Hôpital Beaujon, Hôpitaux Universitaires Paris Nord Val de Seine, Paris, France (Prof C Paugam MD, P-S Abback MD); CHU de Clermont-Ferrand, Department of Perioperative Medicine, GReD, UMR/CNRS6293, University Clermont Auvergne, INSERM U1103, Clermont-Ferrand, France (Prof E Futier MD, Prof J-M Constantin MD); CHU de Nîmes, Département Anesthésie et Réanimation, University of Montpellier-Nîmes, Nîmes, France (Prof J-Y Lefrant MD, L Muller MD); CHU d’Angers, Réanimation Chirurgicale, Angers, France (Prof S Lasocki MD); AP-HP, Département Anesthésie et Réanimation, Hôpital Saint Antoine, Paris, France (Prof T Lescot MD); Hôpitaux Universitaires de Strasbourg, Service d’Anesthésie-Réanimation Chirurgicale–Université de Strasbourg, Fédération de Médecine Translationnelle de Strasbourg, Strasbourg, France

IntroductionAcute acidaemia is frequently observed during crit-ical illness, with a reported incidence varying from 14% to 42%.1–5 Persistent acidaemia has been associated with poor prognosis,1–3,6 with a mortality rate as high as 57% when the pH stays below 7⋅20.5 Along with case-specific treatment, improvement of tissue perfusion and supportive measures such as mechanical ventilation and renal-replacement therapy are the cornerstones of

severe metabolic acidaemia management in critically ill patients.2,3,7 Because an acidotic cellular environment can cause cellular dysfunction, intravenous sodium bicarbonate administration to increase the pH might also be beneficial. In a survey done in North America, more than two-thirds of the programme directors in nephrology or intensive care units (ICUs) declared that they used sodium bicarbonate for the treatment of acidaemia with hyperlactataemia.8

4 8 2 4:847 9 0.

odNf C U E N 1 Mb IuiNo 4ntU E ntL r ID h M P Nc N L]N eN aUp

[ N

n , .• h s = - 2• e e = 2

n , .• h I V s = a2• n V

n , .• h I V s P2

= 2• 0 e=

n , .• g• s e8

g =

0 27i c dob

4363.,4 3,4 7 6 5 0

g eh N M a N M m N M n lT C

�l 0- i lL � % Cl mLp A / � 47-3 � /8/-3 � 1C

l 53. � � 1

�l / H 27l 0- NIl m % 1

O U i lI a% 1 Cl % i

� l 27

� nl S F lL Pl S F nL% L roL g

2 85n

• > > > rR I> 1 PRo P C> g c

S > s S S D >2 5 > g c>af g c>

R g c> n P 2 >rR 2 I OC

n Uo 45 5 54 5 S T RoDR >o N Te d

t K>

S iS T

�l 0- i lL � % Cl mLp A / � 47-3 � /8/-3 � 1C

l 53. � � 1

�l / H 27l 0- NIl m % 1


� l 27


17

Figures

Figure S1: Experimental Design

6 3l 2l 5l

6

A <

215

0 3

2 3

0 3

<

215

0 3

2 3

0 3*-

To

n K• p ../ <h

r 46

• R k> ../ hg 3 20 7 5 7

n L 7 L m H >

n FG 05 9 > i• C . rj G I M o R

nI

n F R a 0689 >• S < 51 9 251

1 > 33 % oR P O

>

�l 0- i lL � % Cl mLp A / � 47-3 � /8/-3 � 1C

l 53. � � 1

�l / H 27l 0- NIl m % 1


� l 27


n M NFM• 557 29 o y K

n 4.,UTS Cn r aO RA Cn 0 VM d p• + 1 + <> 3 +

n 0 I Hn 0

s

n NK I in i akm

n po c k Hg Al

n e n

n 7 6n % 7 6

= =n 2 .-n , 0 ,

n 2 1 58 1 34

��

Articles


We excluded patients who met the following main exclusion criteria: respiratory acidosis, proven digestive or urinary tract loss of sodium bicar bonate (volume loss ≥1500 mL per day), stage IV chronic kidney disease,16 ketoacidosis, and sodium bicarbonate infusion (including renal-replacement therapy) within 24 h before screening. The appendix provides the full exclusion criteria.

We obtained written informed consent from the patient or a relative upon study inclusion. However, considering the severity of the illness, the fact that most of these patients would be unable to consent (sedation or potential delirium)17–19 and that their proxies might not be contactable at the time of inclusion, a deferred consent process for emergency situations was enabled. When deferred consent was used, written permission to pursue the research was obtained from the patient or proxy as soon as possible. If this consent was not obtained, the patient’s data were not used and they were withdrawn from the trial.

Randomisation and maskingWe randomly assigned eligible patients within 48 h of ICU admission in a one-to-one ratio to receive either no sodium bicarbonate infusion (control group) or sodium bi-carbonate infusion (bicarbonate group). We randomly assigned patients by stratified randomisation with mini-misation using a computer-generated allocation sequence accessible from each centre through a secured-dedicated website with stratifi cation according to study site and three prespecified factors: age with a cutoff of 65 years, presence or absence of suspected sepsis,20,21 and presence or absence of Acute Kidney Injury Network (AKIN) score of 2 or 3. Because sodium bicarbonate infusion influences arterial pH levels and because routine arterial blood gases must be done in critically ill patients, masking of the physicians and nurses was not feasible.

ProceduresIn the intervention group, 4⋅2% sodium bicarbonate was intravenously infused with the aim of achieving an arterial pH of 7⋅30 or more during the 28-day ICU admission or ICU discharge because preliminary results suggested that arterial pH in survivors approximated 7⋅30 by day 1 although persistent severe acidaemia was observed in non-survivors.5 Our protocol recommended that the volume of each sodium bicarbonate infusion should be within the range of 125–250 mL in 30 min, with a maximum of 1000 mL within 24 h after inclusion, and that measurement of arterial blood gas should be done 1–4 h after the end of each infusion (appendix p 17). Hypertonic 4⋅2% sodium bicarbonate was chosen according to the scarce literature available,10,11,22 the current practice when sodium bicarbonate is used as reported by Jung and colleagues,5 and the objective of titrating a pH target of 7⋅30 or more.

In both groups, indications for renal-replacement therapy were standardised. Upon admission, urgent

renal-replacement therapy was strongly recommended in the event of kalaemia that was more than 6⋅5 mmol/L with electrocardiogram signs or cardiogenic pulmonary oedema with no urine output, or both. At 24 h after inclusion, renal-replacement therapy was recommended when two of three criteria were present: urine output less than 0⋅3 mL/kg per h for at least 24 h, arterial pH less than 7⋅20 despite resuscitation, and kalaemia more than 6⋅5 mmol/L. Each study site chose the method of renal-replacement therapy according to the local guidelines, which provided additional base because the dialysis included sodium bicarbonate as a buffer.23 In both groups, initiation of invasive mechanical ventilation was indicated if patients had one major or two minor predefined clinical events (appendix).24

942 patients were assessed for eligibility

400 were randomly assigned

542 excluded109 already received sodium bicarbonate

87 were in terminal decline76 had treatment limitation69 had chronic renal failure47 had immediate RRT indication 41 had ketoacidosis37 had digestive loss of sodium bicarbonate21 were eligible but not enrolled18 were included in another clinical study13 had hyperkalaemia with heart signs13 declined to participate 11 were under guardianship protection

201 assigned to control group

7 withdrew consent

22* violated inclusion or non-inclusion criteria

47* received sodium bicarbonate41 salvage therapies

6 misinterpretations of the protocol

194 included in 28-day follow-up and the intention-to-treat analysis

132 included in the per-protocol analysis

199 assigned to bicarbonate group

4 withdrew consent

15 violated inclusion or non-inclusion criteria

1 did not receive sodium bicarbonate (misinterpretations of the protocol)



Figure 1: Trial profileRRT=renal-replacement therapy. *Seven patients both violated inclusion or non-inclusion criteria and received sodium bicarbonate.

��

Articles


We excluded patients who met the following main exclusion criteria: respiratory acidosis, proven digestive or urinary tract loss of sodium bicar bonate (volume loss ≥1500 mL per day), stage IV chronic kidney disease,16 ketoacidosis, and sodium bicarbonate infusion (including renal-replacement therapy) within 24 h before screening. The appendix provides the full exclusion criteria.

We obtained written informed consent from the patient or a relative upon study inclusion. However, considering the severity of the illness, the fact that most of these patients would be unable to consent (sedation or potential delirium)17–19 and that their proxies might not be contactable at the time of inclusion, a deferred consent process for emergency situations was enabled. When deferred consent was used, written permission to pursue the research was obtained from the patient or proxy as soon as possible. If this consent was not obtained, the patient’s data were not used and they were withdrawn from the trial.

Randomisation and maskingWe randomly assigned eligible patients within 48 h of ICU admission in a one-to-one ratio to receive either no sodium bicarbonate infusion (control group) or sodium bi-carbonate infusion (bicarbonate group). We randomly assigned patients by stratified randomisation with mini-misation using a computer-generated allocation sequence accessible from each centre through a secured-dedicated website with stratifi cation according to study site and three prespecified factors: age with a cutoff of 65 years, presence or absence of suspected sepsis,20,21 and presence or absence of Acute Kidney Injury Network (AKIN) score of 2 or 3. Because sodium bicarbonate infusion influences arterial pH levels and because routine arterial blood gases must be done in critically ill patients, masking of the physicians and nurses was not feasible.

ProceduresIn the intervention group, 4⋅2% sodium bicarbonate was intravenously infused with the aim of achieving an arterial pH of 7⋅30 or more during the 28-day ICU admission or ICU discharge because preliminary results suggested that arterial pH in survivors approximated 7⋅30 by day 1 although persistent severe acidaemia was observed in non-survivors.5 Our protocol recommended that the volume of each sodium bicarbonate infusion should be within the range of 125–250 mL in 30 min, with a maximum of 1000 mL within 24 h after inclusion, and that measurement of arterial blood gas should be done 1–4 h after the end of each infusion (appendix p 17). Hypertonic 4⋅2% sodium bicarbonate was chosen according to the scarce literature available,10,11,22 the current practice when sodium bicarbonate is used as reported by Jung and colleagues,5 and the objective of titrating a pH target of 7⋅30 or more.

In both groups, indications for renal-replacement therapy were standardised. Upon admission, urgent

renal-replacement therapy was strongly recommended in the event of kalaemia that was more than 6⋅5 mmol/L with electrocardiogram signs or cardiogenic pulmonary oedema with no urine output, or both. At 24 h after inclusion, renal-replacement therapy was recommended when two of three criteria were present: urine output less than 0⋅3 mL/kg per h for at least 24 h, arterial pH less than 7⋅20 despite resuscitation, and kalaemia more than 6⋅5 mmol/L. Each study site chose the method of renal-replacement therapy according to the local guidelines, which provided additional base because the dialysis included sodium bicarbonate as a buffer.23 In both groups, initiation of invasive mechanical ventilation was indicated if patients had one major or two minor predefined clinical events (appendix).24

942 patients were assessed for eligibility

400 were randomly assigned

542 excluded109 already received sodium bicarbonate

87 were in terminal decline76 had treatment limitation69 had chronic renal failure47 had immediate RRT indication 41 had ketoacidosis37 had digestive loss of sodium bicarbonate21 were eligible but not enrolled18 were included in another clinical study13 had hyperkalaemia with heart signs13 declined to participate 11 were under guardianship protection

201 assigned to control group

7 withdrew consent

22* violated inclusion or non-inclusion criteria

47* received sodium bicarbonate41 salvage therapies

6 misinterpretations of the protocol



199 assigned to bicarbonate group

4 withdrew consent

15 violated inclusion or non-inclusion criteria

1 did not receive sodium bicarbonate (misinterpretations of the protocol)



Figure 1: Trial profileRRT=renal-replacement therapy. *Seven patients both violated inclusion or non-inclusion criteria and received sodium bicarbonate.

1

I 74

Articles

4 www.thelancet.com Published online June 14, 2018 http://dx.doi.org/10.1016/S0140-6736(18)31080-8

Demographic characteristics, including Simplified Acute Physiology Score II and McCabe class, were collected at enrolment. Electrolytes were collected for the study’s purposes using each site’s local laboratory facility at enrolment, 12 h, 24 h, 48 h, and day 7.

OutcomesThe primary outcome measure was a composite of death from any cause by 28 days after randomisation and the presence of at least one organ failure at 7 days after randomisation. A-priori secondary outcome measures were the use, duration, and number of days alive free of life-support interventions (such as renal-replacement therapy, mechanical ventilation, and vasopressors); the SOFA score at enrolment and at 1 day, 2 days, and 7 days after enrolment; the total fluid intake between enrolment and day 2; the adverse events of electrolytes that occurred during the ICU stay (plasma pH >7⋅45, kalaemia >5 mmol/L or <3⋅2 mmol/L; natraemia >145 mmol/L, and ionised calcaemia <0⋅9 mmol/L); the occurrence of ICU-acquired infections; and the length of stay in the ICU.

Statistical analysisThis trial was planned with an interim analysis after the observation of the primary outcome of 200 patients (appendix). On the basis of a previous study,5 we calculated that a total of 376 patients would be needed for an 80% statistical power to show an absolute between-group difference of 15% in the primary outcome at a two-sided α level of 0⋅03 (0⋅02 for the interim analysis and 0⋅03 for the final analysis), assuming that the administration of sodium bicarbonate would be associated with a decrease from 45% to 30% in the primary endpoint. Assuming less than 8% non-analysable patients (loss to follow-up or consent withdrawal), we planned to randomly assign 400 patients.

All analyses were done on data from the intention-to-treat population, which included all patients who underwent randomisation. In the per-protocol analysis, we excluded patients with protocol violations (appendix). We made no imputation for missing values. Baseline characteristics in each study group were analysed as frequencies and percentages for categorical variables and as means and SDs or medians and IQRs for continuous variables, as appropriate. We used an unadjusted χ² test for the primary outcome analysis and its two components (ie, death from any cause by 28 days and the presence of at least one organ failure at 7 days). We did a multiple logistic regression for the primary outcome. The survival time was described by means of Kaplan-Meier method and compared with a log-rank test. A Cox proportional-hazards model was used to calculate hazard ratios (HRs) for death. For this analysis, data from all patients were censored at the time of death or at day 28. Logistic and Cox regression models were adjusted on relevant baseline covariates. Covariates were defined as binary variables and continuous variables dichotomised according to their median tested in the model, and were selected in a backward selection procedure if p<0⋅15 in the univariate analysis and then presented as adjusted odds ratios (ORs) or HRs with 95% CIs (appendix pp 28–34). An adjusted χ² test was done to compare day 28 mortality proportion in each group. For multiple comparisons in each prespecified stratum, a

Control group (n=194) Bicarbonate group (n=195)

Age

Median age (years) 65 (55–75) 66 (55–75)

≥65 100 (52%) 104 (53%)

<65 94 (48%) 104 (47%)

Sex

Men 123 (63%) 115 (59%)

Women 71 (37%) 80 (41%)

Body-mass index (kg/m²) 27 (23–30) 26 (23–29)

Simplified Acute Physiology Score II* 60 (48–73) 59 (49–73)

Pre-existing conditions†

Alcohol abuse 46 (24%) 38 (19%)

Current smoker 61 (31%) 53 (27%)

Diabetes mellitus 43 (22%) 54 (28%)

Chronic hypertension 90 (46%) 88 (45%)

Ischaemic heart disease 28 (14%) 31 (16%)

Chronic heart failure 7 (4%) 8 (4%)

Chronic kidney disease 15 (8%) 14 (7%)

Severe liver insufficiency 17 (9%) 11 (6%)

Cirrhosis 29 (15%) 23 (12%)

Chronic respiratory insufficiency 8 (4%) 9 (5%)

Chronic obstructive pulmonary disease 29 (15%) 18 (9%)

Immunocompromised‡ 36 (19%) 30 (15%)

McCabe class§

0 83 (53%) 94 (56%)

1 62 (39%) 50 (30%)

2 13 (8%) 24 (14%)

Sepsis 115 (59%) 123 (63%)

AKIN status¶

AKIN 0–1 104 (54%) 103 (53%)

AKIN 2–3 90 (46%) 92 (47%)

Source of admission

Medical 112 (58%) 110 (56%)

Surgical 82 (42%) 85 (44%)

Main condition associated with acidaemia at enrolment

Cardiac arrest 18 (9%) 18 (9%)

Septic shock 98 (51%) 107 (55%)

Haemorrhagic shock 40 (21%) 45 (23%)

Others 38 (20%) 25 (13%)

SOFA score|| at enrolment

Cardiovascular 4 (4–4) 4 (4–4)

Respiratory 2 (1–3) 2 (1–3)

Neurologic 1 (0–4) 0 (0–3)

Renal 2 (1–3) 2 (0–2)

Hepatic 0 (0–2) 1 (0–2)

Haematological 0 (0–1) 0 (0–2)

Total 10 (7–13) 10 (7–13)

(Table 1 continues on next page)

2576

% 0/. .

%

��

%,0 8: %

L KOM HC

25 % %PV %

. 46-3 %.-3% % 167

AI %

Articles


Holm-Bonferroni method was done to compute an adjusted p value. A mixed regression model was used to model repeated measures (appendix pp 18–21). Interactions between variables and time were tested. We also did all the analyses described above among prespecified strata of the randomisation. Tests for all outcomes were two-sided.

We did all analyses with SAS (version 9.2) or R (version 3.2.3). An independent data and safety monitoring committee, who were masked to the group allocation, supervised the conduct of the study and reviewed safety data, with interim analyses done after the inclusion of 100 and 200 patients. This trial is registered with ClinicalTrials.gov, number NCT02476253.

Role of the funding sourceThe funders of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. SJ, HB, NM, and BJ had full access to all the data. The corresponding author had final responsibility for the decision to submit for publication.

ResultsFrom May 5, 2015, to May 7, 2017, a total of 942 patients with severe metabolic acidaemia were assessed for trial eligibility (figure 1). Of these patients, 542 were excluded and 400 were randomly assigned to the study groups (201 in the control group and 199 in the bicarbonate group). After secondary exclusion of 11 patients who withdrew consent, a total of 389 patients were included in the intention-to-treat analysis (n=194 in the control group and n=195 in the bicarbonate group). The characteristics of the patients were well balanced between the two groups (table 1; appendix pp 28, 29). At randomisation, sepsis was present in 238 (61%) of 389 patients and acute kidney injury with AKIN scores of 2 or 3 in 182 (47%) patients. Invasive mechanical ventilation was used in 324 (83%) of 389 patients and vasopressors in 310 (80%) patients. Data for the primary outcome were available for all patients.

Overall, 341 (88%) of 389 patients adhered to the planned treatment in their randomisation group. Sodium bicarb-onate was infused in 47 (24%) of 194 patients in the control group, starting at a median of 7 h (IQR 3–27) after randomisation, and in 194 (99%) of 195 patients in the bicarbonate group, starting at a median of 0·2 h (0⋅1–0⋅4) after randomisation (table 2; appendix p 17). The proportion of patients in whom the targeted pH of 7⋅30 was reached and maintained for at least 36 h from enrolment to day 2 was 50 (26%) of 194 patients in the control group and 117 (60%) of 195 patients in the bicarbonate group, taking into account patients who died in the first 48 h (p<0⋅0001; appendix pp 18, 19). Overall, fluid intake from enrolment to day 1 and from day 1 to day 2 was not different between the two groups (table 2; appendix p 30).

Follow-up data were available for all patients for the primary composite outcome (ie, death by day 28 or at least one organ failure at day 7). The primary outcome occurred in 138 (71%) of 194 patients in the control group and

128 (66%) of 195 in the bicarbonate group (absolute difference estimate –5⋅5%, 95% CI –15⋅2 to 4⋅2; p=0⋅24; table 2) without significant effect of the treatment group (crude OR 0⋅775, 95% CI 0⋅505–1⋅190; p=0⋅24; appendix pp 33, 34). The Kaplan-Meier method estimate of the probability of survival at day 28 between the control group and bicarbonate group was not significant (46% [95% CI 40–54] vs 55% [49–63]; p=0⋅09; figure 2A). After multivariate analysis, sodium bicarbonate treatment was significantly associated with fewer deaths than no sodium bicarbonate treatment at day 28 (crude HR 0⋅783, 95% CI 0⋅0589–1⋅040; p=0⋅091; and adjusted HR 0⋅727, 95% CI 0⋅540–0⋅979; p=0⋅0356; appendix pp 35, 36).

In the a-priori defined stratum of patients enrolled with acute kidney injury with AKIN scores of 2 or 3, the primary outcome occurred in 74 (82%) of 90 patients in the control group and 64 (70%) of 92 patients in the bicarbonate group (absolute difference estimate –12⋅3%, 95% CI –26⋅0 to –0⋅1; p=0·0462; table 2). The Kaplan-Meier method estimate of survival by day 28 between the control group and bicarbonate group was significant (63% [95% CI 52–72] vs 46% [35–55]; p=0⋅0283; figures 2B, 2C). Univariate and multivariate analysis showed that sodium bicarbonate treatment was significantly associated with better outcome than no sodium bicarbonate treatment at day 28 (primary composite endpoint crude OR 0⋅494,


(Continued from previous page)

Physiological support†

Invasive mechanical ventilation 160 (82%) 164 (84%)

Vasopressor support 156 (80%) 154 (79%)

Laboratory results

Arterial pH 7·15 (7·11–7·18) 7·15 (7·09–7·18)

PaO2-to-FIO2 ratio (mm Hg) 229 (142–355) 264 (144–403)

PaCO2 (mm Hg) 37 (32–42) 38 (33–42)

Serum bicarbonate (mmol/L) 13 (10–15) 13 (10–15)

Serum lactate (mmol/L) 5·3 (3·4–9·0) 6·3 (3·6–9·7)

Serum lactate ≥2 mmol/L at enrolment 152 (78%) 168 (86%)

Serum creatinine (mg/dL) 1·76 (1·21–2·48) 1·67 (1·11–2·33)

Blood urea nitrogen (mg/dL) 31 (20–48) 28 (20–45)

Data are median (IQR), mean (SD), or n (%). FIO2=fractional concentration of oxygen in inspired air. AKIN=Acute Kidney Injury Network. SOFA=Sequential Organ Failure Assessment. *The Simplified Acute Physiology Score II25 is based on 17 variables; score ranges from 0 to 163, with higher scores indicating more severe disease. †Patients could have more than one pre-existing condition or physiological support, respectively. ‡Defined as >1 mg/kg per day prednisone for 30 days or more, HIV infection, biotherapy, or ongoing chemotherapy. §The McCabe score can range from 0 to 3, with higher scores indicating more severe underlying conditions. McCabe scores were available for 158 patients in the control group and for 168 in the bicarbonate group. ¶AKIN7,26 stages: stage 1 is serum creatinine increase ≥0·3 mg/dL (≥26·5 μmol/L), increase to 1·5–2·0-times from baseline, or urine output <0·5 mL/kg per h for 6 h; stage 2 is serum creatinine increase >2·0–3·0-times from baseline or urine output <0·5 mL/kg per h for 12 h; stage 3 is serum creatinine increase >3·0-times from baseline or serum creatinine ≥4·0 mg/dL (≥354 μmol/L) with an acute increase of at least 0·5 mg/dL (44 μmol/L), the need for renal replacement-therapy, or urine output <0·3 mL/kg per h for 12 h. AKIN zero means no kidney injury. To convert values for creatinine to μmol/L, multiply by 88·4. To convert values for blood urea nitrogen to mmol/L, multiply by 0·357. ||SOFA27 includes subscores ranging from 0 to 4 for each of six components (cardiovascular, respiratory, neurological, renal, hepatic, and haematological). Aggregated scores range from 0 to 24, with higher scores indicating more severe organ failure. The SOFA scores at days 1 and 2 after enrolment are shown in the appendix (p 24).

Table 1: Baseline characteristics of the intention-to-treat population

Articles







Age

Median age (years) 65 (55–75) 66 (55–75)

≥65 100 (52%) 104 (53%)

<65 94 (48%) 104 (47%)

Sex

Men 123 (63%) 115 (59%)

Women 71 (37%) 80 (41%)




Alcohol abuse 46 (24%) 38 (19%)








Cirrhosis 29 (15%) 23 (12%)




McCabe class§

0 83 (53%) 94 (56%)

1 62 (39%) 50 (30%)

2 13 (8%) 24 (14%)

Sepsis 115 (59%) 123 (63%)

AKIN status¶

AKIN 0–1 104 (54%) 103 (53%)

AKIN 2–3 90 (46%) 92 (47%)

Source of admission

Medical 112 (58%) 110 (56%)

Surgical 82 (42%) 85 (44%)



Septic shock 98 (51%) 107 (55%)


Others 38 (20%) 25 (13%)




Neurologic 1 (0–4) 0 (0–3)

Renal 2 (1–3) 2 (0–2)

Hepatic 0 (0–2) 1 (0–2)


Total 10 (7–13) 10 (7–13)


Articles







Age

Median age (years) 65 (55–75) 66 (55–75)

≥65 100 (52%) 104 (53%)

<65 94 (48%) 104 (47%)

Sex

Men 123 (63%) 115 (59%)

Women 71 (37%) 80 (41%)




Alcohol abuse 46 (24%) 38 (19%)








Cirrhosis 29 (15%) 23 (12%)




McCabe class§

0 83 (53%) 94 (56%)

1 62 (39%) 50 (30%)

2 13 (8%) 24 (14%)

Sepsis 115 (59%) 123 (63%)

AKIN status¶

AKIN 0–1 104 (54%) 103 (53%)

AKIN 2–3 90 (46%) 92 (47%)

Source of admission

Medical 112 (58%) 110 (56%)

Surgical 82 (42%) 85 (44%)



Septic shock 98 (51%) 107 (55%)


Others 38 (20%) 25 (13%)




Neurologic 1 (0–4) 0 (0–3)

Renal 2 (1–3) 2 (0–2)

Hepatic 0 (0–2) 1 (0–2)


Total 10 (7–13) 10 (7–13)


��

18

Figure S2: Arterial pH level in patients in the control and in the bicarbonate groups at baseline and after enrolment (overall population).

Figure S2 shows the arterial pH through day 7 for patients in the control and the bicarbonate groups respectively. Although arterial pH was not significantly different among treatment groups at H0, it differed significantly between groups afterwards. The horizontal line in the boxes indicates the median, the top and bottom of the box the 25th and 75th percentiles, and I bars the 5th and 95th percentiles. The open circle and the arrow represent the mean value in the control and the bicarbonate groups respectively. A mixed regression model was used to modelize repeated measures. Interactions between variables and time were tested.

19

Figure S3: Arterial bicarbonate level in patients in the control and in the bicarbonate groups at baseline and after enrolment (overall population).

Figure S3 shows the arterial bicarbonate concentration through day 7 for patients in the control and the bicarbonate groups respectively. Although arterial bicarbonate concentration was not significantly different among treatment groups at H0, it differed significantly between groups afterwards. The horizontal line in the boxes indicates the median, the top and bottom of the box the 25th and 75th percentiles, and I bars the 5th and 95th percentiles. The open circle and the arrow represent the mean value in the control and the bicarbonate groups respectively. A mixed regression model was used to modelize repeated measures. Interactions between variables and time were tested.

H %C 6 30. 94% 521 4 5

H %C O >7

0. .-/

Articles


95% CI 0⋅246–0⋅995; p=0⋅0483; and adjusted OR 0⋅387, 95% CI 0⋅163–0⋅918; p=0⋅0312; and mortality by day 28 crude HR 0⋅648, 95% CI 0⋅435–0⋅966; p=0⋅0332; and adjusted HR 0⋅592, 95% CI 0⋅392–0⋅895, p=0⋅0132; appendix pp 37–40).

In the per-protocol analysis, the primary outcome occurred in 88 (67%) of 132 patients in the control group and 117 (65%) of 179 in the bicarbonate group (p=0·81).

100 (52%) of 194 patients in the control group and 68 (35%) of 195 in the bicarbonate group underwent

Control group (n=194) Bicarbonate group (n=195) Absolute difference estimate (95% CI) p value

Primary outcome

Overall population (n=389)

Composite outcome 138 (71%) 128 (66%) –5·5 (–15·2 to 4·2) 0·24

Day 28 mortality 104 (54%) 87 (45%) –9·0 (–19·4 to 1·4) 0·07

At least one organ failure at day 7 134 (69%) 121 (62%) –2·8 (–15·4 to 9·8) 0·15

Patients with AKIN scores of 2–3* (n=182)

Composite outcome 74/90 (82%) 64/92 (70%) –12·3 (–26·0 to –0·1) 0·0462

Day 28 mortality 57/90 (63%) 42/92 (46%) –17·7 (–33·0 to –2·3) 0·0166

At least one organ failure at day 7 74/90 (82%) 61/92 (66%) –15·9 (–28·4 to –3·4) 0·0142

Secondary outcomes

Renal replacement therapy


Use of renal replacement therapy during ICU stay 100 (52%) 68 (35%) –16·7 (–26·4 to –7·0) 0·0009

Time from enrolment to initiation of renal replacement therapy (h) 7 (3–18) 19 (7–82) 8·8 (3·9 to 15·6) <0·0001

Renal replacement therapy-free days during ICU stay 8 (0–28) 19 (1–28) 0 (0·0 to 1·0) 0·015

Renal replacement therapy-free days during ICU stay in survivors 28 (25–28) 28 (25–28) 0 (0 to 0) 0·47

Dependence on dialysis at ICU discharge 11/32 (34%) 7/32 (22%) –12·5 (–34·3 to 9·3) 0·26


Use of renal replacement therapy during ICU stay 66/90 (73%) 47/92 (51%) –22·2 (–36·0 to –8·5) 0·0020


Renal replacement therapy-free days during ICU stay 1 (0–22) 10 (1–28) 1·0 (0·0 to 5·0) 0·0040

Renal replacement therapy-free days during ICU stay in survivors 24 (22–28) 28 (19–28) 1·0 (0·0 to 3·0) 0·45

Dependence on dialysis at ICU discharge 10/21 (48%) 5/25 (20%) –27·6 (–54·1 to –1·1) 0·0465

Other secondary outcomes


Cumulative fluid intake from enrolment to 24 h (mL) 3500 (1500–5250) 3350 (1800–5250) 34·0 (–450 to 500) 0·835

Cumulative sodium bicarbonate volume intake from enrolment to 24 h (mL) 0 (0–0) 500 (250–750) 500 (375 to 500) <0·0001

Cumulative sodium bicarbonate intake from enrolment to 24 h (mmol) 0 (0–0) 250 (1255–375) 250 (187 to 250) <0·0001

Cumulative fluid intake from 24 h to 48 h (mL) 1050 (0–2000) 1000 (0–2250) 0 (0 to 250) 0·53

Cumulative sodium bicarbonate volume intake from 24 h to 48 h (mL) 0 (0–0) 0 (0–0) 0 (0 to 0) 0·57

Cumulative sodium bicarbonate intake from 24 h to 48 h (mmol) 0 (0–0) 0 (0–0) 0 (0 to 0) 0·57

Duration of invasive mechanical ventilation (days) 3 (2–8) 3 (2–10) 0·0 (0·0 to 1·0) 0·17

Invasive mechanical ventilation-free days 0 (0–24) 4 (0–24) 0 (0 to 0) 0·48

In survivors 24 (17–26) 23 (14–26) –1·0 (–2·0 to 0·0) 0·13

Duration of vasopressor therapy (days) 2 (1–3) 2 (1–4) 0 (0 to 0) 0·36

Vasopressor-free days 9 (0–26) 19 (0–26) 0·0 (0·0 to 1·0) 0·10

In survivors 26 (24–27) 26 (23–27) 0·0 (–1·0 to 0·0) 0·34

ICU-acquired infections

Overall 43 (22%) 48 (25%) 2·4 (–6·1 to 10·8) 0·58

Pneumonia 23 (12%) 29 (15%) 3·0 (–3·8 to 9·8) 0·39

Urinary 10 (5%) 3 (2%) –3·7 (–7·3 to –0·1) 0·0461

Catheter 6 (3%) 8 (4%) 1·0 (–2·8 to 4·7) 0·61

Unexplained bloodstream infection 14 (7%) 13 (7%) –0·6 (–5·7 to 4·5) 0·81

Length of ICU stay (days) 4 (1–13) 5 (2–16) 1·0 (–2·7 to 0·0) 0·10

ICU-free days 0 (0–18) 0 (0–18) 0·0 (–1·1 to 0·0) 0·77

In survivors 20 (8–24) 16 (1–24) –1·0 (–4·0 to 0·0) 0·12


7 8 K 3 NA 2

3 I -

Articles


The lower mortality in the bicarbonate group among patients with acute kidney injury might have resulted from the combination of more vasopressor-free days and more renal-replacement therapy-free days than for patients in the control group. Sodium bicarbonate infusion in patients with severe metabolic acidaemia and acute kidney injury remains controversial,9 and a recent meta-analysis28 from the Cochrane group concluded that there is an inadequate number of randomised clinical trials that have assessed this question. In our trial, almost none of the patients were given sodium bicarbonate in the 24–48 h period, either because of recovery from severe acidaemia or because of death. Although the study was not designed specifically to assess the sodium bicarbonate timing issue, we hypothesise that early sodium bicarbonate infusion is of importance as illustrated by the rapid recovery from severe acidaemia in survivors in our previous observational study,5 which evaluated severe acidaemia (pH ≤7·20) in the critically ill and its management.

Number at riskControl group

Bicarbonate group

0 7 14 28

194195

115131

103117

89108

p=0·090

20

40

60

80

100Cu

mul

ativ

e pro

babi

lity o

f sur

viva

l (%

) Bicarbonate groupControl group

A

C

0 7 14 28

9092

4458

4052

3350

Days since inclusionDays since inclusion

p=0·0283

B

Control group

(n/N)

AKIN score0–12–3Age (years)<65≥65Sepsis statusNoYesAll patients

47/104 (45%)57/90 (63%)

42/94 (45%)62/100 (62%)

39/79 (49%)65/115 (57%)

104/194 (54%)

45/103 (44%)42/92 (46%)

32/89 (36%)55/106 (52%)

30/72 (42%)57/123 (46%)87/195 (45%)

–1·5 (–16·0 to 13·0)–17·7 (–33·0 to –2·3)

–8·7 (–24·0 to 6·5)–10·1 (–24·5 to 4·3)

–7·7 (–24·9 to 9·5)–10·2 (–23·7 to 3·3)–9·0 (–19·4 to 1·4)

0·830·0167

0·230·14

0·340·120·08

0·830·0334

0·230·28

0·340·24

0·0226

Bicarbonate group

(n/N)

Absolute differenceestimate (95% CI)

p value Adjustedp value

p value forheterogeneity

Favours sodiumbicarbonate treatment

Favours no sodium bicarbonate treatment

0–10 10–20–30

0·21

0·0031

Figure 2: Time to death in the overall population (A) and patients with prespecified acute kidney injury (B), and the 28-day mortality risk difference in the overall population and in the three prespecified strata (C)AKIN=Acute Kidney Injury Network. *A χ² test was done to compare day 28 mortality proportion in each group. For multiple comparisons in each prespecified stratum, a Holm-Bonferroni method was done to compute an adjusted p value.


Bicarbonate group

0 7 14 28

194195

6798

6087

5776

p<0·00010

20

40

60

80

100

Cum

ulat

ive u

se o

f ren

al–r

epla

cem

ent t

hera

py (%


Days since inclusion

Figure 3: Cumulative use of renal-replacement therapy from enrolment to day 28 in in the overall population

C 32 (

14 % - . 02 % % %)14 % ( . 02 % ) % .-

14 % . 02 % ))5% .14 % . . 02 % (.5% .%

C A14 I A 9 8 H R14AK 7 32 ( C A K 7

Articles




Bicarbonate group

0 7 14 28

194195

115131

103117

89108

p=0·090

20

40

60

80

100

Cum

ulat

ive p

roba

bilit

y of s

urvi

val (

%) Bicarbonate group

Control group

A

C

0 7 14 28

9092

4458

4052

3350


p=0·0283

B

Control group

(n/N)


47/104 (45%)57/90 (63%)

42/94 (45%)62/100 (62%)

39/79 (49%)65/115 (57%)

104/194 (54%)

45/103 (44%)42/92 (46%)

32/89 (36%)55/106 (52%)

30/72 (42%)57/123 (46%)87/195 (45%)

–1·5 (–16·0 to 13·0)–17·7 (–33·0 to –2·3)

–8·7 (–24·0 to 6·5)–10·1 (–24·5 to 4·3)

–7·7 (–24·9 to 9·5)–10·2 (–23·7 to 3·3)–9·0 (–19·4 to 1·4)

0·830·0167

0·230·14

0·340·120·08

0·830·0334

0·230·28

0·340·24

0·0226

Bicarbonate group

(n/N)






0–10 10–20–30

0·21

0·0031



Bicarbonate group

0 7 14 28

194195

6798

6087

5776

p<0·00010

20

40

60

80

100

Cum

ulat

ive u

se o

f ren

al–r

epla

cem

ent t

hera

py (%




N- 3

A 8 2 I K -

Articles






Primary outcome



Day 28 mortality 104 (54%) 87 (45%) –9·0 (–19·4 to 1·4) 0·07




Day 28 mortality 57/90 (63%) 42/92 (46%) –17·7 (–33·0 to –2·3) 0·0166


Secondary outcomes
























In survivors 24 (17–26) 23 (14–26) –1·0 (–2·0 to 0·0) 0·13



In survivors 26 (24–27) 26 (23–27) 0·0 (–1·0 to 0·0) 0·34


Overall 43 (22%) 48 (25%) 2·4 (–6·1 to 10·8) 0·58

Pneumonia 23 (12%) 29 (15%) 3·0 (–3·8 to 9·8) 0·39

Urinary 10 (5%) 3 (2%) –3·7 (–7·3 to –0·1) 0·0461

Catheter 6 (3%) 8 (4%) 1·0 (–2·8 to 4·7) 0·61



ICU-free days 0 (0–18) 0 (0–18) 0·0 (–1·1 to 0·0) 0·77

In survivors 20 (8–24) 16 (1–24) –1·0 (–4·0 to 0·0) 0·12


Articles






Primary outcome



Day 28 mortality 104 (54%) 87 (45%) –9·0 (–19·4 to 1·4) 0·07




Day 28 mortality 57/90 (63%) 42/92 (46%) –17·7 (–33·0 to –2·3) 0·0166


Secondary outcomes
























In survivors 24 (17–26) 23 (14–26) –1·0 (–2·0 to 0·0) 0·13



In survivors 26 (24–27) 26 (23–27) 0·0 (–1·0 to 0·0) 0·34


Overall 43 (22%) 48 (25%) 2·4 (–6·1 to 10·8) 0·58

Pneumonia 23 (12%) 29 (15%) 3·0 (–3·8 to 9·8) 0·39

Urinary 10 (5%) 3 (2%) –3·7 (–7·3 to –0·1) 0·0461

Catheter 6 (3%) 8 (4%) 1·0 (–2·8 to 4·7) 0·61



ICU-free days 0 (0–18) 0 (0–18) 0·0 (–1·1 to 0·0) 0·77

In survivors 20 (8–24) 16 (1–24) –1·0 (–4·0 to 0·0) 0·12


U RU

C I T R

41

Table S8: Renal Replacement Therapy in the overall population

Control Group

(N = 194)

Bicarbonate Group

(N =195)

Absolute difference (95% CI)

p value

Number of patients with at least one Renal Replacement Therapy session — no. (%) 100 (52) 68 (35) -16⋅7 (-26⋅4 to -7) 0⋅0009

Reason for Renal Replacement Therapy Initiation — no. (%)

Cardiopulmonary Edema 5/100 (5) 3/68 (4) -0⋅5 (-7⋅1 to 6⋅0) 1⋅00 Persistent urinary output lower than 0.3ml/kg/h 74/100 (74) 48/68 (71) -1⋅3 (-15⋅0 to 12⋅5) 0⋅63 Serum potassium 37/100 (37) 14/68 (21) -15⋅8 (-29⋅5 to -2⋅1) 0⋅0309 Persistent blood pH lower than 7.20 63/100 (63) 32/68 (48) -15⋅2 (-30⋅5 to 0⋅0) 0⋅05 Serum creatinine or blood urea nitrogen 45/100 (45) 44/68 (65) 20⋅7 (5⋅7 to 35⋅6) 0⋅0086

Continuous Renal Replacement Therapy as first modality — no. (%) 88/100 (88) 60/68 (88) -0⋅7 (-10⋅5 to 9⋅2) 1⋅00

Data are displayed as number of patients and percentage and median and 95% confidence interval (95% CI). The number of subjects/denominator is displayed if necessary.

• C% ,

• % 8 %

Articles




Bicarbonate group

0 7 14 28

194195

115131

103117

89108

p=0·090

20

40

60

80

100

Cum

ulat

ive p

roba

bilit

y of s

urvi

val (

%) Bicarbonate group

Control group

A

C

0 7 14 28

9092

4458

4052

3350


p=0·0283

B

Control group

(n/N)


47/104 (45%)57/90 (63%)

42/94 (45%)62/100 (62%)

39/79 (49%)65/115 (57%)

104/194 (54%)

45/103 (44%)42/92 (46%)

32/89 (36%)55/106 (52%)

30/72 (42%)57/123 (46%)87/195 (45%)

–1·5 (–16·0 to 13·0)–17·7 (–33·0 to –2·3)

–8·7 (–24·0 to 6·5)–10·1 (–24·5 to 4·3)

–7·7 (–24·9 to 9·5)–10·2 (–23·7 to 3·3)–9·0 (–19·4 to 1·4)

0·830·0167

0·230·14

0·340·120·08

0·830·0334

0·230·28

0·340·24

0·0226

Bicarbonate group

(n/N)






0–10 10–20–30

0·21

0·0031



Bicarbonate group

0 7 14 28

194195

6798

6087

5776

p<0·00010

20

40

60

80

100

Cum

ulat

ive u

se o

f ren

al–r

epla

cem

ent t

hera

py (%




.-2 001N /3 487 986 %

001I C 53 -. % 8

Articles






Primary outcome



Day 28 mortality 104 (54%) 87 (45%) –9·0 (–19·4 to 1·4) 0·07




Day 28 mortality 57/90 (63%) 42/92 (46%) –17·7 (–33·0 to –2·3) 0·0166


Secondary outcomes
























In survivors 24 (17–26) 23 (14–26) –1·0 (–2·0 to 0·0) 0·13



In survivors 26 (24–27) 26 (23–27) 0·0 (–1·0 to 0·0) 0·34


Overall 43 (22%) 48 (25%) 2·4 (–6·1 to 10·8) 0·58

Pneumonia 23 (12%) 29 (15%) 3·0 (–3·8 to 9·8) 0·39

Urinary 10 (5%) 3 (2%) –3·7 (–7·3 to –0·1) 0·0461

Catheter 6 (3%) 8 (4%) 1·0 (–2·8 to 4·7) 0·61



ICU-free days 0 (0–18) 0 (0–18) 0·0 (–1·1 to 0·0) 0·77

In survivors 20 (8–24) 16 (1–24) –1·0 (–4·0 to 0·0) 0·12


Articles






Primary outcome



Day 28 mortality 104 (54%) 87 (45%) –9·0 (–19·4 to 1·4) 0·07




Day 28 mortality 57/90 (63%) 42/92 (46%) –17·7 (–33·0 to –2·3) 0·0166


Secondary outcomes
























In survivors 24 (17–26) 23 (14–26) –1·0 (–2·0 to 0·0) 0·13



In survivors 26 (24–27) 26 (23–27) 0·0 (–1·0 to 0·0) 0·34


Overall 43 (22%) 48 (25%) 2·4 (–6·1 to 10·8) 0·58

Pneumonia 23 (12%) 29 (15%) 3·0 (–3·8 to 9·8) 0·39

Urinary 10 (5%) 3 (2%) –3·7 (–7·3 to –0·1) 0·0461

Catheter 6 (3%) 8 (4%) 1·0 (–2·8 to 4·7) 0·61



ICU-free days 0 (0–18) 0 (0–18) 0·0 (–1·1 to 0·0) 0·77

In survivors 20 (8–24) 16 (1–24) –1·0 (–4·0 to 0·0) 0·12


��

��#�!�� "� ��

Articles


renal-replacement therapy during their ICU stay (absolute difference estimate –16⋅7 days, 95% CI –26⋅4 to –7⋅0; p=0⋅0009; figure 3); and when indicated, renal-replacement therapy was started earlier in the control group than in the bicarbonate group (table 2). The number of days alive free from renal-replacement therapy was also significantly lower in the control group than in the bicarbonate group (table 2). Hyperkalaemia and acidaemia were the main reasons for initiation of renal-replacement therapy in the control group (appendix p 41). Sodium bicarbonate treatment was associated with less hyper-kalaemia (appendix p 23) and less persistent acidaemia (appendix p 18) than no sodium bicarbonate treatment at day 7. Serum creatinine and serum blood urea nitrogen were the main reasons to start renal-replacement therapy in the bicarbonate group (appendix p 41).

The findings were similar between the overall population and the AKIN 2–3 stratum, with more patients dependent on dialysis at ICU discharge in the control group than in the bicarbonate group (table 2). The number of days free from mechanical ventilation was not different between the two groups for both the overall population and the AKIN 2–3 stratum; and in the AKIN 2–3 stratum, the number of days free from vasopressor was higher in the bicarbonate group than in the control group (table 2).

Length of ICU stay did not differ significantly between the treatment groups for the overall population (table 2); length of hospital stay was 12 days (IQR 1–28) in the control group and 15 days (2–28) in the bicarbonate group.

Metabolic alkalosis, hypernatraemia, and hypocal-caemia were observed more frequently in the bicar bonate group than in the control group, with no life-threatening complications reported (appendix p 31). The appendix (pp 31, 32) provides the full details of the metabolic adverse events observed.

DiscussionIn this multicentre randomised trial involving critically ill patients with severe metabolic acidaemia (pH ≤7⋅20), the infusion of sodium bicarbonate, compared with no infusion, to reach and maintain a targeted pH of 7⋅30 did not significantly decrease the primary composite out-come of mortality by day 28 or the presence of at least one organ failure at day 7 in the overall population. However, sodium bicarbonate infusion decreased the need for renal-replacement therapy during the ICU stay. Moreover, in the a-priori stratum of patients with acute kidney injury at enrolment, infusion of sodium bicarbonate resulted in fewer deaths by day 28 than no infusion of sodium bicarbonate.


(Continued from previous page)

Patients with AKIN scores of 2–3† (n=182)

Cumulative fluid intake from enrolment to 24 h (mL) 3150 (1450–5250) 3400 (1500–5500) 150 (–560 to 880) 0·67


Cumulative sodium bicarbonate intake from enrolment to 24 h (mmol) 0 (0–125) 250 (150–500) 250 (187–250) <0·0001



Cumulative sodium bicarbonate intake from 24 h to 48 h (mmol) 0 (0–0) 0 (0–0) 0 (0–0) 0·99



In survivors 22 (18–28) 20 (12–25) –2 (–5·0 to 1·0) 0·18

Duration of vasopressor therapy (days) 2 (1–3) 2 (1–4) 0·0 (0·0 to 1·0) 0·46

Vasopressor-free days 1 (0–24) 18 (0–26) 1·0 (0 to 4) 0·022

In survivors 25 (24–27) 25 (22–27) –0·5 (–2 to 1) 0·73

ICU-acquired infections ·· ··

Overall 18/90 (20%) 19/92 (21%) 0·9 (–10·9 to 12·6) 0·88

Pneumonia 13/90 (14%) 14/92 (15%) 0·9 (–9·4 to 11·3) 0·86

Urinary 4/90 (4%) 1/92 (1%) –3·4 (–8·1 to 1·4) 0·21

Catheter 3/90 (3%) 3/92 (3%) 0·0 (–5·3 to 5·2) 1·00

Unexplained bloodstream infection 9/90 (10%) 3/92 (3%) –6·7 (–13·9 to 0·5) 0·07

Length of ICU stay (days) 4 (1–11) 7 (1–18) 1·0 (0·0 to 4·0) 0·06

ICU-free days 0 (0–13) 0 (0–14) 0 (0 to 0) 0·40

In survivors 17 (12–22) 13 (0–19) –3·0 (–8·0 to 0·0) 0·10

Data are n (%), n/N (%), or median (IQR). AKIN=Acute Kidney Injury Network. ICU=intensive care unit. *AKIN7,26 stages: stage 1 is serum creatinine increase ≥0·3 mg/dL (≥26·5 μmol/L), increase to 1·5–2·0-times from baseline, or urine output <0·5 mL/kg per h for 6 h; stage 2 is serum creatinine increase >2·0–3·0-times from baseline or urine output <0·5 mL/kg per h for 12 h; stage 3 is serum creatinine increase >3·0-times from baseline or serum creatinine ≥4·0 mg/dL (≥354 μmol/L) with an acute increase of at least 0·5 mg/dL (44 μmol/L), the need for renal replacement-therapy, or urine output <0·3 mL/kg per h for 12 h. AKIN zero means no kidney injury. To convert values for creatinine to μmol/L, multiply by 88·4. To convert values for blood urea nitrogen to mmol/L, multiply by 0·357.

Table 2: Primary and secondary outcomes of the intention-to-treat population

31

Table S3A: Metabolic adverse events in the overall population

Control Group

(N = 194)

Bicarbonate Group

(N = 195)

Absolute difference (95% CI)

p value

Number of patients with at least one Metabolic adverse events — no. (%) 150 (77) 154 (79) 1.7 (-6.6 to 9.9) 0⋅69

Number of patients with at least one blood pH higher than 7.45 — no. (%) 17 (9) 31 (16) 7.1 (0.6 to 13.6) 0⋅0324

Number of patients with at least one serum potassium higher than 5 mmol/l — no. (%)

95 (49) 62 (32) -17.2 (-26.8 to -7.6) 0⋅0006

Number of patients with at least one serum potassium lower than 3.2 mmol/l — no. (%)

61 (31) 70 (36) 4⋅5 (-4.9 to 13.8) 0⋅35

Number of patients with at least one serum sodium higher than 145 mmol/l — no. (%)

57 (29) 96 (49) 19.9 (10.4 to 29.4) 0⋅0001

Number of patients with at least one serum calcium lower than 0.9 mmol/l — no. (%)

29 (15) 48 (24) 9.7 (1.8 to 17.5) 0⋅0167

Number of patients with at least one blood transfusion — no. (%) 56 (29) 68 (36) 6.7 (-2.8 to 16.1) 0.17

Data are displayed as number of patients and percentage or median and 95% confidence interval (95% CI).

� ��

. H720 2. 2+ 9< C 429< 451 > 9 51

��

n a 5• 3 % C

C 5 1% 2n 8

A K N UA - 8A 8 8

3 I I A A 7 CA aA

A

��

nT R AT aN A

•

A K I

AB CKD 0 N BI 7090 % . % , R,,963 % % ,% ,

Copyright 2016 American Medical Association. All rights reserved.

uremic symptoms developed. No differences in kidneyrecovery or mortality were observed. In line with theseresults, a recently published multicenter trial investigatingaccelerated vs standard initiation of RRT in 101 critically illpatients with AKI also demonstrated no mortality differencebetween both groups.10 However, this was a feasibility trial,and the trial was not powered to investigate mortality.Finally, 1 small randomized clinical trial demonstrated thatearly initiation of RRT was associated with a reduced mor-tality compared with late initiation of RRT.18 In this study,the authors evaluated the role of early RRT in 28 patientswith AKI following cardiac surgery. Fourteen patients werestarted on continuous hemodialysis when their urine vol-ume decreased to less than 30 mL/h for 3 hours. In patientsin the “late” group (n = 14), RRT was delayed until urineoutput had fallen to less than 20 mL/h for 2 hours. Survivalwas significantly better in the group of patients who startedRRT earlier. There were no differences between the 2 groupswith respect to age, sex, Acute Physiology and ChronicHealth Evaluation (APACHE) II score, and serum creatininelevel at the time of initiation of RRT.

The results of a recently published meta-analysis suggestthat earlier initiation of RRT in critically ill patients with AKImay have beneficial association with survival (OR, 0.45 [95%CI, 0.28-0.72]).7 However, this conclusion is based on hetero-geneous studies of variable quality. Therefore, more random-ized trials are required to answer this question. This researchpriority has been articulated by the KDIGO clinical practiceguidelines,5 and the Acute Kidney Injury Network26 has pri-oritized this research topic.

Potential benefits of earlier initiation are attributable tomore rapid metabolic or uremic control and more effective pre-vention and management of fluid overload.27 Some data alsosuggest that RRT before the onset of severe AKI may attenu-ate kidney-specific and non–kidney organ injury from acide-mia, uremia, fluid overload, and systemic inflammation and

could potentially translate into improved survival and earlierrecovery of kidney function.28,29 The counterargument is thata strategy of early initiation of RRT might subject patients whowould recover renal function with conservative treatmentalone to the potential risks associated with RRT. However, AKIconfers a substantial increased risk of death even in patientsnever treated with RRT.30 As such, although there may be arisk of “unnecessary” RRT, there could be an even greater riskassociated with not providing it. To avoid treating patients withRRT who may have otherwise spontaneously recovered kidneyfunction, biomarkers in addition to the KDIGO classification

Figure 2. Mortality Probability Within 90 Days After Study Enrollmentfor Patients Receiving Early and Delayed Initiation of RenalReplacement Therapy (RRT)

100

80

60

40

20

00

112119

10

9290

20

8279

30

7870

40

7563

50

7362

60

6959

70

6958

80

6654

90

5548

Over

all M

orta

lity

Prob

abili

ty, %

Days Since Randomization

No. at riskEarly RRTDelayed RRT

Early RRT

Delayed RRT

Inverse normal log-rank test, P = .03; HR = 0.66 (95% CI, 0.45-0.97)

KDIGO indicates Kidney Disease: Improving Global Outcomes. In the delayedgroup, 18 patients had an absolute indication for RRT. The median (quartile 1[Q1], quartile 3 [Q3]) duration of follow-up was 90 days (Q1, Q3: 90, 90) in theearly group and 90 days (Q1, Q3: 90, 90) in the delayed group. The vertical ticksindicate censored cases.

Table 2. Patient Characteristics at the Time of Renal Replacement Therapy (RRT) Initiation

Early(n = 112)

Delayed(n = 119)

Absolute DifferenceEarly vs Delayed(95% CI) P Value

Received RRT, No. 112 108

Time from meeting eligibility criteria torandomization, median (Q1, Q3), h

2.0(1.0, 3.0)

2.0 (1.0, 3.0) 0.0 (0.0 to 0.0) .36

Time from KDIGO 2 to RRT, mean (SD), h 5.4 (2.2) 40.0 (54.5) −34.5(−45.0 to −24.0)

<.001

Time from KDIGO 2 to RRT,median (Q1, Q3), h

6.0(4.0, 7.0)

25.5(18.8, 40.3)

−21.0(−24.0 to −18.0)

<.001

Urinary output, median (Q1, Q3), mL 445.0(175.0, 807.5)

270.0(112.5, 670.0)

115.0(25.0 to 220.0)

.01

Serum creatinine, mean (SD), mg/dL 1.9 (0.6) 2.4 (1.0) −0.5(−0.7 to −0.3)

<.001

Blood urea nitrogen, mean (SD), mg/dL 38.5 (15.5) 47.5 (21.6) −9.0(−14.1 to −3.9)

.001

Potassium, mean (SD), mEq/L 5.1 (0.9) 5.1 (0.9) 0.0(−0.2 to 0.3)

.69

Bicarbonate, mean (SD), mEq/L 20.9 (3.6) 20.7 (3.7) 0.1(−0.9 to 1.1)

.79

Hemoglobin, mean (SD), g/dL 8.6 (1.3) 8.6 (1.4) −0.1(−0.4 to 0.3)

.74

White blood cells, mean (SD), ×109/L 16.2 (9.8) 16.5 (9.5) −0.3(−2.9 to 2.3)

.83

Abbreviations: KDIGO, KidneyDisease: Improving Global Outcomes,Q, quartile.SI conversion factor: To convertcreatinine to μmol/L, multiply by88.4; urea nitrogen to mmol/L,multiply by 0.357.

Research Original Investigation Early vs Delayed RRT in Critically Ill Patients With AKI

2196 JAMA May 24/31, 2016 Volume 315, Number 20 (Reprinted) jama.com

Copyright 2016 American Medical Association. All rights reserved.

Downloaded From: by a HIROSHIMA UNIVERSITY User on 07/09/2018

t 2+ 12

8 308365; MNBG %a lnuo8 308365; MNBG a %lnuo

K y b

r ,%U , PM ,U

t 2+ , 9 =

8 30e Zc y

8 30k8S s g Tg S vS1 : / % y

n engl j med 375;2 nejm.org July 14, 2016 127

Renal-Replacement Ther apy Initiation in the ICU

Supplementary Appendix). Sepsis was present in 494 patients (80%), and 389 patients (63%) had been exposed to nephrotoxic agents.

Renal-Replacement TherapyThe patients in the early-strategy group under-went their first renal-replacement therapy session within a median of 2 hours (interquartile range, 1 to 3) after randomization and within a median of 4.3 hours (interquartile range, 2.7 to 5.9) after documentation of stage 3 acute kidney injury and of the fulfillment of other inclusion criteria. In this group, six patients did not receive renal-replacement therapy (see the Supplementary Ap-pendix).

A total of 157 patients (51%) received renal-replacement therapy in the delayed-strategy group within a median of 57 hours (interquartile range, 25 to 83) after randomization (Fig. 1). The me-dian interval between the occurrence of at least one criterion mandating renal-replacement ther-apy and its initiation was 4.7 hours (interquartile range, 1.7 to 10.0). Five patients received renal-replacement therapy without meeting the initia-tion criteria. Persistence of oliguria or anuria for more than 72 hours after randomization and a blood urea nitrogen level higher than 112 mg per deciliter (40 mmol per liter) were the two most common reasons for renal-replacement ther-apy (Table S3 in the Supplementary Appendix).

Patient characteristics at the time of initiation of renal-replacement therapy are provided in Table S4 in the Supplementary Appendix. Meta-bolic abnormalities were more marked in the delayed-strategy group than in the early-strategy group. Details regarding the methods used for renal-replacement therapy, the changes in blood urea nitrogen and serum creatinine levels during follow-up, and the baseline predictors of renal-replacement therapy in the delayed-strategy group are provided in the Supplementary Appendix.

Primary and Secondary OutcomesFollow-up data at 60 days were available for 614 patients (99%); a total of 303 deaths had been observed by day 60 (150 in the early-strategy group and 153 in the delayed-strategy group). The mortality rates in our analysis were estimated with the use of the Kaplan–Meier method. The Kaplan–Meier estimate of the overall mortality at day 60 was 49.1% (95% confidence interval [CI], 45.0 to 52.9). Mortality did not differ sig-nificantly between the two study groups: 48.5%

(95% CI, 42.6 to 53.8) in the early-strategy group and 49.7% (95% CI, 43.8 to 55.0) in the delayed-strategy group (P = 0.79) (Fig. 1). Further stratifi-cation according to center and adjustment for important prognostic factors did not materially change the results.

A post hoc exploratory analysis was performed to compare patients who never received renal-

Figure 1. Probability of Survival and Timing of Renal-Replacement Therapy.

Panel A shows Kaplan–Meier curves of the probability of survival from ran-domization to day 60. Panel B shows the time from randomization to the initiation of renal-replacement therapy, stratified according to study group. Some patients in the early-strategy group received renal-replacement thera-py after 6 hours because of other emergencies resulting in postponement of the initiation of therapy by the medical team, a lack of availability of the renal-replacement therapy machine, or difficulties with catheter insertion. Other reasons for delay included situations such as a surgical procedure or radiologic examinations that needed to be performed before the initiation of renal-replacement therapy.

Prop

ortio

n S

urvi

ving

1.0

0.8

0.9

0.7

0.6

0.4

0.3

0.1

0.5

0.2

0.00 7 14 21 28 42 60

Days

B

A

P=0.79

No. at RiskEarly strategyDelayed strategy

311308

207204

241239

194191

179178

172165

161156

35

167161

157155

56

158156

49

Early strategy

Delayed strategy

Prop

ortio

n Fr

ee fr

om R

enal

-Re

plac

emen

t The

rapy

1.0

0.8

0.9

0.7

0.6

0.4

0.3

0.1

0.5

0.2

0.00 1 2 3 4 6 28

Days

P<0.001

No. at RiskEarly strategyDelayed strategy

311308

4192

4229

7268

4153

4135

3118

139

392

161

028

5

3105

013

24

021

7 8 12 16 20

Early strategy

Delayed strategy

The New England Journal of Medicine Downloaded from nejm.org at HIROSHIMA UNIV CENTRAL LIBRARY on July 10, 2018. For personal use only. No other uses without permission.

Copyright © 2016 Massachusetts Medical Society. All rights reserved.

cd bU , PM U

5BOE > N BI : 4JGI 7 9 E % . %%R963 %

��

n d-

• A dI

n f e K• a -

��

n .

n 4 .Nn % - m 3n N N 5 5n an N N / 4 1

2 .N 0 L 5

��

n Cn I + 1n

n 1•

n l C AK A on a N

n Ci

n I c C AA C A

n % a

n /2 7% 8• s

n /2 7 8• s %

n 001 % 43

n 001 T v A• 001 A

n N R I % 7 8

n K 9 5 . A

n H p 9720• mpLRO I

n a 6 % 5 >3 .FH A

n - C66 %

n 41 L88 % C66• SNTU 5C%

n RL/,, ? 1322 5

n 4 0

n T• % . 8 2 % ?

pin . 1 <546• sH. 1 TU IU 2 -• H 6 ED 3.5• m aH v rPq R

n o 60 +• sH08 A3 - 6 -• H 6 2B C S MNO E• m aH 7 0

M JJ 8: 1 8 I 20 J 8C ; KD 9 :8H9 8J IK C D JI H JH 8J 8:KJ; N AKHN : H8 ,8J898I NIJ 4 , 31.,

f 0.lb cU gnm28 hdU 4 6S T kj 4 6 a T Rie P

Cochrane Database of Systematic Reviews

Sodium bicarbonate supplements for treating acute kidneyinjury (Review)

Hewitt J, Uniacke M, Hansi NK, Venkat-Raman G, McCarthy K

Hewitt J, Uniacke M, Hansi NK, Venkat-Raman G, McCarthy K.

Sodium bicarbonate supplements for treating acute kidney injury.

Cochrane Database of Systematic Reviews 2012, Issue 6. Art. No.: CD009204.

DOI: 10.1002/14651858.CD009204.pub2.

www.cochranelibrary.com

Sodium bicarbonate supplements for treating acute kidney injury (Review)

Copyright © 2017 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.

8a 28 1!2 1# # !1 ' ! 18# c8 b ! 128 181 818 # 8# 8b 1 a 8 ... · chloremia. (e) lactatemia....

Documents