Journal Pre-proof

Risk of incident bleeding after acute kidney injury: A

retrospective cohort study

Farah Zarka, Alexandre Tayler-Gomez, Thierry Ducruet, Anatolie

Duca, Martin Albert, Amélie Bernier-Jean, Josée Bouchard

PII: S0883-9441(20)30562-1
DOI: https://doi.org/10.1016/j.jcrc.2020.05.003
Reference: YJCRC 53533

To appear in: Journal of Critical Care

Please cite this article as: F. Zarka, A. Tayler-Gomez, T. Ducruet, et al., Risk of incident
bleeding after acute kidney injury: A retrospective cohort study, Journal of Critical Care
(2020), https://doi.org/10.1016/j.jcrc.2020.05.003

This is a PDF file of an article that has undergone enhancements after acceptance, such
as the addition of a cover page and metadata, and formatting for readability, but it is
not yet the definitive version of record. This version will undergo additional copyediting,
typesetting and review before it is published in its final form, but we are providing this
version to give early visibility of the article. Please note that, during the production
process, errors may be discovered which could affect the content, and all legal disclaimers
that apply to the journal pertain.

© 2020 Published by Elsevier.

 Journal Pre-proof

Risk of incident bleeding after acute kidney injury: a retrospective cohort study

Farah Zarka MDa, Alexandre Tayler-Gomez MDa, Thierry Ducruet MScb, Anatolie Duca MDa,

Martin Albert MDa,c, Amélie Bernier-Jean MDa, Josée Bouchard MDa

of
ro
Author affiliations:
-p
re
a. Department of Medicine, Hôpital du Sacré-Coeur de Montréal, 5400 Blvd Gouin West,

Université de Montréal, Montreal, Quebec, Canada, H4J 1C5

lP

b. Applied Clinical Research Unit, Hôpital Ste-Justine, 3175 Chemin de la Cote Ste-
na

Catherine, Montreal, Quebec, Canada, H3T 1C5

c. Equipe de Recherche en Soins Intensifs (ERESI), Centre de Recherche Hôpital du Sacré-

ur

Coeur de Montréal, 5400 Blvd Gouin West, Montreal, Quebec, Canada, H4J 1C5
Jo

Email addresses: farah.zarka@hotmail.com; alexandre.tayler-gomez@umontreal.ca;

thierry.ducruet@gmail.com; anatolie.duca@crhsc.rtss.qc.ca; m.albert@umontreal.ca;

ameliebjean@gmail.com; josee.bouchard.1@umontreal.ca

Running title: Bleeding complications and AKI

1
 Journal Pre-proof

Address for correspondence:

Josée Bouchard MD, FRCPC

Hôpital du Sacré-Coeur de Montréal,

5400 Gouin Blvd West, Montreal, Quebec, Canada, H4J 1C5

E-mail: josee.bouchard.1@umontreal.ca

Phone 514-338-2883

of
Fax 514-338-2182

ro
Conflict of interest statement
-p
re
None.
lP

Financial Disclosure statement

na

This study was supported by research grants from the Nephrology Consortium of the Univers ity

of Montreal and the Sacré-Coeur Hospital Research Center. Josée Bouchard was a scholar of the
ur

Fonds de la Recherche du Québec – Santé. Funding agencies did not have any role in study design;
Jo

collection, analysis, and interpretation of data; writing the manuscript and the decision to submit

the report for publication.

Word count: Abstract: 200

Manuscript: 3092

2
 Journal Pre-proof

Abstract

Purpose. End-stage kidney disease (ESKD) causes bleeding diathesis; however, whether these

findings are extrapolable to acute kidney injury (AKI) remains uncertain. We assessed whether

AKI is associated with an increased risk of bleeding.

Methods: Single-center retrospective cohort study, excluding readmissions, admissions<24 hours,

ESKD or kidney transplants. The primary outcome was the development of incident bleeding

of
analyzed by multivariate time-dependent Cox models.

ro
Results: In 1,001 patients, bleeding occurred in 48% of AKI and 57% of non-AKI patients

(p=0.007). To identify predictors of incident bleeding, we excluded patients who bled before ICU

-p
(n=488). In bleeding-free patients (n=513), we observed a trend toward higher risks of bleeding in
re
AKI (22% vs. 16%,p=0.06), and a higher risk of bleeding in AKI-requiring dialysis (38% vs.
lP

17%,p=0.01). Cirrhosis, AKI-requiring dialysis, anticoagulation, and coronary artery disease were

associated with bleeding (HR 3.67, 95%CI:1.33-10.25; HR 2.82, 95%CI:1.26–6.32; HR 2.34,

na

95%CI:1.45–3.80; and HR 1.84, 95%CI:1.06–3.20, respectively), while SOFA score and sepsis
ur

had a protective association (HR 0.92 95%CI:0.84-0.99 and HR 0.55, 95%CI:0.34–0.91,

respectively). Incident bleeding was not associated with mortality.

Jo

Conclusions: AKI-requiring dialysis was associated with incident bleeding, independent of

anticoagulant administration. Studies are needed to better understand how AKI affects coagulatio n

and clinical outcomes.

Keywords: acute kidney injury, bleeding, hemorrhage, outcomes, time-dependent Cox model

3
 Journal Pre-proof

Abbreviations

AKI: acute kidney injury

CKD: chronic kidney disease

DOA: direct oral anticoagulant drugs

eGFR: estimated glomerular filtration rate

ESKD: End-stage kidney disease

HEME: HEmorrhage MEasurement

of
HR: hazard ratio

ro
ICU: intensive care unit

KDIGO: Kidney Disease Improving Global Outcomes

-p
re
MV: mechanical ventilation

SCr: serum creatinine

lP

SOFA: Sequential Organ Failure Assessment

na

STROBE: Strengthening the Reporting of Observational studies in Epidemiology

TAVI: transcatheter aortic valve implantation

ur
Jo

4
 Journal Pre-proof

Introduction

Acute kidney injury (AKI) is frequent and is independently associated with significant morbidity

and mortality [1-5]. However, the exact mechanisms underlying the increased morbidity and

mortality remain uncertain [6, 7]. In end-stage kidney disease (ESKD), uremia increases bleeding

diathesis by impairing platelet aggregation and platelet–vessel wall interaction [8, 9]. This risk

seems to be proportional to the degree and duration of uremia [8]. In AKI, the risk of bleeding may

also be related to other factors such as acute hepatic failure, thrombopenia, coagulopathy, recent

of
surgery, trauma or infection [8, 10, 11]. On the other hand, critically ill patients also often show

ro
signs of hypercoagulability [10, 11]. To our knowledge, only a few studies have characterized the

-p
risk of bleeding in AKI patients [12-15]. Whether AKI is an independent risk factor for bleeding
re
or a surrogate marker of increased severity of illness remains to be established.
lP

We aim to determine the incidence, sources, risk factors and outcomes of bleeding events
na

according to AKI status, and to determine the association between AKI and bleeding events in

critically ill patients. We hypothesized that AKI is an independent risk factor for incident bleeding
ur

events.
Jo

Material and Methods

Study design and population

We performed a retrospective cohort study of critically ill patients admitted to a tertiary care

academic center between January 1 st and December 31st , 2012. Our center includes a medical

intensive care unit (ICU), general surgery, cardiac surgery, neurosurgery and trauma ICUs. We

included randomly selected patients aged 18 or older according to their chart number using a

5
 Journal Pre-proof

computer-generated random list of numbers, and excluded patients readmittted to the ICU, on

chronic dialysis, admitted for less than 24 hours, and kidney transplant recipients. We determined

the timing of AKI, and the timing of bleeding events in relation to the diagnosis of AKI. In patients

with bleeding preceding AKI, we also characterized the incidence and timing of distinct new

episodes of bleeding. We presented our results in two cohorts. First, a descriptive cohort of all

eligible patients to characterize the epidemiology and prognosis associated with bleeding events

according to AKI status. This cohort allows us to assess the generalizability of our results. We then

of
presented a comparative cohort composed of all patients included in the descriptive cohort with

ro
the exception of those who had a bleeding event before ICU admission to characterize the

-p
predictors of incident bleeding. We followed the Strengthening the Reporting of Observational

studies in Epidemiology (STROBE) guidelines for observational studies [16]. The ethics’
re
committee of our institution approved the study. Written consent was waived given the
lP

retrospective observational nature of the study. The study was performed in accordance with the
na

1964 Declaration of Helsinki and its later amendments.

ur

AKI assessment
Jo

AKI was diagnosed on the day when the Kidney Disease Improving Global Outcomes (KDIGO)

serum creatinine (SCr) criteria were met; urine output criteria were not considered [17]. Patients

who had AKI before ICU admission were categorized as having AKI on day 0 if meeting AKI

criteria. Baseline SCr was defined as the closest value between 3 and 12 months before hospital

admission to reflect chronic kidney disease (CKD) status, and if unavailable, the furthest value

between 3 months before hospital admission and hospital admission. Baseline SCr was considered

missing if no SCr values were available before admission. According to KDIGO guidelines, when

6
 Journal Pre-proof

baseline creatinine was missing, we used a baseline SCr computed from the Modification of Diet

in Renal Disease (MDRD) formula, assuming an estimated glomerular filtration rate (eGFR) of 75

ml/min per 1.73 m2 .

Covariates

Our database includes patient demographics, premorbid conditions, laboratory results, sepsis

status, severity of illness scores and processes of care including surgical status, cumulative fluid

of
balance, administration of antiplatelet agents, anticoagulants, nonsteroidal anti-inflammator y

ro
drugs (NSAIDS), proton pump inhibitors, H2-receptor antagonists, use of central catheters,

-p
mechanical ventilation (MV) and dialysis. Anticoagulant therapy was defined as the use of
re
warfarin, intravenous unfractionated heparin, low molecular weight heparin, fondaparinux,

argatroban, and/or direct oral anticoagulant drugs (DOA) at doses providing therapeutic
lP

anticoagulation, which was assessed on a daily basis. CKD status was defined as an eGFR of less
na

than 60 ml/min per 1.73 m2 . Sequential Organ Failure Assessment (SOFA) and non-renal SOFA

scores were assessed at ICU admission [18]. Cumulative fluid balance was the sum of daily fluid
ur

balances during the first week of ICU admission.

Jo

Outcomes

The primary outcome was the incidence of de novo bleeding events after ICU admission, which

was independently adjudicated by two investigators (A.D. and J.B.). The bleeding episodes were

characterized using standardized definitions from the HEmorrhage MEasurement (HEME) in the

ICU tool, a validated bleeding measurement tool in critically ill patients [19, 20]. Only major

bleeding episodes were considered in this study, defined by any of the 5 following criteria: 1) life

7
 Journal Pre-proof

threatening bleeding with hypovolemic shock; 2) life threatening bleeding into a critical site (e.g.

intracranial or pericardial); 3) other critical bleeding (e.g., epidural, intraocular or intra-articula r);

4) bleeding requiring an invasive procedure (e.g., surgery, embolization); or 5) clinically important

bleeding requiring transfusion of 2 or more units of packed red blood cells, or associated with a

decrease in systolic blood pressure of at least 20 mm Hg or an increase in heart rate of at least 20

beats/minute [19, 20]. Other outcomes included requirement for dialysis and mortality at hospital

discharge.

of
ro
Statistical analyses and sample size calculation

-p
We used t-test or Mann-Whitney to analyze continuous variables when comparing 2 groups, and
re
ANOVA or Kruskall-Wallis when comparing more than 2 groups, where appropriate. We used the

2 -test or Fisher exact test to compare categorical variables, where appropriate. In patients free of
lP

bleeding at ICU admission, we used Kaplan-Meier curves and log-rank test to compare the length
na

of time after ICU admission until occurrence of first incident bleeding event in patients with and

without AKI. In the same population, we used a Cox proportional-hazards regression model with
ur

AKI modelled as a time-dependant variable to identify predictors of incident bleeding after ICU
Jo

admission. Variables with a p-value ≤0.20 by univariate analysis were included in the multivar iate

Cox regression model which was adjusted for age and gender. We also conducted a competing risk

analysis with mortality as the competing event. We verified the assumption of proportional hazards

using interaction terms with time. We used a multivariate Cox regression model to identify the

predictors of mortality at hospital discharge. Statistical tests were two sided and a p-value <0.05

was considered statistically significant for all comparisons. The analyses were performed using

SPSS, version 24.0 (IBM, Armonk, NY) and SAS 9.4 (SAS Institute, Cary, NC).

8
 Journal Pre-proof

Based on a previous study, the incidence of AKI and major bleeding at our center were 25% and

55% respectively [2]. A sample size of 1,000 patients would achieve a study power of 90% at a

type I error rate of 0.05 to detect a hazard ratio of 1.30 for the development of bleeding, assuming

an anticipated event rate of 20%.

Results

of
Overall population – descriptive cohort (n=1,001)

ro
We identified 2,464 patients admitted over a year, and 1,073 (43.5%) were randomly selected for

-p
the study population. After excluding 72 patients (Figure 1), 1,001 patients were included in the
re
descriptive analysis. For the comparative analysis, we excluded patients who experienced bleeding

before or upon ICU admission (n=488), to determine predictors of de novo bleeding (n=513).
lP
na

Patient characteristics, outcomes and bleeding sources are presented according to AKI and

bleeding status (Table 1). Among our population (n=1,001), 26% (n=257) developed AKI and
ur

55% (n=548) had a major bleeding event. AKI occurred at a median time of 1 day (IQR 0-3)
Jo

following ICU admission and 54% were stage 1, 18%, stage 2 and 28%, stage 3. Of the 101 patients

with a bleeding event occurring before AKI, 9 (8.9%) had one or more new bleeding events at a

median time of 2 (IQR 2-5) days after AKI. For the 21 patients who had a new bleeding event after

AKI, the first event occurred at a median of 2 (IQR 1-11) days after AKI. Overall, bleeding events

most often originated from peri and post-operative complications and intracerebral sources. Few

bleeding events originated from the gastrointestinal track (5.3%). Patients who did not develop

9
 Journal Pre-proof

AKI nor bleeding events were younger, suffered from fewer comorbidities, had lower SOFA

scores, were less likely to receive vasopressors and had fewer previous hemorrhagic events.

A total of 33 patients required dialysis. The proportion of patients on dialysis was not significa ntly

different across AKI subgroups (AKI alone, 9.6%; bleeding before AKI, 13.9%; and bleeding after

AKI, 28.6%; p=0.05 between groups). Patients with bleeding after AKI had a 38.1% risk of in-

hospital mortality, compared to 25.7% of patients with AKI after bleeding and 23.7% of patients

of
with AKI alone. Patients without AKI (with and without bleeding) had lower in-hospital mortality

ro
rates (7.5% and 7.9%, respectively) (p<0.001 between groups). Patients with bleeding after AKI

-p
had a hospital length of stay (LOS) of 19 days (IQR 15-40), compared to 14 days (IQR 7-32) in
re
those with AKI after bleeding and 17 days (IQR 8-27) in those with AKI alone. Patients without

AKI (with and without bleeding) had a shorter LOS (10 days (IQR 6-20) and 8 days (IQR 5-15),
lP

respectively) (p<0.001 between groups).

na

Population free of bleeding at ICU admission – comparative cohort (n=513)

ur

After excluding patients with a hemorrhagic event before or on the day of ICU admission, patients
Jo

with AKI had a trend toward a higher risk of bleeding compared to those without AKI that did not

reach statistical significance (22% vs. 16%, p=0.06). Patients with AKI requiring dialysis had a

higher risk of bleeding than other critically ill patients (38% vs. 17%, p=0.01). Twenty-seven

percent of patients on dialysis (9/33) received therapeutic anticoagulation (intravenous heparin)

and 6% (2/33) were on therapeutic anticoagulation while having a bleeding event. Therapeutic

anticoagulation was stopped when bleeding occurred in these 2 patients. The Kaplan-Meier curves

for cumulative incidence of bleeding according to AKI status are shown in Figure 2. In this cohort,

10
 Journal Pre-proof

AKI occurred at a median time of 0 day (IQR 0–3) after ICU admission. The first bleeding event

occurred at a median of 2 days (IQR 1–7) after ICU admission in patients with AKI and 2 days

(IQR 1–8) after ICU admission in patients without AKI. In patients who developed AKI, the first

bleeding episode occurred at a median of 2 days after AKI (IQR 1–11). Most bleeding originated

from peri-operative complications, post-operative complications or intracerebral sources, while

gastrointestinal hemorrhages were rare (1.0%) (Table 2). Only one patient experienced bleeding

from a catheter, which was not installed nor used for dialysis.

of
ro
In univariate analysis, coronary artery disease, chronic liver disease, trauma, cardiac event as the

-p
main diagnosis at ICU admission, provision of surgery, mechanical ventilation, vasopressors,
re
NSAIDS, antiplatelet drugs and anticoagulant therapy during hospitalization were significa ntly

associated with the occurrence of bleeding (Table 2), whereas higher SOFA score at ICU
lP

admission and sepsis as the main diagnosis at ICU admission were inversely associated with the
na

occurrence of bleeding. In multivariate analysis, chronic liver disease, AKI requiring dialys is

anticoagulation therapy, and coronary artery disease (CAD) were associated with a higher risk of
ur

bleeding events, while higher SOFA score and sepsis had a protective association with the risk of
Jo

bleeding (Table 3). Mechanical ventilation and use of vasopressors were not included in the model

since these variables are already included in the SOFA score. There was no interaction between

AKI and CKD status in the multivariate analysis. We also performed a supplementary analysis to

predict the risk of bleeding competed by the risk of mortality. The HR of subsequent bleeding for

AKI requiring dialysis remained similar (HR 2.57 95% CI 1.32–4.96; p=0.005). (Supplementary

Table 1).

11
 Journal Pre-proof

In a multivariate Cox regression model to predict mortality, age, mechanical ventilation, higher

fluid balance and SOFA score were significantly associated with a higher risk of death, while a

recent surgical procedure was associated with a lower risk of mortality (Table 4). The occurrence

of a new bleeding event was not associated with an increased risk of mortality.

Discussion

In this study, we have found that new bleeding events were relatively frequent following severe

of
AKI, occurring twice more often in patients with AKI requiring dialysis than in other critically ill

ro
patients (38% vs 17%; p=0.01). The excess of bleeding events occurred within days after AKI.

-p
This result was not related to anticoagulation administered for dialysis, since only two patients had
re
bleeding while on therapeutic anticoagulation. There was also a trend towards an increased risk of

bleeding in patients with AKI compared to patients without AKI (22% vs. 16%; p=0.06). In a Cox
lP

proportional hazards model where AKI status was modelled as a time-dependent variable, AKI
na

requiring dialysis was an independent risk factor for the development of bleeding, with a HR of

2.82 (95%CI 1.26–6.32). The crude mortality rate associated with bleeding following AKI was
ur

higher than when bleeding preceded AKI (38% vs. 26%), or when AKI occurred without bleeding
Jo

(24%). However, in a Cox proportional hazards model, incident bleedings were not associated with

a higher risk of mortality.

Over the last decade, some studies have attempted to explain the excess mortality and morbidity

associated with AKI [6, 21]. Results from animal experiments suggest that AKI can affect other

organs, contributing to non-renal complications [6, 22-26]. Our study is one of a few on the

epidemiology of bleeding events following AKI [10, 12-15, 27-29]. Our results further support the

12
 Journal Pre-proof

association between severe AKI and the risk of bleeding in a heterogeneous population of critica lly

ill patients.

The finding that AKI requiring dialysis is an independent risk factor for bleeding events is

consistent with prior preclinical studies. Firstly, uremia is responsible for qualitative platelet

defects, and higher levels of uremia make these defects more common and more severe. However,

the risk of bleeding appears to vary greatly across degrees of uremia [8, 9, 30]. In addition, studies

of
have shown that while dialysis improves platelet function, it rarely normalizes coagulatio n

ro
parameters [9].

-p
re
Limited studies have compared the risk of bleeding in patients with and without AKI [14, 15, 29].

In a case-control study of 348 patients undergoing radiocontrast exposure, 15% of the 60 patients
lP

that developed AKI and died had a pre-existing bleeding event and 27% developed a new bleeding
na

event after AKI [14]. No data were available on risk factors of bleeding. Some studies have found

that AKI independently predicts the risk of bleeding in specific populations, such as patients
ur

undergoing paracentesis for decompensated cirrhosis [29] or following transcatheter aortic valve
Jo

implantation (TAVI) [15]. One study has reported 13% of incident hemorrhagic events following

AKI [12].

In our study, we observed an incidence of bleeding after AKI requiring dialysis of 38%, which is

also comparable to what has been reported in the literature. Twenty years ago, the incidence of

bleeding complications in critically ill adults on dialysis ranged from 26% to 47% [27, 28]. In

more recent studies, the reported incidence of major bleeding varied between 23% and 27% [10,

13
 Journal Pre-proof

13]. Regarding risk factors of bleeding, comparable risk factors in critically and non-critically ill

patients have been reported, such as liver disease [12, 31], AKI [12, 31], and anticoagulation [31].

One study has found infectious events to be a predictor of bleeding [31]. However, in our study,

sepsis was associated with a lower risk of bleeding. Another observational study of 83 patients

with decompensated cirrhosis corroborated this finding [29]. In addition, we found non-renal

SOFA score to be inversely correlated with bleeding. However, in the competing risk analysis, the

SOFA score was significantly associated with a higher risk of bleeding. We did not have daily data

of
to ascertain whether this relationship would remain over time. Finally, although uremia is

ro
associated with bleeding, our CKD population included only a limited number of patients with

-p
moderate to severe CKD who are at higher risk of hemorrhage [32, 33]. Therefore, we did not
re
expect that CKD would be associated with bleeding in our study.
lP

We found no evidence that bleeding events were associated with an increased risk of mortality
na

post AKI. This finding is consistent with a previous observational study of 198 patients with

dialysis-dependent acute kidney injury [13]. However, a recent meta-analysis in patients

ur

undergoing TAVI showed that AKI was associated not only with an increased risk for major
Jo

bleeding, but also with life-threatening bleeding events [15]. Similarly, acute gastrointestina l

hemorrhage has been independently associated with an increased risk of death in AKI patients

admitted to a nephrology intermediate care unit [12]. Of note, our study was underpowered to

assess the relationship between bleeding events and mortality, which may have led to the lack of

an association between incident bleeding and mortality. Similarly, our study was underpowered to

detect an increased risk of mortality with AKI-requiring dialysis.

14
 Journal Pre-proof

Our study has several strengths. First, we used a recent database including a large heterogeneous

population of critically ill patients with detailed data on anticoagulation and suffering from various

degrees of AKI severity to reflect current practices and standards of care. Second, two independent

investigators adjudicated all suspected major bleeding events based on standardized definitio ns

from a validated tool in critically ill patients [19, 20]. Third, we performed a Cox proportional

hazards model with AKI modelled as a time-dependant variable and adjusting for the most

important covariates that may confound the association between AKI and the development of

of
bleeding. We also performed a competing risk analysis, the results of which were consistent with

ro
our main findings. We also found similar rates of AKI and bleeding events in our cohort to those

-p
reported in the literature, supporting the generalizability of our findings. Furthermore, the risk
re
factors we have identified are similar to those previously described and plausible from a

pathophysiologic standpoint.
lP
na

The present results, however, should be interpreted in light of some limitations. First, it was a

single-center study. Second, since it was an observational cohort, the possibility of residual
ur

confounding and bias remains. For example, we did not have daily data on SOFA scores to adjust
Jo

our results. Third, our study does not confirm the causal relationship between AKI and bleeding

nor offers a pathophysiological mechanism for how AKI could lead to bleeding. Finally, follow-

up was limited to hospital stay.

Conclusions

We have found that severe AKI requiring dialysis is associated with an increase occurrence of

bleeding events in critically ill patients. The association is independent of other usual bleeding risk

15
 Journal Pre-proof

factors, including anticoagulation. Patients with severe AKI may benefit from a closer monitor ing

of bleeding events. Further studies are needed to understand how AKI may affect coagulation and

platelet function before these results can translate into clinical outcomes.

of
ro
-p
re
lP
na
ur
Jo

16
 Journal Pre-proof

Funding

This study was supported by research grants from the Nephrology Consortium of the Univers ity

of Montreal and the Sacré-Coeur Hospital Research Center. Josée Bouchard was a scholar of the

Fonds de la Recherche du Québec – Santé. Funding agencies did not have any role in study design;

collection, analysis, and interpretation of data; writing the manuscript and the decision to submit

the report for publication.

of
Declarations of interest

ro
None.

Authors’ contributions
-p
re
Conceptualization: JB; data curation/interpretation: FZ, ATG, TD, AD, MA, ABJ, JB; formal
lP

analysis: FZ, TD, JB; writing original draft: FZ, ATG, JB; writing review and editing: TD, AD,
na

MA, ABJ. Each author provided intellectual content of critical importance to the work described

and approved the final version.

ur
Jo

17
 Journal Pre-proof

References

[1] Hoste EA, Bagshaw SM, Bellomo R, Cely CM, Colman R, Cruz DN, et al. Epidemiology

of acute kidney injury in critically ill patients: the multinational AKI-EPI study. Intensive

Care Med 2015;41(8):1411-23.

[2] Bouchard J, Acharya A, Cerda J, Maccariello ER, Madarasu RC, Tolwani AJ, et al. A

Prospective International Multicenter Study of AKI in the Intensive Care Unit. Clinical

of
journal of the American Society of Nephrology : CJASN 2015;10(8):1324-31.

ro
[3] Mehta RL, Burdmann EA, Cerda J, Feehally J, Finkelstein F, Garcia-Garcia G, et al.

-p
Recognition and management of acute kidney injury in the International Society of
re
Nephrology 0by25 Global Snapshot: a multinational cross-sectional study. Lancet

2016;387(10032):2017-25.
lP

[4] Mizota T, Dong L, Takeda C, Shiraki A, Matsukawa S, Shimizu S, et al. Transient acute
na

kidney injury after major abdominal surgery increases chronic kidney disease risk and 1-

year mortality. J Crit Care 2019;50:17-22.

ur

[5] Li DH, Wald R, Blum D, McArthur E, James MT, Burns KEA, et al. Predicting mortality
Jo

among critically ill patients with acute kidney injury treated with renal replacement

therapy: Development and validation of new prediction models. J Crit Care 2019;56:113-

9.

[6] Doi K, Rabb H. Impact of acute kidney injury on distant organ function: recent findings

and potential therapeutic targets. Kidney Int 2016;89(3):555-64.

[7] Woodrow G, Turney JH. Cause of death in acute renal failure. Nephrol Dial Transplant

1992;7(3):230-4.

18
 Journal Pre-proof

[8] Eberst ME, Berkowitz LR. Hemostasis in renal disease: pathophysiology and

management. Am J Med 1994;96(2):168-79.

[9] Castaldi PA, Rozenberg MC, Stewart JH. The bleeding disorder of uraemia. A qualitative

platelet defect. Lancet 1966;2(7454):66-9.

[10] Pschowski R, Briegel S, Von Haehling S, Doehner W, Bender TO, Pape UF, et al. Effects

of dialysis modality on blood loss, bleeding complications and transfusion requirements

in critically ill patients with dialysis-dependent acute renal failure. Anaesth Intensive

of
Care 2015;43(6):764-70.

ro
[11] Schetz MR. Coagulation disorders in acute renal failure. Kidney Int Suppl 1998;66:S96-

101.
-p
re
[12] Fiaccadori E, Maggiore U, Clima B, Melfa L, Rotelli C, Borghetti A. Incidence, risk

factors, and prognosis of gastrointestinal hemorrhage complicating acute renal failure.

lP

Kidney Int 2001;59(4):1510-9.

na

[13] Kreuzer M, Ehrich JH, Pape L. Haemorrhagic complications in paediatric dialysis-

dependent acute kidney injury: Incidence and impact on outcome. Nephrol Dial
ur

Transplant 2010;25(4):1140-6.
Jo

[14] Levy EM, Viscoli CM, Horwitz RI. The effect of acute renal failure on mortality. A

cohort analysis. JAMA 1996;275(19):1489-94.

[15] Ma M, Gao WD, Gu YF, Wang YS, Zhu Y, He Y. Clinical effects of acute kidney injury

after transcatheter aortic valve implantation: a systematic review and meta-analysis.

Intern Emerg Med 2019;14(1):161-75.

[16] von Elm E, Altman DG, Egger M, Pocock SJ, Gotzsche PC, Vandenbroucke JP, et al.

The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE)

19
 Journal Pre-proof

statement: guidelines for reporting observational studies. Ann Intern Med

2007;147(8):573-7.

[17] Kidney Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury Work

Group. KDIGO Clinical Practice Guideline for Acute Kidney Injury. Kidney

International Suppl 2012;2:1-138.

[18] Vincent JL, Moreno R, Takala J, Willatts S, De Mendonca A, Bruining H, et al. The

SOFA (Sepsis-related Organ Failure Assessment) score to describe organ

of
dysfunction/failure. On behalf of the Working Group on Sepsis-Related Problems of the

ro
European Society of Intensive Care Medicine. Intensive Care Med 1996;22(7):707-10.

[19]
-p
Arnold DM, Donahoe L, Clarke FJ, Tkaczyk AJ, Heels-Ansdell D, Zytaruk N, et al.
re
Bleeding during critical illness: a prospective cohort study using a new measurement

tool. Clin Invest Med 2007;30(2):E93-102.

lP

[20] Arnold DM, Lauzier F, Rabbat C, Zytaruk N, Barlow Cash B, Clarke F, et al.
na

Adjudication of bleeding outcomes in an international thromboprophylaxis trial in critical

illness. Thromb Res 2013;131(3):204-9.

ur

[21] Faubel S, Shah PB. Immediate Consequences of Acute Kidney Injury: The Impact of
Jo

Traditional and Nontraditional Complications on Mortality in Acute Kidney Injury. Adv

Chronic Kidney Dis 2016;23(3):179-85.

[22] Grams ME, Rabb H. The distant organ effects of acute kidney injury. Kidney Int

2012;81(10):942-8.

[23] Hassoun HT, Grigoryev DN, Lie ML, Liu M, Cheadle C, Tuder RM, et al. Ischemic acute

kidney injury induces a distant organ functional and genomic response distinguishable

from bilateral nephrectomy. Am J Physiol Renal Physiol 2007;293(1):F30-40.

20
 Journal Pre-proof

[24] Kelly KJ. Distant effects of experimental renal ischemia/reperfusion injury. J Am Soc

Nephrol 2003;14(6):1549-58.

[25] Liu M, Liang Y, Chigurupati S, Lathia JD, Pletnikov M, Sun Z, et al. Acute kidney injury

leads to inflammation and functional changes in the brain. J Am Soc Nephrol

2008;19(7):1360-70.

[26] Park SW, Kim M, Kim JY, Ham A, Brown KM, Mori-Akiyama Y, et al. Paneth cell-

mediated multiorgan dysfunction after acute kidney injury. J Immunol

of
2012;189(11):5421-33.

ro
[27] Ward DM. The approach to anticoagulation in patients treated with extracorporeal

-p
therapy in the intensive care unit. Adv Ren Replace Ther 1997;4(2):160-73.
re
[28] van de Wetering J, Westendorp RG, van der Hoeven JG, Stolk B, Feuth JD, Chang PC.

Heparin use in continuous renal replacement procedures: the struggle between filter
lP

coagulation and patient hemorrhage. J Am Soc Nephrol 1996;7(1):145-50.

na

[29] Hung A, Garcia-Tsao G. Acute kidney injury, but not sepsis, is associated with higher

procedure-related bleeding in patients with decompensated cirrhosis. Liver Int

ur

2018;38(8):1437-41.
Jo

[30] Steiner RW, Coggins C, Carvalho AC. Bleeding time in uremia: a useful test to assess

clinical bleeding. Am J Hematol 1979;7(2):107-17.

[31] Herzig SJ, Rothberg MB, Feinbloom DB, Howell MD, Ho KK, Ngo LH, et al. Risk

factors for nosocomial gastrointestinal bleeding and use of acid-suppressive medication

in non-critically ill patients. J Gen Intern Med 2013;28(5):683-90.

[32] Corapi KM, Chen JL, Balk EM, Gordon CE. Bleeding complications of native kidney

biopsy: a systematic review and meta-analysis. Am J Kidney Dis 2012;60(1):62-73.

21
 Journal Pre-proof

[33] Arnson Y, Hoshen M, Berliner-Sendrey A, Reges O, Balicer R, Leibowitz M, et al. Risk

of Stroke, Bleeding, and Death in Patients with Nonvalvular Atrial Fibrillation and

Chronic Kidney Disease. Cardiology 2020:1-9.

of
ro
-p
re
lP
na
ur
Jo

22
 Journal Pre-proof

Table 1. Descriptive characteristics and outcomes of patients stratified by timing of bleeding and
acute kidney injury (n = 1,001)

With acute kidney injury Without acute kidney P*

All injury
patients
Bleeding No Bleeding With No
before bleeding after AKI bleeding bleeding
AKI
n=1,001 n=101 n=135 n=21 n=426 n=318
Age, years median 66 (56-75) 70 (57-77) 70 (62-78) 70 (62-77) 65 (56-74) 65 (51-74) <0.001
(IQR)
Gender, male (%) 606 (61) 63 (62) 80 (59) 10 (48) 275 (65) 178 (56) 0.12
Race, Caucasian or 826 (83) 83 (8.3) 117 (12) 17 (1.7) 348 (35) 261 (26) 0.75
Asian (%)

of
Comorbidities (%)
Hypertension 583 (58) 61 (60) 88 (65) 18 (86) 245 (58) 171 (54) 0.02
Diabetes 260 (26) 26 (26) 57 (42) 10 (48) 94 (22) 73 (23) <0.001

ro
CAD 381 (38) 38 (38) 55 (41) 13 (62) 194 (46) 81 (26) <0.001
CKD 121 (12) 17 (17) 58 (43) 11 (52) 22 (5.2) 13 (4.1) <0.001
Chronic liver disease 21 (2.1) 9 (8.9) 4 (3.0) 2 (9.5) 6 (1.4) 0 (0) <0.001
History of bleeding
Previous ulcer (%)
NSAIDS on admission
24 (2.4)
28 (2.8)
85 (8.5)
6 (5.9)
6 (5.9)
11 (10.9) -p
6 (4.4)
4 (3.0)
11 (8.1)
2 (9.5)
1 (4.8)
4 (19)
7 (1.6)
12 (2.8)
41 (9.6)
3 (0.9)
5 (1.6)
18 (5.7)
0.003
0.22
0.10
re
(%)
Steroids on admission 92 (9.2) 7 (6.9) 21 (15.6) 6 (29) 27 (6.3) 31 (9.7) <0.001
(%)
lP

Anticoagulation on 146 (15) 25 (25) 30 (22) 7 (33) 51 (12) 33 (10) <0.001

admission (%)
Anti-platelets on 401 (40) 42 (42) 60 (44) 9 (43) 177 (42) 113 (36) 0.36
admission (%)
PPI or anti-H2 on 328 (33) 32 (32) 51 (38) 11 (52) 120 (28) 114 (36) 0.03
na

admission (%)
M ain ICU admission
diagnosis (%)
Cardiological 64 (6.4) 3 (3.0) 19 (14) 7 (33) 22 (5.2) 13 (4.1) <0.001
ur

Hemorrhage 33 (3.3) 12 (12) 0 (0) 0 (0) 21 (4.9) 0 (0) <0.001

Neuro 124 (12) 13 (13) 4 (3.0) 1 (4.8) 71 (17) 35 (11) 0.001
Sepsis/severe 170 (17) 9 (8.9) 66 (49) 7 (33) 8 (1.9) 80 (25) <0.001
Jo

infection
Surgery (%) 434 (43) 46 (46) 20 (15) 2 (9.5) 248 (58) 118 (37) <0.001
Trauma (%) 79 (7.9) 11 (11) 6 (4.4) 1 (4.8) 43 (10) 18 (5.7) 0.07
Other (%) 97 (9.7) 7 (7.0) 20 (15) 3 (14) 13 (3.1) 54 (17) <0.001
Surgery (%) 683 (68) 80 (79) 55 (41) 14 (67) 366 (86) 168 (53) <0.001
M echanical ventilation 564 (56) 81 (80) 73 (54) 18 (86) 285 (67) 107 (34) <0.001
(%)
Use of vasopressors (%) 513 (51) 80 (79) 99 (73) 16 (76) 222 (52) 96 (30) <0.001
SOFA score, median 5 (2-8) 8 (4.5-10) 7 (5-10) 6 (3.5-11) 5 (2-8) 3 (2-5) <0.001
(IQR)
Fluid balance, ml (IQR) 824 (-663 – 1890 (-558 1278 (- 871 (- 725 (-640 585 (-606 0.001
2519) – 5564) 825 – 3261 – – 2231) –1841)
4204) 4316)
PPI or anti-H2 per 829 (84) 82 (82) 116 (88) 20 (95) 352 (84) 259 (83) 0.39
hospitalization (%)
NSAIDS per 289 34 (34) 19 (14) 2 (9.5) 163 (38) 71 (23) <0.001
hospitalization (%) (29)
Antiplatelets drugs per 531 (53) 56 (55) 75 (56) 13 (62) 240 (57) 147 (47) 0.09
hospitalization (%)

23
 Journal Pre-proof

Anticoagulation per 311 (31) 39 (39) 58 (43) 12 (57) 131 (31) 71 (23) <0.001
hospitalization (%)
Platelets levels, median 153 (93- 121 (60- 171 (101- 120 (82- 133 (84- 190 (134- <0.001
(IQR) 216) 167) 233) 218) 186) 254)
INR, median (IQR) 1.20 (1.07- 1.27 (1.14- 1.26 1.32 1.20 1.10 (1.02- <0.001
1.34) 1.40) (1.13- (1.15- (1.09- 1.24)
1.54) 1.81) 1.35)
PTT, median (IQR) 38.8 (34.8- 46.3 (38- 42.8 55 (39.8- 38.4 36.9 (33.4- <0.001
47.6) 59.1) (36.3- 113.3) (34.8- 43.4)
57.3) 43.8)
Bilirubin, umol/L, 9 (6-16) 14 (9-23) 9 (5-18) 15 (7-25) 9 (5-14) 8 (6-12) <0.001
median (IQR)
Anticoagulation therapy 195 (20) 20 (20) 37 (27) 9 (43) 83 (20) 46 (15) 0.001
(%)
Bleeding source (%)
Gastrointestinal 29 (5.3) 12 (12) n/a 3 (14) 14 (3.3) n/a <0.001
Intra-cerebral 132 (24) 18 (18) n/a 2 (9.5) 112 (26) n/a 0.06

of
Peri-operative 280 (51) 44 (44) n/a 6 (29) 230 (54) n/a 0.02
Post-operative 54 (9.9) 14 (14) n/a 3 (14) 37 (8.7) n/a 0.23
Other 53 (9.7) 13 (13) n/a 7 (33) 33 (7.7) n/a <0.001

ro
Requirement for dialysis 33 (3.3) 14 (14) 13 (9.6) 6 (29) n/a n/a 0.05
(%)
M ortality (%) 123 (12) 26 (26) 32 (24) 8 (38) 32 (7.5) 25 (7.9) <0.001

*p for comparison between all 5 groups -p

Abbreviations: AKI, acute kidney injury; anti-H2, H2 antagonists; CAD, coronary artery disease; CKD,
re
Chronic Kidney Disease; ICU, intensive care unit; INR, international normalized ratio; IQR, Interquartile
range; NSAIDS, non-steroidal anti-inflammatory drugs; PPI, proton pump inhibitors; PTT, partial
lP

thromboplastin time; SOFA, Sepsis-related Organ Failure Assessment.

na
ur
Jo

24
 Journal Pre-proof

Table 2. Characteristics of patients free of bleeding at intensive care unit admission (n = 513)

All patients With de novo Without de novo p*

bleeding bleeding
n=513 n=91 n=422
Age, years, median (IQR) 66 (55-75) 67 (60-76) 66 (53-75) 0.24
Gender, male (%) 295 (58) 58 (64) 237 (56) 0.19
Race, Caucasian or Asian 430 (83) 75 (82) 355 (84) 0.69
(%)
Comorbidities (%)
Hypertension 303 (59) 59 (65) 244 (58) 0.22
Diabetes 150 (29) 25 (28) 125 (30) 0.68
CAD 176 (34) 49 (54) 127 (30) <0.001
CKD 88 (17) 19 (21) 69 (16) 0.30
Chronic liver disease 9 (1.8) 5 (5.5) 4 (0.9) 0.003
History of bleeding 12 (2.3) 3 (3.3) 9 (2.1) 0.51

of
Previous ulcer 11 (2.1) 3 (3.3) 8 (1.9) 0.40
NSAIDS on admission (%) 39 (7.6) 11 (12) 28 (5.5) 0.08
Steroids on admission (%) 63 (12) 13 (14) 50 (12) 0.52

ro
Anticoagulation on 75 (15) 14 (15) 61 (15) 0.82
admission (%)
Anti-platelets on admission 207 (40) 45 (50) 162 (38) 0.05
(%)
PPI or anti-H2 on admission
(%)
186 (36)
-p
33 (36) 153 (36) 1.00
re
M ain ICU admission
diagnosis (%)
Cardiological 60 (12) 30 (33) 30 (7.1) <0.001
Neurologic 44 (8.6) 6 (6.6) 38 (9) 0.46
lP

Sepsis/severe infection 145 (28) 9 (9.9) 136 (32) <0.001

Surgery 154 (30) 25 (28) 129 (31) 0.56
Trauma 30 (5.8) 10 (11) 20 (4.7) 0.02
Other 80 (16) 11 (12) 69 (16) 0.31
na

Surgery (%) 277 (54) 80 (88) 197 (47) <0.001

M echanical ventilation (%) 234 (46) 67 (74) 167 (40) <0.001
Use of vasopressors (%) 239 (47) 56 (62) 183 (43) 0.002
SOFA score, median (IQR) 4 (2-7) 3 (1-6) 5 (2-7) <0.001
ur

Non-renal SOFA score, 4 (2-6) 2 (1-5) 4 (2-6) <0.001

median (IQR)
PPI or anti-H2 per 430 (86) 79 (90) 351 (85) 0.21
Jo

hospitalization (%)
NSAIDS per hospitalization 110 (22) 27 (30) 83 (20) 0.04
(%)
Antiplatelets drugs per 265 (52) 59 (66) 206 (50) 0.006
hospitalization (%)
Anticoagulation per 176 (35) 54 (59) 122 (29) <0.001
hospitalization (%)
Anticoagulation therapy (%) 119 (23) 41 (45) 78 (19) <0.001
Bleeding source (%)
Gastrointestinal 5 (1.0) 5 (5.5) n/a n/a
Intra-cerebral 8 (1.6) 8 (8.8) n/a n/a
Peri-operative 50 (9.7) 50 (55) n/a n/a
Post-operative 17 (3.3) 17 (19) n/a n/a
Other 11 (2.1) 11 (12) n/a n/a
Acute kidney injury (%) 0.17
No acute kidney injury 352 (69) 55 (60) 297 (70)
Stage 1 99 (19) 24 (26) 75 (18)
Stage 2 23 (4.5) 3 (3.3) 20 (4.7)
Stage 3 including dialysis 39 (7.6) 9 (9.9) 30 (7.1)

25
 Journal Pre-proof

*p for comparison between all groups

Abbreviations: anti-H2, H2 antagonists; CAD, coronary artery disease; CKD, Chronic Kidney Disease;
ICU, intensive care unit; IQR, Interquartile range; NSAIDS, non-steroidal anti-inflammatory drugs; PPI,
proton pump inhibitors; SOFA, Sepsis-related Organ Failure Assessment.

of
ro
-p
re
lP
na
ur
Jo

26
 Journal Pre-proof

Table 3. Predictors of bleeding using a multivariable Cox proportional-hazards model (n = 513)

Crude Hazard Ratio Adjusted Hazard Ratio*

HR 95% CI P HR 95% HR p

Age (years) 1.01 0.99-1.02 0.37 1.00 0.98-1.01 0.41

Gender (male) 1.11 0.72-1.71 0.65 1.29 0.82-2.04 0.28

Coronary artery disease 2.19 1.44-3.33 <0.001 1.84 1.06-3.20 0.03

Chronic liver disease 2.52 1.02-6.23 0.04 3.67 1.33-10.25 0.01

of
Surgical procedure 0.56 0.35-0.90 0.02 0.84 0.50-1.42 0.52

ro
Anticoagulant therapy 2.81 1.84-4.30 <0.001 2.34 1.45-3.80 <0.001

Non-renal SOFA score 0.83 0.77-0.90 <0.001 0.92 0.84-0.99 0.04

NSAIDS per 1.37 0.87-2.13 -p 0.17 1.26 0.76-2.08 0.38

re
hospitalization
lP

Antiplatelets drugs per 1.79 1.15-2.76 0.009 1.18 0.65-2.14 0.58

hospitalization
na

Sepsis 0.47 0.31-0.72 <0.001 0.55 0.34-0.91 0.02

AKI status (time-

ur

dependent)
Jo

No AKI -- -- -- -- -- --

AKI 1.52 0.96-2.40 0.07 1.01 0.60-1.71 0.96

AKI requiring dialysis 3.22 1.53-6.79 0.002 2.82 1.26-6.32 0.009

Abbreviations: AKI, acute kidney injury; SOFA, Sepsis-related Organ Failure Assessment.

27
 Journal Pre-proof

Table 4. Predictors of mortality using a multivariable Cox proportional- hazards model (n = 513)

Crude Hazard Ratio Adjusted Hazard Ratio*

HR 95% CI p HR 95% HR P

Age (years) 1.02 1.01-1.04 <0.001 1.03 1.01-1.04 <0.001

Gender (male) 0.95 0.67-1.34 0.75 1.00 0.68-1.46 0.99

Chronic liver disease 2.31 1.20-4.41 0.01 1.81 0.80-4.08 0.16

Surgical procedure 0.25 0.14-0.42 <0.001 0.17 0.09-0.31 <0.001

of
Anticoagulation therapy 0.83 0.58-1.19 0.30 0.95 0.63-1.44 0.82

Mechanical ventilation 2.45 1.61-3.73 <0.001 1.68 1.01-2.78 0.04

ro
(time-dependent)

Non-renal SOFA score 1.21 1.15-1.27 -p

<0.001 1.12 1.04-1.19 0.001
re
Fluid balance (L/day) 2.02 1.79-2.27 <0.001 1.65 1.44-1.90 <0.001
lP

Incident bleeding (time- 1.65 1.09-2.49 0.02 1.37 0.76-2.47 0.29

dependent)
na

AKI status (time-

dependent)
ur

No AKI -- -- -- -- -- --
Jo

AKI 2.38 1.65-3.45 <0.001 1.34 0.89-2.03 0.17

AKI requiring dialysis 4.39 2.63-7.35 <0.001 1.56 0.82-2.98 0.18

*adjusted for all other included variables

Abbreviations: AKI, acute kidney injury; SOFA, Sepsis-related Organ Failure Assessment.

28
 Journal Pre-proof

Figure legends

Figure 1. Study population.

Figure 2. Kaplan-Meier estimates for the development of bleeding, by severity of acute kidney

injury status in critically ill patients (n = 513, p=0.01 between groups).

of
ro
-p
re
lP
na
ur
Jo

29
 Journal Pre-proof

Author Statement

Farah Zarka: data curation/interpretation, formal analysis, writing original draft

Alexandre Tayler-Gomez: data curation/interpretation, writing original draft

Thierry Ducruet: data curation/interpretation, formal analysis, methodology, writing review and

editing

Anatolie Duca: data curation/interpretation, writing review and editing

of
Martin Albert: data curation/interpretation, writing review and editing

ro
Amélie Bernier-Jean: data curation/interpretation, writing review and editing

-p
Josée Bouchard: conceptualization, data curation/interpretation, formal analysis, funding
re
acquisition, methodology, writing original draft, writing review and editing
lP
na
ur
Jo

30
 Journal Pre-proof

 AKI can affect distant organ function and increase non-renal complications

 AKI requiring dialysis is associated with incident in-hospital bleeding

 Incident bleeding after AKI is not independently associated with an increased risk of

death

of
ro
-p
re
lP
na
ur
Jo

31