CLINICAL PRACTICE

Preoperative estimated glomerular ltration rate and the

risk of major adverse cardiovascular and cerebrovascular
events in non-cardiac surgery

A. Mases
1
*
, S. Sabate
2
, N. Guilera
4
, M. Sadurn
1
, R. Arroyo
1
, M. Fau
1,5
, A. Rojo
1,6
, J. Castillo
1
, J. Bover
3
, P. Sierra
2
and J. Canet
7
on behalf of the ANESCARDIOCAT Group

1
Department of Anaesthesiology, Parc de Salut Mar, Pg. Mar tim 25-29, 08003 Barcelona, Spain
2
Department of Anaesthesiology and
3
Department of Nephrology, Fundacio Puigvert (IUNA), Barcelona, Spain
4
Department of Anaesthesiology, Hospital de Sabadell, Sabadell, Spain
5
Department of Anaesthesiology, Vivantes Klinikum Hellersdorf, Berlin, Germany
6
Department of Anaesthesiology, Centre Hospitalier Saint Palais, Sant Palais, France
7
Department of Anaesthesiology, Hospital Germans Trias i Pujol, Badalona, Spain
* Corresponding author. E-mail: amases@parcdesalutmar.cat
Editors key points
Chronic kidney disease is an
important risk factor for
perioperative
complications.
In a post hoc analysis of a
previous outcomes study,
estimated glomerular
ltration rate (eGFR) was
assessed as a predictor of
complications after
non-cardiac surgery.
Major adverse
cardiovascular and
cerebrovascular events
correlated inversely with
preoperative eGFR.
This index of kidney
function is useful for
cardiovascular risk
assessment in non-cardiac
surgery.
Background. Chronic kidney disease is an independent predictor of perioperative
cardiovascular morbidity and mortality. We analysed the preoperative estimated
glomerular ltration rate (eGFR) as a risk factor for perioperative major adverse
cardiovascular and cerebrovascular events (MACCE) in non-cardiac surgery.
Methods. In a post hoc analysis of the ANESCARDIOCATdatabase, patients were classied into
six stages of eGFR calculated with the abbreviated Modication of Diet in Renal Disease Study
and the Chronic Kidney Disease Epidemiology Collaboration equations: .90 (1), 6089.9 (2),
4559.9 (3a), 3044.9 (3b), 1529.9 (4), and ,15 (5) ml min
21
1.73 m
22
. We analysed
differences in MACCE, length of hospital stay, and all-cause mortality between eGFR stages.
Results. The eGFR was available in 2323 patients. Perioperative MACCE occurred in 4.5% of
patients and cardiac-related mortality was 0.5%. Five hundred and forty-three (23.4%)
patients had an eGFR of ,60 ml min
21
1.73 m
22
and 127 (5.4%) had an eGFR below 45 ml
min
21
1.73 m
22
. Logistic regression analysis showed that MACCE increased with eGFR
impairment (P,0.001), with a marked increase from stage 3b onwards (odds ratio 1.8 vs
3.9 in 3a and 3b, respectively, P0.047). All-cause mortality was not related to eGFR
(P0.071), but increased substantially between stages 3b and 4. The length of stay
correlated with eGFR (P,0.001).
Conclusions. Perioperative MACCE increase with declining eGFR, primarily when ,45 ml
min
21
1.73 m
22
. We recommend the use of preoperative eGFR for cardiovascular risk
assessment.
Keywords: cardiovascular system/complications; glomerular ltration rate; perioperative
complications; renal insufciency, chronic; risk assessment
Accepted for publication: 18 February 2014
The prevalence of chronic kidney disease (CKD) in surgical
patients is increasing with ageing of the population and
advances in less invasive surgical techniques.
13
CKD has been
shown to be related to cardiovascular morbidity and mortality
in the general population
4
and in the perioperative setting.
5
Major adverse cardiac and cerebrovascular events (MACCE)
after non-cardiac surgery are also an important cause of
serious perioperative morbidity and mortality.
68
In 1999, Lee
and colleagues
8
identied a preoperative serum creatinine of
177 mmol litre
21
(2 mg dl
21
) or greater as an independent risk

The ANESCARDIOCAT investigators are listed in the Appendix.

Presented in part at the Euroanaesthesia 2010 Congress, Helsinki, Finland, June 1215, 2010.
British Journal of Anaesthesia 113 (4): 64451 (2014)
Advance Access publication 13 June 2014
.
doi:10.1093/bja/aeu134
&The Author 2014. Published by Oxford University Press on behalf of the British Journal of Anaesthesia. All rights reserved.
For Permissions, please email: journals.permissions@oup.com

a
t

8
5
6
8
4
9
7

o
n

S
e
p
t
e
m
b
e
r

2
9
,

2
0
1
4
h
t
t
p
:
/
/
b
j
a
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

factor for perioperative cardiovascular events. In the multi-
centre ANESCARDIOCAT study, we also determined that
a preoperative serum creatinine level of 177 mmol litre
21
or
greater was an independent risk factor for MACCE in a
regression model.
9
SincetheNational KidneyFoundationPracticeGuidelines for
Chronic Kidney Disease and subsequent guides showed that
serum creatinine alone is not an accurate index of glomerular
ltration rate (GFR), estimated GFR (eGFR) has been regarded
as the best index for assessment of kidney function in health
anddisease.
2 4 10 11
Theequationmost widelyusedtoestimate
GFR is the Modication of Diet in Renal Disease (MDRD) study
equation. However, the Chronic Kidney Disease Epidemiology
Collaboration proposed an alternative equation (CKD-EPI) to
calculate eGFR that has been shown to be more accurate and
predictive than the MDRD study equation and has been
gaining acceptance in recent years.
2 12 13
The KDIGO (Kidney
Disease Improving Global Outcomes) 2012 Clinical Practice
Guidelines recommend that clinical laboratories report eGFR
using the 2009 CKD-EPI creatinine equation.
11
To date, most
studies in the surgical population evaluating poor renal func-
tion and perioperative cardiovascular morbidity and mortality
haveusedserumcreatinine alonetodenerenal impairment.
5
It is well known that serum creatinine concentration can be
affected by extrarenal factors and that normal creatinine
levels do not always rule out the presence of CKD.
4
The litera-
ture evaluating the relationship between eGFR and the occur-
rence of MACCE in non-cardiac surgery is limited. Even less is
known about the possibility of a graded relationship between
renal diseasestageandtheriskof perioperativecardiovascular
events. The objective of the current study was to further
analyse the relationship between preoperative eGFR and the
occurrence of perioperative MACCE in non-cardiac surgery
with a goal of dening an eGFR cut-off value for identifying
patients at higher risk for MACCE.
Methods
This study is a secondary data analysis of a large prospective
multicentre cohort study (ANESCARDIOCAT study) that was
carriedout toidentify risk factors for MACCEinnon-cardiacsur-
gical patients. Themethods andresults of theANESCARDIOCAT
study have been published elsewhere.
9
The current study was
not planned when ANESCARDIOCAT was designed, and was
carried out afterwards.
Thestudywas approvedbytheresearchethicscommitteeof
Hospital Germans Trias i Pujol (approval reference number
EO-07-027) on behalf of all centres and was conducted in ac-
cordance with the Declaration of Helsinki. Signed patient
consent was waived because no care interventions were man-
dated and no protected health information was collected.
Interventions other than routine care were not carried out.
The study included patients from23 participating hospitals
in Catalonia recruited between October 2007 and June 2008
during six randomized weeks. Data collection was carried out
in all hospitals simultaneously.
Participants
Werecruitedall middle-agedtoelderlypatients (40yr of age)
undergoing scheduled or emergency non-cardiac operations
of intermediate-to-high surgery-specic risk according to the
guidelines of the American College of Cardiology (ACC) and
the American Heart Association (AHA).
14
All enrolled patients
received general or spinal epidural anaesthesia. Hospital
stay for surgery-related reasons was expected to be longer
than 24 h. Exclusion criteria were age ,40 yr, childbirth or
any obstetrical procedure related to pregnancy, exclusive use
of local or peripheral nerve anaesthesia, procedures outside
theoperatingtheatre, surgical procedures relatedtoaprevious
postoperative complication, or ambulatory surgery.
Variables and data collection
The main outcome was occurrence of a perioperative MACCE,
dened as any of the following: non-fatal cardiac arrest,
acute myocardial infarction, angina, congestive heart failure,
newcardiac arrhythmia, stroke, cardiovascular death, or cere-
brovascular death.
9
We collected patient characteristic data and potential pre-
operative risk factors for MACCE. These included the following
active cardiac conditions and clinical risk factors: unstable cor-
onary syndromes, decompensated heart failure, signicant
arrhythmias, severe valvular disease, history of coronary
artery disease or congestive heart failure, history of cerebro-
vascular disease, history of CKD, diabetes mellitus, abnormal
ECG, cardiac rhythm other than sinus, hypertension, and
other preoperative variables, among them preoperative
serum creatinine concentration.
Trainedmembers of thelocal researchteamineachhospital
were responsible for obtaining preoperative information and
for surveillance of perioperative complications. Management
of complications was left to the judgement of the attending
physicians ineachcentre. Datawere recordedinanonline cen-
tralized database developed in a secure protocol with quality-
control algorithms to validate data entry as described.
9
For the current study, eGFR for each patient was calculated
from routine serum creatinine measurements with both
the abbreviated isotope dilution mass spectrometry (IDMS)-
traceable MDRD
4 15 16
and the CKD-EPI
11 12
equations (Fig. 1).
To evaluate the distribution and risk relationship of eGFR and
MACCE, weclassiedpatientsintosixgroupsfor bothequations
according to eGFR (ml min
21
1.73 m
22
): stage 1, eGFR.90;
stage 2, eGFR6089.9; stage 3a, eGFR4559.9; stage 3b,
eGFR3044.9; stage 4, eGFR1529.9; stage 5, eGFR,15,
in accordance with staging criteria set forth in several guide-
lines, including the recent global KDIGO guidelines.
11 13 17
This current classication further acknowledges the import-
ance of dividing stage 3, based on data supporting different
outcomes and risk proles, into categories 3a and 3b.
11
De-
pendent variables were MACCE occurring either in the operat-
ing theatre or after operation up to hospital discharge. We
rst analysed the distribution of patients and MACCE for both
equations in search of possible differences between them
and sought to determine whether or not any differences
GFR and perioperative cardiovascular events BJA
645

a
t

8
5
6
8
4
9
7

o
n

S
e
p
t
e
m
b
e
r

2
9
,

2
0
1
4
h
t
t
p
:
/
/
b
j
a
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

found were relevant for use in risk prediction. All other results
werebasedonthemost recentlyrecommendedCKD-EPI equa-
tion only.
11
Statistical analyses
Categorical variables were expressed as the number of cases
and percentages, and continuous variables as the median
and 25th75th percentile. We used the Fisher exact test to
assess statistically signicant differences between the MDRD
and CKD-EPI equations in the rate of MACCE in each GFR
stage. The measure of agreement between MDRD and
CKD-EPI classications was performed with Cohens k coef-
cient. The Mantel Haenszel test was used to analyse and
compare trends in MACCE and length of hospital stay
between the eGFR subgroups. The odds ratio (OR) for all-cause
mortalityandMACCE was calculatedusingalogistic regression
analysis where the independent variable was eGFR. Statistical
analyses were performed using the SPSS software package
(version 20.0; IBM Corp., Chicago, IL, USA).
Results
Of 3519 surgical patients recruited, 132 were lost to follow-up
for outcome, and 1064 were excluded from the present study
due tomissingdata neededtocalculate eGFR; eGFRwas calcu-
lated in the remaining 2323 patients (Fig. 2).
Patient and procedure characteristics are given in Table 1.
Only 85 patients (3.7%) had a preoperative serum creatinine
concentration level .177 mmol litre
21
, whereas 543 (23%)
patients had an eGFR ,60 ml min
21
1.73 m
22
and 127
(5.4%) aneGFR ,45ml min
21
1.73m
22
. Theprevalenceof pre-
operative clinical risk factors increased with progressively
lower preoperative eGFR. The prevalence of .2 preoperative
clinical risk factors, excluding renal disease, increased in the
initial stages of renal impairment and remained stable from
stage 3a onwards (Table 2).
Perioperative MACCE occurred in 4.5%(104) of patients and
cardiac-related mortality was 0.5% (11 of 2323 patients).
When we compared the distribution of patients into the six
eGFR groups with both equations, the estimated k coefcient
was 0.78, showing good agreement between the two equa-
tions (Supplementary Table S1). Supplementary Table S2
shows the incidence of perioperative MACCE for each group
according to each equation. The mean eGFR was 74 ml
min
21
1.73 m
22
(6091) [median (25th75th percentile)] for
the MDRD study equation and 78 ml min
21
1.73 m
22
(6193)
for the CKD-EPI equation. Compared with the MDRD equation,
for the higher values of eGFR, the CKD-EPI equation led to
an upward redistribution of patients to higher stages of
eGFR, resulting in a decrease in the number of patients in
stages 2 and 3 and an increase in the number of patients in
stage 1. There were no statistically signicant differences
between the two equations in the incidence of MACCE within
each specic stage of renal impairment. With the exception
of cerebrovascular events, the incidence of all other MACCE
increasedas preoperativeeGFRdecreased(Table3). Logisticre-
gression analysis showed that the occurrence of any MACCE
increased signicantly with impairment of renal function
(P,0.001), with a marked increase from stage 3b (eGFR ,45
ml min
21
1.73 m
22
) downwards (OR 1.8 vs 3.9 in 3a and 3b, re-
spectively; P0.047). While we did not nd a signicant rela-
tionship between perioperative all-cause mortality and eGFR
stage (P0.071), we did observe a considerable increase in
mortality between stages 3b and 4. Length of hospital stay
also correlated with decline in eGFR (Table 4).
Abbreviated MDRD (IDMS) study equation
CKD-EPI equation expressed for specific sex and serum creatinine level
GFR (ml min
1
1.73 m
2
)=175 (standardized SCr)
1.154
(age)
0.203
(0.742 if female)

(1.210 if black)
Gender Serum creatinine
0.7 mg dl
1
(62 mol litre
1
)
0.9 mg dl
1
(80 mol litre
1
)
>0.9 mg dl
1
(>80 mol litre
1
)
Equation for estimating GFR
>0.7 mg dl
1
(>62 mol litre
1
)
144 (SCr/0.7)
0.329
0.993
Age
(1.159 if black)
144 (SCr/0.7)
1.209
0.993
Age
(1.159 if black)
144 (SCr/0.9)
0.411
0.993
Age
(1.159 if black)
144 (SCr/0.9)
1.209
0.993
Age
(1.159 if black)
Female
Female
Male
Male
Fig 1 Equations to estimate GFR.
11 13 27
SCr, serum creatinine concentration in mg dl
21
. Age is expressed in years.
BJA Mases et al.
646

a
t

8
5
6
8
4
9
7

o
n

S
e
p
t
e
m
b
e
r

2
9
,

2
0
1
4
h
t
t
p
:
/
/
b
j
a
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

Discussion
Our study describes new insights in the evaluation of car-
diovascular risk in the perioperative setting for non-cardiac
surgery in patients with CKD. We analysed the effect of pre-
operativeeGFR, calculatedaccordingtothenewCKD-EPI equa-
tion, on MACCE in a broad spectrumof non-cardiac operations.
We determined an eGFR cut-off of 45 ml min
21
1.73 m
22
with
which to identify patients at higher risk for such events. These
data support routine use of preoperative eGFR instead of cre-
atinine for perioperative cardiovascular risk assessment.
In a broad spectrum of non-cardiac procedures of
intermediate-to-high surgery-specic risk, we found that the
incidence of MACCE increased with the degree of impairment
of preoperative eGFR. The increase in complications was par-
ticularlyevident for eGFR ,45ml min
21
1.73m
22
, correspond-
ingtostage3bof thecurrent CKDclassication. Whilewefound
a non-signicant association between decreasing eGFR and
all-cause mortality, we observed a large increase in all-cause
mortality for an eGFR ,30 ml min
21
1.73 m
22
, mainly due to
non-cardiac death.
To our knowledge, ours is the rst cohort study to analyse
the relationship between eGFR and perioperative MACCE in
a large variety of non-cardiac surgical procedures. While
Mooney and colleagues
18
meta-analysis found that eGFR
was strongly associated with short- and long-term prognosis,
this meta-analysis was based mainly on studies carried out in
cardiac and vascular surgery patients. The only two studies
performed in non-cardiac and non-vascular procedures
included inthis report involved only a small number of patients
and used the old CockcroftGault equation to estimate GFR;
one of them evaluated the association with postoperative
renal failure exclusively.
19 20
Although the CockcroftGault
formula is still widely used and is probably better than isolated
creatinine alone, it was developed before standardization of
creatinine assays and thus cannot be reexpressed for use
with current assays.
11
Ackland and colleagues,
21
using the
MDRD study equation, found an increase in perioperative mor-
bidity(all-causecomplications) andanincreaseinthelengthof
hospital stayafter electiveorthopaedic surgeryinpatients with
a preoperative eGFR ,50 ml min
21
1.73 m
22
.
Our results in a non-cardiac surgery-specic risk population
areconsistent withprevious ndings inthegeneral community
population. Goandcolleagues
22
foundanindependent graded
association between eGFR and risk of death, cardiovascular
events, and hospitalization in a large community-based popu-
lation, with a higher increase in event rates for eGFR ,45 ml
Surgical population available for enrolment
(intermediate-to-high surgery-specific risk of non-cardiac interventions)
Excluded for missing data
(lack of intervention date, outcome, demographic
information, clinical history)
Lost to follow-up for outcome
(lack of discharge date, more than 10 missing
data per case, inconsistencies within variables)
Eligible patients
Participants
3387 (96.2%)
Excluded for missing data necessary to
estimate GFR
3519
7 (0.2%)
125 (3.6%)
1064 (30.2%)
2323 (66.01%)
Fig 2 Recruitment owchart.
GFR and perioperative cardiovascular events BJA
647

a
t

8
5
6
8
4
9
7

o
n

S
e
p
t
e
m
b
e
r

2
9
,

2
0
1
4
h
t
t
p
:
/
/
b
j
a
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

min
21
1.73 m
22
. Although the perioperative period is quite dif-
ferent fromthe non-surgical setting, it is plausible that the risk
of major events shows a similar trend. This could be explained
by the fact that the prevalence of cardiovascular disease
increases with severity of renal impairment,
2225
which in
turn is consistent with our ndings of a higher prevalence of
preoperative clinical risk factors as kidney disease progresses.
Furthermore, kidney disease has been described as an inde-
pendent cardiovascular risk factor and even as an ischaemic
heart disease risk equivalent.
26
Of particular interest inour studywas theuseof theCKD-EPI
equation to calculate eGFR. To the best of our knowledge, ours
is the rst study to use this equationto analyse eGFR and its re-
lationship to perioperative cardiovascular disease in the non-
cardiac surgical population. One of the major limitations of
the MDRD study equation is its imprecision and underestima-
tion of measured GFR at higher values (GFR .60 ml min
21
1.73 m
22
).
27
In contrast, the CKD-EPI equation is as accurate
as the MDRD study equation for the subgroup of patients with
eGFR ,60 ml min
21
1.73 m
22
, whereas it is substantially
more accurate for a higher eGFR, thus leading to an upward re-
distribution of patients to higher stages of renal disease and
decreasing the prevalence of stage 3 CKD (eGFR3059 ml
min
21
1.73 m
22
). The net effect is a decline in the prevalence
of CKD. In the general population, this improved calculation
of eGFRusingtheCKD-EPI equationalsoseems toimproveclin-
ical long-termtermrisk prediction for adverse outcomes (such
as end-stage renal disease and cardiovascular and all-cause
mortality) than the MDRD study equation in patients with
CKDand in patients witha higher eGFR.
13 28 29
Thus, it is gener-
allyconsideredthat theadvantages of theCKD-EPI equationat
a higher GFR make it more applicable than the MDRD study
equation for general practice and public health, at least
in North America, Europe, and Australia where this formula
has been validated.
11
The disagreement that we observed
between the two equations in the distribution of patients are
consistent with previous ndings initially obtained in larger
populations. However, wewereunable toshowanydifferences
in their ability to predict outcome, probably due to the limited
number of patients included in our cohort. This limitation is in-
herent to secondary data analysis of a previous large cohort
study where sample size was not calculated specically for
this purpose. Another obvious possible explanation is our
short-term follow-up period, since we analysed complications
only up to hospital discharge.
Current AHA/ACCguidelines onperioperativecardiovascular
evaluation and care for non-cardiac surgery integrate the
clinical risk factors of the Lee Revised Cardiac Risk Index and
dene a serum creatinine of 177 mmol litre
21
(2 mg dl
21
) or
higher as an independent predictor of perioperative cardiovas-
cular events.
14
However, it is well known that GFR must decline
to approximately half normal before serumcreatinine concen-
tration increases above the upper limit of the normal range.
For this reason, serum creatinine is currently dismissed as a
reliable method to detect kidney dysfunction due to several
factors. For instance, age-related decline in muscle mass
reduces creatinine generation; consequently, creatinine
serum concentration does not reect age-related decline in
GFR and normal creatinine values could correspond to low
eGFR.
4
Although the age-related decrease in renal function is
considered part of the normal ageing process, this decline
in the elderly has been shown to be an independent predictor
of complications such as cardiovascular disease and
death.
3032
Using eGFR, there is a 17% prevalence of patients
aged 60 yr or older with eGFR ,60 ml min
21
1.73 m
22
, which
corresponds to stage 3 of the previous classication.
4
It is
thus not surprising that we diagnosed a greater number of
patients with renal disease when using eGFR rather than
serum creatinine alone. This result is consistent with previous
Table 1 Patient and procedure characteristics
Gender, male [n (%)] 1159 (49.9%)
Age [yr, median (25th75th percentile)] 67 (5776)
BMI [kg m
22
(25th75th percentile)] 27.2 (24.430.5)
ASA physical status [n (%)]
I 171 (7.4)
II 1229 (52.9)
III 807 (34.7)
IV 116 (5.0)
Non-urgent surgery [n (%)] 2165 (93.2%)
Surgical risk [n (%)]
Intermediate 2185 (94.1)
High 138 (5.9)
Type of anaesthesia [n (%)]
General 1173 (50.5)
Neuraxial block 830 (35.7)
Combined (general+neuraxial block) 309 (13.3)
Plexus block+general 11 (0.5)
Duration of surgery [min, median (25th75th
percentile)]
120 (75180)
Surgical speciality [n (%)]
Orthopaedics 735 (31.7)
General and digestive 700 (30.1)
Urology 289 (12.4)
Gynaecology 179 (7.7)
Vascular 172 (7.4)
Thoracic 91 (3.9)
Neurosurgery 85 (3.7)
Earnosethroat 48 (2.1)
Maxillofacial 24 (1.0)
Revised cardiac risk index [n (%)]
I 1702 (73.3)
II 450 (19.4)
III 114 (4.9)
IV 57 (2.4)
Serum creatinine .124 mmol litre
21
(1.4 mg
dl
21
) [n (%)]
221 (9.5%)
Serumcreatinine .177mmol litre
21
(2mgdl
21
)
[n (%)]
85 (3.7%)
Postoperative length of stay [days, median
(25th75th percentile)]
6 (39)
In-hospital mortality [n (%)] 48 (2.1)
BJA Mases et al.
648

a
t

8
5
6
8
4
9
7

o
n

S
e
p
t
e
m
b
e
r

2
9
,

2
0
1
4
h
t
t
p
:
/
/
b
j
a
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

studies in which the prevalence of preoperative eGFR ,60 ml
min
21
1.73 m
22
was 2227%.
20 21
In viewof these considera-
tions and based on our own and previous studies,
18 21
we
believe that current guidelines onperioperative cardiovascular
evaluation for non-cardiac surgery
14 33
should switch from
using serum creatinine alone to using eGFR to evaluate the
risk of perioperative cardiovascular events. Complications in-
crease for eGFR ,60 ml min
21
1.73 m
22
,
4 18 21
but we also
suggest that an eGFR ,45 ml min
21
1.73 m
22
might be a
better cut-off value for predicting perioperative MACCE, as we
found that the rate of complications doubled from stage 3a
to 3b.
A limitation of our study is that eGFR was calculated retro-
spectively from a single measurement of serum creatinine
concentration. An isolated measurement of kidney function
is not enough to diagnose CKD,
4
so we acknowledge the pos-
sibility that we might have included patients with some
degree of preoperative acute kidney dysfunction. In this
regard, it would have been interesting to analyse albumin-
uria, as the degree of albuminuria has been shown to be a
signicant risk factor for both CKD progression and cardio-
vascular disease.
2 11 17
Serum creatinine concentrations
were obtained from preoperative laboratory testing and we
do not know whether or not these creatinine measurements
Table 2 Preoperative clinical risk factors according to preoperative eGFR values (CKD-EPI). Absolute number of cases (%)
eGFR 1 2 3a 3b 4 5
n 703 1077 320 127 57 39
History of ischaemic heart disease 39 (5.5) 95 (8.8) 32 (10) 26 (20.5) 11 (19.3) 9 (23.1)
History of compensated or prior heart failure 24 (3.4) 55 (5.1) 32 (10) 20 (15.7) 11 (19.3) 7 (17.9)
History of cerebrovascular disease 26 (3.7) 70 (6.5) 39 (12.2) 20 (15.7) 8 (14) 4 (10.3)
Diabetes mellitus 93 (13.2) 208 (19.3) 67 (20.9) 38 (29.9) 13 (22.8) 11 (28.2)
Greater than one clinical risk factor
(other than renal disease)
22 (3.1) 55 (5.1) 23 (7.2) 9 (7.1) 4 (7.0) 3 (7.7)
Greater than two clinical risk factors
(other than renal disease)
5 (0.7) 8 (0.7) 6 (1.9) 4 (3.1) 0 (0) 2 (5.1)
Age [median (25th75th percentile)] 56 (4864) 70 (6077) 75 (6880) 77 (6981) 73 (67.582.5) 64 (5571)
Table 3 Description of perioperative MACCE. Absolute number of cases (%). *Mantel Haenszel test for trends. AV, atrioventricular
eGFR (CKD-EPI) 1 2 3a 3b 4 5 P-value*
n 703 1077 320 127 57 39
Cardiac death 1 (0.1) 5 (0.5) 1 (0.3) 2 (1.6) 2 (3.5) 0 (0) 0.006
Non-cardiac death 9 (1.4) 7 (0.6) 9 (2.8) 3 (2.4) 6 (10.6) 3 (7.7) ,0.001
Non-fatal cardiac arrest 1 (0.1) 1 (0.1) 1 (0.3) 3 (2.4) 0 (0.0) 0 (0.0) ,0.001
Angina 6 (0.9) 7 (0.6) 2 (0.6) 4 (3.1) 2 (3.5) 1 (2.6) 0.023
Acute myocardial infarction 0 (0.0) 2 (0.2) 0 (0.0) 2 (1.6) 1 (1.8) 1 (2.6) ,0.001
Congestive heart failure 6 (0.9) 12 (1.1) 5 (1.6) 3 (2.4) 4 (7.0) 0 (0.0) 0.003
Arrhythmia or AV block 6 (0.9) 35 (2.3) 8 (2.5) 6 (4.7) 2 (3.5) 2 (5.1) 0.031
Acute cerebrovascular event 3 (0.4) 3 (0.3) 1 (0.3) 1 (0.8) 0 (0.0) 0 (0.0) 0.931
Table 4 OR for all-cause mortality, any cardiovascular or cerebrovascular event, and length of hospital stay according to GFR. *Logistic regression
for all-cause mortality, P0.071;

Logistic regression for MACCE, P,0.001;

Kruskal Wallis test for comparing means, P,0.001
GFR (ml min
21
1.73 m
22
) All-cause mortality*, OR (95% CI) Any MACCE

, OR (95% CI) Length of hospital stay

[days, median
(10th90th percentile)]
Stage 1 (.90) 1 (ref) 1 (ref) 5 (215)
Stage 2 (6089.99) 0.8 (0.31.8) 1.5 (0.92.5) 6 (215)
Stage 3a (4559.99) 2.2 (0.95.4) 1.8 (0.93.5) 7 (218)
Stage 3b (3044.99) 2.8 (0.98.5) 3.9 (1.98.0) 8 (221.2)
Stage 4 (1529.99) 11.3 (4.329.9) 4.8 (1.911.8) 8 (232.8)
Stage 5 (,15) 5.8 (1.521.9) 3.9 (1.312.0) 8 (336)
GFR and perioperative cardiovascular events BJA
649

a
t

8
5
6
8
4
9
7

o
n

S
e
p
t
e
m
b
e
r

2
9
,

2
0
1
4
h
t
t
p
:
/
/
b
j
a
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

were calibrated to an assay traceable to IDMS, which is
considered the standard method.
13 17
Non-standard
methods can slightly overestimate creatinine concentration
and hence induce errors in GFR estimates.
4
However, the
current Spanish guidelines, published in 2006, strongly rec-
ommend that laboratories standardize creatinine calibration
to be traceable to IDMS.
34
Another potential limitation of our study is the lack of mea-
surement of prospective cardiac biomarkers for all patients as
recommended by other authors.
35
As a result, the rate of
MACCE might have been underestimated, as silent cardiac
events would not have been detected. It is worth mentioning
that TnT measurements (baseline and prospective) have a dif-
ferent signicance inCKDthan inthe general population, since
TnT levels are inuenced by decreased renal clearance.
36 37
In
any case, the denition of MACCE used for the study was broad
enoughtoincludemost clinically relevant cardiac andcerebro-
vascular complications.
9
In conclusion, we found that preoperative eGFR predicts
perioperative MACCE ina broadspectrumof non-cardiac surgi-
cal procedures. The occurrence of MACCE increases with a de-
clining eGFR which is particularly evident for eGFR ,45 ml
min
21
1.73 m
22
. We recommend that anaesthesiologists use
eGFR in place of creatinine in preoperative cardiovascular
evaluation for non-cardiac surgery.
Supplementary material
Supplementary material is available at British Journal of
Anaesthesia online.
Authors contributions
A.M.: conceptionanddesignof thestudy, dataacquisition, ana-
lysis and interpretation of data, and writing and nal approval
of manuscript; S.S.: conception and design of the study, data
acquisition, analysis andinterpretationof data, statistical ana-
lysis, and writing and nal approval of manuscript; N.G., J.C.:
conception and design of the study and data acquisition;
M.S., R.A., M.F., A.R.: patient recruitment and data acquisition;
J.B.: critical review of manuscript; P.S.: patient recruitment
anddataacquisitionandreviewof manuscript; J.C.: conception
of the study and review of manuscript for nal approval.
Acknowledgements
The authors thank all participating anaesthetists for their col-
laborationandIrwinTemkinfor his revisionof Englishlanguage
usage in some versions of the manuscript.
Declaration of interest
None declared.
Funding
This study was supported by the Catalan Public Health Service
(Departament de Salut de la Generalitat de Catalunya) within
the framework of World Alliance of Patient Safety in Catalonia,
Spain. Thefundingsources hadnoroleinthe designor conduct
of the study; the collection, management, analysis, or inter-
pretation of the data; or the preparation, review, or approval
of the manuscript. Presented in part at the Euroanaesthesia
2010 Congress, Helsinki, Finland, June 1215, 2010.
References
1 Coresh J, Selvin E, Stevens LA, et al. Prevalence of chronic
kidney disease in the United States. J Am Med Assoc 2007; 298:
203847
2 Levey AS, Coresh J. Chronic kidney disease. Lancet 2012; 379:
16580
3 Arulkumaran N, Annear NMP, Singer M. Patients with end-stage
renal disease admitted to the intensive care unit: systematic
review. Br J Anaesth 2013; 110: 1320
4 LeveyAS, CoreshJ, Balk E, et al. National KidneyFoundationpractice
guidelines for chronickidneydisease: evaluation, classication, and
stratication. Ann Intern Med 2003; 139: 13747
5 MathewA, Devereaux PJ, OHareA, et al. Chronic kidneydiseaseand
postoperative mortality: a systematic review and meta-analysis.
Kidney Int 2008; 73: 106981
6 Devereaux PJ, Yang H, Yusuf S, et al. Effects of extended-release
metoprolol succinate in patients undergoing non-cardiac surgery
(POISE trial): a randomised controlled trial. Lancet 2008; 371:
183947
7 Kheterpal S, OReilly M, Englesbe MJ, et al. Preoperative and intra-
operativepredictors of cardiac adverse events after general, vascu-
lar, and urological surgery. Anesthesiology 2009; 110: 5866
8 Lee TH, Marcantonio ER, Mangione CM, et al. Derivation and pro-
spective validation of a simple index for prediction of cardiac risk
of major noncardiac surgery. Circulation 1999; 100: 10439
9 SabateS, MasesA, GuileraN, et al. Incidenceandpredictors of major
perioperative adverse cardiac and cerebrovascular events in non-
cardiac surgery. Br J Anaesth 2011; 107: 87990
10 National Kidney Foundation. K/DOQI clinical practice guidelines for
chronic kidney disease: evaluation, classicationand stratication.
Am J Kidney Dis 2002; 39: S1266
11 Kidney Disease: Improving Global Outcomes (KDIGO) CKD Work
Group. KDIGO 2012 Clinical Practice Guideline for the evaluation
and management of chronic kidney disease. Kidney Int Suppl
2013; 3: 1150
12 Levey AS, Stevens LA, Schmid CH, et al. Anewequation to estimate
glomerular ltration rate. Ann Intern Med 2009; 150: 60412
13 MatsushitaK, Mahmoodi BK, Woodward M, et al. Comparison of risk
prediction using the CKD-EPI equation and the MDRD study equa-
tion for estimated glomerular ltration rate. J Am Med Assoc
2012; 307: 194151
14 Fleisher LA, Beckman JA, Brown KA, et al. ACC/AHA2007 guidelines
onperioperativecardiovascular evaluationandcarefor noncardiac
surgery: a report of the American College of Cardiology/American
Heart AssociationTaskForceonPracticeGuidelines(WritingCommit-
tee to Revise the 2002 Guidelines on Perioperative Cardiovascular
Evaluation for Noncardiac Surgery) developed in collaboration with
the American Society of Echocardiography, American Society of
Nuclear Cardiology, Heart RhythmSociety, Society of Cardiovascular
Anesthesiologists, Societyfor Cardiovascular AngiographyandInter-
ventions, Society for Vascular Medicine and Biology, and Society for
Vascular Surgery. J Am Coll Cardiol 2007; 50: e159241
15 Levey AS, Coresh J, Greene T, et al. Using standardized serum cre-
atinine values in the modication of diet in renal disease study
equation for estimating glomerular ltration rate. Ann Intern Med
2006; 145: 24754
BJA Mases et al.
650

a
t

8
5
6
8
4
9
7

o
n

S
e
p
t
e
m
b
e
r

2
9
,

2
0
1
4
h
t
t
p
:
/
/
b
j
a
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

16 Stevens LA, Coresh J, Greene T, Levey AS. Assessing kidney
functionmeasured and estimated glomerular ltration rate.
N Engl J Med 2006; 354: 247383
17 Crowe E, Halpin D, Stevens P. Early identication and management
of chronic kidney disease: summary of NICE guidance. Br Med J
2008; 337: a1530
18 Mooney JF, Ranasinghe I, Chow CK, et al. Preoperative estimates of
glomerular ltrationrateas predictors of outcomeafter surgery: asys-
tematic reviewandmeta-analysis. Anesthesiology 2013; 118: 80924
19 Browner WS, Li J, ManganoDT. In-hospital andlong-termmortalityin
male veterans following noncardiac surgery. The Studyof Periopera-
tive Ischemia Research Group. J Am Med Assoc 1992; 268: 22832
20 Aveline C, Leroux A, Vautier P, Cognet F, Le Hetet H, Bonnet F. Risk
factors for renal dysfunction after total hip arthroplasty. Ann Fr
Anesth Reanim 2009; 28: 72834
21 Ackland GL, Moran N, Cone S, Grocott MP, Mythen MG. Chronic
kidney disease and postoperative morbidity after elective ortho-
pedic surgery. Anesth Analg 2011; 112: 137581
22 Go AS, Chertow GM, Fan D, McCulloch CE, Hsu CY. Chronic kidney
disease andthe risks of death, cardiovascular events, andhospital-
ization. N Engl J Med 2004; 351: 1296305
23 Weiner DE, Krassilnikova M, Tighiouart H, Salem DN, Levey AS,
Sarnak MJ. CKD classication based on estimated GFR over three
years and subsequent cardiac and mortality outcomes: a cohort
study. BMC Nephrol 2009; 10: 26
24 Astor BC, Hallan SI, Miller ER III, Yeung E, Coresh J. Glomerular ltra-
tionrate, albuminuria, andriskof cardiovascular andall-causemor-
tality in the US population. Am J Epidemiol 2008; 167: 122634
25 SharaNM, Resnick HE, LuL, et al. DecreasedGFRestimatedbyMDRD
or CockcroftGault equationpredictsincident CVD: thestrongheart
study. J Nephrol 2009; 22: 37380
26 Sarnak MJ, Levey AS, Schoolwerth AC, et al. Kidney disease as a
risk factor for development of cardiovascular disease: a statement
fromthe American Heart Association Councils on Kidney in Cardio-
vascular Disease, HighBloodPressureResearch, Clinical Cardiology,
and Epidemiology and Prevention. Circulation 2003; 108: 215469
27 Stevens LA, Coresh J, Feldman HI, et al. Evaluation of the modica-
tion of diet in renal disease study equation in a large diverse popu-
lation. J Am Soc Nephrol 2007; 18: 274957
28 Stevens LA, Li S, KurellaTamuraM, et al. Comparisonof the CKDEpi-
demiology Collaboration (CKD-EPI) and Modication of Diet in
Renal Disease (MDRD) study equations: risk factors for and compli-
cations of CKDandmortalityintheKidneyEarlyEvaluationProgram
(KEEP). Am J Kidney Dis 2011; 57: S916
29 MatsushitaK, SelvinE, BashLD, Astor BC, CoreshJ. Risk implications
of the new CKD Epidemiology Collaboration (CKD-EPI) equation
compared with the MDRD Study equation for estimated GFR: the
Atherosclerosis Risk in Communities (ARIC) Study. Am J Kidney Dis
2010; 55: 64859
30 Hallan SI, Matsushita K, Sang Y, et al. Age and association of kidney
measures with mortality and end-stage renal disease. J Am Med
Assoc 2012; 308: 234960
31 Shlipak MG, Fried LF, Crump C, et al. Cardiovascular disease risk
status in elderly persons with renal insufciency. Kidney Int 2002;
62: 9971004
32 ManjunathG, Tighiouart H, CoreshJ, et al. Level of kidneyfunctionas
a risk factor for cardiovascular outcomes in the elderly. Kidney Int
2003; 63: 11219
33 Poldermans D, Bax JJ, BoersmaE, et al. Guidelines for pre-operative
cardiac risk assessment and perioperative cardiac management in
non-cardiac surgery. Eur Heart J 2009; 30: 2769812
34 Gracia S, Montanes R, Bover J, et al. Recommendations for the
use of equations to estimate glomerular ltration rate in
adults. Spanish Society of Nephrology. Nefrologia 2006; 26:
65865
35 Devereaux PJ, ChanMT, Alonso-Coello P, et al. Association between
postoperativetroponinlevels and30-daymortalityamongpatients
undergoing noncardiac surgery. J Am Med Assoc 2012; 307:
2295304
36 Tsutamoto T, Kawahara C, Yamaji M, et al. Relationship between
renal function and serum cardiac troponin T in patients with
chronic heart failure. Eur J Heart Fail 2009; 11: 6538
37 DeFilippi C, Seliger SL, Kelley W, et al. Interpreting cardiac troponin
results from high-sensitivity assays in chronic kidney disease
without acute coronary syndrome. Clin Chem 2012; 58: 134251
Appendix
ANESCARDIOCAT study group
Pilar Paniagua, MD, Beatriz Martin, MD, Maite Rivilla, MD, Marta
Gine, MD (Fundacio de Gestio Sanita`ria de lHospital de la
Santa Creu i Sant Pau); Guillermina Fita, MD, Elisenda Pujol
Rosa, MD, Irene Rovira, MD, Amalia Alcon, MD (Hospital Clinic,
Barcelona); Antoni Sabate, MD, Marta Lacambra, MD, Albert Pi,
MD, Diana Campello, MD (Hospital Universitari de Bellvitge,
LHospitalet de Llobregat); Ana Arnal, MD [Fundacio Puigvert
(IUNA), Barcelona]; Covadonga Llorente, MD, Valentin Mazo,
MD (Hospital Universitari Germans Trias i Pujol, Badalona);
Anna Rodr guez, MD, Silvia Lopez, MD, Novella Calo, MD, M
Jesu s Laso, MD (Hospital de Sabadell); M. Angels Subirana, MD,
Janete Andrade, MD (Hospital Comarcal de lAlt Penedes, Vila-
francadel Penedes); GuillemBrugal, MD(Hospital ArnaudeVila-
nova, Lleida); Olga Ramiro, MD (Hospital Universitari de
Tarragona Joan XXIII); Teresa Vilalta, MD, Gra`cia Cardenas, MD
(Fundacio Hospital-Asil de Granollers); Fina Parramon, MD,
Carmen Hernandez, MD, Xavier March, MD, Alfred Mun oz, MD
(Hospital Universitari Josep Trueta, Girona); Patricia Ciurana,
MD (Hospital Universitari de la Vall dHebron, Barcelona); Albert
Canadell, MD, Lisette Jimenez, MD, Gentxo Balev, MD (Althaia
Xarxa Assistencial, Manresa); Ester Lomban, MD, Carmen
Mart n, MD (Hospital de Terrassa); Teresa Planella, MD, Jordi
Serrat, MD (Hospital General de Vic); Josep Lluis Casbas, MD,
LauraMahillo, MD(Hospital Sant Rafael, Barcelona); JoanForna-
guera, MD, Lluis Martinez, MD (Hospital Municipal de Badalona);
M Paz Villalba, MD, Dolors del Pozo, MD (Hospital de la Santa
Maria, Lleida); Fabian Iban ez, MD (Hospital de Sant Jaume,
Olot); Antonio Garces, MD (Hospital Sant Joan de Deu, Barce-
lona); Alfonso Alonso, MD (Fundacio Sanita`ria dIgualada);
Carola Orrego [Avedis Donabedian Institute, Autonomous Uni-
versity of Barcelona, and CIBER Epidemiology and Public
Health (CIBERESP), Barcelona].
Handling editor: H. C. Hemmings
GFR and perioperative cardiovascular events BJA
651

a
t

8
5
6
8
4
9
7

o
n

S
e
p
t
e
m
b
e
r

2
9
,

2
0
1
4
h
t
t
p
:
/
/
b
j
a
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m