13 / 64
CARDIOVASCULAR JOURNAL OF AFRICA • Volume 30, No 4, July/August 2019
AFRICA
199
evidence of significant hepatic derangement, e.g. bilirubin more
than twice the upper limit of normal, in association with AST/
ALT/ALP more than three times the upper limit of normal),
stroke (previous history of stroke), bleeding (major bleeding
history or predisposition to bleeding), labile INRs (refers to
unstable/high INRs or poor time in therapeutic range
<
60%),
elderly (age
≥
65 years), drug therapy (concomitant therapy such
as antiplatelet agents, NSAIDs) and alcohol intake (consuming
eight or more alcoholic drinks per week).
The CRUSADE score was calculated using baseline
haematocrit, glomerular filtration rate, heart rate on admission,
systolic blood pressure on admission, prior vascular disease,
diabetes mellitus, signs of congestive heart failure on admission
and gender. The Cockcroft–Gault formula was used to calculate
creatinine clearance rate.
In addition to comparing the mean of HAS-BLED and
CRUSADE scores, patients were divided into groups according to
HAS-BLED score
≥
3 (high risk) or 0–2 (low risk), and CRUSADE
score
>
40 (high risk),
>
30 (medium-high risk) and
≤
30 (low risk).
The risk groups were then assessed in terms of the incidence of
bleeding. The Bleeding Academic Research Consortium (BARC)
classification was used to classify bleeding. In addition, 130
patients undergoing percutaneous coronary intervention (PCI)
were compared with the same tests in a subgroup analysis.
PCIs were performed via the femoral and radial route by an
experienced interventional cardiologist (Siemens Axiom Artis
zee Angiography System, Germany). Non-ionic low osmolality
contrast medium (Omnipaque 350 MG/ml; GE Healthcare,
Cork, Ireland) was used for the procedures. All stented patients
were given 300 mg aspirin and a 600-mg clopidogrel loading
dose during the procedure. After having visualised the arterial
anatomy, 100 U/kg heparin was administered. Glycoprotein IIb/
IIIa use was left to the discretion of the physician. No vascular
closure device was used in any patient. After the sheath was
removed, haemostasis was obtained with direct manual pressure
of the fingertips over the pulse. The pressure was held for
approximately 20 minutes (about three minutes for each French
size) until there was no bleeding.
All patients were transferred to the intensive care unit or
cardiology service after the procedure. Bed rest is generally
required for six hours after a sheath is removed. Stented patients
were continued on 100 mg aspirin and 75 mg clopidogrel. In
patients without stenting, treatment was continued on 100
mg aspirin, but in patients with gastrointestinal intolerance
to aspirin, we used 75 mg clopidogrel instead of aspirin. The
decision for concurrent use of statins, angiotensin converting
enzyme inhibitors, calcium channel blocker and beta-blockers
was made according to the recommendations of the American
College of Cardiology/American Heart Association. No patients
used new oral anticoagulants. The use of non-steroidal anti-
inflammatory drugs was avoided. Patients were followed up with
blood samples and the femoral artery area was checked.
Statistical analysis
Statistical analysis was performed using the SPSS 15.0 for
Windows evaluation version statistical package. Continuous
variables are presented as mean
±
standard deviation. Categorical
variables are summarised as frequencies. Differences between the
two groups according to continuous variables were determined
by the independent samples
t
-test. Categorical variables were
compared with the chi-squared or Fisher’s exact test.
C
-statistics
and receiver operating characteristic (ROC) curve analysis
were used to assess the performance of the HAS-BLED and
CRUSADE bleeding scores. Comparison of ROC curves was
done using the de Long test. A
p
-value of
<
0.05 was considered
statistically significant.
Results
The mean age was higher in the group with in-hospital bleeding
than in the group without bleeding (65.32
±
11.40 vs 60.01
±
13.57 years, respectively,
p
=
0.003). Diabetes mellitus was more
frequent among patients in the non-bleeding group compared
to the in-hospital bleeding group (33.8 vs 8%, respectively,
p
=
0.001). Potassium and haematocrit values were statistically
significantly lower in the in-hospital bleeding group.
The results of the groups according to bleeding status are
shown in Table 1. Bleeding was observed in 65 patients. Major
bleeding (BARC type 3) was observed in four patients, in the
form of gastrointestinal bleeding in one patient and from
femoral artery haemorrhage in the others. Minor bleeding was
observed in the remaining 61 patients (femoral artery bleeding
in 57, bleeding from the nose in two, and bleeding in the gingiva
in two). Erythrocyte suspension replacement was needed in only
four patients with major haemorrhage. In one patient with major
haemorrhage, a haematoma in the groin was evacuated and the
femoral artery was sutured.
The mean HAS-BLED score of the patients with in-hospital
bleeding was significantly higher than that of the group without
bleeding (2.21
±
1.15 vs 1.49
±
0.95,
p
<
0.001). There was no
significant difference between the mean CRUSADE scores of
the two groups (23.69
±
11.37 vs 21.28
±
10.82,
p
=
0.105).
The in-hospital bleeding rate in patients with a HAS-BLED
score
≥
3 was significantly higher than in patients with a
HAS-BLED score
<
3 (49.2 vs 14.1%,
p
<
0.001). Similarly,
the rate of in-hospital bleeding in patients with a CRUSADE
score
>
30 was significantly higher than in patients with a
CRUSADE score
≤
30 (36.9 vs 18%,
p
=
0.001). There was no
significant difference in haemorrhage rate between patients with
CRUSADE scores
>
40 and
≤
40.
In the ROC curve analysis, the HAS-BLED score was
found to be superior to the CRUSADE score in predicting
in-hospital bleeding risk among the whole study population who
underwent elective coronary angiography (AUC
=
0.684 vs 0.569,
respectively,
p
=
0.002) (Fig. 1). According to the Youden index
J
-statistics, the HAS-BLED score predicted in-hospital bleeding
in patients undergoing coronary angiography without ACS with
a sensitivity of 59.09% and a specificity of 89.81%. In this patient
group, the sensitivity of the CRUSADE score was 36.36% and
the specificity was 82.69%.
When patients who underwent PCI only were examined,
there was no significant difference between the groups in terms
of mean CRUSADE scores, although there was a significant
difference with regard to the mean HAS-BLED scores of the
groups (Table 2). In the ROC curve analysis of the patient
subgroup that underwent stent implantation, the HAS-BLED
score was superior in predicting in-hospital bleeding events
compared to the CRUSADE score (AUC
=
0.722 vs 0.520,
respectively,
p
=
0.002) (Fig. 2). According to the Youden index,