CARDIOVASCULAR JOURNAL OF AFRICA • Vol 21, No 5, September/October 2010
264
AFRICA
epidemiological transition of cardiovascular disease,
8
includ-
ing socio-economic factors and exposure to risk factors, health
surveillance and access to healthcare between a developed and a
developing world population. Secondly, although beta-blockers,
statins and aspirin were shown to improve long-term survival in
the Rotterdam study,
6
we know that the medical therapy of South
African vascular patients is wholly inadequate.
2,3
Determination of predictors of intermediate and long-term
survival is important. These predictors may identify patients who
require further aggressive risk-factor modification, therapy and
increased surveillance postoperatively. In some cases, these risk
predictors may even identify patients in whom conservative non-
surgical management is preferable.
9
The aim of this study was therefore to evaluate whether the
clinical risk predictors identified in the Revised Cardiac Risk
Index
7
and our own studies of in-hospital cardiac
2
and all-cause
mortality
3
were associated with intermediate and long-term
mortality in South African patients who underwent elective or
urgent vascular surgery. We also examined hypertension as a
predictor of mortality as it is the second highest ranked risk
factor associated with all-cause mortality in South Africans
(following unsafe sex/sexually transmitted infections).
10
Methods
The Ethics Committee of the Nelson R Mandela School of
Medicine for this study granted ethical approval. The patient
cohort included all vascular surgical patients over 39 years of
age admitted for both elective and emergency vascular surgery
at Inkosi Albert Luthuli Central Hospital (IALCH) between June
2003 and June 2007.
From the hospital’s computerised database, we identified
all patients who survived the surgical procedure and were
discharged from hospital. To determine intermediate and long-
term survival, all subsequent hospital clinic visits or hospital
admissions were identified. For patients who did not return to
the hospital, one of the authors (SN) used the registered contact
details on the hospital database to contact the patient and/or the
next of kin. The outcome of the patient (death or survival) and
the time to the outcome following the surgical procedure were
recorded.
The time to the outcome was grouped in six-month blocks.
In presenting survival time, the six-month blocks were treated as
a continuous variable, after confirmation that the distribution of
the six-month blocks was of a normal distribution. Therefore, for
example, a survival time of eight six-month blocks represents 48
months. ‘Lost to follow up’ was defined as a patient who, follow-
ing discharge, had no further visit or admission to IALCH and
neither the patient nor the next of kin were contactable using the
contact details registered on the hospital database.
For all patients in whom we had intermediate and long-term
outcome data, we extracted demographic data associated with
peri-operative cardiac risk,
7
and intra-hospital cardiac
2
and all-
cause mortality.
3
Data on the following clinical risk factors were
collected: history of ischaemic heart disease (or pathological Q
waves on ECG), history of congestive heart failure, diabetes,
serum creatinine
>
180
µ
mol.l
–1
, history of cerebrovascular
accident, age, gender, history of smoking, and history of hyper-
tension.
2,3,7,11
Data on medical therapy collected included chronic pre-oper-
ative statin therapy, beta-blocker therapy and postoperative beta-
blocker withdrawal.
3
Data on the surgical procedure included
major vascular surgery, and out-of-hours surgery.
3
Physiological
data collected included the mean daily heart rate on the day
before surgery and the third postoperative day, and whether the
mean systolic blood pressure (SBP) was
<
100 or
>
179 mmHg
on the third postoperative day.
3
Statistical analyses
To compare survivors and non-survivors, all categorical data
were analysed using descriptive statistics and either the Fisher’s
exact test or Pearson’s Chi-square test, where appropriate. All
continuous data were analysed using descriptive statistics and
compared using independent samples
t
-test, as all continuous
data were normally distributed.
Bivariate and multivariate analysis was conducted using Cox
regression analysis to determine predictors of intermediate and
long-term mortality. Cases with missing data were excluded
from the analysis. Risk factors with
p
<
0.10 on bivariate
analysis were entered into the multivariate regression analysis.
A backward stepwise modelling technique was used, based on
likelihood ratios with entry and removal probabilities set at 0.05
and 0.1, respectively.
Co-linearity was also investigated. Co-linearity was consid-
ered if Pearson’s correlation coefficient was
>
0.6 or the standard
error of a covariate was
>
5.0.
12
If co-linearity was identified, the
multivariate analysis was repeated after removal of the respon-
sible covariate.
13
Kaplan-Meier survival plots analyses were conducted for risk
factors associated with
p
<
0.10 for bivariate Cox regression
analysis. Both the log-rank and Breslow tests are reported.
The hazard ratio (HR) for intermediate and long-term death
rates and 95% confidence intervals (CI) are reported. SPSS 15.0
for Windows (6 Sept 2006) was used for data analysis.
Results
Over the four-year period, a total of 747 patients over the age
of 39 years were discharged from hospital following success-
ful vascular surgery. Four hundred and sixty-four patients were
lost to follow up; therefore 283 patients were included in this
study. There were 21 intermediate and long-term non-survivors
and 262 survivors. The demographics and clinical, surgical and
physiological risk factors are presented in Table 1. The data set
was complete for all the risk factors listed in Table 1 with the
exception of the serum creatinine in 20 patients (7.1%) and
the mean daily heart rate on the third postoperative day in six
patients (2.1%).
The bivariate Cox regression analysis of survival is presented
in Table 2. Only hypertension and diabetes were associated with
intermediate and long-term mortality at the bivariate level of
analysis with
p
<
0.10. There was no co-linearity between hyper-
tension and diabetes. Entering hypertension and diabetes into
a multivariate Cox regression analysis resulted in hypertension
being the only predictor of intermediate and long-term survival
(HR 3.86, 95% CI: 0.83–15.4,
p
=
0.086) retained in the model.
The Kaplan-Meier survival curves for hypertension and
diabetes are shown in Figs 1 and 2. The survival characteristics
of patients with hypertension and diabetes are shown in Table 3.
