CARDIOVASCULAR JOURNAL OF AFRICA • Vol 22, No 5, September/October 2011
236
AFRICA
was considered sufficient to meet the primary endpoint for the
smallest sub-population, patients with the metabolic syndrome.
Approximately 1 060 000 patients in South Africa are treated
with statins, of which 850 000 (80%) are treated as primary
prevention and 210 000 (20%) as secondary prevention. It was
assumed that in South Africa the proportion of patients with the
metabolic syndrome was similar to the proportion of secondary
prevention patients at around 35%. Based on these assumptions a
total sample size of about 3 000 patients was considered necessary.
Analyses of primary and secondary endpoints were performed
using the same models.
For each patient, the risk category was determined and a
dichotomous variable was calculated indicating whether the
patient had achieved the target LDL-C goal corresponding to the
relevant risk category. The number and percentage of patients
achieving the LDL-C goals according to the NCEP ATP III/2004
updated NCEP ATP III/Fourth JETF/South African guidelines
are presented.
Furthermore, a two-level logistic regression analysis was
performed to determine the prognostic factors of achieving the
LDL-C goals, according to each of the guidelines, with patients
at the first level and physicians at the second level. Prognostic
factors were identified among several patient and physician
independent variables.
First the crude (univariate) association of each of the potential
predictors with the outcome (i.e. achievement of LDL-C goals
according to the corresponding guideline) was investigated as
a univariate analysis using logistic regression. The dependent
variable was achievement of the LDL-C goals according to the
corresponding guidelines (yes or no), the potential predictor was
fixed effects and the physician was random effect.
The association of fixed effects with the dependent variable
was appraised by estimated odds ratio with associated 95% CIs
and
p
-values. All predictors with a
p
-value
<
0.10 (using the
Wald-type test) in this crude (univariate) association analysis
were further included in an adjusted multiple logistic regression
model.
Discrete variables with more than two categories were fitted
in the model using SAS PROC GENMOD through a series of
binary variables by means of the CLASS statement. In cases
where too small a number of answers per category was observed,
categories were pooled.
First, a full model of random intercept variables selected
based on the univariate analysis as fixed effects was evaluated.
At each subsequent step, the least significant independent vari-
able was removed until all variables reached a level of signifi-
cance of at least 0.05.
For the final model, the following results were provided:
parameter estimates and 95% CIs anti-logged to obtain the odds
ratio and 95% CI of the odds ratio, standard error and
p
-value for
each effect. The type III effects Wald-type test was used to assess
the significance of a variable.
Results
Baseline characteristics
In total, 3 001 patients consented to participate in the survey.
Laboratory data were missing for three patients and two
patients withdrew their consent. Therefore the full analysis set
(FAS) comprised 2 996 patients, recruited at 69 study centres.
Demographic characteristics of this survey cohort are summa-
rised in Table 2. The mean age was 59.4 years and 47.5% of the
patients were female. A medical history of arterial hypertension,
diabetes and CHD was reported by 71.6, 47.1 and 35.4% of the
patients, respectively.
When patients were asked about the measures taken by their
physician when they were first diagnosed with high cholesterol
levels, 64.2% had been prescribed lifestyle changes and phar-
macotherapy, 20.2% had been prescribed pharmacotherapy only,
14.4% reported having been advised lifestyle modifications
alone, and 1.2% had received neither pharmacotherapy nor life-
style modification advice.
At the time of their assessment, most patients (95.9%) were
receiving LLD monotherapy, with the majority (98.9%) receiv-
ing statins. Simvastatin, atorvastatin and rosuvastatin were the
statins most frequently used in monotherapy (53.5, 30.3 and
12.9%, respectively). All of the patients taking multiple LLDs
received statins in combination with other drugs, mainly fibrates
(64.2%). Primary CVD prevention was the main reason for LLD
therapy (44.3% of patients).
In total, 149 investigators were eligible to participate in the
physician survey, 120 of them completed and returned the inves-
tigator’s questionnaire, while 101 investigators both returned the
completed investigator’s questionnaire and examined patients.
TABLE 2 SUMMARY OF DEMOGRAPHICS
AND PATIENT CHARACTERISTICS
Patient characteristics
Study cohort
(
n
=
2 996)
Age (years)*
59.4 (11.4)
Gender
Male
1572 (52.5)
Female
1424 (47.5)
Ethnic group
Caucasian
1385 (46.2)
Non-Caucasian
1611 (53.8)
Black
510 (17.0)
Mixed ancestry
481 (16.1)
Indian
576 (19.2)
Asian
44 ( 1.5)
BMI (kg/m
2
)*
30.0 ( 6.0)
Waist circumference (cm)*
101.0 (14.1)
SBP (mmHg)*
133.2 (17.7)
DBP (mmHg)*
80.2 ( 9.9)
Current smoker
445 (14.9)
Diagnosed diabetes
1411 (47.1)
Undiagnosed diabetes
71 ( 2.4)
Diabetes and history of coronary heart disease
494 (16.5)
Arterial hypertension
2144 (71.6)
Family history of premature CVD
863 (28.8)
History of coronary heart disease
1060 (35.4)
History of peripheral artery disease
146 ( 4.9)
History of cerebrovascular atherosclerotic disease
158 ( 5.3)
Results expressed as
n
(%) except where indicated by an asterisk (*)
where reported as mean (standard deviation)
CVD: cardiovascular disease, BMI: body mass index, SBP: systolic
blood pressure, DBP: diastolic blood pressure, age (years): calculated
relative to the subject’s date of visit, percentages are calculated rela-
tive to the total number of subjects with data.
