34 / 48
76
VOLUME 12 NUMBER 2 • NOVEMBER 2015
RESEARCH ARTICLE
SA JOURNAL OF DIABETES & VASCULAR DISEASE
were identified as independent covariates of having ST-segment
abnormality in a multivariate analysis. When T-wave inversion was
the dependent variable in a similar model, the presence of ECG-
LVH [OR 1.89 (95% CI, 0.9−3.9),
p
< 0.05) and age [OR 0.98 (95%
CI, 0.969−1.000),
p
< 0.05) were associated with the presence of
T-wave inversion in a univariate regression analysis.
Discussion
This cross-sectional study was the first to investigate the prevalence
and covariates of ECG-LVH in a group of diabetic patients in Dar
es Salaam, Tanzania. The main findings from this study were a
prevalence of 15.6% of subjects having ECG-LVH among the
African diabetic patients, with median diabetes duration of three
years; and secondly, identifying systolic BP and albuminuria as the
main covariates associated with the presence of ECG-LVH in this
study.
The prevalence of ECG-LVH among diabetic patients in sub-
Saharan Africa is largely unknown; therefore the present study adds
to previous knowledge. Most previous reports on the prevalence
of ECG-LVH in Africans come from West Africa, where prevalence
ranged from 4.2% using the Cornell voltage criterion among civil
workers in Benin,
11
to 22 and 48% using Cornell voltage and
Sokolow-Lyon voltage criteria, respectively among hypertensive
patients in Nigeria.
12
A study from Kenya in newly diagnosed mild
to moderate hypertensive patients found the prevalence of ECG-
LVH to be 31.7%.
27
Studies among African-American patients
also showed high rates of ECG-LVH in hypertensives, 36.2% by
Sokolow-Lyon and 23.4% by Cornell product criteria.
28
The previous
finding that more hypertensive African patients with ECG-LVH were
picked by Sokolow-Lyon voltage than by Cornell product criteria is
in accordance with findings in the present study.
Compared to results from the Eurodiab IDDM Complications
study, the prevalence of ECG-LVH in type 1 diabetic patients in the
present study was unexpectedly high.
29
It is well known that the
Sokolow-Lyon voltage criterion may overestimate the diagnosis of
LVH in young, tall or thin subjects as included among the type 1
diabetic patients in our study. However, epidemiological studies
in general East-African populations in the Republic of Seychelles
reported ECG-LVH by the Sokolow-Lyon voltage criterion in 9.3%
of patients.
30
This study also found that the Sokolow-Lyon voltage
criterion had low specificity for anatomical LVH in East African
populations, suggesting that some race-specific ECG features may
interfere with components of ECG-LVH diagnoses by the Sokolow-
Lyon voltage criterion.
It is well known that CV risk-factor clustering in diabetic
patients is associated with an increased risk for developing renal
impairment and coronary vascular complications.
31
In particular,
patients with three or more risk factors are more likely to develop
CV complications, such as coronary heart disease and stroke.
31
Our finding that the prevalence of two or more cardiovascular risk
factors was higher in type 2 than in type 1 diabetic patients (68.3
vs 15.7%,
p
< 0.001) is in accordance with previous findings, and
underscores the necessity of broad screening for CV risk factors in
type 2 diabetic patients at the time of diagnosis.
32
In multivariate
logistic regression, systolic BP and albuminuria were identified as
the most important covariates of ECG-LVH.
In the current study, 16 (44.4%) of the patients with ECG-LVH
were also hypertensive. ECG-LVH among type 2 diabetics was
associated with higher systolic and mean BP as well as the presence
of hypertension, and all were significant covariates of the presence
of LVH by the Cornell product criterion. However, with multivariate
analysis, systolic BP [OR 1.015 (95% CI, 1.001−1.028),
p
<
0.05)] was the only independent covariate of ECG-LVH, while no
independent association was found between diastolic BP and ECG-
LVH. Similar findings have previously been reported from the LIFE
study that included hypertensive patients with ECG-LVH by Cornell
product or Sokolow-Lyon criteria.
33
Likewise, in the Framingham
study, patients with systolic BP > 180 mmHg had a 50% chance of
developing ECG-LVH over 12 years, while no risk association was
found with diastolic BP.
7
Our finding, that hypertension and albuminuria were the main
covariates of ECG-LVH is in accordance with previous reports in
hypertensive patients and type 2 diabetics.
34,35
Furthermore, Mbanya
et al
. demonstrated a significant correlation between left ventricular
hypertrophy by echocardiogram and urinary albumin excretion rate
among diabetic patients in Cameroun.
36
Table 4.
Covariates of intraventricular conduction defects (IVC) in diabetic patients indentified by logistic regression analysis in the total study
population
Covariates
IVC yes n (%)
Unadjusted (univariate) OR (95% CI)
Adjusted (multivariate) OR (95% CI)
Gender: male
23 (50)
1.0
female
23 (50)
0.835 (0.438−1.591)
Diabetes type 1
8 (9)
1.00
Diabetes type 2
38 (25.7)
3.498 (1.549−7.899)**
Age (years)
46 (19.6)
1.029 (1.011−1.049)**
Diabetes duration (years)
46 (19.6)
1.070 (1.008−1.136)*
BMI (kg/m
2
)
42 (18.6)
1.067 (1.009−1.129)*
Serum cholesterol (mmol/l)
46 (20.1)
1.026 (0.869−1.216)
Serum creatinine (mmol/l)
46 (20.1)
1.007 (0.995−1.018)
SBP/10 mmHg
44 (20)
1.219 (1.077−1.378)**
1.221 (1.080−1.382)***
DBP/5 mmHg
44 (20)
1.180 (1.046−1.330)**
LVH: no
38 (19)
1.0
LVH: yes
8 (21.6)
1.176 (0.498−2.776)
Normal AER
36 (78.3)
1.0
High AER
10 (21.7)
1.636 (0.727−3.680)
QRS duration/40 msec
46 (19.6)
3.526 (1.157−10.743)*
Independent covariates involved in the multivariate analysis are diabetes type, diabetes duration, age, WC, SBP/10 mmHg and QRS/40 msec.
*
p
< 0.05, **
p
< 0.01, ***
p
< 0.001.