22 / 40
RESEARCH ARTICLE
SA JOURNAL OF DIABETES & VASCULAR DISEASE
64
VOLUME 16 NUMBER 2 • NOVEMBER 2019
Thirty-two FAs were measured in fasted plasma samples from
711 participants. Six FAs, i.e. pentadecanoic acid (C15:0), margaric
acid (C17:0), trans vaccenic acid (C18:1
n-
7t), rumenic acid (C18:2
n-
7tt), stearidonic acid (C18:4
n-
3) and eicosatrienoic (C20:3
n-
3)
were below the limit of quantification and therefore not included.
The remaining 26 plasma phospholipid FAs were quantified and
expressed as a percentage of total FAs. Quality of data was assured
with a separate calibration for each FA, monitoring of internal
standard (1,2-diheptadecanoyl-snglycerol-3 phosphorylchloline,
Matreya, Pennsylvania, USA) and Levey Jennings graphs for a
pooled plasma control analysed with each batch.
The MetS was defined according to recommendations by the
Joint Interim Statement of six international associations as the
presence of three or more of the following: (1) fasting plasma
glucose levels ≥ 5.6 mmol/l or the use of oral hypoglycaemic
medication; (2) serum triglycerides ≥ 1.7 mmol/l; (3) serum HDL ≤
1.0 mmol/l for men and ≤ 1.3 mmol/l for women; (4) BP ≥ 130/85
mmHg or the use of BP medication; and (5) WC of ≥ 94 cm for men
and ≥ 80 cm for women.
41
Statistical analysis
Continuous variables were described as medians and interquartile
ranges if data deviated from the normal distribution according to
the Kolmogorov–Smirnov test, whereas categorical variables were
presented as percentages. No
n-
normally distributed data were log
transformed before inclusion in regression models. Participants’
characteristics were compared by gender and BMI categories
using the Mann–Whitney
U
-test or chi-squared test for continuous
and categorical variables, respectively. Differences between
individual FAs and ratios by BMI and gender groups were tested
with the Mann–Whitney test. A BMI < 25 kg/m
2
was considered
as underweight and/or normal-weight or lean, whereas BMI ≥ 25
kg/m
2
was considered overweight and/or obese. The effect size of
the differences between groups was calculated using the Man–
Whitney
U
-value and sample size of the groups.
42
Principal-component-based varimax factor analysis of the
correlation matrix was used to define dietary FA (based on the
QFFQ) and plasma phospholipid FA patterns. The identification
and naming of 11 dietary FAs and 26 plasma phospholipid FAs
used in this study are based on relevant literature and the levels of
specific FAs observed in our population.
43
The number of factors
to retain was established by the Kaiser criterion (eigenvalues > 1)
and scree-plot visual inspection. Loadings with absolute values >
0.5 were considered as relevant for the contribution to each FA
pattern. The associations between FA patterns and outcomes were
evaluated by sequential regression models, logistic regression for
the dichotomous outcome (MetS), and generalised linear models
for continuous outcomes (WC, BMI and WHtR).
The first step of the sequential modelling analyses was based on
models that contained only dietary FAs or plasma phospholipid FA
patterns and was referred to as a crude model. The crude model
was then adjusted for gender and age (adjusted model1). This
model was further adjusted for lifestyle confounders, including the
level of education, physical activity, alcohol and total energy intake,
and self-reported smoking status, creating a fully adjusted model.
We further adjusted this model for contraceptives (adjusted for in
plasma phospholipid FA pattern models only) and dietary factors,
including total fats, carbohydrates, dietary fibre and energy from
added sugar as individual confounders and as combined covariates.
Model fittingwas evaluated using the adjusted
R
-square for linear
regression and maximum re-scaled R-square statistic for logistic
regression. Linear regression results are presented as standardised
β
and 95% confidence intervals (CI) with their significance levels,
and odds ratio and 95% CI with significance levels for logistic
regression. All analyses were performed using SAS Version 9.4 (SAS
Institute, Cary, NC, USA)
44
and
p
< 0.05 was considered significant.
Results
The baseline characteristics of the 711 participants are shown in
Table 1. The majority were women (61.6%) and the median age
was comparable between men and women. Men had higher HDL
levels and were more likely to smoke. By contrast, the women had
higher serum triglycerides, as well as higher levels of measures
Table 1.
Demographics, health and dietary intake data of an apparently
healthy cohort of 711 black South African adults participating in the
PURE study
Men (
n
= 273) Women (
n
= 438)
Variables
Median (Q
1
, Q
3
)
b
Median (Q
1
, Q
3
)
b
p
-value
c
Demographics
Age (years)
52 (46, 60)
52 (45, 59)
0.80
Education (educated),
n
(%)
155 (57.6)
263 (62.2)
0.22
Tobacco use (current
smoker),
n
(%)
163 (59.7)
205 (46.8)
0.0008
Alcohol (g/week)
6.4 (0, 24.9)
0 (0, 3.9)
< 0.0001
Physical activity index
2.8 (2.5, 3.1)
2.8 (2.5, 3.3)
0.71
Waist circumference (cm) 75.4 (69.7, 82.4) 82.0 (71.7, 92.6) < 0.0001
Waist-to-height ratio
0.45 (0.4, 0.5)
0.52 (0.5, 0.6) < 0.0001
Body mass index (kg/m
2
) 20.0 (18.1, 23.2) 26.0 (21.8, 31.9) < 0.0001
Systolic blood pressure
(mmHg)
135 (121, 152)
132 (118, 150)
0.06
Diastolic blood pressure
(mmHg)
88 (78, 98)
88 (70, 97)
0.84
Fasting glucose (mmol/l) 4.8 (4.3, 5.4)
4.9 (4.3, 5.4)
0.53
Total cholesterol (mmol/l) 5.0 (4.1, 6.0)
5.1 (4.4, 6.2)
0.35
High-density lipoprotein
cholesterol (mmol/l)
1.54 (1.2, 2.1)
1.5 (1.2, 1.8)
0.04
Low-density lipoprotein
cholesterol (mmol/l)
3.1 (2.3, 4.0)
3.4 (2.7, 4.2)
0.06
Triglycerides (mmol/l)
1.0 (0.8, 1.5)
1.2 (0.9, 1.8)
0.002
Dietary intake
g
Total energy (kcal/day) 1874 (1377, 2612) 1628 (1189, 2212) 0.001
Total carbohydrate
(g/day)
285.4 (199, 378) 248.8 (180.6, 325.1) 0.01
Total fibre (g/day)
14.8 (25, 30)
17.9 (12.7, 25.2)
0.004
Total protein (g/day)
55.0 (38, 75.7)
46.2 (33.1, 65.0) < 0.0001
Total fat (g/day)
45.3 (28.5, 63.7) 40.5 (26.3, 64.4)
0.10
Total saturated fatty
acids (g/day)
10.5 (6.5, 15.7)
9.5 (5.6, 16.6)
0.13
Total mono-unsaturated
fatty acids (g/day)
11.4 (6.8, 17.8)
10.4 (6.0, 18.3)
0.14
Total polyunsaturated
fatty acids (g/day)
13.6 (8.8, 19.6)
13.1 (7.5, 20.0)
0.47
Total
n
-3 polyunsaturated
fatty acids (g/day)
0.4 (0.2, 0.6)
0.33 (0.20, 0.5)
0.15
Total
n
-6 polyunsaturated
fatty acids (g/day)
13.3 (8.8, 19.2)
12.9 (7.3, 19.6)
0.55
a
Baseline demographic details of participants.
b
Data are presented as median (interquartile range): Q
1
, lower interquartile
range; Q
3
, upper interquartile range.
c
Significance levels of differences in parameters between men and women,
based on Mann–Whitney and chi-squared tests for continuous and
categorical variables, respectively.