Termes les plus recherchés
[PDF](+35👁️) Télécharger Assessing clustering of metabolic syndrome components available at primary care for Bantu Africans using factor analysis in the general population. pdf
This article is from BMC Research Notes, volume 6.
Abstract
Background: To provide a stepbystep description of the application of factor analysis and interpretation of the results based on anthropometric parameters(body mass index or BMI and waist circumferenceor WC), blood pressure(BP), lipidlipoprotein(triglycerides and HDLC) and glucose among Bantu Africans with different numbers and cutoffs of components of metabolic syndrome(MS). Methods: This study was a crosssectional, comparative, and correlational survey conducted between January and April 2005, in Kinshasa Hinterland, DRC. The clustering of cardiovascular risk factors was defined in all, MS group according to IDF(WC, BP, triglycerides, HDLC, glucose), absence and presence of cardiomeTélécharger gratuit Assessing clustering of metabolic syndrome components available at primary care for Bantu Africans using factor analysis in the general population. pdfNasila Sungwacha et al. BMC Research Notes 2013, 6:228
http://www.biomedcentral.eom/17560500/6/228
Research Notes
RESEARCH ARTICLE Open Access
Assessing clustering of metabolic syndrome
components available at primary care for Bantu
Africans using factor analysis in the general
population
John Nasila Sungwacha 1 , Joanne Tyler 2 , Benjamin LongoMbenza 3 *, Jean Bosco Kasiam Lasi On'Kin 4 ,
Thierry Gombet 5,6 and Rajiv T Erasmus 7
Abstract
Background: To provide a stepbystep description of the application of factor analysis and interpretation of the
results based on anthropometric parameters(body mass index or BMI and waist circumferenceor WC), blood
pressure(BP), lipidlipoprotein(triglycerides and HDLC) and glucose among Bantu Africans with different numbers
and cutoffs of components of metabolic syndrome(MS).
Methods: This study was a crosssectional, comparative, and correlational survey conducted between January and
April 2005, in Kinshasa Hinterland, DRC. The clustering of cardiovascular risk factors was defined in all, MS group
according to IDF(WC, BP, triglycerides, HDLC, glucose), absence and presence of cardiometabolic risk(CDM) group
(BMI,WC, BP, fasting glucose, and postload glucose).
Results: Out of 977 participants, 1 7.4%( n = 1 70), 1 1%( n = 107), and 7.7%(n = 75) had type 2 diabetes mellitus(T2DM),
MS, and CDM, respectively. Gender did not influence on all variables. Except BMI, levels of the rest variables were
significantly higher in presence of T2DM than nondiabetics. There was a negative correlation between glucose
types and BP in absence of CDM. In factor analysis for all, BP(factor 1) and triglyceridesHDL(factor 2) explained 55.4%
of the total variance. In factor analysis for MS group, triglyceridesHDLC(factor 1), BP(factor 2), and abdominal
obesitydysglycemia(factor 3) explained 75.1% of the total variance. In absence of CDM, glucose (factor 1) and
obesity(factor 2) explained 48.1% of the total variance. In presence of CDM, 3 factors (factor 1 = glucose, factor 2 = BP,
and factor 3 = obesity) explained 73.4% of the total variance.
Conclusion: The MS pathogenesis may be more glucosecentered than abdominal obesitycentered in not
considering lipidlipoprotein , while BP and triglyceridesHDLC could be the most strong predictors of MS in the
general population. It should be specifically defined by ethnic cutoffs of waist circumference among Bantu Africans.
Keywords: Factor analysis, Metabolic syndrome, Black Africans, Type 2 diabetes
* Correspondence: longombenza@gmail.com
3 Faculty of Health Sciences, Walter Sisulu University, Private Bag XI, Mthatha,
Eastern Cape 5117, South Africa
Full list of author information is available at the end of the article
O© 2013 Nasila Sungwacha et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the
BlOlVlGCl CentrBl Creative Commons Attribution License (http://creativecommons.Org/licenses/by/2.0), which permits unrestricted use,
distribution, and reproduction in any medium, provided the original work is properly cited.
Nasila Sungwacha et al. BMC Research Notes 2013, 6:228
http://www.biomedcentral.eom/17560500/6/228
Page 2 of 10
Background
Metabolic syndrome (MS) is defined by a cluster of
cardiovascular risk factors such as obesity (abdominal
obesity in particular), diabetes mellitus (DM), high blood
pressure (BP)/hypertension, dyslipidemia, insulin resis
tance, and hypercoagulability [15]. Furthermore, MS is
nowadays a major public health problem worldwide [6].
Before the advent of the consensus of the definition of
the MS using International Diabetes Federation(IDF) [6],
Experts from WHO and EGIR required the measurement
of serum insulin levels, diabetes mellitus(DM) or glucose
intolerance, hypertension, triglycerides, HDLCholesterol
(C), body mass index(BMI), and waist circumference(WC)
with different cutoff points [7,8]. However, MS charac
terizes by international cutoff points [69], are limited to
detect efficiently bantu Africans at higher cardiometabolic
(CDM) risk(hypertension, DM, atherosclerosis) in com
parisons with ethnic specific definition of MS [10].
Africans have low or normal lipid profile [11,12] and a
high level of HDLC [13]. LongoMbenza et al. include
low birth weight, coronary heart disease, malnutrition,
elevated fibrinogen, total cholesterol, and urea nitrogen
[1113].
In subSaharan Africa, MS, obesity, dyslipidemia, DM,
hypertension and DM are emerging with cardiovascular
complications [1421] because of urbanization, migra
tion, epidemiologic transition, demographic transition,
and nutrition transition [2224]. Identifying patterns of
Bantu Africans at the primary care level can explain, at
least in part, the differences observed in the prevalence
or incidence of MS and cardiovascular diseases between
different populations [2530].
Several statistical methods can be used to identify
patterns of clustering in cardiovascular diseases such as
DM and hypertension. One such important and useful
technique is factor analysis  a multivariate technique
[3136]. Indeed, Factor analysis is a statistical method
used to describe variability among observed variables in
terms of a potentially lower number of unobserved
variables called factors. Factor analysis searches for
such joint variations in response to unobserved latent
variables. The observed variables are modelled as linear
combinations of the potential factors, plus "error"
terms. The information gained about the interdepen
dences between observed variables can be used later to
reduce the set of variables in a dataset. Furthermore, at
our knowledge, there is no information on the phy
siogenic process including the mechanisms with which
the major components of the MS relate to each other
which could be one of the features to make preventive
startegies and control of emerging cardiovascular dis
eases in Africa [1720]. For that reason, the objective of
this study was to provide a stepbystep description of
the application of factor analysis and interpretations of
the results based on the clustering of anthropometric
parameters, blood pressure, triglycerides, HDLC, and
plasma glucose in all, presence of MS defined by IDF,
absence and presence of CDM(exclusion of triglycerides
and HDLC).
Methods
This study was a crosssectional survey conducted bet
ween January, and April 2005, in Kinshasa Hinterland
with details previously published [13]. This study was
carried out in compliance with the Helsinki Declaration
(59 th WMA General Assembly, Seoul, South Korea,
October 2008. http://www.wma.net/en/30publications/
10policies/b3/index.html). This research was approved
by the Ethics Committee of Lomo Medical Clinic
(Ref000380307) at Kinshasa Limete. Fully informed
and written consent was obtained from all adult
participants.
The survey was specifically and extensively designed
using a statistical multistage and stratified random model
at each level to recruit a study sample with similar and
representative characteristics of Kinshasa Hinterland de
mographic and socioeconomic structure and results com
parable with global data on DM.
Each region contributed with a number of cluster (EDs)
calculated by population number: 185, 112 inhabitants for
the upper urban area of Gombe, 161,410 inhabitants of
the semirural Kisero area, 153,265 inhabitants for the
urban Lukemi area and 146,034 inhabitants for the
deepest rural Feshi area. The sample size was calculated as
Z 2 xPxQx the expected prevalence of DM in each area,
Q = 1P, d is the in the absolute accuracy of 2% ad f = 8.5
to correct the design effect.
The details of collection of weight, height, waist circum
ference (WC), systolic blood pressure (SBP), diastolic
blood pressure (DBP), plasma fasting glucose and plasma
post load glucose have been described elsewhere [3032].
Definitions
Body mass index (BMI) was obtained in dividing weigh
(kg) by height (m) 2 . In our setting with limited resources
and lack of routinely measured insulin resistance (gold
standard), we applied the criteria of MS diagnosis pro
posed by the International Diabetes Federation (IDF) as
follows: raised systolic blood pressure (SBP > 130 mmHg)
and diastolic blood pressure (DBP > 85 mmHg), ele
vated triglycerides (TG > 1.7 mmol/L), low high density
lipoprotein cholesterol (HDL < 1.04 mmol/L in men
and <1.29 mmol/L in women) levels, abdominal obesity
defined by increased waist circumference (WC > 94 cm in
men and >80 cm in women), and fasting plasma glucose
(FPG > 5.6 mmol/L)(6).
CDM was defined by the constellation of 3 components
of WHO  defined MS such as diabetes, hypertension,
Nasila Sungwacha et al. BMC Research Notes 2013, 6:228
http://www.biomedcentral.eom/17560500/6/228
Page 3 of 10
and BMI > =30 kg/m2. However, absence of CDM was
defined in participants without prehypertension, abdo
minal obesity, BMI > =25 kg/m2, and CDM. The defi
nition of diabetes was based on clinical arguments and the
latest WHO/IDF criteria among persons with the fasting
venous plasma glucose level > =126 mg/dL or Postload
venous blood plasma level > =200 mg/dL [7]. This was an
undiagnosed T2DM so that information about HbAlc,
duration of diabetes, and medications was not available
and compulsory.
Statistical analysis
Data were presented as mean ± SD. Factor analysis origi
nated in psychometrics, and is used in behavioral sciences,
social sciences, marketing, product management, opera
tions research, and other applied sciences that deal with
large quantities of data.
Factor analysis is based on the following statistical model
and definitions
Suppose we have a set of p observable random variables,
Xi,...,x^, with means. ftj,...,fi p .
Suppose for some unknown constants L and k unob
served random variables Ft, where i e 1,.,., p and j e 1,...,
where k < p, we have
Xi — m = k\F\ + h ljkF k + £j
Here, the e/s are independently distributed error terms
with zero mean and finite variance, which may not be
the same for all i. Let Var(f,)i^ i , so that we have
Cov{e) = Diag(f 1 , f) = W and E(e) = 0
In matrix terms, we have
xfi = LF + s
If we have n observations, then we will have the di
mensions ~x. pxn , Lp^h and F^ xn . Each column of x and F
denote values for one particular observation, and matrix
L does not vary across observations.
Also we will impose the following assumptions on F.
1. F and e are independent.
2. E(F)=0
3. Cov(f)=/
Any solution of the above set of equations following
the constraints for F is defined as the factors, and L as
the loading matrix.
Suppose Cov(#  u) = E. Then note that from the con
ditions just imposed on F, we have
Cov(xfi) = Cov(LF + s),
or
L = LCov(F)L T + Cov(e),
or
Z = LL T + T.
Note that for any Orthogonal Matrix Q if we set L=LQ
and F=Q T F, the criteria for being factors and factor load
ings still hold. Hence a set of factors and factor loadings
is identical only up to orthogonal transformations.
Common factor analysis, also called principal factor
analysis (PFA) or principal axis factoring (PAF), seeks
the least number of factors which can account for the
common variance (correlation) of a set of variables.
Analogous to Pearson's r, the squared factor loading is
the percent of variance in that indicator variable
explained by the factor. To get the percent of variance
in all the variables accounted for by each factor, the sum
of the squared factor loadings for that factor (column)
was added and divided by the number of variables. This
is the same as dividing the factor's Eigenvalue by the
number of variables.
The Eigenvalue for a given factor measured the vari
ance in all the variables which is accounted for by that
factor. Eigenvalues measure the amount of variation in
the total sample accounted for by each factor.
Extraction sums of squared loadings were performed.
Factor scores were the scores of each case (row) on each
factor (column). To compute the factor score for a given
case for a given factor, the case's standardized score was
taken on each variable, multiplied by the corresponding
factor loading of the variable for the given factor; and
these products were summed.
For determining the number of factors, the Kaiser
criterion was used. The Kaiser rule is to drop all compo
nents with Eigenvalues under 1.0.
The Cattell scree test plotted the components as the X
axis and the corresponding Eigenvalues as the Yaxis. As
one moves to the right, toward later components, the
Eigenvalues drop. When the drop ceases and the curve
makes an elbow toward less steep decline, Cattell's scree
test says to drop all further components after the one
starting the elbow.
Varimax Rotation served to make the output more
understandable and facilitated the interpretation of
factors. This is an orthogonal rotation of the factor axes
to maximize the variance of the squared loadings of a
factor (column) on all the variables (rows) in a factor
matrix, which has the effect of differentiating the ori
ginal variables by extracted factor. This procedure yields
Nasila Sungwacha et al. BMC Research Notes 2013, 6:228
http://www.biomedcentral.eom/17560500/6/228
Page 4 of 10
results which make it as easy as possible to identify each
variable with a single factor. To avoid theoretical sup
posed grounds, we used oblique Promax rotation as
additional alternative to varimax rotation for suited clus
tering characteristics.
A Pvalue < 0.05 was considered as statistically signifi
cant. All analyses were performed using the Statistical
Package for Social Sciences (SPSS) for windows version
18.0 (SPSS Inc) Chicago, II, USA.
Results
Out of the original population (n = 977 with 458 males and
519 females), 170(17.4%), 107(11%), and 75(7.7%) were
diagnosed for new T2DM, MS, and CDM, respectively.
Table 1 describes the mean levels of general characteris
tics according to T2DM status. Except similar(P > 0.05)
values of BMI in presence and absence of T2DM, levels of
age, WC, SBP, DBP, and triglycerides were significantly
(P < 0.05) higher in T2DM participants than no diabetic
participants. However, HDLC values were significantly
(P < 0.05) lower in T2DM presence than diabetes absence.
The mean levels of age, BMI, WC, SBP, DBP, triglycerides,
HDLC, FPG,and postload glucose in men were similar
(P > 0.05) with those from women(results not shown).
In the general population, factor analysis generated 2 fac
tors which were explaining 55.4% of total variance: factor 1
(Blood pressure with variance = 29.6%; DBP = 0.881 and
SBP = 0.872), and factor 2(Dylipidemia with variance =
25.8%; HDLC = 0.886 and triglycerides = 0.872).
In MS participants, factor analysis generated 3 fac
tors with total variance of 75.1%: factor l(Dyslipi
demia with variance = 29.5%; triglycerides = 0.911 and
HDLC = 0.874), factor 2(Blood Pressure with vari
ance = 27.9%; DBP = 0.869 and SBP = 0.837), and factor 3
(Abdominal obesity + Dysglycemia with variance = 18.1%;
WC = 0.836 and fasting plasma glucose = 0.609).
Table 1 Association of age and metabolic syndrome
components with incident type 2 diabetes mellitus in the
study population
Variables of
interest
Participants with
incident types 2 DM
Absence of diabetic
participants
P Values
Age (Years)
53 ±13
36+12
< 0.0001
BMI (Kg/m 2 )
24.4 ±5.6
23.3 ±5.4
0.881
WC (cm)
86 ±14.7
78.7 ± 14.3
< 0.0001
SBP (mm Hg)
131.1 ±31.1
117±17
< 0.0001
DBP (mm Hg)
80 ±16.7
70 ± 11.7
< 0.0001
FPG (mmd/L)
7.5 ±2
5.1 ±1
< 0.0001
Triglycerides
(mmd/L)
3.8 ±0.8
2.7 ± 0.9
< 0.0001
HDLC
(mmd/L)
1 ±0.3
2 ±0.6
< 0.0001
Retrieved from "http://en.wikipedia.org/wiki/Factor_analusis."
Table 2 Characteristics in absence of cardiometabolic risk
Variables Mean±SD
BMI (Kg/m 2 )
24.6 ± 8.80
Waist Circumference (CM)
79.8 ± 1 3.6
SBP (mmHg)
11 3.4 ±12.4
DBP (mmHg)
66.6 ± 07.6
FPG (mg/dL)
82.0 ± 14.0
PostLoad PG (mg/dL)
123.21 ±18.0
Absence of CDM(n = 572)
Table 2 describes the mean values of variables analyzed
in participants without CDM. The correlation matrix in
absence of CDM is presented in Tables 3 and 4. Post
load plasma glucose was significantly and positively
correlated to BMI and WC, but significantly and nega
tively correlated to both SBP and DBP. SBP was signifi
cantly and positively correlated to BMI but significantly
but negatively correlated to FPG. DBP was significantly
and negatively correlated to FPG.
Factor analysis revealed two uncorrelated factors that
cumulatively explained 48.1% of the observed variance
of the absence of CDM. The number of those two
factors was determined by the scree plot according to
Eigenvalue (Figure 1). These two factors could be iden
tified as Blood Glucose Metabolism Disordering (Factor
1; 26.3% of variance) and obesity (Factor 2; 22% of
variance) (Table 5 and Figure 2).
Presence of CDM
The mean values of variables analyzed in participants
with CDM are presented in Table 6. Factor analysis re
vealed three uncorrelated factors that cumulatively
explained 73.6% of the observed variance in the presence
of cardiometabolic risk. The number of these three fac
tors (Components) was determined by the scree plot
according to Eigenvalues (Figure 3).
These three factors could be identified as: Blood Glucose
Metabolism Disordering (Factor 1; 35.6% of variance),
Table 3 Correlation matrix in absence of cardiometabolic
risk
WC FPG Postload PG
BMI 0.132
P = 0.016
SBP 0.114
P = 0.031
DBP 0.125
P = 0.020
WC 0.035 0.146
P = 0.286 P = 0.008
Nasila Sungwacha et al. BMC Research Notes 2013, 6:228
http://www.biomedcentral.eom/17560500/6/228
Page 5 of 10
Table 4 Correlation matrix in absence of cardiometabolic
risk
BMI
SBP
DBP
BMI

0.104
P = 0.046
0.052
P = 0.197
SBP
0.128
P = 0.019
DBP
WC
0.200
0.063
0.021
P < 0.0001
P = 0.286
P = 0.364
FPG
0.086
0.155
0.118
P = 0.080
P = 0.006
P = 0.027
Blood Pressure (Factor 2; 20.3% of variance), and obesity
(Factor 3; 17.7% of variance) (Table 7).
general population, absence of CDM, presence of CDM,
and presence of MS among Bantu Africans living in DR
Congo(Central region).
The extent of T2DM, CDM (concurrent presence of 3
nonlipid components of MS), and MS defined by IDF 5
criteria such as 3 nonlipid components and 2 lipid
lipoprotein components [6] was examined. The present
study also determined the interrelation of the main
CDM factors: BMI, WC, SBP, DBP, FPG, and postload
glucose.
Emerging burden of MS
Contrary to the previous myths, non communicable
diseases (Diabetes, hypertension, MS, atherosclerosis) are
no longer rare in Africa [1014]. The extent is increasing
and it is thought to be due to the shifting from traditional
African customs to the Western lifestyle [1518].
Discussion
The present study identified MS combination for which
factor analysis would be appropriate among Bantu Africans.
For that reason, the steps involved in performing factor
analysis procedure were described. Thus, factor analysis
findings using SPSS software have been interpreted.
However, MS is a complex issue in health care. It does
not have a simple cause, but multiple risk factors. Its
natural course is influenced by genetic factors, personal
(Host) attributes, environmental characteristics, or some
interactions of both.
At our knowledge, this was the first study to
characterize factor analysis of possible risks for cluster
ing of some traditional cardiovascular risk factors in the
MS pattern
The present study sought at identifying the physiogenic
factors responsible for the clustering of cardiometabolic
components. Factor analysis showed marked differences
in the MS pattern between the groups of 3 components
(CDM) and 5 components (MS).
Number of generated factors
In the general adult population, factor analysis identified
3 components for MS. This finding about MS was con
sistent with a study conducted in Asian Indians from the
general population [1]. In India, however, the total vari
ance of 65.3% [1] was lower than the total variance of
75.1% explained in the present study. However, in the
Scree Plot
Component Number
Figure 1 Eigen values among participants without cardiometabolic risk.
Nasila Sungwacha et al. BMC Research Notes 2013, 6:228
http://www.biomedcentral.eom/17560500/6/228
Page 6 of 10
Table 5 Rotated component matrix in absence of
cardiometabolic risk
Factor 1
Factor 2
BMI
0.131
0.738
SBP
0.502
0.340
DBP
0.455
0.114
WC
0.077
0.696
FPG
0.765
0.051
PostLoad PG
0.707
0.400
South African general population, 5 factors could be
identified in factor 1 (Obesity), factor 2(Hypertension),
factor 3(Hyperuricemiahypertriglyceridemia), factor 4
(Hyperglycemia), and factor 5(Hyperinsulinemia) [1]. In
our findings, the first 2 factors cumulatively explained
58% of the total variance for MS. Only considering 3
nonlipid components, affordable in limited resources
areas, factor analysis had identified also 3 factors with
total variance almost 74.5% for CDM and similar with
that for MS. The first 2 factors(Dysglycemia and Hyper
tension) cumulatively explained 56% of the total variance
of CDM.
In considering the entire population and the sub
population without CDM, factor analysis generated only
2 factors. In all participants, the factors revealed such as
hypertension(factor 1) and dyslipidemia(factor 2) cumu
latively explainedb55.4%bof the total variance of the
clustering pattern of atherogenic factors from MS. How
ever, in the absence of CDM, BP was not loaded, while
only dysglycemia(factor 1) and obesity/BMI and WC(fac
tor 2) were revealed the first factors which cumulatively
explained 48.1% of the total variance of the charac
terization of this group by the clustering of nonlipid
components for MS.
The present study showed that no overlapping of vari
ables on more than 1 factor indicated that more than 1
variable was responsible for the ultimate phenotype of
the MS. Our findings demonstrated that factor analysis
confirmed the general results from other factor analyses
of the MS on different ethnic groups that had 35 factors
revealed [14].
Our findings with the clustering of the variables in MS
as a result of multiple factors known modifiable in na
ture raised the following question: would it be more effi
cient to include all participants in one major factor
analysis model? Indeed, factor analysis is practically lim
ited to develop a singleparameter screening tool for MS
in this study as mentioned in the literature [4]. IDF
recommended WC as the most frequently used an
thropometric index to define abdominal obesity [6],
Paradoxically, WC, BP, lipid, and glucose levels were
similar among men and women in this study as reported
in the same general population [10]. However, WC cut
off points differ by ethnic groups and gender worldwide
[1,4]. Older age was associated with T2DM in this study,
while age not considered as a component of MS, is a
confounding factor for anthropometric variables of MS
amon Taiwanese individuals [4].
Factor analysis was applied to see whether there was a
less complex space with fewer than the "n" dimensions
of the variables that had been analyzed. It was found
that a three dimensional space or a mixture of three fac
tors could be used to explain a major part of the data. In
Component Plot in Rotated Space
i.o
0.5
CM
*•
c
0.0
0.5 ■
1 .0
DBP
O
•1.0
— I —
0.5
WC
o
qSObis
O
OFASTINO
O
— I —
0.5
0.0
Component 1
Figure 2 Twocomponent plot in rotated space among participants without cardiometabolic risk.
— i —
1.0
Nasila Sungwacha et al. BMC Research Notes 2013, 6:228
http://www.biomedcentral.eom/17560500/6/228
Page 7 of 10
Table 6 In participants with cardiometabolic risk 2
Variables
Mean±SD
BMI (Kg/m 2 )
PostLoad PG (mg/dL)
SBP (mmHg)
DBP (mmHg)
FPG (mg/dL)
Waist Circumference (CM)
43.40 ± 20.20
80.00±15.80
1 27.60 ± 26.70
80.00 ± 15.80
93.40 ± 1 9.80
1 82.84 ± 101.10
more precise mathematical terms the global and exam
ined variables without dyslipidemia(with paradoxes of
triglycerides and HDLC) could be reduced to three fac
tors with eigenvalues greater than one, which explained
73.4% of the variance in MS Africans. The loadings on
these factors sorted out into three metabolic groupings.
Neither of the variables was loaded on all the three
components. These three factors could be identified as
Glucose Metabolism (Factor 1), Blood Pressure (Factor
2) and Obesity (Factor 3). This suggests that those non
lipid components clustered naturally rather than as a re
sult of chance.
No overlapping of variables on more than one factor
indicated that more than 1 variable is responsible for the
ultimate phenotype of the fats. The present factor ana
lysis confirmed global results from other factor analyses
of fats among different populations that had 3 to 4 fac
tors identified as nonmodifiable/genetic risk factors and
modifiable/ environmental risk factors. The study
attempted to observe among BMI, WC, SBP, DBP, FPG,
and postload PG group  which ones go together and
which ones do not [30]. Variables with a factor loading
of at least 0.3 have generally been considered for inter
pretation although it is suggested that only loadings > 0.4
be used, which therefore shares at least 15% of the vari
ance with a factor, should be used in the study [24].
In many studies, fats play a pivotal role in the occur
rence of the onset of CVD, andT2DM. However, lipid
profile and fasting insulin are not available in the major
ity of health centers in developing countries.
Therefore, identification of nonlipid components of
the metabolic syndrome would be helpful in understand
ing the etiology among Bantu Africans. Virtually no
study has been performed on combination of the evalu
ated variables in Sub Saharan Africa.
Perspectives for Africa
This study highlighted the absence of obesity as a factor
of MS in type 2 diabetic Bantu Africans. Moreover, obes
ity was the third factor of MS with lower variance in
comparisons with variances of factor 1 (Glucose) and fac
tor 2(Blood pressure) among type 2 diabetic Africans
with MS. As reported on the factor analysis of risk vari
ables associated with MS in adult Asian Indians [1], fur
ther studies among larger sizes from Bantu Africans, are
needed to demonstrate the responsibility of more than
one underlying physiogenetic polymorphisms in the
present specific glucosecentered pattern for MS with
lower BMI and smaller WC.
Limitations and strengths
The advantages and disadvantages of factor analysis have
been reported in medical, physical, marketing economic
and environmental researches [31]. There are different
Scree Plot
2.5
2.0
1.5
m 1.0
0.5
0.0
12 3 4
Component Number
Figure 3 Eigen values among participants with cardiometabolic risk.
Nasila Sungwacha et al. BMC Research Notes 2013, 6:228
http://www.biomedcentral.eom/17560500/6/228
Page 8 of 10
Table 7 Rotated component matrix in presence of
cardiometabolic risk
Factor 1
Factor 2
Factor 3
BMI
0.077
0.025
0.760
SBP
0.064
0.881
0.137
DBP
0.224
0.833
0.116
WC
0.025
0.001
0.769
FPG
0.906
0.106
0.096
PostLoad PG
0.894
0.179
0.031
reasons of the limitations of this study, that is, ethnic
and cultural heterogeinity, genetic studies, gender, age
composition, number of risk variables included, sample
size, and cutoff points of MS and CMD[ ]. In Asian
Indians, angiotensin converting enzyme gene polymor
phism(insertion/deletion) with BP was identified factor 3
along lipids and lipoproteins(factor 1) and centripetal fat
and BP(factor 3) associated with MS phenotype [1], In
these Asian Indians, DBP in factor 2 overlapped on
another variable in factor 3 [1],
Advantages of factor analysis
The rotation methods are useful in making the output
more understandable and for ease of interpretation of
the factors. The optimal variance of the squared loadings
of a factor (Column) on all the variables (rows) in a factor
matrix is due to varimax rotation (an orthogonal rotation
of the factor axes). Factor matrix differentiates the original
variables from extracted factors.
Groups of interrelated variables are identified and seen
in their manner to be related to each other.
In multi factorial diseases, it is easy and inexpensive to
perform factor analysis which can be used to identify hid
den dimensions which may not be apparent from analysis.
Disadvantages of factor analysis
It is not possible to pick the proper rotation using factor
analysis alone as all rotations represent different under
lying processes and equally valid outcomes of standard
factor analysis optimization.
Though not a strictly mathematical criterion, there is
much to be said for limiting the number of factors to
those whose dimension of meaning is readily compre
hensible. The same limitation is reported about variance
explained criteria.
The research is requested to choose the solution which
generates the most comprehensive evaluation of data.
The Kraiser criterion is the default in SPSS and most
computer programs but is not recommended when used
as the sole cutoff criterion for estimating the number of
factors.
Certain researchers prefer to keep enough factors to
account for 80%90% of the variation. However, other
researchers explain variance with a few factors, but
lower than 50% (Parsimony).
Factor analysis cannot identify causality as interpreting
factor analysis is based on using a " heuristic" convenient
solution even if not absolutely "true". If important attri
butes (such as lipid components of fats) at primary
health care in developing countries like DRC, the value
of the procedure was reduced for BMI in absence of MS.
It requires strong background knowledge of biology
and Pathophysiology or theory as multiple attributes
may be highly correlated for no apparent reason.
Varimax was an orthogonal rotation of the components to
maximize the variance of the squared loadings (unrotated
output accounted for by the first and subsequent factors)
of a dimension (Column) on all the variables(Rows) in a
factor matrix. Varimax rotation is the easiest and the most
simple and common rotation option used in MS [15].
However, oblique rotations might be more suited and
more preferred with methods inclusive [31]. In search of
underlying dimensions, the use (sometimes an abuse) of
factor analysis in Personnality and Social Psychology
literature [32]. There are also different rotation methods
such as quartimax rotation(an orthogonal alternative),
equimax rotation( a compromise between varimax and
quartimax criteria), direct oblimin rotation(standard me
thod with a nonorthogonal/oblique rotation with higher
eigenvalues but lower interpretability of the factors), and
Promax rotation. In this study, we evaluated Promax
rotation in addition to varimax rotation. Indeed, Promax
rotation was computationally faster alternative non
orthogonal/oblique rotation method than other oblique
methods such as direct oblimin rotation. The potential
limitations such as the inability of the investigators in
collecting sufficient set of product activities, unknown on
reasons of associated dissimilar attributes, and obscured
factors were excluded or minimized.
Implementation of factor analysis
The implementation of Factor analysis is well established
within robust statistical software such as SAS, BMDP
and SPSS and R programming language with the factanal
function (GPA rotations), and Open Opt [33]. This is
evidenced by both analysis and scree plots and the three
dimensional charts.
Conclusion
The factor analysis performed for this study suggests
that the clustering of the nonlipid variables is sufficient
to define CDM in black Africans at including glucose
metabolism, Blood pressure and Obesity. Since 3 factors
in sequencing dyslipidemia, hypertension, and abdominal
obesitydysglycemia were identified for the Bantu Central
African MS phenotype, more one major factor could be
accounted for this specific MS. Early prevention and
Nasila Sungwacha et al. BMC Research Notes 2013, 6:228
http://www.biomedcentral.eom/17560500/6/228
Page 9 of 10
management (diagnosis and proper intervention) strategies
for those modifiable loaded risk variables could reduce the
burden of type 2 DM, MS, and emerging cardiovascular
disease in Central Africa.
Competing interests
All authors declare that they have no competing interests.
Authors' contributions
All the coauthors have seen and approved the final version of the
manuscript and it is not currently under active consideration for publication
elsewhere, has not been accepted for publication, nor has it been personally
and actively involved in substantive work leading to the report, and will hold
themselves jointly and individually responsible for its content. JBKL0 was
responsible for the field work. JNS performed review literature related to
factor analysis. BLM conceived of the study, and participated in the study
design. JNS and BLM performed statistical analyses. JTL, JNS, BLM, JBKLO, and
GT participated in the coordination of writing of the study. All authors read
and approved the final manuscript.
Acknowledgements
We thank the medical officers, interns, social workers, and nursing staffs from
the University of Bandundu, the participants and the professionals of the
laboratory of Lomo Medical for the ultrastructural technical assistance.
Author details
'Department of Statistics, Walter Sisulu University, Mthatha, South Africa,
department of Statistics, University of Forte Hare, Eastern cape, South Africa.
3 Faculty of Health Sciences, Walter Sisulu University, Private Bag XI, Mthatha,
Eastern Cape 51 17, South Africa. 4 Department of Internal Medicine, University
of Kinshasa, Kinshasa, DR Congo. 5 Division of Cardiology and Intensive Care,
University of Maiden Ngouabi, Kinshasa, Congo. Emergency Department,
University Hospital Center, Brazzaville, Congo. 'Division of Chemical
Pathology, Stellenbosch University, Cape Town, Stellenbosch, South Africa.
Received: 23 January 2013 Accepted: 21 May 2013
Published: 12 June 2013
References
1 . Mithun D, Susil P, Arnab G: Factor analysis of risk variables associated
with metabolic syndrome in adult Asian Indians. J Cardiovasc Dis Res
2010,91:8691.
2. Ghosh A: Factor analysis of metabolic syndrome among the middle
aged Bengalee Hindu men of Calcutta, India. Diabetes Metab Res Rev 2005,
21:5864.
3. Gray RS, Fabsitz RR, Cowan LD, Lee ET, Howard BV, Savage PJ: Risk factor
clustering in the insulin resistance syndrome. The strong Heart Study.
Am J Epidemiol 1998, 148:869878.
4. Chang CH, Yen CH, Chen MY: Using principal component analysis to
develop a singleparameter screening tool for metabolic syndrome.
BMC Publ Health 2010, 10:708.
5. Esteghanati A, Zandiah A, Khalilzah 0, Meysamie A, Ashraf H: Clustering of
metabolic syndrome components in a middle Eastern diabetic and non
diabetic population. Diabetol Metab Syndr 201 0, 2:36.
6. Alberti KG, Zimmet P, Swaw J: Metabolic syndrome: a new worldwide
definition: a consensus statement from the international diabetes
federation. Diabetes Med 2006, 23:469480.
7. World Health Organization (WHO) and International Diabetes Foundation (IDF)
Definition and Diagnosis of Diabetes Mellitus and Intermediate Hyperglycemia.
Geneva: WHO: Report of a WHO/IDF Consultation; 2006.
8. he European Group for the Study of Insulin Resistance(EGIR): Frequency of
the WHO metabolic syndrome in European cohorts, and an alternative
definition of an insulin resistance syndrome. Diabetes Metab 2002,
28:364376.
9. Expert panel on detection, evaluation and treatment of high blood
cholesterol in adults. Executive summary of the third report of the
national cholesterol education program(NCEP) expert panel on
detection, evaluation and treatment of high blood cholesterol in adults
(adult treatment panel III). JAMA 2001, 285(19)24862497. HYPERLINK
http://www.ncbi. nlm.nih.gov/pubmed ?term=Expert%20Panel%20on%
20Detection%2C%20Evaluation%2C°/o20and%20Treatment%20of%20High%
20Blood%20Cholesterol%20in%20Adults%5BCorporate%20Author%
5D"Expert Panel on Detection, Evaluation, and Treatment of High Blood
Cholesterol in Adults. Executive Summary of The Third Report of The
National Cholesterol Education Program (NCEP) Expert Panel on Detection,
Evaluation, And Treatment of High Blood Cholesterol In Adults (Adult
Treatment Panel III). HYPERLINK "http://www.ncbi.nlm.nih.gov/pubmed/
11368702" \o "JAMA.
10. LongoMbenza B, Kasiam Lasi On'kin JB, Nge Okwe A, Kangola Kabangu N:
The metabolic syndrome in a Congolese population and its implications
for metabolic syndrome definitions. Diabetes & metabolic syndrome.
Clinical Research & Reviews 201 1,5:1 724.
1 1 . LongoMbenza B, Ngiyulu R, Bayekula M, Vita EK, Nkiabungu FB, Seghers KV,
Luila EL, Mandundu FM, Manzanza M: Low birth weight and risk of
hypertension in African school children. J Cardiovasc Risk 1999, 6:31 1314.
12. LongoMbenza B, Vangu Ngoma D, Mbungu Fuele S: Low birth weight,
metabolic syndrome and their associations with the global crisis of 1930 
1945, rapidly growing economy and coronary heart disease in Central
Africa. Int J Nutr Met 201 0, 2:1 1 0.
13. LongoMbenza B, Kasiam LO JB, Nge OA, Kangola KN, Mbungu FS:
Metabolic syndrome, aging, physical inactivity, and incidence of type 2
diabetes in general African population. Diab Vase Dis Res 2010, 7(1)2839.
14. Motala AA, Pirie FJ, Gouws E, Amod A, Omar MAK: High incidence of type
2 diabetes mellitus in South African Indians: a 10 year for study.
Diabetic Med 2003, 20:2330.
15. Kelliny C, William J, Riesen W, Paccaud F, Bovet P: Metabolic syndrome
according to different definitions in a rapidly developing country of the
African region. Cardiovasc Diabetol 2008, 7:27. doi:l 0.1 1 86/1 4752840727.
16. Kasiam LO JB, LongoMbenza B, Nge Okwe A, Ditu Mpandamadi S,
Okitolonda W, He J: Prevalence and risk factors of diabetes mellitus in
Kinshasa Hinterland. Int J Diabetes Metabolism 2008, 1 6:5059.
1 7. Democratic Republic of Congo: National survey of children and women
situation. MICS2 2001. Say yes for children. 2nd edition. Kinshasa: USAID/
WHO/UNICEFV; 2001:4046.
18. Ngondo A, Pitshandenge S: Demographic perspectives of Zaire during
19841999. Cedas 1984, 1:3132.
19. Balde NM, Diallo I, Balde MD, Barry IS, Kaba L, Diallo MM, Kake A, Camara A,
Bah D, Barry MM, SangareBah M, Maugendre D: Diabetes and impaired
fasting glucose in rural and urban populations in Futa Jallon (Guinea):
prevalence and associated risk factors. Diabetes Metab 2007, 33:1 14120.
20. Elbagir MN, Eltom MA, Elmahadi EMA, Kadam IMS, Berne C: A population
based study of the prevalence of diabetes and impaired glucose
tolerance in adults in Northern Sudan. Diabetes Care 1996, 19:1 1 261 128.
21 . Gu D, Reynolds K, Wu X, Duan X, Reynolds RF, Whelton PK, He J, InterASIA
Collaborative Group: Prevalence of the metabolic syndrome and
overweight among adults in China. Lancet 2005, 365:13981405.
22. Lee JS, Kawakubo K, Mori K, Akabayashi A: Effective cutoff values of waist
circumference to detect factors of metabolic syndrome in Japanese men
and women. Diabetes Vase Dis Res 2007, 4:340345.
23. Bryant FB, Yarnold PR: Principal components analysis and exploratory and
confirmatory factor analysis. In Reading and understanding multivariate
statistics. Edited by Grimm LG, Yarnold PR. Washington, DC: American
Psychological Association; 1995:99136.
24. Sheppard AG: The sequence of factor analysis and cluster analysis:
differences in segmentation and dimensionality through the use of raw
and factor scores. Tour Anal 1996, 1:4957.
25. Sternberg RJ: The geographic metaphor. In Metaphors of mind: conceptions
of the nature of intelligence. Edited by Sternberg RJ. New York: Cambridge;
1990:85111.
26. Stills DL: International encyclopedia of the social sciences: biographical
supplement. New York: Macmillan Publishing Co; 1989.
27. Gorsuch RL: Factor analysis. Hillsdale, NJ: Lawrence Erlbaum; 1983.
28. Velicer WF, Jackson DN: Component analysis versus common factor
analysis: some issues in selecting an appropriate procedure.
Multivar Behav Res 1990, 25(l):128.
29. Kasiam LO JB, LongoMbenza B, Nge OA, Kangola KN, Mbungu FS,
Milongo DG: Classification and dramatic epidemic of diabetes mellitus in
Kinshasa Hinterland: the prominent role of type 2 diabetes and lifestyle
changes among Africans. Niger J Med 2009, 18(3)31 1320.
30. Nunnally JC, Bernstein IH: Psychometric theory. 3rd edition. New York:
McGraw Hill; 1994.
Nasila Sungwacha et al. BMC Research Notes 2013, 6:228
http://www.biomedcentral.eo
Lire la suite
 389.20 KB
 15
Vous recherchez le terme ""
35
34