Heart disease and diabetes risk factors in Pacific Islands communities and associations with measures of body fat.

Abstract

To describe the prevalence of obesity and other coronary heart disease and Type 2 diabetes risk factors by age and ethnic group in Pacific Island communities and to determine the associations between these risk factors and body mass index.
Cross-sectional data from commuity-based intervention projects were combined to provide anthropometric, blood sample and blood pressure data on 1,175 Pacific Islands people (467 men, 708 women) aged 20 years and over from church communities in South, Central and West Auckland. Self-reported data on diabetes status and leisure-time physical activity were also collected.
Based on an ethnic-specific mass index (BMI) cut-off (> 32 kg/m2), 45% of men and 66% of women were obese. The age-standardised prevalence of known diabetes was 12%. Men and women aged 40-60 years had the highest risk factor levels and were the most sedentary. Tongans had higher risk factor levels than Samoans. In men, BMI and waist circumference were associated (p<0.05), in the direction of greater disease risk, with blood pressure and concentrations of total cholesterol, HDL-cholesterol, LDL-cholesterol, triglycerides, and blood glucose. In women, these associations were similar but less consistent.
While these data are not representative for all Pacific people living in New Zealand, they do show an extremely high prevalence of obesity and significant associations between obesity and other cardiovascular risk factors. These communities warrant a very high priority as part of public health efforts to address New Zealand's growing obesity epidemic.

Full text

Page 1

208New Zealand Medical Journal11 May 2001
Pacific populations in New Zealand carry a heavy burden
of coronary heart disease (CHD) and diabetes.1,2 Several
CHD risk factors are more prevalent amongst Pacific
people than they are amongst European New Zealanders.
They have higher mean blood pressures and a higher
prevalence of hypertension,3 microalbuminuria is more
prevalent,4 physical activity levels are lower and smoking
rates are higher.5 Also, Pacific people consume larger
quantities of food than Europeans and their diets contain
more meat and less fruit and vegetables.6 Moreover, Pacific
populations are among the most obese populations in the
world and obesity is a strong independent risk factor for
both CHD and Type-2 diabetes.7,8 In contrast, some CHD
risk factors are not as prevalent as might be expected in
such obese populations. For example, serum cholesterol
levels tend to be lower for Pacific people compared to
Europeans.9 This may be due to genetic differences and to
diet, but also, weak associations between high levels of
body fat and other CHD and diabetes risk factors have
been described.10
Heart disease and diabetes risk factors in Pacific Islands communities and
associations with measures of body fat
A Colin Bell, Postdoctoral Research Fellow; BA Swinburn, Associate Professor, Public Health Nutrition,
Department of Community Health, Faculty of Medicine and Health Science, University of Auckland; D Simmons,
Senior Lecturer in Medicine; W Wang, Data manager, South Auckland Diabetes Project, Department of Medicine,
Middlemore Hospital; H Amosa, Project Coordinator, Department of Community Health, Faculty of Medicine and
Health Science, University of Auckland; B Gatland, Coordinator, South Auckland Diabetes Project, department of
Medicine, Middlemore Hospital, Auckland.
Aims. To describe the prevalence of obesity and other
coronary heart disease and Type 2 diabetes risk factors by
age and ethnic group in Pacific Island communities and to
determine the associations between these risk factors and
body mass index.
Methods. Cross-sectional data from two community-based
intervention projects were combined to provide
anthropometric, blood sample and blood pressure data on
1175 Pacific Islands people (467 men, 708 women) aged 20
years and over from church communities in South, Central
and West Auckland. Self-reported data on diabetes status
and leisure-time physical activity were also collected.
Results. Based on an ethnic-specific body mass index (BMI)
cut-off (> 32 kg/m2), 45% of men and 66% of women were
obese. The age-standardised prevalence of known diabetes
was 12%. Men and women aged 40 - 60 years had the
highest risk factor levels and were the most sedentary.
Tongans had higher risk factor levels than Samoans. In men,
BMI and waist circumference were associated (p<0.05), in
the direction of greater disease risk, with blood pressure and
concentrations of total cholesterol, HDL-cholesterol, LDL-
cholesterol, triglycerides, and blood glucose. In women,
these associations were similar but less consistent.
Conclusions. While these data are not representative for
all Pacific people living in New Zealand, they do show an
extremely high prevalence of obesity and significant
associations between obesity and other cardiovascular risk
factors. These communities warrant a very high priority as
part of public health efforts to address New Zealand’s
growing obesity epidemic.
Abstract
NZ Med J 2001; 114: 208-13

Page 2

New Zealand Medical Journal 20911 May 2001
Much of the information available on the prevalence of
obesity and other CHD risk factors in Pacific populations in
New Zealand is based on the Workforce Diabetes Survey11,12
or the 1997 National Nutrition Survey (NNS97).13
Unfortunately, both surveys have their limitations in
providing the full picture of obesity in Pacific populations.
The NNS97 over-sampled Pacific people but interpretation
is hampered by a poor response rate (less than 50%) and low
sample size (273 with anthropometric data). The workforce
survey had larger numbers (n=650) and was able to examine
for differences in CHD risk between Pacific ethnic groups,
rather than assuming that risk is homogeneous.11 However,
the sampling frame was restricted to the older (40+ years),
employed Pacific workforce in Auckland and Tokoroa.
This present study pools baseline data from two large
community-based intervention projects in Auckland and
describes variations in the prevalence of obesity and other
CHD and diabetes risk factors by age and ethnic group. It
also investigates the associations between these risk factors
and measures of body fatness. While this study also has
limitations of extrapolation to the New Zealand-wide Pacific
communities, the overall aim is to provide information from
these two large studies to build a clearer picture of these
important risk patterns.
Methods
Participants. Participants came from two community-based intervention
programs, the South Auckland Diabetes Project (SADP) and the Samoan
Ola Fa’autuata Project (SOFP). The SADP was established in 1991 as a
multi-faceted program that aimed to reduce the incidence of diabetes in
New Zealand through lifestyle interventions. The Samoan Ola Fa’autauta
or ‘Life-wise’ Project was a similar community-based lifestyle program.14
Both projects worked with Pacific Islands church groups in South,
Central and West Auckland. The selection of churches for interventions
was non-random but all adults from each church community were invited
to take part. There was no specific selection for those who were obese or
those with diabetes. The church community was defined as those people
whose names were on the church roll plus their household members.
From these membership lists, the baseline response rate for the SADP
was 60% and for the SOFP it was 81%.
This analysis includes self-identified Pacific Islands people, aged 20
years and over who had complete anthropometric data at baseline. A total
of 1175 people, 725 (287 men, 438 women) from the SADP and 450 from
the SOFP (180 men, 270 women). Where our data from these Church
Intervention Surveys (CIS) were compared with the National Health and
Nutrition survey, we included BMI data from an additional 117
adolescent’s aged 15 - 19 years. The SOFP was given ethical approval by
the University of Auckland Human Subjects Ethics Committee and the
SADP by the Auckland Area Health Board Ethics Committee.
Data collection. Both projects collected data at a series of health
surveys on church premises between 1991 and 1996. Participants received
information sheets, translated if necessary, and a brief presentation
explaining the project. Consent was obtained from all participants
(interpreters were available). Standardised techniques were used to
measure weight, height, waist, and hip circumference.15 The SOFP used
Seca electronic scales (model 708) with an attached stadiometer to
measure weight and height after removing heavy clothing and shoes. The
SADP used the same standardised techniques, a portable stadiometer
(CMS, London) and Salter spring scales. The scales were calibrated
regularly. Non-stretch fibreglass tapes were used to measure waist and
hip circumference. Waist circumference was measured horizontally
through a point midway between the top of the iliac crest and the bottom
of the ribs. Hip circumference was measured at the largest posterior
extension of the buttocks.
Non-fasting venous blood samples were collected, stored and analysed
for blood lipids, glucose and fructosamine levels by Medlab Ltd. Samples
were measured using Roche Diagnostic protocols. Blood pressure was
measured twice in the sitting position with a standard mercury
sphygmomanometer using Korotkoff phase 1 and 5 sounds for systolic
and diastolic blood pressure respectively. Oversize cuffs were used for
large arms. Participants were considered sedentary if they did no
moderate or vigorous activity during a normal week.
Statistical Analysis. BMI was calculated as weight (kg) divided by
height (m) squared. Pacific specific BMI cutoffs were used to define
overweight (26 kg/m2 ≥ BMI < 32 kg/m2) and obesity (BMI ≥ 32 kg/
m2).16 The data were stratified by gender and analysis of variance was
used to calculate adjusted age- and ethnic-specific means. Age-
standardised means (Table 1) were calculated by the direct method
using Segi’s world population for those aged ≥ 20 years.17 Multiple
regression was used to test for linear associations between age and blood
cholesterol, triglyceride, glucose and blood pressure. A second model
that included a quadratic term (age-squared) was used to test for curvi-
linearity. With the exception of the ethnic group analysis, indicator
variables were used to adjust for confounding by ethnic group and to
adjust for systematic differences in risk factor levels between the two
projects.
Table 1. Association between age and anthropometric, biochemical and blood pressure measurements, mean (SEM), in Pacific Islands men. The
prevalence of obesity, sedentary leisure time activity and known diabetes is also given.
AgeAge group (years)
40-49
P for linear
term†
P for quadratic
term‡
Standardised* 20-2930-3950-59
≥ 60
Anthropometry, n
Weight, kg
Height, cm
Body mass index, kg/m2
Waist, cm
Hip, cm
BMI > 32 kg/m2, %
467
97.6 (8.9)
172.9 (3.0)
32.6 (2.7)
103.9 (6.9)
110.7 (5.6)
45
106
96.7 (l.9)
177.1 (0.6)
30.8 (0.6)
96.7 (l.5)
109.9 (l.2)
32
97
98.3 (2.0)
174.1 (0.7)
32.3 (0.6)
103.0 (1.5)
109.7 (l.3)
37
120
99.9 (l.8)
171.5 (0.6)
33.9 (0.5)
107.2 (l.4)
112.1 (1.1)
59
94
99.0 (2.1)
170.9 (0.7)
33.9 (0.6)
108.2 (l.6)
111.2 (l.3)
56
50
94.7 (2.6)
168.9 (0.9)
33.2 (0.8)
108.3 (2.0)
111.4 (l.6)
47
0.69
<0.0001
0.0004
<0.0001
0.33
-
0.0006
0.02
<0.0001
<0.0001
0.13
-
Non-fasting lipids, n
Total cholesterol, mmol/L
HDL, mmol/L
LDL,§ mmol/L
Total:HDL ratio
Triglyceride,|| mmol/L
352
5.80 (0.49)
1.10 (0.14)
3.67 (0.44)
5.56 (0.78)
2.40 (l.06)
67
5.28 (0.13)
1.14 (0.04)
3.17 (0.12)
4.83 (0.21)
2.17 (0.28)
69
5.87 (0.13)
1.13 (0.04)
3.68 (0.13)
5.67 (0.21)
2.65 (0.28)
94
6.27(0.11)
1.09(0.03)
4.08(0.11)
5.98(0.18)
2.77(0.25)
79
6.08(0.13)
1.05(0.03)
3.94(0.12)
6.03(0.20)
2.51(0.27)
43
5.74(0.16)
1.07(0.04)
3.74(0.14)
5.65(0.25)
2.00(0.34)
0.002
0.09
0.0002
0.003
0.54
<0.0001
0.62
<0.0001
0.002
0.02
Non-fasting glucose, n
Glucose,|| mmol/L
402
6.37 (l.61)
81
4.96 (0.39)
84
5.36 (0.40)
105
6.42 (0.35)
88
7.60 (0.39)
44
8.47(0.50)<0.00010.07
Blood pressure, n
Systolic, mmHg
Diastolic, mmHg
414
136.2 (8.4)
86.8 (5.6)
86
126.4 (2.0)
80.9 (l.3)
87
130.0 (2.0)
84.2 (l.4)
110
134.1(1.8)
87.9 (l.2)
87
145.3 (2.0)
92.2 (l.4)
44
152.3 (2.6)
92.9 (l.8)
<0.0001
<0.0001
0.32
<0.04
Leisure time activity, ¶n
Sedentary, %
342
22
94
14
75
19
75
31
66
24
32
28--
Diabetes prevalence, n
Known diabetes, %
490
12
114
2
104
3
121
7
98
21
53
36--
Age standardised to Segi’s world population. †Model 1. Age was the independent continuous variable and variables for ethnic group and study were included in the model.
‡Model 2. Same as model 1 with the quadratric term (age squared) included. §Total n for LDL = 310 because of triglyceride levels > 4 mmol/L. ||The natural logs of
triglyceride and glucose were used in the models. ¶Leisure time activity data were not collected from 1995 onwards in the South Auckland Diabetes project.

Page 3

210New Zealand Medical Journal 11 May 2001
Associations between body size and lipids, glucose and blood pressure
were tested using analysis of variance. Multiple regression was used where
the variables were treated continuously. Triglyceride and glucose
variables were log transformed to improve the normality of the
distributions. Analyses were carried out in SAS version 6.10 (SAS
Institute, Cary, NC, USA).
Results
Participants in this CIS study were older (mean age 42 years
for men, 41 years for women) and more likely to be female
(60% of participants) than the general Pacific Islands
population (1996).18 Most were Samoan (64%) or Tongan
(26%) whereas approximately 50% of the general Pacific
Islands population identify themselves as Samoan and 14%
as Tongan. Compared to the overall distribution of
occupations in New Zealand (1991),19 the distribution of
occupations for Samoans in the SOFP project was shifted to
the lower income end (p<0.001).
Based on our results, 81% and 86% of Pacific Islands men
and women aged 15 years and over were either overweight (26
kg/m2 ≥ BMI < 32 kg/m2) or obese (BMI ≥ 32 kg/m2), with
44% of men and 58% of women being obese (note: these are
not age-standardised). We compared these percentages with
results from the National Nutrition Survey in Figures 1 and 2.
Compared to the NNS97, we observed a considerably lower
combined prevalence of overweight and obesity for young
Pacific males (87.6% vs 55.8%). On the other hand our
estimates of obesity were higher (44.8% vs 27.4%) for males
aged 25-44 years and overweight lower than for the NNS97.
Differences were also noted for females aged 25-44 years
(61.5% obese vs 34% obese; Figure 2).
Tables 1 and 2 present associations between age and CHD
and diabetes risk factors for men and women aged 20 years
and over. Pacific men had an age-standardised mean BMI of
32.6 kg/m2. Most risk factors were higher in the older age
groups, although mean weight, BMI, obesity prevalence, total
cholesterol, LDL, the total: HDL ratio, triglycerides and
sedentary leisure time activity were lower in the 60+ age
group than the younger two decade groups. Women (Table 2)
had a higher aged- standardised mean BMI (34.8 kg/m2) than
men and a higher prevalence of obesity (60% with a BMI ≥
32kg/m2). As with men, most risk factor levels were lower in
the 60+ age group than the younger age groups. Overall, 12%
of men and women reported having Type-2 diabetes and
approximately one-quarter did no leisure time physical
activity during a normal week.
A comparison of risk factors by ethnic group is given in
Table 3. Tongans had the highest mean BMI and the
highest prevalence of obesity. Their lipid profiles were
significantly more atherogenic than Samoan profiles. Also,
Tongan men had significantly higher mean glucose levels
and Tongan women were significantly more likely to do no
leisure time physical activity.
Table 4 shows relationship between BMI and waist
circumference with other CHD and diabetes risk factors.
For men, each risk factor, with the exception of HDL
(where the association was inverse), was higher at higher
quartiles of BMI. A similar, although less consistent pattern
was observed between these risk factors and quartiles of
waist circumference. Where BMI and waist circumference
were treated continuously, both were positively (p<0.05)
associated with each risk factor. There was an inverse
association with HDL. These models were re-run (not
shown) including squared terms for BMI or waist. None of
the squared terms was significant.
For women, the variation in these risk factors with
quartiles of BMI and waist was not as marked as it was for
men. Mean total and LDL cholesterol and glucose levels
differed little although mean systolic and diastolic blood
pressure was higher with each quartile of BMI. Using
continuous data, only the total:HDL cholesterol ratio, (log)
triglycerides and systolic and diastolic blood pressure were
positively associated with BMI. There was a significant
negative association between HDL and BMI. Total
cholesterol, the total:HDL cholesterol ratio, (log)
triglycerides, glucose, and diastolic blood pressure were all
positively associated with waist circumference in women.
When the squared terms for BMI and waist were included in
these models, negative associations (p<0.05 for BMI) were
observed for cholesterol, the total:HDL cholesterol ratio,
LDL and (log) triglyceride concentrations (not shown).
Discussion
These cross-sectional analyses of CHD and diabetes risk
factors came from church-based Pacific populations and are
not, therefore, a representative sample. However, the sample
size was large (total n = 1175) and the results are probably as
characteristic of New Zealand’s Pacific Islands population as
Figure 1. Overweight and obesity in three age groups as reported by
the Church Intervention Surveys (CIS) and the National Nutrition
Survey 1997 (NNS97): Pacific males.
Figure 2. Overweight and obesity in three age groups as reported by
the Church Intervention Surveys (CIS) and the National Nutrition
Survey 1997 (NNS97): Pacific females.
Males
15-24 yrs 25-44 yrs45+yrs
CIS
(n=113)
NNS97
(n=26)
CIS
(n=221)
NNS97
(n=48 )
CIS
(n=215)
NNS97
(n=31)
Percent
0
10
20
30
40
50
60
70
80
90
100
? = Obese ? = Overweight
Females
15-24 yrs25-44 yrs 45+ yrs
CIS
(n=157)
NNS97
(n=42)
CIS
(n=369)
NNS97
(n=93)
CIS
(n=275)
NNS97
(n=33)
Percent
0
10
20
30
40
50
60
70
80
90
100
? = Obese
? = Overweight

Page 4

New Zealand Medical Journal211 11 May 2001
the National Nutrition Survey which was smaller and had a
lower response rate. The data show that risk factors for
CHD and diabetes are very high, especially in the 40-60 year
age range and amongst the most obese.
The prevalence of obesity in Pacific populations has been
comprehensively reviewed by Hodge et al for populations
living in the Pacific Islands.20 The review indicated that
Pacific populations are amongst the most obese in the world.
Western Samoans, for example, had a higher prevalence of
obesity than native Hawaiians and Pima Indians.21,22 In this
study, we found that Pacific people living in New Zealand
have an even higher prevalence. Our observation that
women had a higher prevalence of obesity than men is
consistent with findings from studies in Western Samoa and
New Zealand. Schaaf et al observed no significant
differences in BMI between Cook Islands Maori, Samoans,
Table 2. Association between age and anthropometric, biochemical and blood pressure measurements, mean (SEM), in Pacific Islands women.
The prevalence of obesity, sedentary leisure time activity and known diabetes is also given.
Age Age group (years)
40-49
P for linear
term†
P for quadratic
term‡
Standardised* 20-2930-39 50-59
≥60
Anthropometry, n
Weight, kg
Height, cm
Body mass index, kg/m2
Waist, cm
Hip, cm
BMI > 32 kg/m2, %
708
91.2 (9.6)
161.5 (2.9)
34.8 (3.3)
102.1(6.9)
115.5 (6.9)
60
187
87.8 (l.6)
164.0 (0.5)
32.1 (0.5)
94.0 (1.1)
111.8 (1.1)
41
169
93.3 (l.6)
163.0 (0.5)
34.9 (0.6)
101.0 (1.2)
115.0 (1.2)
64
164
96.8 (l.7)
161.4 (0.5)
37.0 (0.6)
105.2 (1.2)
118.9 (1.2)
72
111
93.7 (l.9)
160.0 (0.6)
36.4 (0.7)
108.4 (1.4)
119.3 (1.4)
73
77
85.6 (2.3)
157.4 (0.7)
34.2 (0.8)
106.7 (1.7)
114.9 (1.7)
58
0.84
<0.01
<0.01
<0.01
<0.01
-
<0.01
0.06
<0.01
<0.01
<0.01
-
Non-fasting lipids, n
Total cholesterol, mmol/L
HDL, mmol/L
LDL,§ mmol/L
Total:HDL ratio
Triglyceride,|| mmol/L
548
5.46 (0.46)
1.22 (0.15)
3.47 (0.39)
4.70 (0.62)
1.77 (0.53)
130
4.84 (0.09)
1.26 (0.03)
2.96 (0.08)
4.05 (0.12)
1.49 (0.10)
120
5.09 (0.09)
1.19 (0.03)
3.20 (0.08)
4.48 (0.12)
1.63 (0.11)
148
5.44 (0.09)
1.20 (0.03)
3.48 (0.07)
4.70 (0.12)
1.79 (0.10)
94
6.04 (0.10)
1.23 (0.03)
3.83 (0.10)
5.16 (0.14)
2.04 (0.12)
56
6.33 (0.13)
1.19 (0.04)
4.21 (0.12)
5.50 (0.18)
2.08 (0.15)
<0.01
0.41
<0.01
<0.01
<0.01
0.05
0.55
0.06
0.15
0.02
Non-fasting glucose, n
Glucose,|| mmol/L
613
6.67 (2.25)
147
5.28 (0.41)
146
5.89 (0.41)
156
6.62 (0.41)
98
8.86 (0.49)
66
7.81 (0.59)<0.01 0.18
Blood pressure, n
Systolic, mmHg
Diastolic, mmHg
631
132.0 (8.9)
82.6 (5.8)
152
119.5 (l.6)
76.1 (1.0)
150
123.8 (l.6)
79.7 (l.1)
156
133.7 (l.6)
85.6 (l.1)
103
145.9 (l.9)
88.2 (l.2)
70
145.9 (2.3)
87.4 (l.5)
<0.01
<0.01
0.11
<0.01
Leisure time activity, ¶n
Sedentary, %
478
27
149
25
109
29
99
29
70
26
51
29--
Diabetes prevalence, n
Known diabetes, %
732
12
194
1
175
4
167
8
118
25
78
31--
*Age standardised to Segi’s world population. †Model 1. Age was the independent continuous variable and variables for ethnic group and study were included in the model.
Model 2. Same as model 1 with the quadratric term (age squared) included. §Total n for LDL = 514 because of triglyceride levels > 4 mmol/L. || The natural logs of
triglyceride and glucose were used in the models. ¶Leisure time activity data were not collected from 1995 onwards in the South Auckland Diabetes project
Table 3. Ethnic differences in anthropometric, biochemical and blood pressure measurements, mean (SEM). Differences in the prevalence of
obesity, sedentary leisure time activity and known diabetes are also given.
Men
Tongan
Women
Tongan SamoanOther PI* Samoan Other PI*
Anthropometry, n
Weight, kg
Height, cm
Body mass index, kg/m2
Waist, cm
Hip, cm
Obesity, % BMI ≥30 kg/m2
Obesity, % BMI >32 kg/m2
286
94.5(l.1)
171.4(0.4)
32.1(0.3)
103.1(0.8)
108.9(0.7)
63
46
140
101.9 (1.8)b
174.4 (0.6)c
33.5 (0.6)a
105.0(1.4)
111.5(l.1)
68a
49
41
96.7(2.8)
171.7(0.9)
32.7(0.8)
105.9(2.1)
112.2(l.7)
51
42
474
89.4(0.9)
160.9(0.3)
34.5(0.3)
101.0(0.6)
115.0(0.6)
73
60
167
95.9(1.7)b
162.8(0.5)b
36.1 (0.6)a
105.7 (1.3)b
117.4(l.3)
75a
59
67
80.0(2.4)
159.8(0.7)
34.2(0.8)
102.5(l.7)
115.6(l.7)
76
66
Non-fasting lipids, n
Total cholesterol, mmol/L
HDL, mmol./L
LDL,† mmol/L
Cholesterol/HDL ratio
log Triglyceride, mmol/L
218
5.58(0.07)
1.16(0.02)
3.52(0.07)
5.03(0.11)
2.16(0.15)
96
5.96 (0.12)a
1. 03 (0.03)b
3.85 (0.12)a
6.15 (0.19)c
2.82 (0.26)a
38
6.00 (0.17)a
1.09(0.04)
3.80(0.16)
5.72 (0.26)a
2.28(0.35)
374
5.44(0.05)
1.31(0.01)
3.37(0.05)
4.38(0.07)
1.69(0.06)
120
5.61(0.10)
1. 19 (0.03)b
3.62 (0.09)a
4.89 (0.13)b
1.82(0.11)
54
5.59(0.13)
1. 14 (0.04)c
3.62(0.12)
5.06 (0.17)b
1.90(0.14)
Non-fasting gucose, n
Glucose, mmol/L
232
6.37(0.22)
130
7.29 (0.3 5)a
40
6.04(0.53)
406
6.75(0.23)
148
6.79(0.43)
59
7.01(0.59)
Blood pressure, n
Systolic, mmHg
Diastolic, mmHg
242
134.6(l.1)
85.7(0.8)
131
137.2(l.8)
86.2(1.2)
41
141.0 (2.7)a
90.9 (1.8)a
420
131.4(0.9)
82.4(0.6)
149
132.4(l.7)
81.5(l.1)
62
137.5 (2.3)a
8 6.3 (1.5)a
Leisure time acitivity,‡ n
Sedentary, %
231
24
96
11
15
60b
347
25
106
36b
25
24
Diabetes prevalence, n
Known diabetes, %
302
11
145
11
43
12
494
10
171
9
67
13
*Other Pacific Islands (PI) ethnic groups, Cook Islands Maori (n=51), Niuean (n=20), and mixed Pacific Islands ethnic group (n=39). †Total n for LDL = 824 due to
triglyceride levels > 4 mmol/L. ‡Leisure time activity data were not collected from 1995 onwards in the South Auckland Diabetes Project. a,b,c Significantly different from
Samoan ethnic group at ap<0.05, bp<0.01, cp<0.0001 adjusted for age group and study.

Page 5

212 New Zealand Medical Journal11 May 2001
Tongans and Niueans.11 We found that Tongan men and
women were bigger than their counterparts from other
islands.
The lipid profiles of Pacific men and women in the
current study were less atherogenic than those reported in
the National Nutrition Survey,13 and the Workforce
Diabetes Survey.11 The attenuation of risk factor levels in
older Pacific people has previously been described23,24 and
probably reflects selective mortality of high risk individuals
or the cohort effect of a relatively lower risk group of
individuals now reaching older age.2 The age-standardised
prevalence of known diabetes (12%) was similar in these
church communities to the prevalence observed in a
household survey of inner urban South Auckland.25 Systolic
and diastolic blood pressure levels were comparable to those
observed in Pacific members of a Seventh-Day Adventist
church.24 Also, similar increases in blood pressure with age
have been observed in Pacific Islands people in the
workforce.3,26 The number of people who were sedentary
during leisure-time was high in these communities. Previous
New Zealand studies found that Pacific people were less
involved in leisure-time activities than Maori or European
and that Pacific women were less active than men.5
Body size was adversely associated with other CHD and
diabetes risk factors. However, the associations were not as
strong as those observed for European New Zealanders,24 and
other less obese populations.27 Moreover, for Pacific women,
there was evidence of attenuation between BMI and blood
cholesterol and triglycerides at the upper end of the BMI
distribution. Similar findings have been observed in studies of
Samoans in Western Samoa and American Samoa, and
Micronesian Nauruans.28,29 The poor associations observed
were attributed to extreme obesity in these populations. This
suggests that these populations have reached a level of obesity
above which the impact of total fat or intra-abdominal fat on
CHD and diabetes risk factors becomes less apparent.
There are a number of limitations to this study. As mentioned
above, neither the church communities nor the participants were
randomly selected and therefore the results may not readily be
generalised to New Zealand’s wider Pacific population. Numbers
in the studies were lower than they could have been for several
variables because providing a blood sample was voluntary and up
to 30% declined. Finally, combining the results of two separately
conducted studies is not ideal methodology although we tried to
overcome this limitation by adjusting for risk factor differences
between the projects.
The SADP and the SOFP were both designed to reduce
CHD and diabetes risk factors in these church communities
and both had some success.30,31 The high prevalence of
obesity and other risk factors at baseline suggests that these
interventions were not only warranted, but long overdue in
the effort to bring the health status of these communities in
line with that of other New Zealanders. Future efforts to
contain the rising prevalence of obesity in New Zealand
need to give priority to Pacific People.
Acknowledgements. The Samoan Ola Fa’autauta Project was supported by
the Health Research Council of New Zealand and the National Heart
Foundation of New Zealand. Material support for the Pacific Islands Church
program of the South Auckland Diabetes Project was provided by the
Lotteries Board, North Health, South Auckland Health, Boehringer
Mannheim / Roche Diagnostics, ASB Trust, Novo Nordisk, Eli Lilly, Tegal,
New Zealand Dairy Board and Sanitarium. Colin Bell was supported by a
National Heart Foundation Postgraduate Scholarship. We thank the pastors
and their wives for their tremendous contribution to these studies and the
congregations of the churches for their participation.
Correspondence. Boyd Swinburn, School of Health Sciences, Deakin
University, 221 Burwood Hwy, Victoria 3125, Australia. Fax (+61) (3) 9244
6017. Email: swinburn@deakin.edu.au
1.Simmons D, Gatland B, Flemming C, Scragg R. Prevalence of known diabetes in a
multiethnic community. NZ Med J 1994; 107: 219-22.
Bell AC, Swinburn BA, Stewart AW et al. Ethnic differences and recent trends in coronary
heart disease incidence in New Zealand. NZ Med J 1996; 109: 66-8.
2.
Table 4. Association between non-fasting blood lipids, blood glucose and blood pressure with quartiles of body mass index (BMI) and waist for
Pacific Islands men and women. Multivariate regression co-efficients (β) and the variation explained by each model (R2) are also included.
Cholesterol
mmol/L
Chol:HDL
Ratio
HDL
mmol/L
LDL
mmol/L
Triglyceride
mmol/L*
Glucose
mmol/L*
Systolic BP
mm Hg
Diastolic BP
mm Hg
Men: BMI, n
Quartile (mean kg/m2)
1(26.1)
2(31.3)
3(35.0)
4(41.2)
Linear, β (R2 %)
348345 346 307348397411411
5.56
5.82
5.961
6. 191,2
0.04‡ (3.2)
4.92
5.621
5.851
6.381
0.10‡ (5.3)
1.17
1.11
1.051
1.021
-0.01‡(1.9)
3.60
3.67
3.84
3.95
0.02† (1.2)
1.87
2.49
2.60
3.101
0. 04† (8.8)
5.74
6.41
6.39
7.141
0.01‡ (2.5)
128.8
135.51
137.21
142. 81,2
0. 83‡ (8.3)
81.0
86.11
88.51
94.11,2,3
0.81‡ (2.5)
Men: Waist, n
Quartile (mean, cm)
1(84.90)
2(97.10)
3(106.3)
4(120.5)
Linear, β (R2 %)
351348 349309 351400 414414
5.51
5.71
6.021
6.111
0.02‡ (2.1)
4.73
5.31
5.951
6.191
0.04‡ (5.3)
1.21
1.13
1.051
1.041
-0.004‡ (2.0)
3.53
3.70
3.84
3.83
0.01† (1.3)
1.44
2.441
2.671
3.011
0.02‡ (5.0)
5.76
5.75
6.75
6.86
0.005‡ (1.8)
129.9
132.1
138.81,2
139.51,2
0.30‡ (7.0)
81.2
84.0
88.41,2
91.81,2
0.30‡ (1.9)
Women: BMI, n
Quartile (mean kg/m2)
1(26.2)
2(31.3)
3(35.1)
4(42.5)
Linear, β (R2 %)
544541 543516 544608627 627
5.21
5.48
5.47
5.44
0.01(0.3)
4.21
4.731
4.741
4.801
0.02† (1. 3)
1.30
1.22
1.201
1. 181
-0.005† (1.4)
3.28
3.50
3.45
3.43
0.005(0.1)
1.43
1.76
1.851
1.881
0.01‡ (3.8)
6.39
6.62
6.66
6.69
0.004(3.2)
127.6
128.6
130.6
136.61, 2,3
0.53‡ (7.7)
78.3
80.5
82.21
86.51,2,3
0.49‡ (0.8)
Women: Waist, n
Quartile (mean cm)
1(82.60)
2(96.70)
3(105.9)
4(119.7)
Linear, β (R2 %)
540 537539506540 602621621
5.23
5.44
5.551
5.44
0.01†(0.8)
4.21
4.711
4.941
4.781
0.02‡ (3.4)
1.31
1.191
1.161
1.191
-0.004‡ (2.8)
3.3
3.45
3.51
3.44
0.004 (0.4)
1.33
1.851
2.011
1.831
0.01‡(6.0)
5.55
7.241
6.47
7.151
0.004‡ (2.4)
126.9
129.0
134.11
135.51
0.25‡ (4.6)
78.8
81.7
84.01
85.41
0.18‡ (2.2)
*The natural logs of triglyceride and glucose were used for the regression models. 1,2,3Significantly different from quartile 1,2 or 3 respectively at p<0.01 adjusted for age,
ethnic group and study. †‡Significant association with BMI or waist circumference at †p <0.05 or ‡p <0.001 adjusted for age, ethnic group and study.