American Journal of Epidemiology
Copyright O 1996 by The Johns Hopkins University School of Hygiene and Public Hearth
ATI rights reserved
Vol. 144, No. 9
Printed In U.S.A
Effect of Migration and Related Environmental Changes on Serum Lipid
Levels in Southwestern Chinese Men
Jiang He,1'2 Michael J. Klag,1 Zhenglai Wu, 2 Ming-Chu Qian,3 Jun-Yun Chen,3 Pei-Sheng Mo,4
Guan-Qing He,2* and Paul K. Whelton1
To determine migration-related differences in serum lipid levels and to estimate the contribution of environmental factors to these differences, the authors conducted a cross-sectional study among 1,222 residents
of southwestern China in 1988 and 1989: 521 Yi fanners living in a remote rural area, 340 Yi migrants to an
urban setting, and 361 Han people living in the same urban setting. Serum total and high density lipoprotein
cholesterol and serum triglyceride were measured after a 14-hour fast. Information on age, cigarette smoking,
alcohol consumption, and physical activity was obtained by questionnaire. Diet was assessed by use of three
24-hour diet recalls. The Yi farmers had the lowest age-adjusted serum lipid levels among the three groups,
while the Yi migrants and their Han counterparts had similar lipid levels. After adjustment for age, body mass
index, cigarette smoking, alcohol drinking, and dietary nutrient intake, the difference in serum lipid profile
between the Yi migrants and the Yi farmers was substantially reduced. These results indicate that rural-urban
migration is associated with higher serum lipid levels. This migration-related difference in serum lipids can be
partially explained by changes in lifestyle and dietary nutrient intake. The authors conclude that environmental
factors are important determinants of population serum lipid levels in southwestern Chinese. Am J Epidemiol
1996;144:839-48.
alcohol drinking; cholesterol; diet; life style; lipoproteins; smoking; transients and migrants
Mortality from cardiovascular disease is the leading
cause of death in both industrialized nations, like the
United States, and many economically developing
countries (1). A high blood cholesterol level is an
important risk factor for cardiovascular disease in observational studies, and lowering serum cholesterol
reduces the risk of coronary heart disease (2-5). According to the third National Health and Nutrition
Examination Survey, conducted in 1988-1991, 20
percent of adults in the United States have a high
blood cholesterol level (>240 mg/dl), and 31 percent
have a borderline-high level (200-239 mg/dl) (6).
In contrast, cardiovascular disease is virtually absent
in isolated populations who live in remote areas and
maintain a traditional lifestyle. Studies carried out in
such populations have demonstrated that the prevalence of cardiovascular disease risk factors, such as
high blood pressure and overweight, is much lower
among these persons than in urbanized populations
(7-10). Migrants from remote areas to more urban
locations tend to develop higher blood pressures and a
higher prevalence of hypertension and overweight
(11-14). However, little information exists on the effect of rural-urban migration on serum lipid values.
Studies conducted in southwestern China have provided researchers with a unique opportunity to explore
the basis for differences in serum lipid levels between
an unacculturated group (the Yi people) living in their
native environment, migrants (Yi) to an urbanized
environment, and urban dwellers (Han people) with a
different ethnic background. The purposes of the
present study were 1) to determine whether migration
is associated with differences in serum lipid levels and
2) to estimate the contribution of environmental factors to these differences.
Received for publication December 13, 1995, and in final form
April 19, 1996.
Abbreviations: HDL, high density lipoprotein; LDL, low density
lipoprotein.
1
The Welch Center for Prevention, Epidemiology, and Clinical
Research, The Johns Hopkins Medical Institutions, Baltimore, MD.
2
Department of Epidemiology, Peking Union Medical College
and Chinese Academy of Medical Sciences, Beijing, People's Republic of China.
3
Uangshan Yi People Autonomous Prefecture Arrti-Epldemic
Station, Xichang City, Sichuan Province, People's Republic of
China.
4
National Center for Clinical Laboratories, Ministry of Public
Health, Beijing, People's Republic of China.
•Deceased.
Reprint requests to Dr. Jiang He, The Welch Center for Prevention, Epidemiology, and Clinical Research, 2024 East Monument
Street, Suite 2-600, Baltimore, MD 21205-2223.
839
840
He et al.
MATERIALS AND METHODS
Study participants
Details on the study population and our study methods have been published previously (13, 14). Briefly,
the study sample consisted of 521 Yi farmers, 340 Yi
migrants, and 361 Han people who were randomly
selected from the community as part of the Yi People
Study and the Yi Migrant Study. Because local traditions prevented collection of urine specimens from
women, only men were included in this study.
The Yi people are an ethnic minority in China who
are distinct from the majority Han people and are
thought to have originated in Mongolia. They live
principally in a remote mountainous area in southwestern China and are mainly engaged in subsistence agriculture. Yi farmers are relatively isolated from the
outside world and have preserved their own language
and lifestyle. Traditionally, their main crops are potatoes, rice, corn, oats, and buckwheat. Consumption of
meat is limited to weddings, funerals, and semiannual
celebrations. Animal fat and vegetable oil are not used
for daily cooking. According to a review of medical
records in local hospitals, hypertension, coronary heart
disease, stroke, and diabetes are rare among Yi farmers (15). The main causes of death in Yi farmers are
chronic obstructive lung disease, malignant tumors,
pneumonia, pulmonary tuberculosis, and injuries, including suicide (15).
Starting in the 1950s, Yi farmers began to migrate to
Xichang City, the capital of Liangshan Yi People
Autonomous Prefecture, and to the county seats of
counties in the prefecture. The traditional residents of
Xichang City and the county seats are the Han people.
Yi migrants and Han residents live in the same communities and consume the same dietary staples: rice,
meat, and fresh vegetables.
Data collection
The study was conducted during the spring of 1988
and 1989. Blood samples were obtained using the
same protocol in all participants. Participants were
instructed to fast for 14 hours prior to their visit. Blood
was drawn into a Vacutainer (Becton Dickinson, Rutherford, New Jersey) and left at room temperature for
30 minutes to allow the blood to clot. Serum was
separated in the field using a battery-powered centrifuge and was sent to the county hospital laboratory on
the same day. Specimens were stored at — 20°C until
they were airmailed to the National Center for Clinical
Laboratories in Beijing. Serum total cholesterol, high
density lipoprotein (HDL) cholesterol, and triglyceride
levels were measured using an enzymatic method (16,
17) (TL Monarch 2000; Instrumentation Laboratory,
Lexington, Massachusetts). HDL cholesterol was separated from serum by precipitation after the addition of
phosphotungstic acid and magnesium ion (18). These
analytic procedures were standardized and met the
performance requirements of the Lipoprotein Standardization Program of the Centers for Disease Control and Prevention (Atlanta, Georgia). Low density
lipoprotein (LDL) cholesterol was estimated by application of the formula devised by Friedewald et al. (19).
Blood pressure was measured on 3 consecutive
days. Before the blood pressure measurement, participants refrained from strenuous activity, smoking, and
eating for at least 30 minutes. After a minimum of 5
minutes of quiet sitting, seated blood pressure was
measured on the right arm and recorded to the nearest
2 mmHg by specially trained physicians using standard mercury sphygmomanometers. Systolic pressure
was recorded at the appearance (phase 1) and diastolic
pressure at the disappearance (phase 5) of Korotkoff
sounds. Three blood pressure measurements were
made each day for a period of 3 days. For each
participant, the mean value of the nine blood pressure
readings obtained over the 3-day period was reported.
Body height and weight were measured, and body
mass index (weight (kg)/height (m)2) was calculated
as an index of obesity (20).
Age, race, sex, educational level, physical activity,
cigarette smoking, and alcohol use were ascertained
by local physicians who were fluent in both the Yi and
Chinese languages. Three consecutive 24-hour diet
recall interviews were conducted by trained interviewers using food models. Intakes of total energy, protein,
total fat, saturated fat, monounsaturated fat, polyunsaturated fat, cholesterol, and total fiber were calculated using the Sichuan Province section of the Chinese food composition tables (21). Nutrient intakes
were reported as both absolute values and nutrient
densities. The latter measure was expressed as the
percentage of total calories for all nutrients, except
cholesterol and fiber, which were expressed as mg/
1,000 kcal and g/1,000 kcal, respectively. As a measure of the combined effect of dietary saturated and
polyunsaturated fats and cholesterol on serum lipids,
the Keys score was calculated as 1.35 (2S — P) +
1.5Z, where S is the percentage of dietary calories for
saturated fat, P is the percentage of dietary calories for
polyunsaturated fat, and Z is the square root of dietary
cholesterol (mg/1,000 kcal) (22). Assessment of usual
physical activity was based on occupational activity,
because leisure-time activity was almost nonexistent.
Am J Epidemiol
Vol. 144, No. 9, 1996
Migration and Lipids
Statistical analysis
Group differences in age-adjusted mean serum lipid
and dietary nutrient levels were examined by analysis
of covariance. The prevalence of hyperlipidemia,
smoking, and alcohol drinking were standardized using the age distribution of the Liangshan Yi People
Autonomous Prefecture population in 1982. Univariate and multivariate linear regression analysis, adjusted for age and body mass index, was used to
explore the relation between serum lipids and cigarette
smoking, alcohol consumption, and dietary nutrients
by group. Data from the Yi migrants and Han people
were pooled in this analysis because they had a similar
serum lipid profile, as well as similar levels of independent variables. Furthermore, a final multiple linear
regression model for each lipid component was established using data from all participants. All possible
interactions between area of residence (urban residents
vs. farmers) and dietary nutrients and other covariates
were tested. Significant interaction terms were included in the final models. For selection of a final
model, a series of nested models were built and compared with a full model containing all of the variables
of potential interest. The model with the fewest covariates but similar predictive power was selected as
the "best" model. Because of collinearity between
Keys scores and dietary total fat (r = 0.89), saturated
fat (r = 0.94), and polyunsaturated: saturated fat ratio
(r = 0.85), they were not included in the same model
simultaneously.
The contribution of differences in dietary nutrients
and other covariates to the impact of migration on
serum lipids was estimated by adjusting the difference
in serum lipids between Yi migrants and Yi farmers
for these variables. The Han city dwellers were excluded from this analysis. Only covariates which were
selected in the final models were used for the adjustment. Because physical activity was strongly related to
migration status, it was not included as an adjustment
variable. In addition, the contribution of dietary nutrients and other factors to the differences in serum lipids
between the Yi migrants and Han people was examined in a multiple Linear regression model. Yi farmers
were excluded from this analysis. All analyses were
performed using the SAS statistical analysis package
(23).
RESULTS
The general characteristics of the three study populations are shown in table 1. On average, the Yi
migrants were 6 years older than the Yi farmers, but
nearly 5 years younger than the Han people. Because
of the substantial difference in age between the three
TABLE 1. Characteristics of study participants, by study group: Uangshan, People's Republic of China,
1988-1989*
YI farmers
(n-521)
Continuous variables (mean ±
Age (years)
Body mass Index}
Systolic Wood pressure (mmHg)
Dlastolic blood pressure (mmHg)
Total cholesterol (mg/dl)
HDLt cholesterol (mg/dl)
HDL:total cholesterol ratio
LDLt cholesterol (mg/dl)
TrlgJycerkle (mg/dl)
31.7 ±0.6
20.6 ± 0.1
105.2 ± 0.6
65.7 ± 0.4
140.2 ± 1.7
54.4 ± 0.8
0.40 ± 0.01
61.2 ±1.6
135.9 ± 4.7
YI mJQrante
(n = 340)
Han people
(n-381)
set;
37.8 ± 0.7
21.2 ±0.1
112.4 ±0.7
70.8 ± 0.5
178.4 ±2.0
60.4 ± 0.9
0.36 ± 0.01
88.5 ± 1.9
152.4 ±5.7
42.4 ± 0.7
21.2 ±0.1
112.4 ±0.7
70.4 ± 0.5
171.7 ±2.0
56.7 ± 0.9
0.34 ± 0.01
87.7 ± 1.9
143.8 ± 5.7
Categorical variables (%)
Physical activtty
Light
Moderate
Heavy
Alcohol use
Cigarette smoking
High total cholesterol (£240 mg/dl)
Low HDL cholesterol (<35 mg/dl)
High LDL cholesterol (£160 mg/dl)
0.0
1.6
98.4
73.3
21.1
0.5
8.9
0.5
61.4
31.7
6.9
63.5
69.3
10.4
1.7
7.1
* AD estimates were age-adjusted, except for age.
t SE, standard error; HDL, high density llpoprotein; LDL, low density llpoproteln.
* Weight (kg)/helght (m)*.
Am J Epidemiol
Vol. 144, No. 9, 1996
841
56.8
35.8
7.4
47.1
71.5
3.9
6.6
2.7
842
He et al.
groups, all of the analyses were adjusted for age.
Compared with the Yi farmers, the Yi migrants had a
higher age-adjusted mean body mass index and blood
pressure, a lower level of physical activity, a higher
prevalence of cigarette smoking, and a lower prevalence of alcohol consumption. The Han people had a
similar level of physical activity and prevalence of
cigarette smoking as the Yi migrants but a lower
prevalence of alcohol consumption.
Compared with the Yi farmers, the Yi migrants had
higher age-adjusted mean serum total, HDL, and LDL
cholesterol and triglyceride levels and a lower ratio of
HDL cholesterol to total cholesterol (table 1). The Yi
migrants also had higher levels of mean serum total
and HDL cholesterol than the Han people but similar
mean levels of serum LDL cholesterol and triglyceride. The rise in total and LDL cholesterol with age was
similar and statistically significant in all three groups
(0.71, 0.84, and 0.90 mg/dl per year for total cholesterol and 0.40, 0.60, and 0.47 mg/dl per year for LDL
220
HDL Cholesterol (mg/dl)
Total Cholesterol (mg/dl)
100
15-24
25-34
35-44
cholesterol in the farmers, Yi migrants, and Han people, respectively). A significant age-related rise in
HDL cholesterol was noted in the Yi fanners up to age
45, but this was not the case for the other two groups
(figure 1). Mean serum triglyceride levels rose significantly with age in the Yi migrants and Han people but
not in the Yi farmers. In every age category, the Yi
migrants had significantly higher total and LDL cholesterol levels than the Yi fanners. The Yi migrants
also had a significantly higher mean HDL cholesterol
level than the Yi farmers before age 45 years and a
significantly higher mean triglyceride level after age
34 years.
The prevalences of high total cholesterol (S240
mg/dl), high LDL cholesterol (>160 mg/dl), and low
HDL cholesterol (<35 mg/dl) levels were 0.4 percent,
0.4 percent, and 9.6 percent for the Yi farmers; 10.9
percent, 8.0 percent, and 2.1 percent for the Yi migrants; and 5.3 percent, 3.4 percent, and 6.7 percent
for the Han people, respectively. After adjustment for
45-54
55 +
25-34
Age (years)
Yi Farmers
Yi Migrants
25-34
35-44
Age (years)
45-54
55 +
Han People
Triglyceride (mg/dl)
LDL Cholesterol (mg/dl)
40
15-24
35-44
Age (years)
45-54
55 +
25-34
35-44
45-54
55 +
Age (years)
FIGURE 1. Mean levels of serum lipids among Yi farmers, Yi migrants, and Han people, by age: Uangshan, People's Republic of China,
198&-1989. HDL, high density lipoprotein; LDL, low density llpoprotein.
Am J Epidemiol
Vol. 144, No. 9, 1996
Migration and LJpids
age, the Yi migrants still had a higher prevalence of
high total cholesterol and LDL cholesterol levels and
a lower prevalence of low HDL cholesterol levels than
the Yi farmers (table 1).
The Yi farmers followed a diet that was low in total
fat and cholesterol and high in carbohydrate, polyunsaturated fat, and fiber content (table 2). The Yi migrants and the Han people consumed more total fat and
cholesterol but less carbohydrate, polyunsaturated fat,
and fiber than the Yi farmers. The age-adjusted mean
Keys score was much lower in the Yi farmers than in
the Yi migrants and Han people.
Relations between environmental factors and
serum llpids
Univariate and age- and body mass index-adjusted
regression coefficients are presented by group in table
3. While cigarette smoking was associated with a
higher serum total cholesterol level in the urban residents, it was associated with lower total cholesterol in
the Yi farmers. However, this inverse association disappeared after adjustment for the effects of age and
body mass index. Alcohol consumption, dietary saturated fat, cholesterol intake, and Keys score were
associated with higher serum total cholesterol in both
groups. The polyunsaturated: saturated fat ratio was
associated with lower serum total cholesterol in both
groups, while polyunsaturated fat was associated with
lower serum total cholesterol only in the Yi farmers.
Alcohol consumption was consistently associated
with a higher HDL cholesterol level in both the Yi
farmers and the urban residents (table 3). Dietary total
fat, saturated and polyunsaturated fats, cholesterol,
and Keys score were positively associated with HDL
cholesterol in the Yi farmers, while the polyunsaturated : saturated fat ratio was negatively associated with
HDL cholesterol in the urban residents.
Dietary polyunsaturated fat was associated with a
lower LDL cholesterol level in the Yi farmers, while
dietary total fat, saturated fat, and Keys score were
associated with a higher LDL cholesterol level in the
urban residents. Alcohol consumption was positively
associated with triglyceride levels in both groups. In
addition, polyunsaturated fat was negatively associated and Keys score was positively associated with
LDL cholesterol in the Yi farmers.
The best predictive models for serum lipid levels are
shown in table 4. All covariates in the final model had
p values less than 0.1. Age, body mass index, cigarette
smoking, alcohol consumption, and Keys score were
all significantly and positively associated with serum
total cholesterol (p < 0.05). Together, these covariates explained about 31.8 percent of the variation in
total cholesterol. Age, alcohol consumption, and dietary fiber were significantly and positively associated
TABLE 2. Ag*-ad]usted mean daily Intakes of various nutrients, by study group: Uangshsn, People's
Republic of China, 1988-1989
Energy (kcal/day)
Carbohydrate (g/day)
%ofkcal
Alcohol (g/day)
%ofkcal
Protein (g/day)
%ofkcal
Total (at (g/day)
%ofkcal
Polyunsaturated fat (g/day)
%ofkcal
Monounsaturated fat (g/day)
%ofkcal
Saturated fat (g/day)
%ofkcal
Polyunsaturated:saturated fat ratio
Cholesterol (mg/day)
mg/1,000kcal
Keysscoret
Dietary fiber (g/day)
mg/i,000kcal
Yl termers
(n = 521)
3,400 ± 72*
566 ±7.7
68.3
35.5 ± 2.0
5.9
103.9 ± 5.2
12.3
52.4 ± 3.5
13.5
17.7 ±1.6
4.6
18.3 ±1.3
4.7
12.2 ± 0.9
3.1
2.23 ± 0.03
50.8 ± 10.2
15.6
5.7 ± 0.6
19.4 ±0.3
5.8
YI migrants
(n = 340)
Han people
(n = 361)
2,946 ± 8 6
349 ±9.3
49.8
30.8 ± 2.5
6.8
84.2 ± 6.3
10.9
109.9 ±4.2
32.5
18.9 ±2.0
5.1
53.0 ±1.6
16.0
34.6 ±1.1
10.4
0.60 ± 0.04
313.7 ±12.3
114.2
35.8 ± 0.7
6.1 ± 0.4
2.1
2,851 ± 86
350 ±5.3
51.1
12.8 ±2.5
2.7
82.8 ± 6.3
11.2
114.1 ±4.2
35.0
19.6 ± 2.0
5.6
56.9 ±1.6
17.7
34.1 ± 1.1
10.6
0.58 ± 0.04
343.6 ± 12.3
131.1
37.4 ± 0.7
6.2 ± 0.4
£1
* Mean ± standard error.
t Keys score: 1.35 (2S - P) + 1.5Z, where S Is % of kcal for saturated fat, P is % of kcal for polyunsaturated
fat, and Z l s the square root of dietary cholesterol (mg/100 kcal).
Am J Epidemiol
Vol. 144, No. 9, 1996
843
844
He et a).
TABLE 3. Coefficients from linear regression of serum lipids on dietary nutrient Intakes and other lifestyle factors In 1,222 men:
Uangshan, People's Republic of China, 1988-1989
Yifarmers(n= 521)
Untwartate
Urban residents (n = 701)
MutUvartatet
P
SE*
-0.646
0.617
0.141
0.626
-1.887
-2.884
0.068
0.259
-0.784
0290*
0.163*"
0.166
0.394
0.713**
1.306*
0.045*
0.107*
0.744
P
SE
Untvartate
Mutttvariatet
P
SE
0380
0.730
0.045
0265
-1.098
-8.829
0.030
0233
-1.329
0.163*
0.198"*
0.147
0.379
0395
4.437*
0.014*
0.116*
1.672
0.438
0.750
0246
0.808
-0.564
-12.515
0.019
0.315
-2.566
0.157**
0.186"*
0.143
0365*
0.853
4209"
0.014
0.111"
1.596
-0.007
0.321
-0.083
0.019
-0.711
-3.254
-0.003
0.014
0228
0.066
0.060"*
0.059
0.153
0.359*
1.785
0.006
0.047
0.672
0.014
0.309
-0.075
0.041
-0.651
-3.555
-0.002
0.026
0.271
0.068
0.080*"
0.061
0.156
0.363
1301*
0.006
0.047
0.682
0.324
0.038
0215
0.460
0.251
-3.465
0.024
0215
-1.567
0.152*
0.187
0.137
0.353
0.829
4.117
0.013
0.108*
1.558
0.284
0.068
0.347
0.847
0.505
-6.339
0.015
0.273
-2.546
0.150
0.179
0.135*
0.347*
0323
3.982
0.013
0.105"
1314
0.438
0.524***
0389
0.999
2.382
11.709
0.038
0.306
4.420
0.693
1.597
0.145
0.587
-1.319
-15.419
0.023
0.286
-0.381
0.426
0.501"
0382
0.976
2273
11252
0.037
0296
4269
P
SE
Serum total cholesterol (mgAS)
SmoMng (ctgarettes/day)
Alcohol (% of teal)
Total tat (% of kcal)
Saturated tat (% of kcal)
Polyunsaturated tat (% of kcal)
R S * ratio
Cholesterol (mg/1,000 kcal)
Keys score§
Dtetary fiber (g/1,000 kg)
-0.114
0.524
0218
0359
-2.118
-3.965
0.097
0.330
-0.762
0294
0.158*"
0.160
0.380*
0.687"
1259"
0.044*
0.103"
0.715
Serum HDLt cholesterol (mgHl)
SmoMng (cigarettes/day)
Alcohol (% of teal)
Total tat (% of teal)
Saturated tat (% of kcal)
Polyunsaturated fat (% of kcal)
RS ratio
Cholesterol (mg/1,000 kcal)
Keys score
Dietary fiber (g/1,000 kg)
-0.151
0.289
0.297
0.495
1.179
-0353
0.071
0.138
0.360
0.148
0.084***
0.084***
0.200*
0.364**
0.672
0.O23**
0.055*
0.380
SmoMng (cigarettes/day)
Alcohol (% of teal)
Total tat (% of kcal)
Saturated fat (% of teal)
Poryunsaturatad fat (% of kcal)
P:S ratio
Cholesterol (mg/1,000 teal)
Keys score
Dtetary fiber (g/1,000 kg)
-0.534
0.090
-0.287
-0.268
-2.738
-0.625
-0.021
-0.007
-0.455
0568*
0.158
0.153
0367
0.663"*
1237
0.042
0.100
0.697
-0.012
0251
0.304
0.529
1.054
-1.090
0.071
0.150
0.42O
0.153
0.083*
0.083*
0.198*
0.360*
0.665
0.022*
0.054*
0.373
Serum LDL% cholesterol (mg/<f)
-0217
0.045
-0222
-0.090
-2.779
-1.397
-1.012
0.043
-0.513
0.279
0.156
0.153
0.365
0.657*"
1232
0.042
0.100
0.688
Servm trigtyceride (mg/O)
SmoMng (cigarettes/day)
Alcohol (% of teal)
Total tat (% of kcal)
Saturated tat (% of teal)
Polyunsaturated fat (% of kcal)
P:S ratio
Cholesterol (mg/1,000 kcal)
Keys score
Dietary ffoer (g/1,000 kg)
-0.363
1.772
0.730
2.132
-1.402
-9.067
0.127
0.660
-3.821
0.826
0.465***
0.471
1.122
2.045
3.716*
0.130
0.308*
2.114
-0.399
1.624
0.620
1.945
-2.065
-8.903
0.110
0.622
-3.306
0.881
0.465*"
0.471
1.121
2.037
3.720*
0.129
0.305*
2.102
0257
1338
-0.241
-0.334
-2.542
-11.791
0.035
0.164
0272
• p < 0.05; " p < 0.01; • • • p < 0.001.
t Adjusted for age and body mass Index
t SE, standard error; P:S, polyunsaturated tatsaturated fat; HDL, high density tpoprotaln; LDL, low density Itpoprotein.
§ Keys score: 1.35 (2S - P) + 1 .SZ, where S is % of kcal for saturated fat, P is % of kcal for polyunsaturated tat, and Z Is the square root of dietary cholesterol
(mg/100 kcal).
with HDL cholesterol. Dietary total fat was positively associated with HDL cholesterol in the Yi
farmers but negatively associated in the urban residents (p value for interaction < 0.001). However,
these covariates explained only about 8 percent of
the variation in HDL cholesterol. Age, body mass
index, and dietary total fat were positively associated with LDL cholesterol, while dietary polyunsaturated fat was negatively associated with LDL
cholesterol. Age, body mass index, alcohol consumption, and Keys score were all positively associated with serum triglyceride.
Contribution of environmental factors to
differences in serum lipid levels between Yi
migrants and Yi farmers
Adjusted differences in mean serum lipid levels
between the Yi migrants and the Yi farmers are shown
in table 5. Adjustment for each covariate, except for
dietary polyunsaturated fat, reduced the difference in
serum total cholesterol between the migrants and the
farmers. After adjustment for all covariates in the final
model, the difference in total cholesterol between the
Yi migrants and the Yi farmers was reduced by 17.3
Am J Epidemiol
Vol. 144, No. 9, 1996
Migration and LJpids
i : .
•*fr •*• m t p
•s
(i
i
a
3
•
*
•
If ^
1*121
cg^-oooooo
§
845
mg/dl (45.1 percent), but total cholesterol was still
21.1 mg/dl higher in the Yi migrants than in Yi farmers. With the exception of adjustment for dietary total
fat, which made the difference in HDL cholesterol
between the migrants and the farmers nonsignificant,
adjustment for body mass index, alcohol consumption,
and dietary fiber had little effect on the difference in
HDL cholesterol between the Yi migrants and the Yi
farmers. After adjustment for body mass index, cigarette smoking, dietary total fat, and polyunsaturated
fat, the difference in LDL cholesterol between the
migrants and the farmers was reduced by 9.4 mg/dl
(34.4 percent), but it was still 17.9 mg/dl greater in the
Yi migrants than in the Yi farmers. Differences in
body mass index, alcohol consumption, and Keys
score explained all of difference in serum triglyceride
between migrants and farmers.
Because migration was strongly associated with
physical activity (97.5 percent of the Yi farmers reported heavy physical activity as compared with only
7.4 percent of the Yi migrants), physical activity was
not included in the multivariate models. However,
when the 25 Yi migrants who reported engaging in
heavy physical activity were compared with the 508
Yi farmers with a similar level of physical activity, the
age-adjusted differences were very small: 8.7 mg/dl
(p = 0.2) for total cholesterol, 5.6 mg/dl (p = 0.09)
for HDL cholesterol, 0.9 mg/dl {p = 0.9) for LDL
cholesterol, and 13.3 mg/dl (p = 0.5) for triglyceride.
After adjustment for all of the variables listed in table
5, these differences were reduced almost to zero.
^
a S
io ^ co
o^^^ 3t
co
i d d d d
Contribution of environmental factors to
differences in serum lipid levels between Yi
migrants and Han people
S3? ONOO'-O
The differences in serum total and HDL cholesterol
between the Yi migrants and the Han people were not
statistically significant after adjustment for age, body
mass index, and alcohol intake (difference for total
cholesterol, 5.1 mg/dl (p = 0.12); difference for HDL
cholesterol, 2.3 mg/dl (p = 0.09)). The Yi migrants
and Han people had similar mean levels of serum LDL
cholesterol and triglyceride. After additional adjustment for the covariates listed in table 4, the differences
in serum lipid levels between the Yi migrants and the
Han people were reduced even further and were still
nonsignificant.
DISCUSSION
The three groups in this study provided an unusual
opportunity to compare the effects of environmental
factors and ethnic heritage on serum lipid levels. The
Yi migrants and Yi farmers had the same ethnic (and
Am J Epidemiol
Vol. 144, No. 9, 1996
846
He et al.
TABLE 5. Mean differences In serum lipld levels tn linear regression analysis comparing YI migrants
and YI farmers, after adjustment for dietary nutrients and other lifestyle factors: Uangshan, People's
Republic of China, 1988-1989f
Adjustment
factors
Age only
Age and:
Booty mass Index
Cigarette smoking
Alcohol drinking
Total dietary fat
Dietary PUFA*
Keys scored
Dietary fiber
All of the above factors
Difference (migrants - farmers)
H D I 4 cholMterol
(mo/dl)
L O U cholesterol
(mg/dT)
TMBtycerlde
(mp/dl)
38.4 (2.7)«*,§
5.95 (1.18)***
27.3 (2.6)***
18.0 (7.0)**
35.9 ( 2 . 7 ) " *
31.2 (3.2)'**
37.9 (2.7)***
5.78(1.21)***
25.3 (2.6)***
22.0 (3.0)***
12.5 (6.9)
Total cholesterol
(mgftJ)
5.70 (1.18)***
2.49(1.60)
38.9 ( 2 . 7 ) " *
28.6 ( 3 . 9 ) " *
21.1 (4.1)'**
14.7 (6.9)*
26.3 (3.5)***
28.3 (2.6)***
1.7(10.0)
6.45 (1.76)***
5.55(1.87)**
17.9 (3.9)***
-2.7 (9.7)
• p < 0.05; **p < 0.01; *«p < 0.001.
t Differences shown are values for migrants minus those for farmers. Covarlables Included were selected on
the basis of the best multiple linear regression models given In table 4.
t HDL, high density llpoproteln; LDL, low density lipoproteln; PUFA, potyunsaturated fat
§ Numbers In parentheses, standard error.
H Keys score: 1.35 (2S - P) + 1.5Z, where S Is % of kcal for saturated fat, P is % of kcal for potyunsaturated
fat and Z is the square root of dietary cholesterol (mg/100 kcal).
presumably genetic) background but different environmental exposures. In contrast, the Yi migrants and the
Han people had a similar lifestyle but different ethnic
backgrounds.
Yi farmers are one of the most isolated populations
in China. In addition to having mean serum total and
LDL cholesterol and triglyceride levels that were well
below those of Yi migrants and Han people, elevations
in total and LDL cholesterol levels were almost nonexistent in the Yi farmers. Moreover, the lipid levels
noted in the Yi farmers were much lower than those
reported from the general Chinese population. In the
PRC-USA Collaborative Study, mean total, HDL, and
LDL cholesterol and triglyceride levels were 181.5,
49.8, 97.9, and 112.1 mg/dl, respectively, in southern
Chinese men (24). The serum total cholesterol and
triglyceride levels we observed in the Yi farmers were
similar to those that have been noted in other isolated
populations (25-30).
In contrast to most previous reports from isolated
populations (25-28), we noted an age-related increase
in serum total and LDL cholesterol levels. This agerelated increase in serum cholesterol was not due to a
higher Keys score, because Keys score was not associated with age. An age-related increase in serum
cholesterol has also been seen in other unacculturated
populations, such as the Yanomamo Indians of Brazil
(29) and the Tarahumara Indians of Mexico (30).
These data suggest that higher cholesterol levels at
older ages are not solely a reflection of westernization
or acculturation.
The Yi migrants in our study had serum lipid levels
that were significantly higher than those of their Yi
farmer counterparts but similar to those noted in the
Han people. The slightly higher average total cholesterol level noted in the Yi migrants as compared with
the Han people was due solely to the higher HDL
cholesterol level in the Yi migrants, and appeared to
be due to the greater prevalence of alcohol drinking
among the Yi migrants. Most of the difference in
serum lipid levels that we noted between the Yi migrants and the Yi fanners could be explained by differences in dietary and lifestyle factors.
Diet and serum lipid levels
The differences in dietary nutrient intake between
the Yi farmers and the Yi migrants were striking. The
Yi farmers followed a vegetarian diet that was low in
total fat, saturated fat, and cholesterol and high in
polyunsaturated fat and fiber content. Many studies
conducted in westernized societies have found that
vegetarians have lower serum total cholesterol and
triglyceride levels than their counterparts in the general population (31-37).
Our study confirmed the observation that dietary fat
and cholesterol intakes are important determinants of
serum lipid levels, even in a group with very low
dietary fat intake and serum lipid levels. Keys score, a
combined measure of dietary intake of saturated and
polyunsaturated fats and cholesterol, was significantly
and positively associated with serum total cholesterol
in both the Yi farmers and the urban residents. This
Am J Epidemiol
Vol. 144, No. 9, 1996
Migration and Lipids
fairly strong relation between Keys score and serum
total cholesterol might be due to a relatively large
person-to-person variation in Keys score in the study
population. Keys scores alone explained at least 25
percent of the difference in total cholesterol between
the Yi migrants and the Yi fanners. Keys score was
also significantly associated with higher serum triglyceride levels. This relation was independent of age,
body mass index, and alcohol intake. Keys score alone
explained more than 90 percent of the difference in
triglyceride levels between Yi migrants and Yi farmers.
Our study found that intakes of dietary total fat and
cholesterol were significantly associated with higher
HDL cholesterol levels in the Yi farmers but not in the
urban residents. Dietary total fat intake explained almost 60 percent of the difference in HDL cholesterol
levels between the Yi migrants and the Yi farmers.
Several studies have suggested that diets low in total
fat and cholesterol content and high in polyunsaturated: saturated fat ratio lower serum HDL cholesterol
levels (31-33, 38, 39).
Dietary fiber was positively related to HDL cholesterol in multivariate analysis. Several studies have
demonstrated that dietary fiber selectively reduces
LDL cholesterol and increases the HDL: total cholesterol ratio (40, 41). Our study also showed that dietary
total fat was positively associated and polyunsaturated
fat was inversely associated with levels of LDL cholesterol.
In our study, dietary nutrient intakes were estimated
by an average of three 24-hour diet recalls. These
estimates differed from "true" customary nutrient intake values because of day-to-day variation in dietary
nutrient intake. This random error in nutrient intake
measurements tends to bias regression coefficients
toward zero (42). Thus, our study findings almost
certainly underestimated the true effect of an alteration
in diet on migration-related changes in serum lipids.
Other lifestyle factors and serum lipid levels
Body mass index was significantly associated with
total and LDL cholesterol and triglyceride in multivariate models. Differences in body mass index also explained some of the differences in serum lipid levels
between the Yi migrants and the Yi farmers. Cigarette
smoking was positively associated with total cholesterol in the urban residents but not in the Yi farmers.
However, the difference in the smoking-cholesterol
relation between the urban residents and the Yi farmers was not statistically significant. Cigarette smoking
was significantly associated with serum total cholesterol in the multivariate regression model, and it explained almost 20 percent of the migration-related
Am J Epidemiol
Vol. 144, No. 9, 1996
847
difference in total cholesterol. Alcohol consumption
was positively associated with total and HDL cholesterol and triglyceride in both the urban residents and
the Yi farmers. Body weight, cigarette smoking, and
alcohol consumption have been related to serum total
cholesterol and lipoprotein levels in other populations
(43-45).
Physical inactivity is an important risk factor for
hyperlipidemia (44,45). In our study, large differences
between the Yi farmers and the Yi migrants precluded
adjustment for physical activity. However, ageadjusted mean serum lipids were not significantly different between the Yi migrants and the Yi farmers
after data were stratified by physical activity level.
This suggests that changes in physical activity play an
important role in the migration-related rise in serum
lipid levels.
Given the cross-sectional design of the present
study, the possibility that persons with higher lipid
levels are more likely to migrate cannot be excluded.
In other longitudinal migration studies of cardiovascular risk factors, however, there has been no evidence
of such selection (11).
In summary, the present study indicates that migration from an isolated area with very low population
serum lipid levels to an urban environment is associated with an increase in serum lipid levels. Diets
higher in fats, especially saturated fat and cholesterol,
and lower in polyunsaturated fat and fiber appear to
contribute to the migration-related changes in serum
lipid levels. In addition, cigarette smoking, alcohol
consumption, and physical inactivity also explain
some of the migration-related differences in serum
lipids. These findings suggest that lifestyle modifications should be an important component of the overall
approach to lowering blood lipid levels in acculturated
populations. The subsequent favorable changes in
lipid levels are likely to result in a substantial reduction in the societal burden of cardiovascular diseases
related to hyperlipidemia.
ACKNOWLEDGMENTS
This study was supported by the Ministry of Public
Health, People's Republic of China, and by Outpatient
General Research Center grant 5M01RR00722 from the US
National Institutes of Health (NIH). Computational assistance was received through NTH grant RROOO35.
Dr. Jiang He was supported by training grant
5T32HL07024-21 from the National Heart, Lung, and
Blood Institute, NIH. Dr. Michael J. Klag is an Established
Investigator of the American Heart Association.
848
He et al.
REFERENCES
1. World Health Organization. World health statistics annual,
1991. Geneva, Switzerland: World Health Organization, 1992.
2. Castelli WP, Anderson K, Wilson PWF, et al. Lipids and risk
of coronary heart disease: The Framingham Study. Ann Epidemiol 1992;2:23-8.
3. Klag MJ, Ford DE, Mead LA, et al. Serum cholesterol in
youth and subsequent cardiovascular disease. N Engl J Med
1993;328:313-18.
4. Stamler J, Wentworth D, Neaton JD. Is relationship between
serum cholesterol and risk of premature death from coronary
heart disease continuous and graded? Findings in 356,222
primary screenees of the Multiple Risk Factor Intervention
Trial (MRFIT). JAMA 1986;256:2823-8.
5. Muldoon MF, Manuck SB, Matthews KA. Lowering cholesterol concentrations and mortality: a quantitative review of
primary prevention trials. BMJ 199O;301:3O9-14.
6. Sempos CT, Cleeman JI, Carroll MD, et al. Prevalence of high
blood cholesterol among US adults: an update based on guidelines from the second report of the National Cholesterol Education Program Adult Treatment Panel. JAMA 1993;269:
3009-14.
7. Shaper AG, Wright DH, Kyobe J. Blood pressure and body
build in three nomadic tribes of northern Kenya. E Afr J Med
1969;46:273-81.
8. Zimmet P, Jackson L, Whitehouse S. Blood pressure studies in
two Pacific populations with varying degrees of modernisation. N Z Med J 1980;91:249-52.
9. Sever PS, Gordon D, Peart WS, et al. Blood-pressure and its
correlates in urban and tribal Africa. Lancet 1980;2:60-4.
10. Reed D, Labarthe D, Stallones R. Health effects of westernization and migration among Chamorros. Am J Epidemiol
1970;92:94-112.
11. Prior IAM. Migration, hypertension, and Pacific perspectives
for prevention. In: Lovenburg W, Yamori Y, eds. Nutritional
prevention of cardiovascular disease. Orlando, FL: Academic
Press, Inc, 1984:137-53.
12. Poulter NR, Khaw KT, Hopwood BE, et al. Determinants of
blood pressure changes due to urbanization: a longitudinal
study. J Hypertens Suppl 1985;3(suppl 3):S375-7.
13. He J, Tell GS, Tang YC, et al. Effect of migration on blood
pressure: The Yi People Study. Epidemiology 1991;2:88-97.
14. He J, Klag MJ, Whelton PK, et al. Migration, blood pressure
pattern, and hypertension: The Yi Migrant Study. Am J Epidemiol 1991;134:1085-101.
15. He J. Epidemiology of blood pressure in Yi People, Liangshan, China. (PhD dissertation). Beijing, People's Republic of
China: Peking Union Medical College and Chinese Academy
of Medical Sciences, 1994.
16. Allain CC, Poon LS, Chan CSG, et al. Enzymatic determination of total serum cholesterol. Clin Chem 1974;20:470-5.
17. McGowan MW, Artiss JD, Strandbergh DR, et al. A peroxidase-coupled method for the colorimetric determination of
serum triglycerides. Clin Chem 1983;29:538-42.
18. Lopes-Virella MF, Stone P. Ellis S, et al. Cholesterol determination in high-density lipoproteins separated by three different methods. Clin Chem 1977;23:882-4.
19. Friedewald WT, Levy RI, Fredrickson DS. Estimation of the
concentration of low-density lipoprotein cholesterol in
plasma, without use of the preparative ultracentrifuge. Clin
Chem 1972; 18:499-502.
20. Keys A, Fidanza F, Karvonen MJ, et al. Indices of relative
weight and obesity. J Chronic Dis 1972;25:329-43.
21. Institute of Nutrition, Chinese Academy of Medical Sciences.
Chinese food composition tables. 3rd ed. Beijing, People's
Republic of China: People's Health Press, 1985.
22. Keys A, Anderson JT, Grande F. Serum cholesterol response
to changes in the diet. I-IV. Metabolism 1965; 14:747-87.
23. SAS Institute, Inc. SAS/STAT user's guide, release 6.03.
Cary, NC: SAS Institute, Inc, 1991.
24. Tao S, Li Y, Xiao Z, et al. Serum lipids and their correlates in
Chinese urban and rural populations of Beijing and
Guangzhou: PRC-USA Cardiovascular and Cardiopulmonary
Epidemiology Research Group. Int J Epidemiol 1992;21:
893-903.
25. Mann GV, Roels OA, Price DL, et al. Cardiovascular disease
in African Pygmies: a survey of the health status, serum lipids
and diet of Pygmies in Congo. J Chronic Dis 1962;15:341-71.
26. Bamicot NA, Bennett FJ, Woodburn JC, et al. Blood pressure
and serum cholesterol in the Hadza of Tanzania. Hum Biol
1972,44:87-116.
27. Sinnett PF, Whyte HM. Epidemiological studies in a total
highland population, Tukisenta, New Guinea: cardiovascular
disease and relevant clinical, electrocardiographic, radiological, and biochemical findings. J Chronic Dis 1973;26:265-90.
28. Page LB, Damon A, Moellering RC Jr. Antecedents of cardiovascular disease in six Solomon Islands societies. Circulation 1974;49:1132-46.
29. Connor WE, Cerqueira MT, Connor RW, et al. The plasma
lipids, lipoproteins, and diet of the Tarahumara Indians of
Mexico. Am J Clin Nutr 1978;31:1131-42.
30. Mancilha-Carvalho JJ, Crews DE. Lipid profiles of Yanomamo Indians of Brazil. Prev Med 1990; 19:66-75.
31. Sacks FM, Castelli WP, Donner A, et al. Plasma lipids and
lipoproteins in vegetarians and controls. N Engl J Med 1975;
292:1148-51.
32. Sacks FM, Ornish D, Rosner B, et al. Plasma lipoprotein
levels in vegetarians: the effect of ingestion of fats from dairy
products. JAMA 1985;254:1337-41.
33. Burslem J, Schonfeld G, Howald MA, et al. Plasma apoprotein
and lipoprotein lipid levels in vegetarians. Metabolism 1978;
27:711-19.
34. Ruys J, Hickie JB. Serum cholesterol and triglyceride levels in
Australian adolescent vegetarians. Br Med J 1976;2:87.
35. Thorogood M, Carter R, Benfield L, et al. Plasma lipids and
lipoprotein cholesterol concentrations in people with different
diets in Britain. Br Med J 1987;295:351-3.
36. Burr ML, Bates CJ, Fehily AM, et al. Plasma cholesterol and
blood pressure in vegetarians. J Hum Nutr 1981;35:437—41.
37. Resnicow K, Barone J, Engle A, et al. Diet and serum lipids
in vegan vegetarians: a model for risk reduction. J Am Diet
Assoc 1991 ;91:447-53.
38. Lock DR, Varhol A, Grimes S, et al. ApoA-I/ApoA-U ratios
in plasmas of vegetarians. Metabolism 1983;32:1142-5.
39. Knuiman JT, West CE, Katan MB, et al. Total cholesterol and
high density lipoprotein cholesterol levels in populations differing in fat and carbohydrate intake. Arteriosclerosis 1987;7:
612-19.
40. Hunninghake DB, Miller VT, LaRosa JC, et al. Hypocholesterolemic effects of a dietary fiber supplement. Am J Clin Nutr
1994;59:1050-4.
41. Mackay S, Ball MJ. Do beans and oat bran add to the effectiveness of a low-fat diet? Eur J Clin Nutr 1992;46:641-8.
42. Liu K, Stamler J, Dyer A, et al. Statistical methods to assess
and minimize the role of intra-individual variability in obscuring the relationship between dietary lipids and serum cholesterol. J Chronic Dis 1978;31:399-418.
43. Linn S, Fulwood R, Carroll M, et al. Serum total cholesterol:
HDL cholesterol ratios in US white and black adults by
selected demographic and socioeconomic variables (HANES
II). Am J Public Health 1991;81:1038-43.
44. Koyama H, Ogawa M, Suzuki S. Relationship between total
cholesterol and high-density lipoprotein cholesterol and the
effects of physical exercise, alcohol consumption, cigarette
smoking and body mass index. J Nutr Sci Vitaminol (Tokyo)
1990;36:377-85.
45. Bolton-Smith C, Woodward M, Smith WC, et al. Dietary and
non-dietary predictors of serum total and HDL-cholesterol in
men and women: results from the Scottish Heart Health Study.
Int J Epidemiol 1991;20:95-104.
Am J Epidemiol Vol. 144, No. 9, 1996