Professional Documents
Culture Documents
25-300-306
Melbourne Chinese Health Study 1 Australia: Melbourne (Chinese) 60-80 68 Telephone lists
(Chinese, Australia)
Brazilian National Survey of 1 Brazil national survey 60-108 4419 Household sample
Health and Nutrition, 1989 O
X
Chinese Nationwide Nutrition 1 China national survey 60-94 1764 Household sample H
Survey 1982 X
Mini-Finland Health Survey 1 Finland national survey 60-90+ 2126 Population register; no m
anthropometric collection in O
institutionalized X
H
Nutritional Assessment of 1 Guatemala: one rural 60-103 202 >
Guatemalan Ambulatory Elderly location Z
Census; excludes institutionalized D
(Guatemala) a
Longitudinal Study of Health Hong Kong sample 70-100 977
and Social Support in the Hong
Kong Chinese Elderly Cohort Household and old-age housing
z
Nutrition in Old Age in Italy Italy: 17 locations 60-97 921 Electoral rolls; excludes en
(Italy/17) institutionalized and severely
Italian Nutrition Examination Italy: 5 locations 65-95 1248 Electoral rolls; excludes r
M
Survey of the Elderly (Padova, institutionalized en
Italy)" o
•n
Rotterdam Study Rotterdam, The Netherlands 60-103 3752 Population register O
National Health Examination United States national sample 60-86 3695 Household sample; excludes r
D
Follow-up Survey (U.S. institutionalized m
National White)
Food Habits in Later Life: Australia: Melbourne (Anglo Celtics) 60-79 111 Telephone, population or m
A Cross Cultural Study Australia: Melbourne (Greeks) 70-104 186 electoral registers; excludes v
China: Beijing 60-95 264 institutionalized in
China: Tianjin 70-96 441 O
Z
Greece: Spata 70-94 70 en
Sweden: Johanneberg 69-91 204
International Collaborative Barbados: 1 rural location 60-80+ 192 Population sample from census
Study on Hypertension in Nigeria: 2 rural locations 60-80+ 129 or government data; excludes
Blacks Cameroon: 2 rural locations 60-80+ 283 institutionalized
Body Mass Index (weight in kg/height in m ); SD: Standard deviation; , < 10.
(described more fully in Table I) that used standard methods men (weighted r = 0.79; p < 0.001) and women
to measure weight (to the nearest 0.1 kg to 1.0 kg) and height (weighted r = 0.73; p = 0.01). However, the Mediter-
(to the nearest 0.1 cm). To assess geographic/ethnic differ- ranean samples [Italy (n = 2), Greece (n = 1), and
ences, comparisons were limited to persons aged 70—79 years Greeks living in Australia (n = 1)] showed relatively
old. To assess age trends within and between studies, three
age groups were defined (60-69 years old, 70-79 years old, greater weight for height.
80+ years old). Comparisons between these age groups were BMI: Women had a higher mean BMI and standard
limited to eight studies with complete data for these age strata. deviation than men in most samples (Table II), which
We also compared the BMI distributions of persons aged 70- may reflect greater variability in weight among women.
79 by three levels of reported global health status: poor/fair, The highest mean BMI was generally found in the
good, and very good/excellent. Since not all studies collected samples of Mediterranean origin, as expected from the
data on health status, used the same question, or had adequate data presented in Table II. Mean BMI was highest
numbers (^10 per health status category), comparisons by among Australian women of Greek origin [30.7 (5.1)]
health status were limited to five studies. and women in Barbados [29.2 (6.9)].
Data on height, weight and BMI are presented separately Underweight and overweight: Geographic/ethnic dif-
for men and women. Distributions were compared using the
Underweight (%)
70 60 50 40 30 20 10 0
Nigeria
Cameroon
Guatemala
Chinese Australian
Hong Kong
Tianjin, China
Beijing, China
Brazil (National)
Johanneberg Sweden
Rotterdam, NL
US (National, White)
Italy (17 sites)
Finland (National)
Padova, Italy
Barbados
"B
Anglo Australian
Greek Australian
Spata, Greece
0 10 20 30 40 50 60 70
Overweight (%)
* = prevalence is 0%
Figure 1. Comparison of anthropometric characteristics of older persons from geographically/ethnically diverse samples.
Percentage overweight (BMI ^30) and underweight (BMI < 20) among 70-79-year-old men.
L. J. LAUNER, T. HARRIS ET AL.
Underweight (%)
70 60 50 40 30 20 10 0
Nigeria
Cameroon
Guatemala
Chinese Australian
Hong Kong
Tianjin, China
Beijing. China
Brazil (National)
Johanneberg, Sweden
Rotterdam, NL
US National. White
Italy (17 sites)
Finland (National)
Padova, Italy
Barbados
10 20 30 40 50 60 70
Overweight (%)
+ = prevalence is 0%
* = <10 in the strata
Figure 2. Comparison of anthropometric characteristics of older persons from geographically/ethnically diverse samples.
Percentage overweight (BMI^30) and underweight (BMI < 20) among 70-79-year-old women.
Table III. Comparison of anthropometric characteristics of older persons from geographically/ethnically diverse samples: height
(m) by age strata
Guatemala 1.55 (0.05) 1.43 (0.06) 1.56(0.06) 1.40(0.07) 1.53(0.07) .40 (0.06)
China (National) 1.62 (0.07) 1.51 (0.06) 1.61 (0.08) 1.48(0.06) 1.55 (0.07) .46 (0.07)
Italy (17 sites) 1.64(0.07) 1.53 (0.07) 1.62(0.07) 1.51 (0.07) 1.60(0.07) .48 (0.07)
Brazil (National) 1.65 (0.11) 1.52(0.07) 1.63 (0.10) 1.50(0.08) 1.62(0.09) .49 (0.09)
Padova, Italy 1.66(0.06) 1.53 (0.06) 1.64(0.07) 1.51 (0.06) 1.62(0.07) .51 (0.07)
Finland (National) 1.70(0.12) 1.57(0.06) 1.69(0.06) 1.55(0.06) 1.67(0.06) .53 (0.06)
US (National, White) 1.74(0.05) 1.61 (0.05) 1.72(0.08) 1.59(0.05) 1.71 (0.05) .58 (0.05)
Rotterdam, NL 1.75 (0.06) 1.63 (0.06) 1.73 (0.06) 1.59(0.07) 1.71 (0.06) .57 (0.06)
Table IV. Comparison of anthropometric characteristics of older persons from geographically/ethnically diverse samples: body
mass index by age strata
Table V. Comparison of anthropometric characteristics of older persons from geographically/ethnically diverse samples: body
mass index by strata of self-reported health status
Self-reported health
Hong Kong
Men 13 21.8(3.8) 112 21.7(3.2) 117 20.7 (3.5)
Women 17 22.3 (4.2) 221 22.8 (4.0) 239 22.2 (4.0)
Padova, Italy
Men 31 26.1 (3.4) 305 26.5 (3.7) 165 26.2 (4.0)
height, differences by age were smaller, suggesting the children and adults; little is known about its value for
potential for bias in self-reported data on height from predicting the health status of older people. As a first
older persons (data not shown). step towards evaluating the use of anthropometric
BMI: The mean BMI tended to decrease with age indicators in persons older than 60 years, we undertook
more for women than for men (Table IV). We have a comparison of distributions of weight, height and
shown that height declined with age; for BMI to decline BMI from geographically diverse populations. There
with age, weight must also decline and to a greater are some common findings relating to patterns of
extent than height. In general, this decline resulted in a weight and height by age and sex. However, these
shift in the distribution for the entire sample rather distributions differed widely by geographic region/
than a skewed tail of lower weight. ethnic groups and by health status.
Underweight and overweight: Consistent with the Some caution is needed when interpreting these
shift in BMI distributions by age, in most studies the comparisons. Although most studies randomly
proportion of the sample with a BMI < 20 increased recruited their participants and included more than
with age. For example the prevalence of underweight in 1000 respondents, several studies were based on small
women was 1.5 times higher in the Brazil sample and samples of <15 per age-sex stratum. These small
6.0 times higher in the Italy/17 sample in the 80+ year- samples may not be representative of the older persons
old group than in the 60-69-year-old groups. Patterns living in their community. In addition, the distribu-
were similar for the men (data not shown). tions based on these smaller samples may not be stable.
Therefore, the size of the sample should be taken into
Variation by reported health status in 70— 79-year-old account when evaluating the comparison. For instance,
men and women the regression association of height and weight from
Across studies, no one range of BMI was consistently data in Table II were weighted by sample size.
associated with either excellent/very good health or fair/ Given the limitations of sample size and representa-
poor health (Table V). Rather, the distribution by tiveness, these comparisons highlight the differences in
health status overlapped considerably within studies height and BMI distributions by geographic region/
and related to the underlying distribution of BMI in the ethnic group and by age, as well as in the prevalence of
sample. underweight and overweight. Sources of these differ-
ences reflect multiple factors including the effects of
genetic potential, early growth and nutritional status,
Discussion differences in socio-economic status and health beha-
Anthropometry is often used as an indicator of viours, biological changes in body composition that
nutritional and health status, in particular for infants, accompany ageing, the high prevalence of chronic
3 o6 L. J. LAUNER, T. HARRIS ET AL.