AMERICAN JOURNAL OF HUMAN BIOLOGY 18:112–122 (2006) Original Research Article Farming and Adiposity in Samoan Adults EMBER D. KEIGHLEY,1 STEPHEN T. MCGARVEY,1* PASA TURITURI,2 AND SATUPAITEA VIALI3 1 International Health Institute, Department of Community Health, Brown University, Box G-B495, Providence, Rhode Island 2 Diabetes Control Program, Department of Health, American Samoa Government, Pago Pago, American Samoa 96799 3 Tupua Tamasese Meaole Hospital, Ministry of Health, Government of Samoa, Apia, Samoa ABSTRACT Samoans are experiencing some of the highest prevalences of obesity and associated health conditions in the world. Sustainable interventions are needed to prevent further increases in obesity. This study describes the cross-sectional association between farm work and adiposity among 754 adults residing in American Samoa in 2002 and 957 adults residing in Samoa in 2003. Adiposity was measured by body mass index (BMI) and percent body fat (% BF), based on bioelectrical impedance. Regression models adjusted for the effects of age, education, occupation (in women), and material lifestyle (MLS), and the clustering within households due to the family design of the parent study. After controlling for these variables, participation in farm work was associated with a significantly lower BMI and % BF in men of all ages residing in American Samoa, women 45 years residing in American Samoa, and women 18–44 years residing in Samoa, and a significantly lower BMI in men 18–44 years residing in Samoa. These results suggest that farm work plays an important role in regulating body size and fatness of adult Samoans and may be ideal for interventions in the Samoan archipelago. Am. J. Hum. Biol. # 2005 Wiley-Liss, Inc. 18:112–122, 2006. Over the past one hundred years, modernization has had profound effects on the way of life, biology and health of Samoans (Baker et al., 1986). Obesity and its associated health conditions have reached prevalences among the highest in the world and appear to have continued steadily increasing through the last three decades (Keighley et al., 2006; McGarvey, 1991, 2001). Between 1976 and 2003 in both American Samoa and Samoa there were sharp increases in the prevalence of obesity, defined here as a BMI > 32 kg/m2, based on body composition studies among Polynesians (Swinburn et al., 1999). Among men, 25–74 years, living in American Samoa, 32.2% were obese in 1976, while 63.1% were obese by 2002 (Keighley et al., 2006). Among women living in American Samoa the prevalence of obesity increased from an already high 58.0% in 1976 to 75.0% in 2002. While the prevalence of obesity is considerably lower among residents of Samoa (formerly Western Samoa), it also increased rapidly over the past 30 years. Between 1979 and 1982, 10.8% of men and 31.5% of women were obese but by 2003, 34.1% of men and 58.7% of women were obese. ß 2005 Wiley-Liss, Inc. Furthermore, there were temporal increases in the mean BMI and the proportion of those within the highest BMI classes. In American Samoa, between 1976 and 2002, the percent of men, 25–74 years, with BMI 40 kg/m2 almost tripled, from 5.0% to 14.4%, and the percent of women more than doubled, reaching 32.8%. Among adults residing in Samoa, the percent of people with BMI 40 kg/m2 has risen sharply from 1.8% of men and 3.8% of women in 1979–1982 to 6.8% and 15.9% respectively in 2003. Longitudinal data also indicate that despite pre-existing excess adiposity, Samoans are able to continue to gain weight (McGarvey, 1991). Contract grant sponsor: NIH; Contract grant number: DK59642; Contract grant sponsor: Brown University Undergraduate Teaching and Research Assistantship. *Correspondence to: Stephen T. McGarvey, Ph.D., M.P.H., International Health Institute, Brown University, Box G-B495, Providence, RI 02912. E-mail: Stephen_ McGarvey@brown.edu Received 29 June 2005; Revision received 16 September 2005; Accepted 27 September 2005 Published online in Wiley InterScience (www.interscience. wiley.com). DOI: 10.1002/ajhb.20469 113 FARMING AND ADIPOSITY IN SAMOAN ADULTS Samoans are not alone in this rapid nutrition and health transition. Throughout the Pacific, obesity has become widespread. Currently 8–10% of all deaths in the Pacific region are attributable to obesity (WHO, 2000). These trends are ultimately due to the economic, social and nutritional changes in the Pacific, but may be interacting with a putative increased susceptibility to obesity among Polynesians (McGarvey, 1994; McGarvey et al., 1989). As the Pacific islands rapidly modernize, it is crucial that we find effective means of avoiding obesity and implement interventions to help prevent further increases in adiposity. Farm work is a traditional subsistence activity for both men and women throughout the Samoan archipelago (Greksa et al., 1986). In American Samoa, participation in farm work has decreased dramatically in the last 40 years. On Tutuila, the main island of American Samoa, the primary occupation among adult males in 1960 was agriculture, but by 1974 only 8% of economically active males were still involved in plantation work (Greksa and Baker, 1982). According to the American Samoa 1990 census 3% of all adults and in the 2000 census 7% of all adults in American Samoa did primarily subsistence work (U.S. Department of Commerce, 1992, 2004). In Samoa, over time, professional and less physically demanding occupations have increased. In the 1991 census, among adults 24% of men and 14% of women worked in wage or salary jobs (Western Samoan Department of Statistics, 1995). In 1991, 67% of men and 63% of women in the economically active group still worked primarily in subsistence farming and fishing. By the 2001 census in Samoa, 44% of men in the economically active group were engaged primarily in subsistence work (Samoa Department of Statistics, 2003). The shift away from agriculture may have a 2-fold impact on the health of Samoans. First, people may be less physically active. On average agriculture requires a 65-kg Samoan man to expend 6–7 kcal/min (Greksa and Baker, 1982). In contrast, office employees only expend 1.7–1.9 kcal/min. A 1982 study in Samoa found greater energy expenditure and lower adiposity among young rural men and active urban workers compared to sedentary urban workers (Pelletier, 1987). Second, people who are not participating in farm work may experi- ence a dietary shift decreasing the consumption of locally produced fruits and vegetables, and increasing the consumption of imported processed foods. While the consumption of fat from imported foods increases the risk of obesity by 2.2-fold and diabetes by 2.4-fold, diets composed primarily of locally produced foods have been shown to prevent and reduce obesity (WHO, 2003). A 1990–1991 dietary study in American Samoa and Samoa showed greater consumption of imported foods with increasing socioeconomic status regardless of residence (Galanis et al., 1999). Population variation in patterns of physical activity with the physical environment, modes of subsistence and with economic modernization has been described and reviewed (Ulijaszek, 1995; Weitz et al., 1989). A fuller multilevel exploration of the impacts of modernization and urbanization on occupational changes, work intensity, dietary patterns and physical activity patterns remains an ambitious and crucial goal for human population biology. Here we do not focus on the ultimate and proximate factors that determine physical activity variations in modernizing Samoans, or an exploration of the retention of the seemingly ‘‘traditional’’ physical activity, farm work. That is one of our future objectives. In this report we concentrate on the effect of a specific type of physical activity on adiposity from a human biological and public health perspective. The purpose of this study is to examine the cross-sectional association between farm work and adiposity, measured by BMI and percent body fat (% BF), in adult Samoans residing in American Samoa in 2002 and Samoa in 2003. METHODS The study population derives from the U.S. territory of American Samoa and the independent nation of Samoa (formerly Western Samoa). In 2000, the population of American Samoa was 57,291, of which 88.2% are ethnic Samoans (U.S. Department of Commerce 2004). The government offices and tuna canneries in the Pago Pago harbor offer many job opportunities in wage-labor. In 2000 most adults, 52%, are employed in the paid labor force, while only 6.7% of adults participate primarily in subsistence activities. In 2001, Samoa had a population 114 E.D. KEIGHLEY ET AL. of 176,848 (Samoa Department of Statistics, 2003). Although Samoa is located just 100 km from American Samoa, fewer adults hold wage-labor jobs and the economy is largely based on agriculture, with tourism becoming increasingly important (Western Samoa Department of Statistics, 1995). In 1991, approximately 15% of adults of both sexes were employed in wage-labor professions, but by 2001 36% of men and 26% of women were in wage or salary jobs (Samoa Department of Statistics, 2003). All participants took part in the Samoan Family Study of Overweight and Diabetes in 2002–2003. Protocols for this study were approved by the Brown University Institutional Review Board, American Samoan Institutional Review Board, and the Government of Samoa, Ministry of Health, Health Research Committee. Written informed consent was obtained from all participants. Recruitment in American Samoa in 2002 was based on random selection of probands seen in the 1990– 1994 cohort study in American Samoa, and the presence of at least two adult siblings alive and residing in American Samoa. Recruitment in Samoa in 2003 was first based on finding named individuals residing in Samoa who were members of American Samoa pedigrees. After that short period of 4–6 weeks, we selected villages to represent geographic and socioeconomic diversity across the nation. Then, selection of families was based on discussions with the village leaders about the size of the families and the presence of at least two adult siblings. Participants were recruited based on availability of large extended families and probands were not selected for obesity or type 2 diabetes. Participants resided in a range of rural to urban villages in both Samoa and American Samoa and had varying degrees of education, occupation and material way of life (Table 1). Participants were classified into two age groups: 18–44 years and 45 years and above. All data analyses were stratified by nation and sex. Participants were assigned to the farm work or no farm work group based on selfreported participation in farm work, and/or self-reported subsistence work as their primary occupation. Farm work was used as a dichotomous variable. Measurements of stature and weight were taken with participants wearing light tropical clothing, and BMI was calculated. Impedance measures of participants were taken with the Quantum II Bioelectrical Body Composition Analyzer. Percent body fat (% BF) was calculated using resistance in the equation developed by Swinburn et al. (1999) for use in Polynesians. We discovered that this equation contained a mistake in publication and after communication with the author we used the following correct equation: Fat Mass ¼ 11.58 þ 0.79(weight) 0.15(height) þ 0.04(age) 6.0(sex) 0.39 (height2/resistance), where weight is measured in kg, height in cm (instead of m as previously reported), age in years, resistance in ohms, and sex ¼ 1 for males and sex ¼ 0 for females. Swinburn et al. (1999) developed this equation for Polynesians between the ages of 20 and 70 years, but here we extended this age range to 18–90 years. While it is not ideal to extend the age range, we believe that this is the best equation available for measuring body fat in Polynesian adults. All % BF results from the Swinburn et al. (1999) equation when applied to the extended age-range were cross-validated with % BF estimated from equations developed specifically for multiethnic U.S. adolescent girls (Philips et al., 2003), Polynesian adolescents (Rush et al., 2003), and European elderly (Dey and Bosaeus, 2003). The % BF results from the different equations were quite similar and are available upon request. Education was classified into two groups: less than secondary education and secondary education and above. In American Samoa, where schools are run on a U.S.based system, secondary education is the equivalent of 12 years of education. In Samoa, the schools use a New Zealand system and secondary education is equivalent to 13 years. Occupation was divided into three categories: unemployed (including housewives and retirees), subsistence work (farming and fishing), and non-subsistence work (including wage-labor occupations and traditional village leadership). Those reporting two jobs were assigned the occupation with a higher socioeconomic status. For example, people reporting both a wage-labor occupation and farm work were assigned the wagelabor occupation. Material lifestyle (MLS) was defined using a 10-point index (Galanis et al., 1999). The index was based on ownership of amenities including: flooring, refrigerator, stereo, TV, 115 FARMING AND ADIPOSITY IN SAMOAN ADULTS TABLE 1. Descriptive statistics of the study sample Education Less than secondary % n/N Men in American Samoa, 18–44 years 13.4% 27/201 Men in American Samoa, 45þ years 34.7% 43/124 Women in American Samoa, 18–44 years 9.2% 24/262 Women in American Samoa, 45þ years 38.7% 55/142 Men in Samoa, 18–44 years 74.0% 188/254 Men in Samoa, 45þ years 90.0% 162/180 Women in Samoa, 18–44 years 70.5% 208/295 Women in Samoa, 45þ years 90.0% 199/221 Material lifestyle Occupation Secondary % n/N Unemployed % n/N Subsistence work % n/N Non-subsistence work % n/N Low % n/N High % n/N 86.6% 174/201 9.8% 18/183 4.9% 9/183 85.2% 156/183 36.5% 72/197 63.5% 125/197 65.3% 81/124 48.7% 58/119 5.0% 6/119 46.2% 55/119 44.0% 55/125 56.0% 70/125 90.8% 238/262 43.0% 107/249 – 57.0% 142/249 39.6% 106/268 60.4% 162/268 61.3% 87/142 71.4% 100/140 – 28.6% 40/140 37.4% 55/147 62.6% 92/147 26.0% 66/254 8.6% 22/255 64.3% 164/255 27.1% 69/255 55.3% 140/253 44.7% 113/253 10.0% 18/180 17.0% 31/182 63.7% 116/182 19.2% 35/182 68.3% 123/180 31.7% 57/180 29.5% 87/295 75.8% 225/297 0.3% 1/297 23.9% 71/297 56.7% 165/291 43.3% 126/291 10.0% 22/221 91.0% 202/222 0.9% 2/222 8.1% 18/222 62.2% 138/222 37.8% 84/222 VCR, vehicle, electricity, stove toilet, and drinking water. Only people who completed at least 7 out of 10 questions were given a score. Any missing values for these people were imputed based on the average for their nation, sex, and 10-year age group. Those who scored an 8 or higher were assigned relative high MLS while those with less than an 8 were assigned relative low MLS. Whenever possible, household was defined as the fale, house, where participants resided at time of measurement. In American Samoa, household was determined by the telephone number and by families living in the same village when there was no phone. In Samoa, household was defined by each participant’s reported fale when data was available and by families living in the same village when no individual household information was available. Bivariate analysis between farm work and education, MLS, and occupation were done with categorical methods. Multivariate analysis of BMI, % BF, and farm work were done with general linear models. Education, occupation, and MLS were controlled for as confounders. We adjusted for household clustering due to the participation of individuals from the same house, which violates the assumption of independence of observations. Age was controlled for in all analyzes with BMI and % BF. All analyzes were done using SPSS. RESULTS The study sample includes 332 men and 422 women from American Samoa and 438 men and 519 women from Samoa between the ages of 18 and 90 years (Table 2). Participants lived in 426 separate house- 116 E.D. KEIGHLEY ET AL. TABLE 2. Percent of people participating in farm work and average number of hours per week spent on the farm American Samoa Males Females Age (years) n/N % 18–44 45þ Total 18–44 45þ Total 101/204 67/128 168/332 58/271 36/151 94/422 49.5% 52.3% 50.6% 21.4% 23.8% 22.3% Samoa X (SD) (h/week) 4.1 4.4 4.2 4.1 2.6 3.5 (4.2) (5.1) (4.6) (5.4) (1.9) (4.5) n/N % X (SD) (h/week) 215/256 150/182 365/438 91/297 67/222 158/519 84.0% 82.4% 83.3% 30.6% 30.2% 30.4% 8.1 6.7 7.5 3.3 3.9 3.6 (8.8) (8.4) (8.7) (3.5) (5.5) (4.4) TABLE 3. Mean BMI and % BF of the study sample American Samoa 2 Age (years) Males 18–44 45þ Females 18–44 45þ Samoa 2 BMI (kg/m ) n X (sd) %BF n X (sd) BMI (kg/m ) n X (sd) %BF n X (sd) n ¼ 204 33.3 (7.5) n ¼ 128 33.7 (7.3) n ¼ 271 36.7 (8.3) n ¼ 151 36.2 (8.1) n ¼ 204 32.8 (7.2) n ¼ 128 34.8 (5.9) n ¼ 271 41.1 (6.5) n ¼ 151 42.2 (6.1) n ¼ 256 27.9 (5.0) n ¼ 182 31.5 (5.8) n ¼ 297 32.0 (6.7) n ¼ 222 34.0 (7.7) n ¼ 152 26.0 (6.4) n ¼ 95 33.3 (6.3) n ¼ 150 37.6 (6.9) n ¼ 125 40.7 (6.2) holds: 201 in American Samoa and 225 in Samoa. The study sample is marked by high levels of BMI and % BF (Table 3). In American Samoa, mean BMI for both men and women is above the Polynesian criteria for obesity. Percent body fat ranges from 32.8% to 34.8% in men and 41.1% to 42.2% in women. In Samoa, mean BMI is significantly lower than in American Samoa for all age and sex groups. Percent body fat in Samoa ranges from 26.0% to 33.3% in men and 37.6% to 40.7% in women. About 50% of men in American Samoa and over 80% of men in Samoa participate in farm work (Table 2). Fewer women, 22% in American Samoa and 30% in Samoa, participate in farm work. Young men and women from American Samoa, 18–44 years, spend about 4 h per week doing farm work (Table 2). In Samoa, young men spend twice as much time working on the farm, 8.1 h, and young women spend slightly less, 3.3 h. In American Samoa, older men, 45 and above, spend more time per week on the farm than their younger counterparts, 4.4 h, and older women spend the least time on the farm, 2.6 h. In Samoa, older men continue to spend considerable time on the farm, 6.7 h, and older women spend more time than the younger age group, 3.9 h. Most of the groups sampled here average above the recommended minimum number of hours of physical activity per week, 30 min daily (WHO, 2005), on the farm, with the exception of older women in American Samoa and youn ger women in Samoa. The study sample in American Samoa is characterized by high levels of education, especially among the younger age group where 86.6% of men and 90.8% of women completed a secondary education or above (Table 1). In American Samoa, the majority of young men and women have non-subsistence occupations, while almost 50% of men and over 70% of women 45 and above are unemployed or retired. The majority of men and women in all age groups in American Samoa have a relatively high material lifestyle status. In Samoa, the study sample is characterized by comparatively lower levels of education than in American Samoa (Table 1). Less than 30% of younger men and women and only 10% of older men and women completed a secondary education or above. The majority of men in Samoa perform subsistence work as their primary occupation, while the majority of women are unemployed. Most men and women in Samoa have a relatively low material lifestyle status. 117 FARMING AND ADIPOSITY IN SAMOAN ADULTS Among men of all ages from American Samoa, older women from American Samoa, and older men and women from Samoa, there are no significant differences in education level between the farm work and no farm work groups. While among young women from American Samoa, and young men and women from Samoa, those who participate in farm work are less likely to have completed a secondary or higher education (P ¼ 0.043, 0.038, and 0.030, respectively). Men of all ages from American Samoa and all men and women from Samoa show significant differences in occupation between the farm work and no farm work groups. This is primarily due to people who report participating in subsistence work all being in the farm work group. Men from American Samoa and Samoa, and older women from Samoa who participate in farm work are less likely to report being unemployed. Younger women from Samoa who participate in farm work are more likely to report being unemployed. Among women from American Samoa, there are no significant differences in occupation in the farm work group. There are no significant differences in material lifestyle between the farm work and no farm work groups in either polity. Education level alone is not significantly related to BMI or % BF in either nation or sex. Occupation is only directly related to BMI in older men from American Samoa (P ¼ 0.041). Those men who are unemployed have an average BMI of 35.7 kg/m2 compared to 32.1 kg/m2 in those who participate in subsistence work as their primary occupation, and 32.2 kg/m2 in the employed group. Occupation is not directly related to % BF in either nation or gender. MLS is significantly related to BMI in younger men and older women from American Samoa, as well as younger men from Samoa (P < 0.001, P ¼ 0.003, P ¼ 0.001, respectively). In each of these age groups, being in the higher MLS category is associated with an increase in BMI of 1.8–4.0 kg/m2. MLS is also significantly related to % BF in the same groups (P ¼ 0.004, P ¼ 0.010, P ¼ 0.015, respectively), with an increase in average % BF in the higher MLS group. Farm work is significantly associated with a lower age-adjusted BMI in men and older women from American Samoa, as well as young men and young women residing in Samoa (Table 4). Within these groups, those who participate in farm work have age-adjusted BMI 2.0–3.6 kg/m2 lower than those who do not participate. Farm work is also significantly associated with a lower % BF in men and older women living in American Samoa, as well as younger women living in Samoa (Table 5). To measure the direct effect of farm work on BMI, in women multivariable analyses were done controlling for age, education, MLS, household, and occupation, and in men analyses were done controlling for age, education, MLS, and household. A large number of men reported subsistence work as their primary occupation. Therefore, multivariable models in men that adjust for occupation would lead to over-controlling for the effect of farm work on adiposity. This is not a problem in women because few women report subsistence work as their primary occupation. In the multivariable models with adjustments for age, education, MLS, and occupation (in women), participation in farm work is significantly associated with reduced BMI TABLE 4. Association between farm work and BMI (in kg/m2) in adults* American Samoa Age (years) Male 18–44 45þ Female 18–44 45þ Mean (SEM) N No farm work Farm work No farm work Farm work No farm work Farm work No farm work Farm work *Controlling for age; NS, nonsignificant. 34.7 32.0 35.4 32.2 36.9 35.9 37.1 33.5 (0.7) (0.7) (0.9) (0.8) (0.6) (1.1) (0.7) (1.3) N N N N N N N N ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ 103 101 61 67 213 58 115 36 Samoa t P 6.9 0.009 7.1 0.009 0.7 NS 5.9 0.017 Mean (SEM) N 29.5 27.5 31.8 31.4 32.7 30.4 34.1 34.0 (0.7) (0.3) (1.0) (0.5) (0.4) (0.6) (0.6) (1.0) N N N N N N N N ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ 41 215 32 150 206 91 155 67 t P 6.8 0.010 0.2 NS 9.2 0.003 <0.1 NS 118 E.D. KEIGHLEY ET AL. TABLE 5. Association between farm work and % BF in adults* American Samoa Mean (SEM) N Age (years) Male 18–44 45þ Female 18–44 45þ No farm work Farm work No farm work Farm work No farm work Farm work No farm work Farm work 34.5% 31.1% 36.4% 33.4% 41.3% 40.5% 42.9% 40.0% (0.7) (0.7) (0.7) (0.7) (0.4) (0.9) (0.5) (1.0) N N N N N N N N ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ 103 101 61 67 212 57 115 36 Samoa t P 12.6 <0.001 8.6 0.004 0.6 6.7 NS 0.011 Mean (SEM) N 27.4% 25.7% 34.2% 33.0% 38.6% 35.5% 40.3% 41.5% (1.1) (0.5) (1.5) (0.7) (0.6) (0.8) (0.7) (1.0) N N N N N N N N ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ 25 127 20 75 102 48 86 39 t P 2.1 NS 0.5 NS 9.4 .003 0.9 NS *Controlling for age; NS, nonsignificant. and % BF in the same age and sex groups as the age-adjusted results (Tables 6 and 7). In young men from American Samoa, adjusted mean BMI is 1.8 kg/m2 lower in the farm work group, and % BF is 2.9% lower (Tables 6a and 7a). Older men from American Samoa who participate in farm work have an adjusted mean BMI 2.6 kg/m2 lower and an adjusted mean body fat 2.9% lower (Tables 6a and 7a). Older women from American Samoa doing farm work have a BMI 4.9 kg/m2 lower and a body fat 4.0% lower (Tables 6b and 7b). There is not a significant association between farm work and % BF in young men from Samoa, but the mean adjusted BMI is 1.9 kg/m2 lower in the farm work group (Tables 6c and 7c). Among young women living in Samoa, those who participate in farm work have an average BMI 2.0 kg/m2 lower and a body fat 3.0% lower (Tables 6d and 7d). DISCUSSION The results presented here show consistent cross-sectional evidence that farm work is associated with lower BMI and % BF among Samoan men and women living throughout the Samoan archipelago. These findings remain significant after adjusting for household clustering, and the influences of age, level of education, MLS, and occupation (in women). The results are consistent with those found in an earlier cross-sectional comparison of physical activity and adiposity among urban and rural men in Samoa (Pelletier, 1987), as well as ecological studies from Samoa and other Pacific nations (Baker et al., 1986; Wessen et al., 1992). Our findings show a distinct age and gender pattern. It is not surprising that farm work and BMI are related in young men from American Samoa and young men and women from Samoa. Within Samoan culture, it is often the younger and untitled men and women who perform the majority of the physical labor for the village (Greksa et al., 1986; Mead, 1969; O’Meara, 1990). While young women from American Samoa who participate in farm work also have a lower average BMI and % BF, these results are not statistically significant. These young women from American Samoa are participating in the same number of hours of farm work as the young men, but they may not be participating in tasks that are as physically demanding. While younger men and women are responsible for the majority of the physical farm labor, older men and women, especially titled members of the village, often take more managerial roles (O’Meara, 1990). Despite these age patterns, it is noteworthy that in American Samoa, older men and women who continue to participate in farm work also maintain a lower average BMI and % BF. This may be due to the daily activity from farm work, or it may be due to a lifetime effect of consistently living a more active lifestyle and perhaps eating a diet richer in fruits and vegetables. For the over 50% of older men from American Samoa who participate in farm work, doing less than 1 h of work per day, on average, is associated with a BMI 2.5 kg/ m2 lower, regardless of age, education, and MLS (Tables 6a and 7a). Among women 45 and above from American Samoa, participating in as few as 2.6 h per week of farm work, less than the recommended weekly minimum (WHO, 2005), is associated with a decrease in BMI of 5.1 kg/m2 and a decrease in body fat of 4.1%, after controlling for age, 119 FARMING AND ADIPOSITY IN SAMOAN ADULTS TABLE 6. Results of linear models of farm work and BMI adjusted for age, education, occupation, and MLS 18–44 years (a) Men in American Samoa Age Farm work Education Occupationa MLS Household Adjusted means of BMI (SEM) Farm work No Yes (b) Women in American Samoa Age Farm work Education Occupation MLS Household Adjusted means of BMI (SEM) Farm work No Yes (c) Men in Samoa Age Farm work Education Occupationa MLS Household Adjusted means of BMI (SEM) Farm work No Yes (d) Women in Samoa Age Farm work Education Occupation MLS Household Adjusted means of BMI (SEM) Farm work No Yes 45þ years F P F P 16.6 4.2 0.4 — 11.1 1.4 <0.001 0.041* 0.531 — 0.001 0.237 12.1 5.7 0.7 — 0.5 0.4 0.001 0.019* 0.389 — 0.488 0.553 N ¼ 97 N ¼ 97 x ¼ 33.8 kg/m2 (0.6) x ¼ 32.0 kg/m2 (0.6) N ¼ 59 N ¼ 62 x ¼ 34.7 kg/m2 (0.8) x ¼ 32.1 kg/m2 (0.8) 10.0 0.5 0.2 0.1 0.6 0.5 <0.002 0.485 0.696 0.805 0.447 0.502 8.4 10.6 0.1 1.2 10.3 1.0 0.004 0.001** 0.786 0.275 0.002 0.319 N ¼ 185 N ¼ 53 x ¼ 37.5 kg/m2 (0.6) x ¼ 36.6 kg/m2 (1.2) N ¼ 95 N ¼ 33 x ¼ 38.1 kg/m2 (0.8) x ¼ 33.2 kg/m2 (1.3) 61.7 6.0 0.2 — 10.0 0.1 <0.001 0.015* 0.672 — 0.002 0.794 19.4 0.3 <0.1 — 2.7 0.1 <0.001 0.588 0.921 — 0.105 0.775 N ¼ 40 N ¼ 210 x ¼ 29.5 kg/m2 (0.7) x ¼ 27.6 kg/m2 (0.3) N ¼ 31 N ¼ 148 x ¼ 32.0 kg/m2 (1.0) x ¼ 31.4 kg/m2 (.5) 63.0 6.6 0.5 0.6 <0.1 3.7 <0.001 0.010* 0.497 0.444 0.886 0.055 16.9 <0.1 0.6 0.2 1.0 0.6 <0.001 0.995 0.451 0.640 0.331 0.454 N ¼ 199 N ¼ 90 x ¼ 32.6 kg/m2 (0.4) x ¼ 30.6 kg/m2 (0.6) N ¼ 154 N ¼ 67 x ¼ 34.0 kg/m2 (0.6) x ¼ 34.0 kg/m2 (1.0) a Occupation is left out due to over controlling (for men’s values only). *For farmwork, P < 0.050. **For farmwork, P < 0.010. education, occupation, and MLS (Tables 6b and 7b). Levels of obesity among older women in American Samoa are so high that participation in some form of physical activity, even if it is only for a minimal amount of time each week and at a lower intensity, appears to influence adiposity. While those women who do not participate in farm work have an adjusted mean BMI in the range of class II obesity, those women who do have an adjusted mean BMI in the class I obesity range. A decrease from class II to class I obesity carries important health benefits, including a decrease from severe to moderate risk of hypertension, type 2 diabetes, and other aspects of metabolic syndrome (SPC, 2002). Furthermore, physical activity increases insulin sensitivity and has been shown to be effective in controlling type II Diabetes (Barnard et al., 1994). Despite these positive health benefits, less than 25% of older women in American Samoa participate in farm work. 120 E.D. KEIGHLEY ET AL. TABLE 7. Results of linear models of farm work and % BF adjusted for age, education, occupation, and MLS 18–44 years (a) Men in American Samoa Age Farm work Education Occupationa MLS Household Adjusted mean % BF (SEM) Farm work No Yes (b) Women in American Samoa Age Farm work Education Occupation MLS Household Adjusted mean % BF (SEM) Farm work No Yes (c) Men in Samoa Age Farm work Education Occupationa MLS Household Adjusted mean % BF (SEM) Farm work No Yes (d) Women in Samoa Age Farm work Education Occupation MLS Household Adjusted mean % BF (SEM) Farm work No Yes 45þ years F P F P 22.9 9.3 0.3 — 7.3 3.0 <0.001 0.003** 0.600 — 0.008 0.087 1.0 7.6 0.6 — 0.2 1.7 0.310 0.007** 0.456 — 0.665 0.189 N ¼ 97 N ¼ 97 x ¼ 34.0% (0.7) x ¼ 31.1% (0.7) N ¼ 59 N ¼ 62 x ¼ 36.2% (0.7) x ¼ 33.3% (0.7) 0.001 0.593 0.850 0.373 0.159 0.790 4.0 12.3 <0.1 0.1 7.5 1.3 0.049 0.001** 0.882 0.720 0.007 0.262 x ¼ 41.7% (.5) x ¼ 41.1% (.9) N ¼ 95 N ¼ 33 x ¼ 43.5% (.6) x ¼ 39.5% (1.0) <0.001 0.250 0.486 — 0.015 0.257 4.6 1.7 0.4 — 0.3 <0.1 x ¼ 27.2% (1.1) x ¼ 25.8% (0.5) N ¼ 19 N ¼ 73 48.5 7.8 1.0 0.2 0.6 0.3 <0.001 0.006** 0.311 0.661 0.456 0.566 1.7 1.0 0.4 <0.1 0.4 3.1 N ¼ 97 N ¼ 48 x ¼ 38.6% (0.6) x ¼ 35.6% (0.9) N ¼ 85 N ¼ 39 10.6 0.3 <0.1 0.8 1.0 <0.1 N ¼ 184 N ¼ 53 42.7 1.3 0.5 — 6.0 1.3 N ¼ 24 N ¼ 123 0.034 0.200 0.540 — 0.559 0.879 x ¼ 35.0% (1.5) x ¼ 32.9% (0.7) 0.200 0.311 0.533 0.971 0.530 0.083 x ¼ 40.3% (0.7) x ¼ 41.6% (1.1) a Occupation is left out due to over controlling (for men’s values only). *For farmwork, P < 0.050. **For farmwork, P < 0.010. These results have important human biological and applied public health implications. Because this study is part of genetic epidemiologic research on determinants of adiposity, the results indicate that we must adjust for this type of physical activity as part of the search to detect putative genetic effects on Samoan adiposity. It also suggests that we should consider gene-by-physical activity analyses to determine if different genetic factors are linked to adiposity depending on levels of physical activity. From a public health perspective, in a population with high prevalence of obesity, farming for less than 1 h per day appears to be associated in both men and women with some relative protection from reaching the extreme levels of obesity of their nonfarming peers. This difference remains after controlling for household, education, MLS, and occupation (in women). These findings suggest that farm work intervention may be FARMING AND ADIPOSITY IN SAMOAN ADULTS feasible among adult residents of both Samoan polities. An intervention could have a profound impact on the levels of adiposity, even if it only engages people in minimal amounts of physical activity. Lowering adiposity, especially among this older age group, may have extensive health benefits for the people and ease the burden of chronic disease on the healthcare system. Limitations and future directions This study does not take into account actual levels of physical activity. Such a study measuring METS or some other index of exercise intensity might provide more ability to detect the direct effects of physical activity from farm work on adiposity. We are currently developing such an index. Future studies of the impact of farm work on BMI and % BF should also incorporate measures of diet. Here we have not explored if the impact of farm work on adiposity is due entirely to physical activity or if it is confounded with variation in diet. Controlling for the MLS categories we created allows us to conclude that farm work is not tied to income level at least with the measure we used, although there are certainly other methods of measuring and categorizing income which may show different results. Additionally, while controlling for household income suggests that the relationship between farm work and adiposity is not principally due to either shared dietary intake or genetics (most households contain two to three generations of family members and share meals), future studies should look at dietary intake and genetic effects. Furthermore, there may be additional issues of selectivity that we have not controlled for in our farm work group. Because this is a cross-sectional study, we cannot determine if the effect of farming, especially in the older age group, is due primarily to the direct effect of farming at the time of measurement and in the recent past, or if it is due to a lifetime effect of participating in physical activity from farming. It is possible that older adults who have continued to remain active are benefiting from a lifetime of activity, and not just the daily energy expenditure which they were participating in when they were measured. If this older age group is benefiting from lifetime effects of participating in farm work, that would 121 suggest that in order to improve Samoans’ health it is most important to implement intervention in younger age groups. Otherwise, if the act of farming is beneficial to older adults regardless of whether they farmed continuously since childhood, then implementing interventions in older adults may also have important health benefits. The findings also indicate the need for a more detailed analysis of the sociocultural, economic, and behavioral determinants of the patterns of physical activity in modernizing adults Samoans. Why are some adults able to retain an evidently healthy pattern of physical activity amidst an obesogenic environment? While the results of this cross-sectional study have strong implications, we are limited in the conclusions we can draw due to the study design. Further understanding of the socioeconomic structural barriers and individual psychosocial factors related to physical activity are required before definitive public health recommendations can be made. Prospective studies of physical activity in the Samoas should also evaluate other traditional activities which provide considerable forms of physical activity, including fishing, paddling, and dancing. Such studies should also take a community-based approach to evaluate participation in physical activities relative to available facilities and other environmental conditions influencing lifestyle choices. Further research is needed to identify incentives and barriers to increasing physical activity in rapidly modernizing societies (WHO, 2005). CONCLUSIONS The results of this study show that regardless of place of residence, sex, age, education level, household of residence, occupation (in women), and MLS, participating in farm work is associated with lower adiposity. In the face of the rapid changes in lifestyle and modernization throughout the Samoan archipelago, maintenance of traditional farming activities may be having a relative protective health effect through reduced adiposity. These results reveal important suggestions for future interventions. ACKNOWLEDGMENTS of We thank the members of the Department Health, American Samoa, and the 122 E.D. KEIGHLEY ET AL. Ministry of Health, Government of Samoa, for their advice and guidance in data collection, as well as local political officials for their cooperation and help. Thanks to the research assistants from American Samoa and Samoa for their hard work and dedication to the project. Also, to the 2002–2003 field teams and field directors staffed by students and alumni from Brown University and other schools. Above all, we are grateful to the families, individual participants, and people of Samoa and American Samoa. LITERATURE CITED Baker PT, Hanna J, Baker TS. 1986. The changing Samoans: behavior and health in transition. New York: Oxford University Press. p 275–296. Barnard RJ, Jung T, Inkeles SB. 1994. Diet and exercise in the treatment of NIDDM: the need for early emphasis. Diabetes Care 17(12):1469–1472. Dey DK, Bosaeus I. 2003. Comparison of bioelectrical impedance prediction equations for fat-free mass in a population-based sample of 75y olds: The NORA Study. Nutrition 19:858–864. Galanis DJ, McGarvey ST, Quested C, Sio B, AfeleFa’amuli SA. 1999. Dietary Intake of modernizing Samoans: implications for risk of cardiovascular disease. J Am Diabetic Assoc 99:184–190. Greksa LP, Baker PT. 1982. Aerobic capacity of modernizing Samoan men. Hum Biol 54:777–788. Greksa LP, Pelletier DL, Gage TB. 1986. Work in contemporary and traditional Samoa. In: Baker PT, Hanna J, Baker TS, editors. The changing Samoans: behavior and health in transition. New York: Oxford University Press. p 351–393. Keighley ED, McGarvey ST, Quested C, MuCuddin C, Viali S, Maiava T. 2006. Nutrition and health in modernizing Samoans: evolutionary and adaptive perspectives. In: Ohtsuka R, Ulijaszek SJ, editors. Nutrition and health changes in the Asia-Pacific region. Cambridge, England: Cambridge University Press (in press). Mead M. 1969. Social organization of Manu’a. Honolulu, Hawaii: Bishop Museum Press. McGarvey ST. 1991. Obesity in Samoans and a perspective on the etiology in Polynesians. Am J Clin Nutr 53:1586S–1594S. McGarvey ST. 1994. The thrifty gene concept and adiposity studies in biological anthropology. J Polynes Soc 103:29–42. McGarvey ST. 2001. Cardiovascular disease (CVD) risk factors in Samoa and American Samoa, 1990–1995. Pac Health Dialog 8(1):157–162. McGarvey ST, Bindon JR, Crews DE, Schendel DE. 1989. Modernization and adiposity: causes and consequences. In: Little MA, Haas JD, editors. Human population biology: a trans-disciplinary science. New York: Academic Press. p 263–279. O’Meara JT. 1990. Samoan planters: tradition and economic development in Polynesia. Fort Worth, TX: Holt, Rinehart and Winston, Inc. Pelletier DL. 1987. The relationship of energy intake and expenditure to body fatness in Western Samoan men. Ecol Food Nutr 19:185–199. Phillips SM, Bandini LG, Compton DV, Naumova EN, Must A. 2003. A longitudinal comparison of body composition by total body water and bioelectrical impedance in adolescent girls. J Nutr 133(5):1419–1425. Rush EC, Puniani K, Valencia ME, Davies PS, Plank LD. 2003. Estimation of body fatness from body mass index and bioelectrical impedance: comparison of New Zealand European, Maori and Pacific Island Children. Euro J Clin Nutr 57:1394–1401. Samoa Department of Statistics. 2003. Census of Population and Housing 2001. Apia, Samoa: Government Printing House. SPC (Secretariat of the Pacific Community). 2002. Obesity in the Pacific too big to ignore. Secretariat of the South Pacific, Noumea, New Caledonia. Swinburn BA, Ley SJ, Carmichael HE, Plank LD. 1999. Body size and composition in Polynesians. Intl J Obes 23:1178–1183. U.S. Department of Commerce. 1992. Census of Population and Housing, American Samoa, 1990. Washington, D.C.: Government Printing Office. U.S. Department of Commerce. 2004. Census of Population and Housing, American Samoa, 2000. Washington, D.C.: Government Printing Office. Ulijaszek SJ. 1995. Human energetics in biological anthropology. Cambridge, U.K.: Cambridge University Press. Weitz CA, Greksa LP, Thomas RB, Beall CM. 1989. An anthropological perspective on the study of work capacity. In: Little MA, Haas JD, editors. Human population biology: a trans-disciplinary science. New York: Academic Press. p 113–131. Wessen AF (editor). 1992. Migration and health in a small society: the case of Tokelau. Oxford, England: Clarendon Press. Western Samoa Department of Statistics. 1995. Western Samoa Census 1991. Apia, Samoa: Government Printing House. WHO. 2000. World Health Organization Regional Office for the Western Pacific. Workshop on obesity prevention and control strategies in the Pacific. Apia, Samoa. WHO. 2003. World Health Organization Regional Office for the Western Pacific. Diet, food supply and obesity in the Pacific, WHO, Manila. WHO. 2005. Physical activity fact sheet. Online: http:// www.who.int/dietphysicalactivity/publications/facts/ pa/en/. WHO, Geneva.