Blood Pressure Studies Among Amazonian Native Populations: a Review from an Epidemiological Perspective

BLOOD PRESSURE STUDIES AMONG AMAZONIAN NATIVE POPULATIONS: A REVIEW FROM AN EPIDEMIOLOGICAL PERSPECTIVE

MILLICENT FLEMING-MORAN’ and CARLOS E. A. COIMBRAJR’ ‘Center on Aging/SPEA, Indiana University, Indianapolis, IN 46201, U.S.A. and *Department of Epidemiologia. Escola National de Saude Publica, Fiocruz, Rio de Janeiro, Brazil

Abstract-Nine publications are reviewed concerning blood pressure (BP) levels among Amerindian tribes of the Amazon Basin. The lifestyle of these lowland peoples includes most known protective factors against hypertension, and relative isolation from Western society. The latter, however, is rapidly changing. Sampling, data, and analysis problems make current blood pressure data difficult to interpret, and provide a questionable baseline from which to document pressure change during rapid culture change for these groups. Ethnographic and epidemiological perspectives are offered for future blood pressure and health studies among native Amazonians.

Key words-blood pressure, hypertension, Amerindian, Amazon, cultural change, epidemiology, blood pressure measurement, male/female differences in diet

INTRODUCIION relative social homogeneity among tribal members [7, 12, 131. Numerous researchers have now compared ‘tradi- However, even the most isolated Amazonian tional’ societies with industrial nations, and docu- groups face highways, mining, agricultural coloniz- mented differences in age/sex trends in mean blood ation and other development projects in the pressure levels, or differential rates of hypertension Basin, particularly since the 1970s [16, 18, 191. These prevalence [ 11. These studies often contradict Western societies now represent a continuum of increasing experience, where blood pressure increases with age, contact and integration with the larger Western and mean pressure levels for adult males normally society [20,21]. Most now live in large territorial exceed that of adult females [2,3]. Similar trends in reservations. Increased access to Western medical age/sex increases in blood pressure occasionally have care has been a part of this process, including been noted for traditional societies undergoing social health surveillance and blood pressure measurement change, or in subjects who migrate from pre- [e.g. 9, 10, 12, 19. Biomedical researchers, health industrial settings to modem, urban societies [4-6]. officials and social scientists continue to collaborate These patterns have fostered on-going epidemio- in the difficult work of recording changes in logical debate. Does blood pressure necessarily Amerindian health status during this period [e.g. increase with advancing age? Are there always signifi- 9, 10, 13, 16,22-251. cant differences in mean pressure levels between the There have been numerous suggestions of an asso- sexes when age and weight factors are considered? ciation between greater mean blood pressure levels Finally, is there a consistent relationship between and the increase in social complexity of a population socio-cultural disruption in a society, and long- [3-7,26-291. As a formally isolated and homoge- term blood pressure alteration for its individual neous group becomes acculturated, or its members members? migrate to a larger, more complex society, greater Certain reviewers do not attempt comparisons of social stratification and specializaton occur. This blood pressure data from nonindustrial groups due to poses several cultural, physical and psychological methodological problems [e.g. 7,8], and the diversity challenges and changes for the group’s members. of historical experiences of each traditional society as The terms ‘acculturation’ and ‘culture change’ have it encounters another culture [l]. Unfortunately this been used somewhat interchangably in the social- is also true for research among the native peoples of medicine literature, implying certain assumptions the Amazon Basin, even though studied groups are have been made [30]. Most common is the notion isolated and small, and much of the research has been that the ‘pre-contact’ or ‘pre-migrational’ society conducted within the last 15 yr [7]. is relatively homogenous, its cultural values are Amerindian populations of the Amazon Basin unanimously accepted, and given its social integra- offer an opportunity to study a way of life which tion, the society is considered to be relatively ‘stress- incorporates most current maxims for preventing and free’ [30]. Post-contact acculturation, whether controlling hypertension. These include a regular proceeding to complete social assimilation, or new physical exercise, low sodium/high potassium diets cultural configurations, conversely, is taken to imply [9-111, low prevalence of obesity or diabetes extended social disintegration, economic change and [9, IO, 12-151, moderate use of low-alcohol or psychosocial distress. While social scientists today caffinated beverages in traditional diets [17] and would question such a static and undifferentiated

593 594 MILLKENT FLEMING-MORANand CARLOS E. A. COIMBU JR

view of Amerindian societies, the general hypothesis We sampled only 124 of the 800 of the estimated popu- of Western-contact/social disintegration may be lation and the studied groups, the Kalapalo and the generally applicable to lowland groups of South Kamaiura, are those with which the researchers have had America. the longest contact.. other, more isolated groups will be studied . in the future (translation by authors). It is difficult, however, to test ‘acculturation’ as a construct in models of health status or disease prevalence. Tangible criteria for these socio-economic It must be recognized that ‘accessible’ villages may changes are needed, such that individual (versus not be representative of the total population. These group) differences in process of cultural change, and sites are more likely to include individuals who are presumably health risk, can be discerned. As one more dependent on wages, medical or other services, example, certain data suggest that process of cultural or who participate in governmental or missionary change affects social roles and health status of the two goals to ‘pacify’ the Indian groups, than are persons sexes differentially, and is associated with different from more isolated villages. health effects for members of various age cohorts Sampling and participation bias represent two of [7,26-28,301. the most severe problems in both the collection This review outlines the methodological hazards and interpretation of health data from Amerindian in this much needed anthropological and epidemio- societies. Although most authors state that measure- logical research on covariates of blood pressure, ments were attempted ‘. . . in all inhabitants’ [e.g. 91, as seen in previously published data for Amazonian age/sex pyramids for these studies indicate certain Amerindians. Sample selection, blood pressure and age groups, particularly males aged 2540 are strik- other data collection, data analysis and reporting, ingly underrepresented, even for a nonindustrial and research design will each be addressed. Ethno- population structure. Neel and co-workers, for graphic examples will illustrate common field con- example, provide useful age/sex ratios comparing ditions, and highlight specific variables which require their Xavante subjects, the Kaingang Amerindians of careful evaluation, if the mechanisms underlying southern Brazil, and the general 1950 Brazilian blood pressure response to cultural change are to be national census [12, p. 921. Taken together with the understood. Thus, the relative impact of modified sample sizes for the blood pressure data, the reader dietary and health habits, obesity, access to medical may deduce that 27% of Xavante under the age care and tangible cultural changes may be assessed of 15 were included in the study sample, as were for their independent (or interactive) effects on blood about 40% of the adults. As blood pressure data pressure level. were only a secondary objective of a larger genetic study, selection criteria for the blood pressure sub- SAMPLING PROCEDURES jects are not specified. However, as the representativeness of the blood pressure data cannot be . . . (A)pproximately 12-15,000 Yanomamo are distributed evaluated, the reader can only speculate about ‘aver- among some 150 villages in an area of approximately age’ Xavante blood pressure levels, or comparisons 100,000. square miles. . . At present there are some 16 small with other Amerindian populations. The Xavante missions. . . as well as government health stations. . . for the most part located on navigable rivers. (How- study is but one of several which presents this ever,). . . there are probably villages yet to be visited by problem. non-indian[s]. The selected subjects consisted of all available Oliver and colleagues note that “. . . several inhabitants of sixteen. . . villages. In a few instances several wanomamo] . . . were absent on a hunting trip.. . ” Indians were absent . . . on a hunting trip.. . . Selection of [9]; a common feature of Amerindian life. Groups villages was determined by a cross-section of widely dis- may also leave the village for gardening or collect- persed living sites . . . , as remote as possible, but still ing tasks, and women become inaccessible due to accessible by small plane or boat. . . Blood pressures ritual isolation during their menstrual periods, or were.. . performed on 506 Indians during the expeditions of post-partum seclusion [32]. Long term absences 1966, 1967, and 1968.. . [9]. may include seasonal expeditions, villages in the This is one of the best methodological descriptions cyclical process of relocation, and/or individuals of sampling in this literature, listing the selection engaged in wage-labor in the local cash economy. criteria for research sites and study participants. In some cases wage earners may be absent several Oliver and coworkers [9] have the advantage of a long months [ 131. history of ethnographic and governmental contact Prior knowledge of the seasonal round of economic with the Yanomamo, and accurate maps from which and social activities of the village(s) is not only to enumerate villages and their members. Seven essential in planning the study sample, but ethno- separate field trips were required to accomplish the graphic and genealogical data may identify missing medical objectives of their 1975 study, and related members and group demographic characteristics, studies are still in progress. The reader can appreciate relate absentees to others in the group, and explain the extreme physical difficulties in accessing these the latter’s absence. This is probably most feasible isolated populations and determining an appropriate where the household is used as the sampling unit. study sample. Some knowledge of mortality, morbidity, migration One can also identify potential sources of sampling and labor patterns of the group, enhances the likeli- and/or participation bias in work with lowland hood of locating missing subjects. Thus the latter’s Amerindian groups. Opportunistic sampling of effect on the representativeness of the sample can be villages linked to air or water transportation is determined. In short, demographic details, including characteristic of most of the reports reviewed here. As an age/sex pyramid for the study sample, and the noted by Pedro de Lima [31, p. 7881. estimated total population, are not only essential Blood pressure studies among Amazonians 595 tools for data collection, but provide a base for NO publication mentions how (or whether) the sub- interpreting changes in health status. Identifying jects were prepared for medical examinations. members with regular employment or contact with In small samples the influence of random error is the larger society may also indicate pertinent indices magnified in obscuring the detection of true differ- of individual acculturation. ences in blood pressure level. The number of inter- viewers and blood pressure instruments therefore BLOOD PRESSURE MEASUREIMENT should be minimized, and pre-survey training offered to reduce both observer error and bias [40,42]. Even Most publications follow the accepted norms in use of medically-trained observers is not failsafe. reporting the type/make of sphygmomanometer used, Korotkoff phases are often unconsciously rounded the position of the subject, which Korotkoff phase is up or down to a ‘preferred’ or clinically significant used for diastolic readings, the arm used for pressure digit, particularly in diastolic readings [43]. Arbitrary reading(s), and the time of day when the measure- rounding reduces the chance of detecting a true ments were recorded. One report addressed inter- difference in comparing mean blood pressure instrument variation [9], but none addressed levels [43]. Readings of blood pressure in nonclinical inter-reader variability, when more than one observer situations demand particular attention to the proper took pressure readings. More recent studies have size of cuff [44], support for the arm and allowing opted for two sequential readings; some recording the subject time to sit quietly prior to the readings [42]. average of the two readings [IO, 13,221, others use The field data should record the reader(s) identity, only the lowest reading [34]; while in some it is time of day and location, and any noted problems in unclear which reading is presented [35]. cuff fit or detection of Korotkoff phases. Similarly, Both casual and multiple blood pressure readings subjects who are obviously febrile, pregnant or in have their value in epidemiological studies. Although whom other tests show significant parasitic loads casual readings have been used in longitudinal studies such as hookworm, should be noted by field staff as of heart disease (361, the average of two (sometimes all these factors may influence pressure readings [45]. three) readings of both systolic and diastolic press- Such information allows the researcher to monitor ures, respectively, are now used in the analysis of the quality of data in the field, and to expurge any large national blood pressure studies [e.g. 21, and questionable values during analysis, if necessary. are recommended for epidemiologic investigations [37-391. The latter approach allows for greater accli- mation by the subject to the process of blood pressure OTHER RELEVANT COVARIATES OF BLOOD PRESSURE measurement for baseline prevalence surveys [40,41]. A minimum of two readings, taken some 15 minutes Age apart is considered ideal [38], and may be feasible All studies reviewed record the sex and estimated where the researchers can pursue medical history age of their subjects. However, use of nonstandard and sociological questions during blood pressure intervals, unreported standard deviations, readings. However, if members of the group are age means, range and number of subjects in each age suspicious of the proceedings, and/or are inordinately category inhibits both internal and external com- curious, field workers may be forced to sacrifice ideal parisons between groups. J&speaking Amerindian acclimating readings, for the need to quickly evaluate peoples recognize social-grades, particularly for available and willing subjects. males, as initiation ceremonies mark the transition In many instances blood pressure readings were from one age grade to another (see [12] for a detailed only one of several medical or anthropometric description from the Xavante). Subjects from non-J6 measurements taken of the study population. groups, or those over the age of 40 may represent More than one author alludes to ‘resistant’ or ‘un- a special challenge in determining accurate ages. cooperative’ subjects [lo, 12,341. While language In many cases, however, household genealogies problems pose an obvious barrier to cooperation, (and comparison of known age-mates and siblings), insensitivity to cultural norms, particularly where significant historical events and or evaluation of ‘clinic’ patient management is imposed, threaten social and physical characteristics may be needed information and participation bias in the study. In to estimate age. Thus while exact birthdates are one site visit, Coimbra attempted such bio-medical indeterminable, an accurate age estimate may be evaluations in each household, but found the trans- feasible. Presentations of the demographic and distri- porting of chairs, scales, calipers and other equip- butional characteristics of blood pressure is most ment, often by flashlight, to be so cumbersome that helpful when the closest age estimate is given for each it was more feasible to arrange for family groups to individual. appear at a central location. The whole adult population was thus sampled, but Coimbra attributes this to a long and careful introduction of his aims to the Weight subjects. Indeed, many Amerindians have cultural Individual, or age-specific mean weight is often, prohibitions about the ‘loss’ of blood, or other body but not always reported in the literature [see fluids [32], and may resist blood and urine sampling. 9, 10, 12, 13, 161, and mean heights are occasionally If Western subjects exhibit anticipatory rises in published [ 12, 13,161. In Western populations, cur- pressure prior to experimental tests [42], Amerindian rent weight, weight gain and ponderosity (weight for subjects are equally prone to altered blood pressure height) explain a great deal of blood pressure vari- during unfamiliar procedures, and should be appro- ance and future hypertension risk, second only to the priately briefed prior to collection of physical data. effect of age [36,45-48]. It is unclear what these 596 MILLKENT FLEMING-MORANand CARL~S E. A. COIMBRAJR

weight-blood pressure relations are among, Amazo- major exception may be Munduruku males, where nian Amerindians, particularly as there are few longi- both measures increase after the age of 50 [7l. How- tudinal studies of these groups. None of the reviewed ever, as that author notes, the number of elders in his, articles uses relative weight, e.g. where sex-specific and most study samples, is too small to test the mode or median figures serve as a reference value. If significance of this trend [A. Also, in cross-sectional published at all, weights appear as raw data [34], or studies such as these one cannot discount the possible as mean values in age/sex stratified tables to illustrate impact on selective mortality among those with the weight status of compared groups [e.g. higher pressures [7,54], even though the mean pres- 7, 10, 12, 161. Only one report calculates age/sex spe- sure levels are very low by Western standards. cific mean Quetelet Index (weight/height’ x 100) [13]. Lowenstein interprets the greater blood pressure As field-worthy instruments become more avail- level of Munduruku versus Karaja men in terms of able, the collection of both height and weight mea- the formers’ contact and acculturation through Fran- sures is desirable for at least two reasons. First, while ciscan missionaries, and their change of status from obesity is rare in isolated Amerindian groups, in- savanah/forest warriors to hunter-agriculturalists [7]. creased Western contact is likely to first promote He states, “ . . . In comparison to the Carajas, the differences in diet and daily activities [4, 13-15261. Mundurucus are less primitive, . . . and their whole These in turn, are more prone to affect ponderosity way of life has been changed. . . [including] changes (i.e. fat/lean ratio) than is height among surviving in their life habits, such as the regular use of table adults. Ponderosity, as measured by some ratio of salt . . . and the use of tobacco” [7, p. 3911. weight to height, or direct skinfold measurements, Other studies on blood pressure levels in changing has been clearly established in explaining differences traditional societies note a steeper slope in the rela- in both systolic and diastolic pressure [SO-521, espe- tionship of blood pressure with increasing age, partic- cially among females of minority groups in modem ularly among male subjects [3]. In a Samoan example, societies [4,26,51]. the difference in slope of the blood pressure trends It must be noted, however, that it is still uncertain was most pronounced when comparing men with whether composit height/weight indices convey simi- little contact with Western culture and economy, and lar information about relative obesity, and its relation those with wage-labor jobs in a major industrial port to blood pressure in all populations. For example, [26]. Among Samoan migrants to Hawaii, only males while the Quetelet Index is strongly correlated with emigrating from the traditional areas of Samoa ex- anthropometric measures of lean/fat ratios in Amer- hibited significant post-migration increases in dias- ican Caucasian and black populations [Sl], simple tolic blood pressures. Blood pressure data for female height and weight were better correlates of ponderos- Samoans, however, showed the same shape and basic ity in models for Mexican-American subjects [53]. slope in the age-blood pressure relationship, regard- Such questions can be explored only when compara- less of level of integration, and any ‘modernization’ ble data on height, weight, blood pressure and ide- effect toward increased blood pressure appeared to be ally, anthropometric measures, are available for limited to migrant women over the age of 50 [26]. diverse ethnic groups. Mens’ greater involvement in the modem settings’ Most Amerindian data demonstrate a similarity in cash economy is postulated to expose them to greater mean blood pressure levels between male and female psychosocial stress [26]. Similar conclusions have subjects, and while age/sex specific mean weights may been reached in other South Pacific studies [5]. be reported, no Amerindian data explores the role of Weight and dietary changes are one of the earliest relative weight in sex-specific blood pressures. Neel signals of cultural change, and thus are relevant to the and colleagues for example, report that on average, discussion of blood pressure alteration [4]. These Xavante women weigh 18% less than males, while aspects of culture change are probable confounders height differences between the sexes appear minimal of pressure change have been alluded to in some [12]. This may explain the slightly higher pressure Amerindian research [e.g. 7, 10,291. These also repre- levels for Xavante males. Similar differences were sent further interpretive problems when samples are noted in systolic (but not diastolic) readings in a biased toward villages which are accessible to Indian recent study of subjects on the Xingu Indian Park posts or mission sites. In one case, Carvalho and [13]. While such clearly is simplistic, and sometimes colleagues [29] turned this to advantage. They offer can be inferred from published stratified data [e.g. an excellent comparison of two Yanomamo groups: 10, 131, comparisons of sex-specific blood pressure one with significant contact with Brazilian society are facilitated when height and weight differences (N = 105); the other more isolated from contact between the sexes are accounted for. (N = 149). The authors surveyed all adults, and while The issue of weight-change-with-age is also rele- no age/sex stratification is available in their published vant to the discussion of blood pressure change with account, no significant variation of blood pressure age. Both pressure measures appear to be negatively with age was noted for either sex. Interestingly, the associated with age in several Amerindian groups contact-group showed greater mean body weight and [lo, 22,341, and systolic pressure decreases with age spot-urine sodium concentrations than the more iso- among Munduruku women [A. In these settings, lated group, but no meaningful difference in mean older subjects reportedly gained little weight with blood pressure level. In a linear regression model of advancing age [7, IO], and may weigh less than these data, each additional kg of weight increased younger members if the former have few relatives to systolic pressure by 0.4 mm Hg, and diastolic read- provide food for them [16,22]. Negligible change in ings by 0.7 mm Hg, similar to the relationship seen in mean blood pressure by age cohorts has also been Western samples [36]. However, only in the contact recorded (7 for the Karajas group) [12,13,31]. The group were body weight and sodium retained as Blood pressure studies among Amazonians 597 significant covariates in models for both blood pres- a threat to their health, nor exclude them from the sure measures [29]. workforce. While differences in mean blood pressure levels Future studies of Amazonian populations must be between contact and noncontact groups were not concerned with selective bias due to migration, mor- significant, members of the contact group who tality and poor study participation in this developing showed facility with Portuguese were reported to region. Greater socio-economic differentiation is be- show significantly higher mean blood pressure levels coming more common among Amerindian groups, as (SBP + 8.6 mm and DBP + 6.0 mm Hg) than mem- a result of rapid socio-economic change and assimila- bers who only used the native language. Analysis of tion of their members into the national labor force. possible confounding effects due to age, weight or In some cases whole families migrate permanently to salt-use differences among the former individuals was a non-Indian environment; in others, single members not available in the brief report, but as such data were may alternate residence between the new locale and collected, these questions can be addressed by the home village, for varying lengths of time. Dietary authors. More importantly, the authors have iden- changes must be presumed as these individuals gain tified a marker (language) which distinguishes indi- access to new subsistence technologies [7, 12,351, and vidual exposure to culture change in this society, and to nontraditional food items [e.g. 12,661. can be weighed against the influences of other estab- Migrants often show greater relative weight in lished covariates in a multivariable model for mean comparison to their home populations [26,60], but blood pressure level. this effect need not be uniform for all age/sex groups [16,26,58]. In some cases, the presumed effect of Migration migration toward increased blood pressure was as The influence of migration on blood pressure status hypothesized, and was independent of the effect of of traditional societies has also been alluded to above weight [28]. In other research, control of weight and [e.g. 526,581. While ethnographic accounts exist of age differences greatly attenuated the effect of migra- migration and urbanization in Brazilian Amerindian tion on blood pressure [5,26,58]. Amazonian groups groups (Z&57], measurement of blood pressure has experiencing frequent relocations [ 16,22,23] and dec- not been presented in relation to these social changes, imation by disease [16,24], exhibited severe weight and has been rarely studied among South American loss and/or clinical malnutrition in children, female Indians. Cruz-Coke and coauthors note a ‘significant’ and elderly members [16]. Multiple relocations in this difference in the prevalence of elevated blood pres- era of more accessible transportation has been prob- sures, and in the relationship between age and blood lematic in other studies of changing societies [58]. pressure, in comparing Andean Aymara pastoralists Amazonian cases may represent an extreme effect, with Aymara migrants to the agricultural lowland and comparative data from other relocated Amazo- area near urban Arica, Chile [35, p. 561. This study of nian tribes is not available for weight and blood migration effects showed systolic pressures to be pressure change. Conversely, Vieira-Filho identifies relatively constant across age groups. Diastolic levels Xavante and Bororo individuals whose increased declined slightly for highlanders over age 50, but caloric intake and lower energy expenditure has increased significantly among similarly aged migrants resulted in the first cases of obesity in these groups [35, p. 581. The latter in fact accounts for the reported [ 151. Significant deviation in individuals’ weight significant difference in hypertension prevalence status from the group norm can be readily observed among the migrants (defined as 150/90 mm Hg or and recorded, and are most easily interpreted in more). modeling blood pressure. However, possible confounding by age-specific It has also been proposed that effect of migration weight differences, among other possibilities, are not per se on blood pressure level, apart from that of addressed in this Andean study. At least one longitu- weight change, may be temporal in nature, depending dinal study of migrants from a traditional to modem on members’ perceived stress about moving, the setting notes that when age, body mass, initial pres- frequency of travel between old and new locations, sure levels and length of residence in the new setting family support and shifts in body weight over time were controlled, the contribution of ‘migrant’ status [26, 581. It has also been noted that severe social and to percent of blood pressure variance explained was economic changes in the ‘traditional’ setting occurred insignificant for women, and for systolic measures in prior to the migration event [58]. Clearly age and sex men [58]. Diastolic readings for males initially de- differences in weight first must be accounted for, creased, then increased (+ 3.5 mm Hg), by the end of before secular and cohort effects on blood pressure the 14yr study [58]. can be understood in changing societies [8]. Problems of selection bias are of.particular con- cern in any study of the health effects of migration. Use of additional dietary sodium As a first caveat, selective mortality of persons with Other lifestyle changes may affect blood pressure, elevated pressures precludes any definitive state- at least in the short run. For example, the use of ment regarding increased blood pressure level supplemental sodium in Amerindian diets has been particularly among elder subjects [54], and in a mentioned by several authors [7,10,12, 15,611, and cross-sectional research design [59]. Migration to has been specifically studied among the Yanomamo industrial and agricultural labor markets exerts a by Carvalho (291 and by Oliver and colleagues [lo]. strong selective force for younger adults [58]. How- In the latter, the renin-aldosterone-creatinine hor- ever, mild elevations in blood pressure would go monal levels of the ‘salt free’ Yanomamo is compared unnoticed by most individuals, and the more minor to that of the Guaymi; a relatively isolated Pana- symptoms of hypertension would not be perceived as manian group which uses salt in cooking and food 598 MILLKENT FLEMING-MORANand CARLOSE. A. COIMBRAJR

preservation. In addition, the hormonal changes dur- or institutions [30] as markers of acculturation. Scales ing pregnancy and lactation are also investigated [lo]. of access to significant material goods and services In this difficult field situation, Oliver’s group has has been employed, for example, in testing the impact documented chronic elevations of renin and aldos- of ‘modernization’ on blood pressure level in urban terone in the low-sodium using Yanomamo (without Brazilian samples [67]. Thus the relative contribu- accompanying hypertension), in which the stress to tions of change in diet, the type or nature of the sodium homeostasis is further heightened during contact experience, or both, can be explored in pregnancy [lo]. It is proposed that over millenia, explaining blood pressure variance in Amerindian tropical groups with nearly vegetarian diets have populations. However, although dietary changes are survived and thrived in low-sodium environments by of social and medical interest in studies of cultural these hormonal mechanisms [IO]. change, the need for accurate weight and height data In their 1981 publication, however, Oliver and cannot be over-emphasized. This is true whether the coworkers found little direct evidence that, among goal be the comparison of mean blood pressures nonpregnant females, the higher salt/potassium ratio between groups, or longitudinal examination of of the Guaymi’s diet adversely affected their blood blood pressure status within a population. pressure levels [lo]. However, while mean blood pressure levels were similar for women in the two Other health habits cultures, the Guaymi women were on average 10 kg A great deal of the present discussion has been heavier than the Yanomamo women [lo]. Similar devoted to field methods and physical measures. It is trends in mean weight differences, salt intake and on this base that any expanded understanding of blood pressure levels have been documented for social change and hypertension can occur, before contacted and isolated sub-groups of the Yanomamo other variables of interest can be. evaluated. For [29]. Carvalho’s combined data for both sexes sup- example, current data are insufficient to determine port Oliver’s conclusions [21], that is, in the early whether other inter-group and inter-sex differences contact period, increased weight and salt intake may in use of tobacco and alcohol influence the blood not differentiate blood pressure level between tribal pressure findings. One of the authors (Coimbra) groups, but both variables became significant, inde- noted greater alcohol use (including distilled rum) pendent covariates of individual blood pressure level, among men in a multi-tribal Indian post of Western when data were combined to compare members who Amazonia (Rondonia) [see also 57,611. High levels of were, or were not fluent in Portuguese. alcohol use has been documented as an adverse factor Some would argue that increased salt versus caloric on blood pressure level in some studies of industrial intake, weight change, the sodium-potassium balance societies [68-711, and may interact with perceived in the diet and individual hormonal adaptations social disruption or stress in influencing blood make investigation of the salt-blood pressure issue pressure level [30,72]. Alcohol use is often difficult to enormously complex [62-64]. Lowland groups may estimate in such instances [72], but rapid change add table salt or commercially salted foods, but may indicate cultural disruption, if the subjects can be maintain high potassium intake through regular de- persuaded to report their drinking habits. Such pendence on bananas (Musu sp.), fish, locally made behaviors in Amerindian groups may also be influ- high-potassium salt and other foods [l l-13,65,66]. enced by the nature of contact with representatives of In short, the Na/K balance of the diet should be Western society. For example, Coimbra has noted considered. While some contacted groups change the where protestant missionaries are the chief contact, relative emphasis of certain traditional foods (e.g. the both smoking and drinking may be curtailed, as Yanomamo now exploit more fish versus game meat members accept new religious guidelines for their sources; maniac versus corn and beans), others ap- behavior. Thus increased drinking, as well as absti- pear to add carbohydrate sources (e.g. purchased nence represent distinct responses to cultural change, rice, bread, sugar) to the existing local fare [15,66]. which might have variable effects on blood pressure It has been suggested that early contact increases in level. body weight may be associated with lower energy Tobacco use is nearly universal among Amerindian expenditure as well use of labor-saving Western foods men (except as noted above), but is rarely seen among which may increase total caloric intake [13]. women [73,74]. While there is a greater prevalence of The field researcher would do well to simply outline smoking among Western hypertensives [45,70], the relative adherence to ‘traditional’ versus ‘new and nicotine produces transient increases in blood food sources, to estimate household use of certain pressure levels (75-771, it is still debated to what index foods, or to isolate significant changes in degree smoking influences resting blood pressure relevant dietary elements. Minor or recent changes in level. As a precaution, subjects of recent blood pres- sodium intake alone, for example, are difficult to sure studies are encouraged to refrain from smoking evaluate in any multivariable analysis for their spe- prior to blood pressure readings [40]. Reported smok- cific effect on blood pressure, given the limited statis- ing habits can be helpful in the data set, particularly tical power of most studies. Instead, gross dietary if the former appear to be in flux for any portion of changes may serve as a good proxy variable for the the population. broader concept of ‘culture change’ and a logical, more proximate cause of altered weight or blood Blood pressure in female subjects pressure status. Such an approach would be a logical Some authors report great difficulty in acertaining follow-up to studies of blood pressure which rank pregnancy and menopausal status among Amer- Amerindian subjects’ facility with a nonnative lan- indian women [9, lo]. In a noncontracepting popu- guage [29], or acceptance of Western material goods lation, a significant portion of women of child Blood pressure studies among Amazonians 599 bearing age may be pregnant, and consequently, 5. Patrick R. C., Prior I. M., Smith J. C. et al. The blood pressure readings will be altered for this relationship between blood pressure and modernity group [45]. While self-reported reproductive status is among Panopeans. Inr. J. Epidem. 12, 36-44, 1983. considered unreliable in certain lowland Amerindian 6. Cassel J. Studies of hypertension in migrants. In The Epidemiology and Control of Hypertension (Edited by subjects, blood or urine tests may be useful in ascer- Paul 0.). pp. 41-61. Stratton Intercontinental Medical taining pregnancy status [lo]. Female researchers Book Corp., New York, 1975. may facilitate interviews among Amerindian women. 7. Lowenstein F. W. Blood pressure in relation to age and It further has been suggested in certain instances of sex in the tropics and subtropics. Lancer 1, 389-392, cultural change, that womens’ exposure to changes in 1961. family, education and economic spheres may differ 8. Vaughn J. P. A review of cardiovascular diseases in from that of males, thus reflecting differential ‘stres- developing countries. Ann. rrop. Med. Purasir. 72, sor’ relationships with blood pressure, as noted in the 101-109, 1978 above discussion of migration effects [26,30,58]. 9. Oliver W. J., Cohen E. L., Neel J. V. er al. Blood pressure, sodium intake and sodium related hormones in the Yanomamo Indians, a “no-salt” culture. Circu- CONCLUSIONS larion 52, 146-151, 1975. 10. Oliver W. J., Neel J. V., Grekin R. J. er al. Hormonal There is a need to monitor longitudinal blood adaptation to the stresses imposed upon sodium balance pressure, and other health status changes in these by pregnancy and lactation in the Yanomama Indians, small, threatened populations. This is important not a culture without salt. Circulation 63, 110-l 16, 1981. only for the latter’s well-being, but also for our own 11. Sick H. Sobre a extra&o do sal de cinzas vegetais pelos understanding of social-health interactions under indios do Brasil Central. Revfa Museu Paulisra (New conditions of social change. Such patterns are often Series) 3, 381-390, 1949. 12. Neel J. V., Salzano F. M., Junqueira P. C. er al. Studies impossible to interpret in complex, industrial settings of the Xavante Indians of the Brazilian Mato Grosso. and the opportunity to record the early history of Am. J. Hum. Genet. 16, 52-140, 1964. such events is rapidly diminishing. If these investi- 13. Franc0 L. J. Aspectos Metabolicos da PopulacHo Indi- gations are to be useful, they must be as rigorous as gena do Alto Xingli (Brasil Central). Doctoral disser- difficult field conditions will allow, and complete tation. Escola Pauhsta de Medicina, Slo Paula, 1981. enough to address several possible theoretical alter- 14. Vieira-Filho J. P. B. 0 diabetes mellitus entre OSindios natives. This includes not only multiple aspects of dos Estados Unidos do Brasil. Reura Antrop. 21,53-60, cultural change, their differential impact on individu- 1978. als by age, sex and ethnic affiliation, but also consid- 15. Vieira-Filho J. P. B. Problemas da aculturaclo alimen- tar dos Xavante e Bororo. Revtu Antrop. 24, 37-49, eration of the varying contact experiences of 1981. Amerindian groups. 16. Baruzzi R. G., Marcopito L. F., Serra M. L. C. er al. Changes in health habits, such as diet, patterns of The Kren-Akorore: a recently contacted indigenous smoking, alcohol use, as well as access to regular tribe. In He&h and Disease in Tribal Societies, medical care should also be considered, particularly pp. 179-211. Ciba Foundation Symposium 49 (New where comparison of contemporary group blood Series), Elsevier North Holland, New York, 1977. pressure means is the aim. Obviously such an ap- 17. Lima 0. G. de. Pulque, E&hi e Pujuuaru: 1VaErno- proach requires the mutual cooperation of social biologic dar Bebidas e dos Alimentos Fermentudos. scientists, medical researchers, local leaders, national Editdra da Universidade Federal de Pemambuco, Indian agencies and public health officials, but such Recife, 1975. 18. Ramos A. R. Frontier expansion and Indian peoples in effort can be far more rewarding than any previous the Brazilian Amazon. In Frontier Expansion in Amazo- study of lowland Amerindian health. nia (Edited by Schmink M. and Wood C. H.), pp. 83-104. University of Florida Press, Gainesville, REFERENCES FL, 1984. 19. Seiler-Baldinger A. M. Indians and the pioneer-front Page L. Hypertension and atherosclerosis in primi- in the north-west Amazon. In fand, People, und tive and acculturating societies. In International Planning in Contemporary Amuzonia (Edited by Symposium on Hypertension (Edited by Hunt Barbira-Scazzocchio F.), pp. 244-248. Cambridge J. C.), pp. I-12. Health Learning Systems, New York, University Press, 1980. 1974. 20. Ribeiro D. Brazilian Indian cultures and languages. In Vital and Health Statistics. Blood Pressure Level Indians of Brazil in the Twenrierh Cenrury (Edited by in Persons 18-74 Years of Age in 1976-80, and Hopper J. H.), ICR Studies No. 2, Institute for Cross- Trend% in Blood Pressure from 1960 to 1980 in Cultural Research, Washington; DC, 1967. the United States. National Health Survey, Series 21. Ribeiro D. OS Iridios e a Civilizu@o: A Integrqdo aks 11, No. 234. Department of Health and PopuIa@es Indigenas no Brasi Moderno. Civiliza9go Human Services (No. PHS 86-1684). Hyattsville, MD, Brasileira, Rio de Janeiro, 1970. 1986. 22. Avres M. and Salzano F. M. Health status of Brazilian Epstein F. H. and Eckoff R. D. The epidemiology of &yap6 Indians. Trop. Geog. Med. 24, 178-185, 1972. high blood pressure: geographic distributions and etio- 23. Colchester M. (Ed.) The He&h and Surcicul of the logic factors. In The Epidemiology of Hypertension Venezuehm Yunomuma. International Work Group for (Edited by Stamler J. er al.). Grune & Stratton, New Indigenous Affairs, Document No. 53, Copenhagen, York, 1967. 1985. Waldron I., Nowotarski M., Freimer M. ef al. 24. Coimbra C. E. A. Jr, 0 sarampo entre socie- Cross-cultural variation in blood pressure: a quanti- dades indigenas brasileiras e algumas consider- tative analysis of the relationships of blood pressure a96es sobre a pratica de saude ptiblica entre estas to cultural characteristics, salt consumption, and popula@es. Cadernos S&de Ptibl. (Rio de Janeiro) 3, body weight. Sot. Sri. Med. 16, 419430, 1982. 22-37, 1987. 600 MILLICENTFLEMING-MORAN and CARLOSE. A. COIMBRAJR

25. Salzano F. M. Changing patterns of disease among ment of hypertension. Ann. intern. .Med. 67, 48-54, South American Indians. In Disease of Complex Etiol- 1967. ogy in Small Populations: Ethnic Differences and 47. Criqui M. H.. Mebane I., Wallace R. B. ef al. Multi- Research Approaches (Edited by Chakraborty R. and variate correlates of adult blood pressure in nine North Szathmary E. J. E.), pp. 301-324. Liss, New York. 1985. American populations: The Lipids Research Clinics 26. MC Garvey S. T. and Baker P. T. The effects of Prevalence Study. Preu. Med. li, 403416, 1982. modernization and migration on Samoan blood pres- 48. Dver A. R.. Stamler J.. Shekelle R. B. er al. Relative sure. Hum. Biol. 51, 467479. 1979. weight and blood pressure in four Chicago epidemio- 27. Marmot M. G. Affluence, urbanization and coronary logical studies. J. chron. Dis. 35, 897-908, 1982. heart disease. In Disease and Urbanization (Edited by 49. Vieira-Filho J. P. B., Russo E. M. K. and Julian0 Y. As Clegg E. J. and Garhck J. P.), pp. 127-144. Taylor & proteinas glicolisadas dos indios Xikrin e Paracang. France, London, 1980. Arq. Bras. Endocr. Merab. 31, 72-75, 1983. 28. Prior I. A. M. and Stanhope J. M. Blood pressure 50. Khosla T. and Lowe C. R. Indices of obesity derived patterns, salt use, and migration in the Pacific. In The from body weight and height. Br. J. Prer. Sot. Med. 21, Epidemiology of Arterial Blood Pressure. Developmenrs 121, 1967. in Cardiotascular Medicine. Vol. 8 (Edited by Kesteloot 51. Khoury P., Morrison J. A. et al. Relationships of H. and Joosen J. V.), pp..243-262. Mart&s Nijhoff, education and occupation to CHD risk factors in school The Hague, 1980. children and adults in the Princeton School District 29. Carvalho J. J. M., Lima J. A. C. Carvalho J. V. et al. Study. Am. J. Epidem. 113, 378, 1981. Blood pressure is directly related to the degree of 52. Langford H. G. Symposium on obesity and hyper- acculturation among primitive Yanomamo Indians. tension. J. chron. Dis. 35, 875-877, 1982. Circulation Suppl. III 72, 296, No. 1181, 1985. 53. Ross C. and Mirowsky J. Social epidemiology of over- 30. Kunitz S. J. and Levy J. E. The prevalence of hyper- weight: a substantive and methodological investigation. tension among elderly Navajos: a test of the accultura- J. Hlth sot. Behao. 24, 288-298, 1983. tion stress hypothesis. Cult. Med. Psychiar. 10,97-121, 54. Hypertension Detection and Follow-up Program Co- 1986. operative Group. The effects of treatment on mortality 31. De Lima P. E. Niveis tensionais dos fndios Kalapalo e in “mild” hypertension. New Engl. J. Med. 307, Kamaiura. Revta Brasil. Med. 7. 787-788. 1950. 976-980, 1982. 32. Cloutier S. Sang et Inrerdit chez les Zero d’dmazonie 55. De Oliveira R. C. Urbanizagcio e Tribahsmo: A inreg- Bresilienne. Groupe de Recherche sur L’Amerique ragJo dos indios Terena numa sociedade a’e classes. Latine, Universite de Montreal, 1987. Zahar, Rio de Janeiro, 1968. 33. Gregor T. Exposure and seclusion: a study of insti- 56. Figoli L. H. G. Identidad regional y “caboclismo” tutionalized isolation among the Mehinaku Indians of Indios del alto rio Negro en Manaos. Anuario Anrrop. Brazil. Efhnologv 9. 234-250. 1970. 83, 119-154, 1985. 34. De Oliveir H. E O’Estado de sairde dos indios Karaja 57. Vieira-Filho J. P. B. OS indios Caraja na cidade Aruang. em 1950. Reura Museu Paulisla (New Series) 6,489-508, Reuta Anrrop. 22, 151-152, 1979. 1952. 58. Salmond C. E., Prior I. A. M. and Wessen A. F. 35. Cruz-Coke R., Donoso H. and Barrera R. Genetic Blood pressure patterns and migration: a l4-year cohort ecology of hypertension. Clin Sri. Mol. Med. 45, studv of adult Tokelauans. Am. J. Eoidem. 130.37-52, 55s-66s, 1973. 1989-. 36. Kannel W. B., Gordon T. and Schwartz M. J. Systolic 59. Kleinbaum D. G., Kupper L. L. and Morganstem H. versus diastolic blood pressure and the risk of coronary Epidemiologic Research. Lifetime Learning, London, heart disease. Am. J. Cardiol. 27, 335-346, 1967. 1982. 37. Tyroler H. A. Hypertension. In Public Health and 60. Bray G. A. Obesity in America. Inr. J. Obesity 3, Praenrariue Medicine (Edited by Last J. M.), 1Ith edn, 363-375, 1979. ~;80~201-1227. Appleton-Century-Crofts, New York, 61. Vieira-Filho J. P. B. PrevenCHo das molbtias da cultura ocidental ou industrial: Recomenda@es ao Projeto 38. Souchek J. et al. The value of two of three, versus a Carajls. Revta Anfrop. 26, 173-177, 1983. single reading of blood pressure on a first visit. J. chron. 62. Hovel1 M. F. The experimental evidence for weight loss Dis. 32, 197-210, 1979. treatment of essential hypertension. A critical review. 39. World Health organization (WHO). Arterial Hyperren- Am. J. Publ. Hlth 72, 359-368, 1982. sion. WHO Technical Renort Series, No. 628. Geneva, 63. Reisin E. and Froelich E. D. Effects of weight reduction 1978. on arterial pressure. J. chron. Dis. 35, 887-891, 1982. 40. Kirkendall W. H., Feinleib M., Freiss E. D. et al. 64. Reisin E.. Abel R.. Modan M. er al. ‘The effect of weight Recommedations for human blood pressure determi- loss without sodium restriction on the reduction-of nation with sphygmomanometers. Hypertension 3, blood pressure in overweight hypertensives. New Engl. 509A-519A, 1981. J. Med. 298, l-6, 1978. 41. Light K. C. Psychosocial precursors of hypertension: 65. &her H. A importancia da banana entre OS indios experimental evidence. Circulation 76, 167-176, 1987. Suruana e Pakid& Revta Anfrop. 5, 192-194, 1962. 42. Schneiderman N. and Pickering T. G. Cardiovascular 66. Coimbra C. E. A. Jr. Estudos de ecologia humana entre measures of physiologic reaction. In Handbook of OS Surui do parque Indigena Aripuang, Rondonia, Stress, Reactivity, and Cardiovascular Disease (Edited Aspectos Alimentares. Bol. MU. Para. Emilio Goeldi. by Matthews K. A. et al.), pp. 145-186. Wiley (Antropol.) (New Series) 2, 57-87, 1985. Interscience, New York, 1986. 67. Dressier W. W.. Santos J. E.. Gallagher P. N. et al. 43. Wilcox J. Observer factors in the measurement of blood Arterial blood pressure and modemLtion in Brazil. pressure. Nurs. Res. 10, 4-17. 1961, Am. Anrhrop. 89, 398-409, 1987. 44. Maxwell M. H., Waks A. U. et al. Error in blood 68. Aro S. Occupational stress, health related behaviour pressure measurement due to incorrect cuff size for and blood pressure: a five year follow-up. Prev. Med. 13, obese subjects. Lancer 3, 33-36, 1982. 333-348, 1984. 45. Kaplan N. B. Clinical Hyperrension, 3rd edn. Williams 69. Cairns V., Keil U., Kleinbaum D. er al. Alcohol & Wilkins, Baltimore, MD, 1982. consumption as a risk factor for high blood pres- 46. Kannel W. B.. Brand N., Skinner J. J. Jr ef af. The sure: Munich Blood Pressure Study. Hypertension 6, relation of adiposity to blood pressure and the develop- 124-131, 1984. Blood pressure studies among Amazonians 601

70. Croft J. B., Freeman D. S., Cresanta J. et ul. Adverse 74. Wilbert J. Tobacco and Shamanism in South America. influences of alcohol, tobacco, and oral contraceptive Yale University Press, New Haven, CT, 1987. use on cardiovascular risk factors during transition to 75. Cryer P. E., Haymand M. W. er (II. Norepinephrine and adulthood. Am. J. Epidem. 121, 530-546, 1987. epinephrine release and adregenic mediation of smok- 71. Jackson R., Stewart A., Beaglehole R. et al. Alcohol ing-associated hemodynamic and metabolic events. New consumption and blood pressure. Am. J. Epidem. 122, Engl. J. Med. 295, 573-577, 1976. 1037-1044, 1985. 76. Freestone M. B. and Ramsay M. B. Effect of coffee and 72. De Frank R. S., Jenkins C. D. and Rose R. M. A cigarette smoking on the blood pressure of untreated longitudinal investigation of the relationship among and diuretic treated hypertensive patients. Am. J. Med. alcohol consumption, psychosocial factors, and blood 73, 348-535, 1982. pressure. Psychosom. Med. 49, 236249, 1987. 77. Mac Dougall J. M. et al. Selective cardiovascular effects 73. Mendoza N. U. El tabaco entre las tribus indigenas de of stress and cigarette smoking. J. Human Srress 9, Colombia. Reurn Colomb. Anfrop. 5, 11-52, 1956. 13-20, 1983.