Article Text

Download PDFPDF
Daily energy expenditure and cardiovascular risk in Masai, rural and urban Bantu Tanzanians
  1. J A Mbalilaki1,2,3,4,
  2. Z Masesa3,
  3. S B Strømme1,
  4. A T Høstmark1,2,
  5. J Sundquist4,
  6. P Wändell4,
  7. A Rosengren5,
  8. M-L Hellenius4
  1. 1Institute of Sport and Biology, The Norwegian School of Sport Sciences, Oslo, Norway
  2. 2Section of Preventive Medicine and Epidemiology, University of Oslo, Oslo, Norway
  3. 3Muhimbili University of Health Allied Sciences, Dar-Es Salaam, Tanzania
  4. 4Karolinska Institutet, Centre for Family and Community Medicine, Stockholm, Sweden
  5. 5Section of Preventive Cardiology, Department of Medicine, Sahlgrenska University Hospital/Östra, Gothenburg, Sweden
  1. Julia Aneth Mbalilaki, Grünersgt 2, 0552 Oslo, Norway; j.a.mbalilaki{at}


Background Several studies have revealed that the Masai, pastoralists in Tanzania, have low rates of coronary heart disease despite a diet high in saturated fat. It has also been suggested that they may be genetically protected. Recent studies detailing other potential protective factors, however, are lacking.

Methods A cross-sectional investigation of 985 Tanzanian men and women (130 Masai, 371 rural Bantu and 484 urban Bantu) with mean age of 46 (9.3) years. Anthropometric measures, blood pressure, serum lipids, and the reported dietary pattern and physical activity level were assessed.

Results 82% of Masai subjects reported a high fat/low carbohydrate intake, whereas 77% of the rural Bantu subjects reported a low fat/high carbohydrate intake, while a high fat/high carbohydrate intake was the main dietary pattern of the urban Bantu group as, reported by 55%. The most conspicuous finding for the Masai was the extremely high energy expenditure, corresponding to 2565 kcal/day over basal requirements, compared with 1500 kcal/day in the rural and 891 kcal/day for the urban Bantu. Mean body mass index among the Masai was lower than the rural and urban Bantu. Mean systolic blood pressure of the Masai was also lower compared with their rural and urban Bantu counterparts. The Masai revealed a favourable lipid profile.

Conclusion The potentially atherogenic diet among the Masai was not reflected in serum lipids and was offset probably by very high energy expenditure levels and low body weight.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Despite overwhelming evidence of a link between saturated fat and coronary heart disease (CHD), the near absence of CHD in Masai people, despite a diet based on meat and milk, has intrigued researchers for decades.1,,7 The Masai are seminomadic pastoralists and warriors, primarily found in East Africa, in northern Tanzania and in southern Kenya.

Approximately four decades ago, studies among the Masai people provided reports about their diet high in saturated fat2 3 8 9 was assumed to result in a favourable lipid profile. However, several authors instead found low levels of serum cholesterol and low rates of CHD in the Masai.3,,6

More recent studies in the Masai are lacking. Therefore, the aim of this study was to analyse and compare lifestyle and CHD risk factors in Masai with those of rural and urban Bantu Tanzanians.



The study population included samples of Masai (n = 130), rural Bantu (n = 371) and urban Bantu (n = 484). The Masai sample was drawn from three villages in the Monduli district, pastoralists and farmers with very low socio-economic status (fig 1). The rural Bantu sample was drawn from three villages in the Ilala district and three villages in the Handeni district, involving farmers (96%) with very low socio-economic status, and only 4% with skilled occupations. The urban sample included inhabitants from the city of Dar-Es-Salaam and the Morogoro town with different socio-economic statuses. Sixty-two per cent were employed in white-collar jobs and 20% in blue-collar jobs, with the rest (18%) involved in small business. Initially, the study proceeded by the first author meeting with region heads and district leaders. Finally, the head of the local community and workplace leaders cooperated, introducing the survey and identifying participants. Eligibility was based on age (30–60 years), permanently residing or employed in selected areas. There were 950 eligible people in the rural (Masai, 350 and rural Bantu, 630) and 1050 in the urban community. All data were computerised, and a random list of 160 Masai, 440 rural Bantu and 600 urban Bantu were selected from this sampling frame to participate in the survey. Selected individuals were visited by the first author 1 month before the survey. In additional to verbal instructions, each participant was given a formal written invitation for the survey signed by the head of the local community and workplace leaders. Participation was voluntary. The response rates in the rural and urban areas were 81%, 84% and 81%, respectively.

Figure 1

A young Masai warrior.

The individual testing survey’s data collections were held at the local administration offices, the villages’ primary schools or workplaces. Blood samples were collected and analysed for blood glucose and later on serum lipids. Physical activity level, diet, tobacco use and alcohol consumption were assessed by an interview administered by the first author.

Physical activity, alcohol consumption and tobacco use

Physical activity at work, at home, during leisure time and transport were estimated by interview-administered questionnaires as modified from validated questionnaires 10 11 and translated into Swahili, slightly adjusted for Tanzanian conditions. Physical activity was calculated and classified in metabolic equivalents (METs).12 Energy expenditure values were calculated as kcal/kg/day. Diet, social habits and occupational status were recorded. Dietary habits were based on 24 h dietary recall (Appendix 1), and based on the response, each subject was assigned to one of four diet groups (table 1). The participant was asked whether or not he or she consumed alcohol or was a current cigarette smoker.

Table 1

Percentage of subjects in the Masai, rural and urban Bantu reporting various dietary patterns of fat and carbohydrate intake


Height and weight were measured to the nearest 0.5 cm and 0.5 kg. Body mass index (BMI) was calculated as: weight (kg)/(height (m))2. Waist circumference was measured at the narrowest point or at the umbilicus if no narrowest point existed. Hip circumference was measured at the widest part of the buttocks. The waist-to-hip ratio (WHR) was expressed as the ratio of waist to hip circumference. All measurements were made with light clothing and no shoes. Blood pressure (in the right arm) was measured three times by trained and certified staff, using a pull-on Velcro cuff for either thin or fat adults, after 5 to 10 min rest in a sitting position using an oscillometric minicomputer with an automatic measuring sequence (Tensoplus OSZ 2, Germany). The average of the second and third blood pressure readings was used in the analysis for the study.

Non-fasting blood samples (10 ml) were drawn after the interview and kept in a cold box until centrifuged. The serum was kept at -45°C before transportation to Norway. Serum total cholesterol, triglycerides and HDL cholesterol (HDL-C), were analysed using reagents manufactured by Boehringer Mannheim (Germany), at the Section of Preventive Medicine and Epidemiology, University of Oslo, Norway. The standard serum Seronorm Lipid (Sero AS, PO Box 24, N-1361 Billingstad, Norway) served as control. Serum triglycerides were assessed enzymatically.13 14 Determination of HDL-C was made after precipitation of chylomicrons, VLDL cholesterol and LDL cholesterol (LDL-C), using phosphotungstic acid and magnesium ions.15 LDL-C was calculated by using the Friedewald formula: LDL-C = TC–HDL-C–(TG/2.2).16 Apolipoprotein A-1 and B were determined with a commercial immunochemical kit (Turbiquant, Behring).

Statistical methods

Characteristics of the subjects are described as mean (SD) and percentages. Because some of the data were skewed, the Wilcoxon rank-sum test was used to assess differences in continuous variables. The χ2 test was used to investigate differences in the categorical variables.

Multiple linear regression analyses with total cholesterol/high-density lipoprotein ratio and ApoB/ApoA-1 ratio as the dependent variables and ethnic groups as exposure was performed. The results are shown as β coefficients with 95% confidence intervals (CI). Three models were used and crude model also adjusted for energy expenditure model; full model also adjusted for BMI. The interactions between the other independent variables were also tested (data not shown). Explained variance (R2) was used as a measure of goodness-of-fit. All models met the requirements in these test. A p value <0.05 was considered statistically significant. Stata V.9.2 was used for analyses.17


Characteristics of the study populations

Characteristics of the three study populations are given in table 2. The urban Bantu were somewhat younger than Masai and rural Bantu. There was a higher percentage of tobacco users among the Masai compared with the rural and urban Bantu, respectively. Alcohol use was more common in the Masai and urban Bantu compared with the rural Bantu. The Masai had the lowest BMI values compared with rural and urban Bantu. Corresponding differences were found for WHR. Mean blood pressure among the Masai was also lower than among rural and urban Bantu. The urban group had higher lipid levels and ratios than the Masai and rural populations, whereas the difference between Masai and rural Bantu was small, with slightly higher total and LDL-C among the Masai. LDL/HDL and ApoB/ApoA-1 ratios were lowest among the Masai.

Table 2

Characteristics of the study populations in mean (SD) or percentage (%)

Daily energy expenditure and diet

The most conspicuous finding for the Masai was an extremely high reported mean physical activity, corresponding to daily energy expenditure; 45.22 (14.3) kcal/kg/day = 2565 kcal/day, compared with 24.65 (12.4) kcal/kg/day = 1500 kcal/day in the rural Bantu and 16.52 (8.6) kcal/kg/day = 891 kcal/day for the urban Bantu (fig 1). In addition to basal requirement, the mean energy expenditure for the Masai subjects was approximately twice that of the rural Bantu and tree times as high as for the urban Bantu.

A majority of Masai subjects reported a high fat/low carbohydrate diet. In contrast, a high proportion of rural Bantu reported a low fat/high carbohydrate diet, whereas high fat/high carbohydrate was the predominant diet pattern of the urban Bantu (table 1).

Association between lifestyle and lipids

Table 3 shows results by multiple linear regressions (β-coefficient and 95% CI) by ethnic groups (Masai, rural and urban Bantu) for total cholesterol/high-density lipoprotein ratio and ApoB/ApoA-1 ratio, respectively. Three models presented, model 1, crude model; model 2, adjusted for energy expenditure; and model 3, also adjusted for BMI. Adjustment for age did not change the results and was omitted. In the crude model, the total cholesterol/HDL-C ratio was lower among the Masai compared with both the rural and urban Bantu groups, respectively. When also adjusted for energy expenditure and BMI, the difference persisted although on a lower level, compared with the rural and urban Bantu groups, correspondingly. Furthermore, the ApoB/ApoA-1 lipoprotein ratio was lower in the Masai in comparison with both the rural and urban Bantu groups in the crude model. In the energy expenditure adjusted model, the ApoB/ApoA-1 lipoprotein ratio remained lower in the Masai compared with the rural Bantu group, however not compared with the urban Bantu group, while in the BMI-adjusted model, none of the differences were significant, indicating that differences in ApoB/ApoA-1 lipoprotein ratio largely were explained by differences in these factors.

Table 3

Results by multiple linear regression (β-coefficient and 95% confidence interval, CI) for total cholesterol/high-density lipoprotein ratio and ApoB/ApoA-1 ratio, respectively


The most marked differences between the Masai people and the urbanised Bantu were the very high energy expenditure of the Masai, in addition to a markedly low body weight, with very low rates of obesity, with the rural Bantu occupying an intermediate level. Despite their staple diet high in animal fat, lipid levels were only slightly higher than among the rural Bantu, and lower than for the urban Bantu. LDL/HDL-C and ApoB/ApoA-1 ratios were lowest for the Masai. These findings are similar to those reported among the rural and urban populations of Tanzania.18

Whereas cardiovascular diseases (CVD) have long been common in developed countries, mortality rates from these diseases are now falling rapidly. With rapid urbanisation in developing countries, CVD are now rising markedly in many former low-risk countries. It is expected that by 2020, more than 80% of global CVD deaths will be in developing countries.19 The INTERHEART study, involving 15 152 myocardial infarction (MI) cases and 14 820 controls from 52 countries, showed that nine risk factors (smoking, ApoB/ApoA-1 ratio, hypertension, diabetes, obesity, psychosocial factors, lack of daily fruit and vegetables, physical inactivity and lack of alcohol consumption) were associated with more than 90% of the population attributable risk (PAR) of myocardial infarction.20 In a subanalysis of the African population, the risk factors for MI were found to be similar. Five risk factors (smoking, diabetes, hypertension, abdominal obesity and the ratio of ApoB/ApoA-1) provided a population attribution risk of 89%.21

The rapid and marked changes over time in prevalence and mortality from CVD, as well as the great geographical variation, suggest that CVD is strongly related to social changes. In the present study, we observed a gradient from the Masai, across the rural Bantu, with the urban Bantu showing high rates of obesity, and levels of lipids and blood pressure equalling those of many Western communities. In general, low education and income is associated with higher risk of MI,22 but in the sub-analysis of the African population in the INTERHEART study, differences were found with the black African group showing a positive association between high education and MI.21 This is a sign that this group is at a different point of the epidemiological transition and, therefore, the development of the CVD epidemic.23 The results in our study, with a higher rate of abdominal obesity, a higher energy intake and lower energy expenditure in the urban Bantu population, illustrates this change.

One of the main findings from this study of the Masai and rural and urban Bantu Tanzanians is the very high levels of energy expenditure of the Masai, probably contributing to their low levels of risk factors.24 25 These findings support the observations among other African societies.26,,28 It has been recognised that inactive Westerners would have to walk at least 19 km more per day to expend the same amount of the energy as the Masai do.29 Additionally, regular walking or cycling to and from work was common. This type of physical activity has been associated with low levels of coronary risk factors and low mortality from cardiovascular causes.30 31 Indeed, work and commuting physical activities were the main sources of the study subjects’ total physical activity rather than leisure time physical activity. Not only that, those people are used to carrying heavy loads on their backs or heads, for example, babies, water and firewood. Fetching water and firewood was part of their daily activities throughout the year.

In the three populations studied, several factors should be taken into account in understanding their daily life. All Masai and rural residents had a traditional African lifestyle, residing in poor and short-term housing with no electricity or water. They were pastoralists, farmers or fishermen, and very few were employed. Majority of those who were employed were exposed to hard physical labour, in addition to manual farming, fishing or cattle work. In the Masai and rural areas, there was almost no regular public transport. Bicycles were often used for travel, and walking was common. They could walk (continually relaxed) from morning to evening crossing mountains and hills to their destinations, for example, taking animals for grazing, farming, going to the market or hospital, or fetching water and firewood. The findings with respect to risk factors could provide some support for the contention that time spent in light to moderate walking may be more beneficial than a rapid walking pace for coronary risk factors.30 32

With respect to the dietary pattern, there were within-population differences in the prevalence of subjects reporting various combinations of fat and carbohydrate intake. The dietary assessment used in the study was simple and did not allow for detailed calculation of consumption of fat or carbohydrate. However, the major proportion of Masai subjects reported high fat/low carbohydrate intake, rural Bantu low fat/high carbohydrate and urban Bantu high fat/high carbohydrate intake. The findings indicate that the diet of the Masai is still based on the traditional staple food of milk, dairy products and meat from their animals, with a high fat content. Few Masai individuals reported high amounts of dietary carbohydrate such as beans, vegetables, peas and lentils, which were more commonly consumed in both rural and urban Bantu areas. In spite of the high fat intake, coronary risk factor levels such as body mass index, waist circumference, blood pressure and lipids were favourable among the Masai, possibly because of their high levels of physical activity. Additionally, however, the Masai have a marked seasonal variation in their dietary patterns, and during the dry season, fat diet intake is supplemented with carbohydrate. They also have an irregular meal pattern of overeating followed by fasting. One factor that could contribute to a favourable lipid profile is that the lipid pattern in cattle, sheep and goat grazing in the natural habitat may be different from that of animals bred in industrialised settings. In the urban areas of Tanzania, the western diet seems to influence eating habits, with more fried food than in the rural areas (personal observation).

The mean values for all blood lipids were low, with the total cholesterol concentrations of all groups lower than the normal range of values for African-Americans.33 This finding is in accordance with other studies conducted among rural and urban black Africans.25 26 34 Evidence from cross-sectional studies indicates that physical activities using 1200 to 2200 kcal/week are associated with a favourable cardiovascular risk profile, particularly with respect to lipid levels.32 In fact, this amount of moderately intensive physical activity is within the recommended range for healthy adults in Western communities.31 32 The observed low lipid levels in this study population could be the result of high daily energy expenditure (Masai, 2565 kcal/day = 17 955 kcal/week; rural, 1500 kcal/day = 10 500 kcal/week; urban; 891 kcal/day = 6237 kcal/week).

The low levels of HDL-C among all three groups of the present study, particularly among the rural Bantu, were unexpected, in view of the high levels of physical activity. Possible explanations may be genetic factors, and for the rural Bantu, their total cholesterol level and dietary patterns; low fat/high carbohydrate intake.35 It might also be that the Masai reveal lower levels but more antiatherogenic HDL particles, that is, larger.36 This observation could suggest that a low HDL-C concentration level may not enhance the risk of CHD, but should probably be considered in relation to other atherogenic lipids, especially ApoB/ApoA-1 ratio. In the INTERHEART study, ApoB/ApoA-1 ratio was revealed as the most powerful risk factor and regular physical activity as a protective factor.18 This is supported by a study done among Norwegians that regular physical activity maintain a favourable level of ApoB/ApoA-1 ratio.37 In our study, the Masai group had considerably higher level of physical activity and lower ApoB/ApoA-1 ratio than the Bantu. By multiple linear regressions, the ApoB/ApoA-1 lipoprotein ratio was higher in both the rural and urban Bantu groups compared with their Masai counterparts. Additionally, the median value for ApoB/ApoA-1 ratio in the INTERHEART study among the controls was 0.80, which puts the values for all three African populations in the present study in the mid-range globally.

Overall, our results are not coherent with past views that the low prevalence of CHD in the Masai, despite their high intake of animal fat, is genetically determined.1,,7 We also found trends towards increasing levels in mean total cholesterol and other risk factors compared with previous reports. The African rural daily lifestyle, with long periods of the day spent walking seems compatible with low rates of CHD, particularly in the Masai. However, rapid changes in many environmental and socio-economic factors have taken place in Tanzania over the past two decades, and naturally, the Masai people have also been affected. Like other rural young people, Masai young men are moving to urban areas searching for jobs, and the effects of this rural–urban transition are as yet unknown.

There are several limitations to this study. First, studies about physical activity and diet are sparse in developing countries, and the data for the study were obtained from participant responses to a questionnaire originally worked out for use in developed populations. The subjects were not always able to ascertain the duration, intensity and frequency of their different activities. Second, the diet analysis and pattern grouping are crude and were obtained from food frequency questions, which could not be translated into actual amounts of carbohydrate and fat. Third, the blood samples were non-fasting. However, apolipoprotein levels, in addition to being a more precise marker of an atherogenic lipid pattern than single lipid concentrations, are not influenced by the fasting status of the individual. Fourth, even though the participants were randomly selected from a list of residents, there is no real personal registration system to ensure that the sample was truly representative of the respective background populations. The strengths of our study are the comparatively large sample size and the fact that the response rate was high, as well as the uniqueness of this sample, because there are very few studies on physical activity, diet and CHD risk factors available in developing countries.

In conclusion, rural Masai people probably are not genetically protected from developing CVD. High energy expenditure and maintenance of low body weight may play an important role in their low risk of CVD. Preventive measures are needed to encourage rural Africans to maintain the benefits of their traditional lifestyle in parallel with adopting some of the good things of a western lifestyle.

Figure 2

Energy expenditure (kcal/kg/d, mean (SD)) in Masai, rural and urban Bantu Tanzanians.

Appendix 1 Dietary and drinks list from study questionnaire

What type of the following food did you eat and alcohol containing drinks did you consume the past 24 h? (Please respond for each food and drink item and write down how much you ate/drank)



  • Funding This study was supported by grants from the Norwegian Centre for International University Cooperation and the Swedish Council for Working Life and Social Research. The technical assistance of Eva Kristensen is gratefully acknowledged. The subjects who willingly gave their time to participate in the survey are appreciated.

  • Competing interests None.