Dietary Intake of the Urban Black Population of Cape Town: The Cardiovascular Risk in Black South Africans (CRIBSA) Study

Introduction: To determine dietary intake of 19 to 64 years old urban Africans in Cape Town in 2009 and examine the changes between 1990 and 2009. Methods: A representative cross-sectional sample (n = 544), stratified by gender and age was randomly selected in 2009 from the same areas sampled in 1990. Socio-demographic data and a 24-h dietary recall were obtained by trained field workers. The associations of dietary data with an asset index and degree of urbanization were assessed. Results: Fat intakes were higher in 19–44-year-old men (32% energy (E)) and women (33.4%E) in 2009 compared with 1990 (men: 25.9%E, women: 27.0%E) while carbohydrate intakes were lower in 2009 (men 53.2%E, women: 55.5%E) than in 1990 (men: 61.3%E; women: 62%E) while sugar intake increased significantly (p < 0.01) in women. There were significant positive correlations between urbanization and total fat (p = 0.016), saturated fat (p = 0.001), monounsaturated fat (p = 0.002) and fat as a %E intake (p = 0.046). Urbanization was inversely associated with intake of carbohydrate %E (p < 0.001). Overall micronutrient intakes improved significantly compared with 1990. It should also be noted that energy and macronutrient intakes were all significant in a linear regression model using mean adequacy ratio (MAR) as a measure of dietary quality in 2009, as was duration of urbanization. Discussion: The higher fat and lower carbohydrate %E intakes in this population demonstrate a transition to a more urbanized diet over last two decades. These dietary changes reflect the nutrition transitions that typically occur as a longer time is spent in urban centers.

The following adults were excluded: Pregnant and lactating women; those who were bedridden; those unable to give consent; those on tuberculosis treatment; those on antiretroviral therapy; cancer patients having received treatment within the last year; and individuals residing in Cape Town for less than three months.

Dietary Intake
Each participant completed a 24-h recall conducted by fieldworkers who had been trained by a dietician with many years of research experience. The participant was required to recall all foods and drinks consumed during the past 24 h using the multiple pass method [15]. The 24-h recall dietary method was used for comparability with the 1990 study, which used the same method. Visual life-size photographs and sketches of foods and measures (such as cups, glasses) were used to identify portion sizes [16]. At the end of the interview, the interviewer summarized and probed for forgotten foods. The data was coded and the South African food composition tables [17] were used to calculate the dietary intake of every person and adjusted for fortification values. In order to determine dietary quality the values of macro-and micronutrients were compared with the dietary reference intakes (DRIs) including the recommended dietary allowances (RDA), the estimated average intakes (EARs), adequate intakes (AIs) and the acceptable macronutrient distribution ranges (AMDRs) [18]. The latter provided information on acceptable dietary ranges for carbohydrate, fat, and protein, as a percentage of energy intake. Dietary quality was also assessed by determining the nutrient adequacy ratio (NAR) of each micronutrient [19]. This was based on the mean % of the EAR value of the nutrient. For example, if a nutrient had a mean NAR of 75% then it met the EAR value by 75%. A 100% met the EAR value completely. The mean adequacy ratio (MAR) of the diet comprised the sum of the NAR values, truncated to 100%, divided by 11. Unfortunately, we were unable to access the MAR of the 1990 data since we did not have access to the original data.

Data Analyses
All dietary questionnaires were checked, coded, computerized and cleaned by an experienced dietician. The dietary data was then re-analyzed by using Goldberg's method to remove the under-reporters [20]. This was done for two reasons: Firstly, the 24-h recall is known to underestimate dietary intake; and, secondly, energy intakes of the study participants were found to be very low, frequently lying well below resting energy expenditure (REE). The Goldberg equation includes dividing energy intake (EI) by estimated energy requirement (EER). Under-reporters are those regarded as falling below 0.8. To ensure comparability of the current study with the 1990 study, which was conducted in 15-64-year-old participants [9,10], the common age categories of 19-64 years were used for this analysis. t tests were used to determine significant differences between mean dietary values in 1990 compared with 2009.
The associations of different nutrient intakes with asset index, a proxy for wealth, and the duration of urbanization were assessed. A principal component analysis of the pooled data, based on the assets that defined wealth, was used to develop an asset index [21]. This was categorized into tertiles with relative wealth classified as poorest, poor and least poor. In lieu of the absence of an internationally agreed definition on what constitutes an urban environment, the proportion of life spent in the city, used in previous studies, defined the degree of urbanization [22]. Length of urban residence was recorded by summing the duration lived in a city, from birth until the date of data collection. We further computed a linear regression model using MAR as the dependent variable and added age, gender, urbanization, asset index and other variables to the model. Correlations between the asset index and urbanization duration were done with energy and nutrient intakes using Pearson's correlations.
The University of Cape Town's Research and Ethics Committee approved the study (REC REF: 224/2006). All participants signed informed consent.

Results
Dietary data was collected on 1097 participants in the CRIBSA study. Of these, 1009 were between 19 and 64 years of age. After removal of the under-reporters 544 participants (214 men and 330 women) aged between 19 and 64 years were included in the final sample for analysis. Participant distribution by area was as follows: Khayalitsha 42.4%, Langa 31.4%, Gugulethu 15.3% and less than 10% from Nyanga and Crossroads. Twelve percent of participants had lived in the city less than 20% of their lives, 44.5% between 20% and 69% of their lives, and 43.3% for 70% to 100% of their lives. By relative wealth, the poorest, poor and least poor participants comprised 30%, 54.8% and 14.8%, respectively, of the study sample.
Energy and macronutrient intakes are shown in Table 1. Mean energy intakes were on the low side in men while they appeared to be more moderate in women when compared with recommended values. Mean protein intakes were above the RDA values with nearly half coming from plant sources. Mean protein intakes and protein %E were significantly lower (p < 0.01) in men and women in 2009 compared with 1990.
Mean fat intake in 2009 (M = 70.3 g; W = 66.4 g) in the younger (19-44 years) group was significantly higher than in 1990 (M = 60 g; W = 49 g; p < 0.01). A similar scenario was found for fat %E. Saturated fat intake was lower in 2009 in men than in 1990. This difference was significant in the older (45-64 years) group (p < 0.01). In women, saturated fat intake was significantly higher in 2009 compared with 1990 (p < 0.0001) in both age groups. Fat %E, while within the recommended range, was towards the upper limit of the fat AMDR of 20%-35% in women and younger men. Saturated fat %E was above the recommended value of <7%E. Polyunsaturated fat intake and polyunsaturated fat %E were significantly higher in men and women in 2009 (p < 0.0001) compared with 1990. Similar results were found for the diet polyunsaturated/saturated ratio. Cholesterol intake, at >300 mg per day, was significantly higher in younger men and women in 2009 compared with 1990 (p < 0.001).
Mean carbohydrate intakes (which included added sugar) were nearly double the EAR values and were significantly lower in young men in 2009 (247.8 g vs. 282 g; p < 0.001) compared with 1990, while in women they were significantly higher in 2009 in younger (232.4 vs. 214 p < 0.01) and older women (221.4 g vs. 198 g; p < 0.01). Mean carbohydrate %E falls within the AMDR of 45%-65% and while not significant were lower in 2009 than in 1990. However added sugar is greater than the WHO [1] recommendation of less than 10%, with the exception of the younger men. Added sugar intake increased significantly in women in 2009 (p < 0.01), while it remained similar in men in the two studies. Mean fiber intakes were considerably lower than the AI of 25 g, in both studies.
Mean calcium intakes were very low in both studies lying below the AI of 1000 mg ( Table 2). Values for men in 1990 were significantly higher than for men in 2009. Mean iron intakes were above the EAR in both studies. Vitamin A values were above the EARs in all groups with the exception of older men in 1990. Thiamin, riboflavin, niacin, vitamin B6, folate and vitamin C fell below the EARs in women in the 1990 study. In men in 1990, vitamin B6, folate and vitamin C were less than the EARs. Significant differences in mean intakes were noted between the two studies, which were generally higher in the 2009 study. This is notable for iron, folate, vitamin B6, niacin, thiamin, riboflavin and vitamin A, where there were significantly higher intakes compared with 1990. Calcium intakes remained low and zinc intakes were lower than the 1990 study. Table 3 presents the NARs for the micronutrients. Generally the NAR values are highest in men in the 2009 study, followed by women in the 2009 study and lowest in women in the 1990 study. The mean NAR for vitamin A was above 100% in all groups with the exception of older men in 1990. Other nutrients that had NARs less than 100% were vitamin C, calcium and folate. In women in the 1990 study, calcium, vitamin B6, vitamin C, folate, riboflavin and thiamin were less than 100%.
The numbers of food portions (based on diabetic exchanges) eaten per day from five groups are shown in Table 4. The highest number of portions consumed in the 2009 study was from the cereal group with men and women having 9.45 and 8.73 mean portions per day, respectively. Cereals were followed by the fat group, with 3.30 and 4.20 mean portions per day for men and women, respectively.
Vegetables and fruit are next with 2.78 and 2.90 portions respectively, followed by the meat group and lastly the dairy group with both men and women consuming less than one portion a day. For both men and women the portions in the 2009 study were significantly lower than those of the 1990 study for milk products, meat group, legumes, cereals (men only), saturated fats and brick margarine. However, the 2009 study had significantly higher intakes of eggs, vitamin C rich fruits and vegetables, cereals (women only), and polyunsaturated sources.
Correlations of nutrients with duration of urbanization (and the asset index score) were done to test for significant associations) ( Table 5). Total energy intake and carbohydrate intake were not associated with duration of urbanization or with the asset index. However, total protein and fat intake were significantly associated with both. Significant positive correlations were found between most sub-groups of protein and fat with urbanization and the asset index except for polyunsaturated fat and cholesterol. The polyunsaturated fat/saturated fat ratio, carbohydrate %E, added sugar and sugar %E were inversely correlated with urbanization as well as with the asset index, except for added sugar. Sodium, zinc, thiamine, niacin and vitamin B6 were significantly associated with urbanization while calcium, zinc, thiamine and niacin were significantly associated with the asset index.
A linear regression model is presented in Table 6 with MAR as the dependent variable. Kilojoules and total fat were found to be highly correlated with total protein, and if kilojoules and total fat are included in the regression model, it results in multicolinearity, which lead to insignificant relationships between MAR and kilojoules and total fat, respectively. It is more useful to delete them from the model. Similarly, urbanization duration and asset index are highly correlated, and only one of the two variables need to be included in the model. For every BMI unit increase the mean MAR will increase by 0.23%. For every gram of protein consumed the mean MAR will increase by 0.26%. While for every gram of protein the MAR will increase by 0.03%. However, for every one-gram increase in the intake of added sugar, the mean MAR will decrease by 0.04%. The average MAR is 2.86% higher for females than males. These relationships are true if all other variables remain constant. Energy and macronutrients were all significant in the regression model, as was duration of urbanization.     [24]. * p < 0.05; ** p < 0.01; *** p < 0.0001.   * p < 0.05; ** p < 0.01; *** p < 0.0001; 1 Urbanization duration: Time in years spent in the city; 2 Asset index: A principal component analysis of the pooled data, based on the assets that defined wealth, was used to develop an asset index [20].

Discussion
The findings pertaining to dietary intake in the CRIBSA study are in keeping with the nutritional transition occurring in urban centers in South Africa. Unlike rural areas where commonly, traditional diets that are low in fats (<25% of energy intake) and sugar, and high in carbohydrates are consumed, in urban centers there is a shift with the adoption of more Western diets. These are higher in fat consumption (>25% of energy intake) and lower in carbohydrates [25] as mirrored in this study.
Indeed, this is further illustrated by the significant positive correlations between greater urbanization duration and total fat, saturated fat, monounsaturated fat, and fat as a percent of energy intake. A further reinforcement of this transition is the differences observed in fat intake between 1990 [10] and 2009. This rise is due to an increase in polyunsaturated fats since saturated fat intake remained similar in the two studies, although rising significantly in the younger women.
Additionally, the significant inverse correlation of carbohydrate (which includes added sugar) as a percent of energy intake with urbanization duration reinforces the shift to a Western diet in urban centers. Additionally, the proportion of carbohydrate intake was significantly lower in men in 2009 compared with 1990 reinforcing the nutritional transition. These higher proportions of fat and lower proportions of carbohydrate intake, highlight that over two decades the dietary intake in this population has transitioned to a more urbanized and Western diet [25].
In addition to urbanization, socio-economic advancements are also known to be associated with westernization of lifestyles and increased uptake of energy-dense foods high in fats [26]. These foods are likely to be more affordable to individuals with higher socioeconomic status (the affluent and better educated) but may still be out of the price range of lower socio-economic groups, particularly in developing regions [27]. This accords with these study findings that showed a significant association between wealth and fat intake.
The overall energy intake of this urban population was within the range of other South African urban reports. It was greater than those of urban participants in a study in Free State Province [28] (M = 7078 kJ; W = 6621 kJ), one of few recent studies in adult blacks, but lower than those of participants in urban North West Province (PURE study) [29] (M = 10,054 kJ; W = 9008 kJ).
Notably, the energy intake in men was similar in 1990 and 2009, which may account for the same obesity prevalence of 9.5% in the two studies [11]. In contrast, energy intake in women increased from 1990 [10] to 2009 as did obesity levels. However, energy intake was within the normal range and does not reflect the extremely high prevalence of obesity among women in this population (1990: 41.7%; 2009: 61.5%). It should also be noted that energy and macronutrient intakes were all significant in the linear regression model using MAR in 2009, as was duration of urbanization.
Mean protein intake in this study was lower than that reported in the Free State [28] (M = 76 g; W = 69 g), the North West [29] (M = 74 g; W = 64 g), and in the 1990 study [10] (M = 77 g; W = 56 g). This may be due to socio-economic reasons since protein-rich foods are generally more costly, particularly animal protein sources and dairy products.
In the 1990 study numerous mean micronutrient values were lower than the DRI values [9,10]. Low micronutrient levels in men included calcium, vitamin A, vitamin B6, folate and vitamin C. In women deficient nutrients included calcium, thiamin, riboflavin, niacin, vitamin B6, folate, and vitamin C. However, most of these improved substantially as evinced by the higher NAR values for iron, zinc, vitamin A, folate, vitamin B6, niacin, thiamin and riboflavin in the 2009 study. This is most probably the result of mandatory fortification of maize meal and wheat flour, which was introduced in 2004. Calcium and zinc intake however remain below the recommended values. Since both studies were undertaken in the city of Cape Town, it is not possible to make comparisons with the rural areas; however, the national food consumption study [30] in South Africa in 1999 did find that micronutrient deficiencies were more prevalent in rural areas.
The lower than recommended intake of fruits and vegetables in this study is of concern because adequate amounts are necessary to lower the risk of IHD, stroke and high blood pressure, as well as stomach and colorectal cancers [5,31,32]. Fruit and vegetables contribute to cardio-metabolic health through numerous phyto-nutrients, potassium and fiber [31,32].
In addition, the low meat group consumption in women could lead to possible low iron and vitamin B12 intakes. The very low consumption of dairy products explains the low mean calcium and riboflavin values in both studies. The portions of food groups consumed are also a reflection of urbanization of diet. As with the 1990 study, the lowest intakes are in the dairy group [9]. The 1990 study found that with increased time spent in the city dairy consumption decreased by 33% and cereal intake by 26% while fruit and vegetable intake increased by 19%, non-basic foods by 17%, fats by 8%, and meat by 14% [9].
The 1990 study determined that overall the dietary pattern of the urban black population generally met the prudent dietary recommendations of the American Heart Association (AHA) [33] namely that fat intake should be less than 30% of energy intake, saturated fats less than 10% energy intake and sugar less than 10% energy intake. To a large degree, the 1990 population was considered to be in a transition phase regarding their dietary intake since many participants had only lived in the city for a short period of time.

Limitations
Limitations of the study include the use of a single 24 h recall, which is known to under-estimate dietary intakes [34]. Nonetheless, this method was used to ensure comparability with the 1990 study, which used only one 24 h recall. Another limitation was the fact that we did not have access to the original data of the 1990 data and had to rely on published data. Removal of under-reporters using the Goldberg method [20] could also be regarded as a limitation since the sample was nearly halved, which means the findings are not generalizable. This limitation was particularly noticeable when interpreting the mean energy intakes of the women. These values are within the normal to lower range and do not actually reflect the high prevalence of obesity found in women in this study. However, it has been shown that obese women do tend to under-report their intake [35]. It is also likely that low physical activity levels may have contributed in this regard since the national survey [8] did find a low prevalence of physical activity in African women.

Conclusions
The nutrient intakes demonstrate that while certain changes have taken place between 1990 and 2009, the dietary pattern regarding foods eaten remains poor. The diet has become more urbanized and atherogenic with regard to distribution of fat and carbohydrate intake while the consumption of certain food groups have remained low, such as the poor consumption of dairy products and low intake of fruit and vegetables. On the other hand, low micronutrient intakes have increased over the two time periods as a result of mandatory fortification of staple foods in 2004. Interventions are urgently needed to combat the shift towards the diet continuing to become more atherogenic and to improve the consumption of priority food groups, such as dairy products and fruit and vegetables.