1. Introduction
Type 2 diabetes (T2D) prevalence and the burden it places on populations is increasing globally, thereby making it a public health challenge which requires urgent attention [
1]. For instance, the global prevalence of T2D among women increased from 5% to 7.9% from 1980 to 2014 [
2]. The diabetes prevalence in Africa is projected to have the largest increase of 109% compared to other regions in the world by 2035 [
3]. Urban black South Africans have not been spared from the T2D burden. The highest prevalence of T2D in Sub Saharan Africa was reported among urban black South Africans, with 60% of these cases being reported amongst women [
4]. Black South African women also have the highest prevalence of obesity, which has been reported to be rising together with T2D in this population [
5].
The adoption of Westernised lifestyles by urban dwellers is suggested to be among the leading factors resulting in the increase in non-communicable diseases (NCDs) such as T2D in developing countries [
6]. The migration of populations from rural to urban areas has been accompanied by an increase in meat consumption as well as an increase in the consumption of sugary foods in South Africa [
7]. Both of these foods are recognised as dietary risk factors for T2D [
6]. However, improvements in micronutrient intake among black South African women in urban areas is suggested to be a result of the increased consumption of fruits and vegetables which are protective of T2D risk [
8,
9,
10]. Thus, the role of diet as a risk factor for T2D is complex among the black South African women. In addition, people eat meals with a variety of nutrients which have interactive and synergistic effects on health [
11]. Therefore, it is difficult to determine the separate effect of a food or nutrient on disease development as it is highly interrelated with other nutrients [
11]. There is need of dietary pattern analysis methods which are able to evaluate the diet as a whole and clarify the effects of the consumption of sugary foods and meat products, together with improved intakes of fruits and vegetables, related to T2D risk amongst this population group.
Dietary pattern (or food pattern) approaches comprise data driven methods such as factor analysis, which allow the dietary information (or food intake) at hand to determine the unique dietary pattern for the population group being evaluated [
10]. Although dietary patterns have been associated with disease risk, their effect is considered to be through nutrient intake; therefore, it is pivotal to determine nutrient patterns that are associated with T2D risk [
12]. This information will aid in the understanding of the aetiology of T2D. The foods that people eat are governed by their cultural norms and beliefs, which vary amongst ethnic groups, making dietary patterns limited and not applicable across divergent population groups [
12]. Evidence exists that the Western dietary pattern association with fasting glucose varies among different ethnic groups [
13]. However, nutrients are universal, thereby making nutrient patterns associations with disease risk applicable for multiple population groups [
14]. To the best of our knowledge, no study has reported on the association of the nutrient patterns with fasting glucose and glycated haemoglobin levels among apparently healthy individuals. T2D is a complex and multifactorial diseases, however glycated haemoglobin and fasting glucose levels are proxies for the development of this disease which are affected by diet [
15]. Fasting glucose levels are indicative of the short term changes in glucose metabolism, while glycated haemoglobin depicts the long term changes [
15]. This study seeks to evaluate the association of nutrient patterns derived by factor analysis with fasting glucose and glycated haemoglobin levels among apparently healthy black South Africans.
4. Discussion
We set out to determine the nutrient patterns associated with fasting glucose and glycated haemoglobin levels among apparently healthy volunteers. The principal component analysis method enabled the extraction of three nutrient patterns among a black South African population which explained about 73% of the variation of the nutrient factors in the urban/rural and gender stratifications. The magnesium, phosphorus and plant protein driven nutrient pattern was associated with a trend of increasing fasting glucose and glycated haemoglobin levels per 1 SD increase in the pattern in the rural women while the thiamine, zinc and plant protein nutrient pattern was associated with a positive trend of increasing glycated haemoglobin among urban men. Notably, the starch, dietary fibre and B vitamin nutrient pattern was associated with decreases in glycated haemoglobin and fasting glucose levels, −0.175% ((−0.303; −0.047); p = 0.007) and −0.236 mmol·L−1 ((−0.458; −0.014); p = 0.037), respectively, among rural women. The thiamine, zinc and plant protein driven nutrient pattern was associated with significant reductions in fasting glucose and glycated haemoglobin of −0.382 mmol·L−1 (−0.752; −0.012; p = 0.043) and −0.288% (−0.543; −0.033; p = 0.027) in rural men. These associations were significantly maintained after adjusting for age, BMI, log total energy intake, smoking, physical activity, alcohol intake, seasonality and education level attained thus indicating an independent association of the starch, dietary fibre and B vitamin nutrient pattern and the thiamine, zinc and plant protein driven nutrient pattern with fasting glucose and glycated haemoglobin in the rural women and rural men strata’s, respectively.
Comparable studies evaluating the associations of nutrients patterns with fasting glucose and glycated haemoglobin are scarce. However, evidence exists of dietary patterns which have evaluated this phenomenon [
10]. The Health/Prudent dietary pattern has been associated with decreases in fasting glucose and glycated haemoglobin levels while the Western dietary pattern has been associated with increases in these biomarkers of T2D [
27,
28]. However, the association of the Western dietary pattern, which is characterised with high intakes of animal proteins and snacks, with fasting glucose levels was noted to vary among ethnic groups [
13]. The Western dietary pattern was reported in one study to be only significantly associated with high fasting glucose levels and glycated haemoglobin levels among Dutch and not among the Moroccans and Turkish groups [
13]. Dietary patterns are known to vary among different ethnic groups, therefore nutrient patterns that were considered in our study are considered helpful in indicating a non-ethnic insight into this phenomenon.
In our study, we noted that fat and animal protein driven nutrients were also not associated with fasting glucose and glycated levels, as had been depicted in dietary pattern analysis studies among different ethnic groups [
13]. This disparity from the Western populations where the animal based/Western dietary patterns are associated with increasing fasting glucose levels has been explained by the realisation that ethnic groups such as Asians may adopt the Western diets but their intake of animal products will remain lower as they also continue to consume traditional cereals and vegetables [
29]. Similarly, in this study population, the intake of fat and animal protein nutrients were lower in the plant driven nutrients patterns that accounted for the greatest variance among the three nutrients patterns extracted per stratum as illustrated in
Table S3. In other local studies, the protein and fat intakes as percentage of total energy intake for urban women from 1975 to 2005 did not change drastically though evidences of the adoption of Western dietary patterns were being noted and the greater proportion of the energy intake was still being contributed by the carbohydrate intake as had been noted in the Asians [
6]. From 1975 to 2005, the percentage of energy of protein intake changed from 14% to 13%, while fat intake as percentage of energy changed from 21% to 30% and carbohydrate intake changed from 67% to 57% of total energy intake [
6]. Thus, the intake of fat and animal protein driven nutrients in this population group might have been lower and thus not associated with increases in fasting glucose and glycated haemoglobin as was expected.
The magnesium, phosphorus and plant protein driven nutrient pattern indicated a positive trend association with increases in fasting glucose and glycated haemoglobin levels among rural women, while the thiamine, zinc and plant protein driven nutrient pattern was associated with a trend of increasing glycated haemoglobin levels in the urban men (
Table S7). However, plant protein and zinc, which were high in these nutrient patterns, have been reported to independently lower fasting glucose levels [
30,
31,
32,
33]. In addition, the thiamine, zinc and plant protein driven nutrient pattern was associated with a trend of decreasing glycated haemoglobin and fasting glucose levels in the rural men. In view that the study population comprised people who consumed diets with both animal and plant based nutrients and not purely vegetarians, the discrepancies in the associations of the plant driven nutrients can be explained by the varied proportions of animal protein, saturated fat, mono-saturated fat, cholesterol and sugar among these plant driven nutrient patterns, as illustrated in
Table S3. The magnesium, phosphorus and plant protein driven nutrient pattern in rural women and the thiamine, zinc and plant protein driven nutrient pattern in urban men had higher loadings of animal protein, saturated fat, mono-saturated fat, cholesterol and sugar compared to the other nutrient patterns discussed below such as the starch, dietary fibre and B vitamins driven nutrient pattern among rural women and the thiamine, zinc and plant protein nutrient pattern among rural men which associated with low fasting glucose and glycated haemoglobin levels. Animal protein, saturated fat, cholesterol and sugar are high in Western dietary patterns which have been previously associated with increased risk of T2D and this helps clarify the positive trend of association of the magnesium, phosphorus and plant protein driven nutrient pattern in rural women and the thiamine, zinc and plant protein driven nutrient pattern in urban men with the study outcome variables [
34]. The comparisons of the varied constituents of the plant driven nutrients as discussed above and illustrated in
Table S3 indicates the attractiveness of the PCA approach of nutrient pattern determination as it allows a whole based approach of the effect of nutrients to particular outcomes to be evaluated.
The starch, dietary fibre and B vitamins driven nutrient pattern was consistently associated with reduced fasting glucose and glycated haemoglobin levels in all the multivariate linear models which were considered in this study among rural women. The thiamine, zinc and plant protein nutrient pattern was also associated with significant reductions in glycated haemoglobin and fasting glucose among rural men. These findings are similar to the results of other studies on plant based dietary patterns and T2D risk [
10,
34]. It has been consistently reported that plant based diets are protective against T2D risk [
10,
34,
35]. A number of mechanisms have been proposed to explain this phenomenon [
10]. Plant based diets which are rich in dietary fibre are suggested to reduce T2D risk by reducing postprandial insulin demand, improving insulin sensitivity and the antioxidants found in these diets may help enhance β-cell function [
36,
37]. Dietary fibre from cereal foods which are also high in starch has been consistently shown across eight European countries to be associated with reduced T2D risk [
38]. Maize meal fortified with iron, zinc and B vitamins which is largely consumed among the black South African population group is known to contain resistant starches that are partially digested and have been associated with improved insulin sensitivity which may then lead to reductions in fasting glucose levels [
39,
40]. Thiamine was also high in this nutrient pattern. Evidence exists that dietary fibre beneficial effects might also be due to its concomitant intake together with thiamine [
41].The results of this study suggest that starch, dietary fibre and B vitamins nutrients, zinc and plant protein consumed together may lead to the lowering of fasting and glycated haemoglobin levels as has been postulated elsewhere [
35].
The strengths of the current study include the use of a validated QFFQ and recruitment criteria of selecting apparently healthy individuals who were not taking chronic medications or suffering from chronic diseases. This might have helped to prevent the confounding of drugs and T2D to the nutrient pattern association results with glycated haemoglobin and fasting glucose levels. The factor analysis approach used to derive the nutrient patterns is known for depicting real-world dietary behaviours [
42]. However, this approach is based on a number of subjective decisions such as naming nutrient patterns, method of rotation and selection of food groups which can lead to an overall measurement error [
10,
42]. Since this approach is a posteriori analysis tool, it derives nutrient patterns based on data at hand and this makes comparisons with other studies difficult [
10,
42]. However, regardless of the differences in the constituents of the nutrient patterns due to the data driven approaches used to derive them, consistent similarities of the dietary factors associated with T2D risk as reported in this study, has been depicted in multiple populations [
10,
34,
37]. It might be probable that the residual confounding effect of total energy intake might have led to distortions in the association of the nutrient patterns with fasting glucose and glycated haemoglobin levels. Residual confounding is a result of measurement error in a confounder included in the model [
43]. Although the nutrient and total energy intake based on the QFFQs are not precisely measured, adjustment for total energy intake should control for total energy intake confounding [
44]. However, evidence exists that the control of total energy intake is not complete in epidemiological studies involving QFFQs thereby leading to residual confounding [
44,
45]. Therefore, there is need to further explore the association of the nutrient patterns in isocaloric clinical trials which are better designed to control for confounding for total energy intake [
45].