The prevalence of obesity has significantly risen both in New Zealand (NZ) and worldwide, with 32% of all females in NZ classified as obese [1
]. It is well-established that obesity is a major risk factor for many health conditions including but not limited to cardiovascular disease, type 2 diabetes, hypertension, insulin resistance, dyslipidemia, osteoarthritis, sleep apnoea, psychological and social problems, and some cancers [2
The most widely used indicator to quickly identify overweight or obesity is the Body Mass Index (BMI), determined by weight/height2
]. In NZ, overweight and obesity are defined as a BMI between 25 kg/m2
and 29.9 kg/m2
, and ≥30 kg/m2
, respectively [2
]. Although routinely used in epidemiological studies and by health professionals, BMI is an imperfect measure of body fatness [5
]. It speculates that at any given height, a higher weight correlates to a larger body fat percentage (BF%) and consequently a higher risk of morbidity and mortality [6
]. Adiposity may be overestimated in individuals who have a high lean muscle mass as they are incorrectly classed as overweight (e.g., athletes), whereas underestimation may occur for others who have a normal BMI but excess body fat (>30%) (e.g., premenopausal women) [7
]. It is estimated that more than half of individuals are misclassified when using BMI categorization [2
]. Air displacement plethysmography (ADP) is the emerging gold standard to accurately measure BF% [8
]. Reviewing both BMI and BF% allows accurate estimation of body composition, and in combination with lifestyle and dietary factors, allows the risk of chronic disease to be assessed and the tailoring of appropriate dietary advice.
Both BMI and body composition (i.e., ratio of fat mass to lean muscle mass) are influenced by lifestyle and dietary factors, such as the types of foods habitually consumed. Younger, premenopausal women are particularly at risk of higher BF% due to poor lifestyle (e.g., physical inactivity) and eating (e.g., dieting, fast foods) behaviors [9
]. Traditionally nutritional epidemiology has focused on the effects of individual foods and nutrients on health outcomes. Although many studies have investigated the effects of specific macronutrients on body composition, dietary patterns have received less attention [10
]. As foods and nutrients are consumed in combination, dietary patterns have recently emerged as an effective way to assess diet-disease relationships [11
Specific dietary patterns such as a “Western” dietary pattern have been associated with weight gain and an increased risk of cardiovascular disease, type 2 diabetes, and cancer [12
]. This pattern is characterized by a habitual intake of refined grains, breads and cereals, red and processed meats, fast-food, sugar-sweetened beverages (SSB), sweets and desserts [13
]. In comparison, a diet composed of fruit, vegetables, whole grain bread and cereals, lean unprocessed meats, poultry, fish and low fat dairy products has been found to be independently associated with a higher dietary adequacy, lower BMI, and lower waist circumference, thus reducing the risk of consequent chronic disease [13
]. Practicing these dietary patterns may also determine body composition, particularly body fatness, which may in turn influence morbidity and mortality risk factors. No studies to date have investigated associations between BF% and dietary patterns in premenopausal women.
Another important aspect of dietary patterns is their macronutrient composition which provides further information regarding the make-up of the dietary pattern. Studies have found that when the BF% of women increased, the percentage of fat derived from their diet also increased (35%) and the carbohydrate content decreased (46%) compared to lean individuals (29% and 53%, respectively) [15
]. However, studies have not concurrently examined dietary patterns and the macronutrient profile of the diet in relation to BF%. Consideration of dietary patterns and body composition may enable the tailoring of dietary interventions appropriate for women with different body fat profiles to improve their morbidity and mortality outcomes. The aims of this study were to identify the dietary patterns of New Zealand European women (NZE) aged 16–45 years and examine their associations with BMI, body fat and macronutrient profile intakes.
Four dietary patterns (“snacking”, “energy-dense meat”, “fruit and vegetable”, “healthy”) were identified in this group of NZE women. Older women were more likely to follow the “snacking” pattern, and younger women were more likely to follow the “energy-dense meat” pattern. After adjusting for energy intake, no dietary patterns were significantly associated with BMI or BF%. Evidence is mixed when investigating the association between dietary patterns and BMI, and two reviews concluded no clear associations [28
]. Some studies have found an inverse association [30
], a positive association [31
], and others, no association between energy-dense patterns and BMI [33
]. One review found dietary patterns in women to appear more weakly associated with BMI and obesity compared to men, indicating other factors may also play a part in determining women’s body composition [28
]. Few studies, we are aware of, have investigated dietary patterns in relation to BF% [35
]. When investigating macronutrient intakes, high and low consumers of all dietary patterns had carbohydrate intakes below and saturated fat intakes above the AMDR. The high saturated fat intakes are consistent with the NZ National Nutrition Survey, where intakes were above the AMDR for females [27
]. The “snacking”, “energy-dense meat”, “fruit and vegetable”, and “healthy” patterns will be discussed in relation to our findings.
The “snacking” pattern was characterized by cakes and biscuits, sweet and savory snack foods (e.g., chocolate and potato chips) and peanut butter. This pattern is similar to the “junk food” pattern identified by Kourlaba et al. [36
] and the “snacky” pattern identified by Aranceta et al. [37
]. A significant difference in age was observed for the “snacking” pattern with older women more likely to follow this pattern (33.9 ± 7.9 years) compared with younger women (31.9 ± 8.7 years). This suggests that women who are older may eat more snack-like foods compared to younger women. McNaughton et al. [24
] did not find a significant association between age and a dietary pattern characterized by chocolate, confectionary, added sugar, and fruit drink in women, however this pattern included additional foods to our snacking pattern (e.g., dairy milk and yoghurt). With an increasing availability of energy-dense food and the higher prevalence of overweight and obesity [1
], the increased snacking in older pre-menopausal women (younger than 45 years) is plausible, as snacking has been shown to significantly contribute to excess energy consumption [9
Women across all tertiles of the “snacking” pattern met the AMDR for percentage total energy intake from protein; however all dietary pattern tertiles were at the top end of the AMDR for total fat, above the AMDR for saturated fat (34.5% ± 5.6% and 13.4% ± 3.0% of total energy intake, respectively), and below the AMDR for carbohydrate (42.6% ± 7.3% of energy intake). Similarly, Pryer et al. [38
] identified a “convenience” pattern followed by women who had the highest total fat intakes out of all patterns with 40.5% of energy coming from total fat. This shows that snack foods may contribute large amounts of fat to dietary intake. However, in the “snacking” pattern there were no significant differences in macronutrient intakes between the tertiles of this pattern, suggesting that the macronutrient intakes of women do not significantly change regardless of whether they follow the “snacking” pattern or not.
The “energy-dense meat” pattern was categorized by red, white, and processed meat, puddings and crumbed and deep fried foods (e.g., crumbed and battered fish). This pattern is similar to the “Western” pattern identified by Schulze [14
] or the “empty calorie” pattern identified by Quatromani et al. [39
]. The “energy-dense meat” pattern had a significant positive association with BMI and an inverse association with age. However, after adjusting for energy intake, the association between the “energy-dense” meat pattern and BMI was no longer significant. Other studies [24
] found younger individuals are more likely to follow the similar “Western” dietary pattern. Adolescents have been shown to have an increased consumption of soft drinks and fast-food, both of which are similar to this “energy-dense meat” dietary pattern [41
]. The “Western” pattern has been associated with an increase in weight gain and a higher risk of being overweight [14
], while Slattery [31
] and Hu [32
] found a “Western” dietary pattern to be associated with a higher BMI. Other studies, have found no association between dietary patterns and BMI [33
]. These controversial findings may be explained by differences in study designs, study populations, and the dietary patterns identified. For example, Beaudry [42
] included men and pre- and post-menopausal women, and Fung [33
] included only male participants. In our study, BF% was significantly associated with the “energy-dense meat” pattern after controlling for age, but not energy intake. We are aware of only one other study to date which has investigated dietary patterns and BF% [35
]. Tucker et al. [35
] found a similar “Meat” dietary pattern which was associated with a significantly higher BF% which remained after adjusting for energy intake. This is similar to our findings and thus further investigation is warranted surrounding the influence of specific dietary patterns on BF% in premenopausal women.
Women with low scores on the “energy-dense meat” pattern had a higher percentage total energy intake from protein compared to those with higher scores. The “energy-dense meat” pattern is characterized by red, white, and processed meats, puddings and deep fried foods and high fat cheese, contributing to both protein and saturated fat. Increased red meat consumption has been associated with increased BMI due to its high energy and saturated fat content [43
]. The high meat consumption likely contributes to the high saturated fat intakes seen in women following this “energy-dense meat” pattern. After adjusting for energy intake, no association was seen between the “energy-dense meat” pattern and BMI. Other studies found overweight and obese individuals are more likely to consume a greater amount of saturated fat compared to those of normal weight [44
], and McCroy et al. [34
] identified percentage of dietary fat to be positively associated with body fatness.
High intakes of all fruits and vegetables, including starchy vegetables but excluding potatoes comprised the “fruit and vegetable” pattern. This pattern is similar to the “healthy” pattern identified by Suliga [45
] and Newby et al. [46
], however both of their patterns included other items such as dairy and fruit juice. The “fruit and vegetable” dietary pattern was not significantly associated with age, BMI or BF%. Prudent patterns which also have high fruit and vegetable intakes have been associated with a lower BMI, however these patterns typically also have additional items such as legumes, wholegrains, fish, and poultry [32
]. These additional food items may account for the lack of association seen between the “fruit and vegetable” pattern and BMI or BF%. Nevertheless, other studies suggest further research is required into which aspects of dietary patterns are preventative of increased BMI [46
]. Additionally, the lack of association suggests additional factors may influence body composition in individuals following this pattern such as lack of physical activity, and socio-economic status, or dietary factors not accounted for using dietary pattern analysis.
The majority of women who scored high on the “fruit and vegetable” pattern had mean ± SD total fat as a percentage of total energy intake lower (34.1% ± 7.3%) and within the AMDR guidelines, compared to the women who consumed only a few items from this pattern who had mean ± SD intakes exceeding the recommendations (36.8% ± 6.5%). Similarly, Schulze [14
] also found individuals with lower fruit and vegetable intakes to have a diet higher in energy and total fat. On average, women less likely to follow the “fruit and vegetable” pattern had significantly higher total and saturated fat intakes and lower fiber intakes compared to those more likely to follow the pattern suggesting that more foods higher in both total and saturated fat may be consumed in the presence of lower fruit and vegetable consumption. Similarly, less fiber rich foods were consumed by those following the “energy dense meat” pattern, and the most was consumed by the highest consumers in the “fruit and vegetable” pattern. This was also observed by Tucker et al. [35
], reporting that women with poor intakes of fruit, starch, fiber, and non-starchy vegetables had high intakes of meat, other carbohydrates (added sugars), and fat. As expected, we found, for all patterns except the “energy dense meat” pattern, that as energy intakes increased, fiber intakes also increased.
The “healthy” pattern comprised fish and seafood, legumes, nuts, and seeds. This pattern was similar to the “bean” pattern identified by Maskarinec et al. [47
]. The “healthy” pattern was not significantly associated with age, BMI or BF%, indicating that other factors; such as those suggested above, may play a role in determining body composition in individuals following this dietary pattern.
Women scoring highly on the “healthy” pattern had significantly higher protein intakes (as a percentage of total energy intake) compared with women who scored low on this pattern. This pattern was associated with protein-rich items such as fish and seafood, legumes, eggs, and soy products which likely contribute to the higher protein content of their diet. Increases in protein intake, especially obtained from meat alternatives have been associated with increased satiety and weight regulation [48
] as shown by Maskarincec et al. [47
] who identified an inverse relationship between a “bean” pattern and BMI. Women following the “healthy” pattern also had significantly higher total fat intakes as a percentage of total energy intake in excess of the AMDR (36.7% ± 7.9%), compared to those less likely to follow this pattern. Despite the higher intake of fats, there was no association between the “healthy” dietary pattern and BMI levels. Foods rich in mono- and poly-unsaturated fatty acids (such as fish, nuts, and seeds) have previously been described as characterizing healthy dietary patterns, such as the Mediterranean diet, which are associated with a good metabolic profile due to its high content of such “healthy” fats [49
]. It can be speculated that this may be due to the high mono- and poly-unsaturated fatty acid content of many of the foods associated with this pattern such as fish, nuts, and seeds. The “healthy” pattern also had significantly different carbohydrate intakes between the tertiles with women scoring high on this dietary pattern consuming less carbohydrate. The carbohydrate intakes in all tertiles of the “healthy” pattern were however below the AMDR and this trend; although not significant between the tertiles, was also present among the “snacking”, “energy-dense meat”, and “fruit and vegetable” patterns.
Several factors may explain the lower carbohydrate intakes seen among participants in this study. Firstly, a low carbohydrate diet may be eaten to compensate for the consumption of foods with undesirable nutrient profiles. For example, consuming fewer carbohydrates means this energy needs to be replaced with protein or fat to maintain satiety and body weight [15
]. The macronutrient distributions discernable from the women’s dietary patterns suggest that they compensate the low carbohydrate intakes with a higher consumption of foods higher in saturated fat. Secondly, a low carbohydrate diet has been highly publicized as a healthy diet to maintain or lose weight [50
], with dietary trends currently promoting “high fat low carbohydrate” and “paleo” diets [51
]. Individuals concerned about their health or weight often follow these trends and this may be why the macronutrient intakes in these women reflect this way of eating. Finally, social desirability and social approval bias have the potential to severely skew participant reporting [52
]. Scagliusi et al. [53
] found that women were more likely to report their energy intake in a socially desirable way and food items which are considered “fattening” or “unhealthy” such as confectionary, fried foods, and refined breads were more likely to be under reported.
This study has several strengths. Firstly, the use of a homogenous group of women allows clear associations to be drawn between premenopausal women aged 16 to 45 years and the dietary patterns identified. This reduces variation and simplifies analysis by bringing focus to this particular group of women. We can be confident that these associations are relevant as other factors have been eliminated such as ethnicity. Secondly, the study addressed BMI and objectively measured BF% using ADP, which enabled both factors to be investigated in regard to their relationship with the dietary patterns. Investigating them separately enabled the relationship between body fatness and dietary patterns to be investigated as opposed to general BMI which may misclassify some individuals, especially those with normal weight obesity [54
This study also has several limitations. Firstly, the study used retrospective reporting of dietary intake. It is difficult to assess dietary intake and measurement error can arise from a number of sources which may contribute to the lack of association between the dietary patterns and body composition [55
]. Secondly, social desirability and social approval bias may have influenced the reporting of the NZE women with women reporting higher consumption of foods associated with the “fruit and vegetable” and “healthy” dietary patterns if they perceive these foods to be beneficial. This is a concern in all self-report dietary assessments [55
]. In an attempt to reduce this bias, a validated FFQ was used which was shown to have good validity in ranking nutrient intake in relation to a food record over the last month of consumption [19
]. Thirdly, this study did not adjust for confounding factors such as physical activity, socioeconomic status, and health behaviors such as smoking. This could have biased the analysis given the well-established relationship between physical activity and BMI and body fat mass [56
]. In addition, the relatively small amount of total variance explained by the principal component factor analysis is also a limitation. The small amount of variance explained means that other dietary factors not captured by the dietary patterns may also influence the body composition (BMI and BF%) of these women. Finally, as this study involved volunteer participants, the study is not generalizable to all NZ women.