This study investigates the relationship between diet quality and weight gain in young women. Young women (
Recently, there has been a focus on evaluating the association between the nutritional quality of dietary intake and health outcomes [
A recent study demonstrated, in a nationally representative sample in the United States, that younger adults have poorer diet quality when compared with both children and older adults [
In this study we are analysing the relationship between three different approaches of diet quality indices including: index based on the food groups, which is the Australian Recommended Food Score (ARFS), and nutrients-based approach, the Diet Quality Index (DQI). In addition, we developed a new brief index that, based on consumption frequency and variety of fruits and vegetables items, is called the Fruit and Vegetables Index (FAVI). This tool can help to reduce the burden to both participants and researchers in terms of measuring diet quality. It can be used to predict weight change and therefore weight gain prevention or treatment interventions. Evidence suggests that greater consumption of fruit and vegetables in adults is associated with lower weight gain in longitudinal studies [
Notably, two studies exploring the association between diet quality and weight gain among middle-aged women have shown mixed results. A longitudinal study, conducted in an American middle-aged population, demonstrated that those who achieved the highest score on the DQI had a smaller weight gain (3 pounds) than those who achieved the lowest DQI score (5–8 pounds) during eight years of followup [
Therefore, the aim of this study was to investigate the relationship between diet quality and weight gain in young women from the ALSWH, using three different diet quality indices, ARFS, Australian-DQI (Aus-DQI), and the Fruit and Vegetable Index (FAVI).
The population is a subset from the ALSWH cohort study. ALSWH recruited women into three cohorts according to age at baseline (young, middle-aged, and older). Further details of the cohort are published elsewhere [
Flow chart of participant selection for analyses.
Weight was self-reported at baseline (2003) and at followup (2009), in stones or kilograms (kg) to the nearest pound or gram, respectively. All data were converted to kilograms. Weight change
Baseline self-reported dietary intake was assessed using a food frequency questionnaire (Dietary Questionnaire for Epidemiological Studies Version 2 (DQES v2), Cancer Council of Victoria). The DQESv2 has been previously validated [
The ARFS is a food-based index adapted to the Australian adult population by Collins et al. (2008) [
The DQI was chosen as studies have shown that higher scores on this index are associated with lower weight gain [
Evidence suggests that greater consumption of fruit and vegetables in adults is associated with lower weight gain in longitudinal studies [
Data were assessed for normality and presented as means and standard deviations. Results were considered statistically significant if
The total number of women included in this analysis, with complete baseline and followup data on weight change and diet, was
Demographic characteristics of young women in the Australian longitudinal study on women’s health (ALSWH) (
Characteristic | Baseline | Follow-up | ||
---|---|---|---|---|
Total sample |
Valid TEI |
Total sample |
Valid TEI |
|
|
||||
Obesity (%) | 15.7 | 11.6 | 20.6 | 16.7 |
Overweight (%); | 22.5 | 19.6 | 25.0 | 23.6 |
BMI; mean ± SD |
|
|
|
|
Weight (kg); mean ± SD |
|
|
|
|
|
||||
ARFS |
|
|
n/a | n/a |
Aus-DQI |
|
|
n/a | n/a |
FAVI |
|
|
n/a | n/a |
|
||||
Age (years); mean ± SD |
|
|
|
|
Total energy intake (kJ); mean ± SD |
|
|
n/a | n/a |
Physical activity in METs (nil/low/moderate/high); (%) | 8.9/35.3/22.8/33.0 | 8.9/9.7/20.4/31.1 | n/a | n/a |
Smoking status (never/ex-smoker/current); proportion (%) | 58.7/18.3/23.0 | 60.1/16.5/33.4 | n/a | n/a |
Residence (urban/rural/remote); proportion (%) | 57.3/39.0/3.7 | 55.3/41.0/3.7 | n/a | n/a |
Highest education (nil/school certificate/trade/university degree); proportion (%) | 1.5/31.0/3.3/64.3 | 1.0/29.9/3.1/66.0 | n/a | n/a |
TEI: total energy intake, ARFS: Australian recommended food score, FAVI: fruit and vegetables index, and Aus-DQI: Australian diet quality index.
There was no significant difference across tertiles of ARFS for mean weight change, but there were significant differences in the means of energy intake (kJ/d), fibre (g/d), carbohydrate (%), and protein (%) intakes total fat (%) and saturated fat (%) intakes observed across ARFS tertiles (Table
Weight change data (2003 to 2009) and baseline macronutrient intakes (2003) for young women in the Australian longitudinal study on women’s health (ALSWH) by tertile of Australian recommended food score (ARFS).
ARFS tertiles (Total sample |
ARFS tertiles (Valid TEI sub-sample |
|||||||
---|---|---|---|---|---|---|---|---|
Number |
1 |
2 |
3 |
|
1 |
2 |
3 |
|
Mean ± SD |
|
|
|
|
|
|
||
ΔWeight (kg); mean ± SD |
|
|
|
0.2 |
|
|
|
0.09 |
Baseline weight (kg); mean ± SD |
|
|
|
0.16 |
|
|
|
0.4 |
Follow-up weight (kg); mean ± SD |
|
|
|
0.44 |
|
|
|
0.7 |
Energy intake (kJ/d); mean ± SD |
|
|
|
<0.0001 |
|
|
|
0.02 |
|
||||||||
Total fat (% energy); mean ± SD |
|
|
|
<0.0001 |
|
|
|
<0.0001 |
Saturated fat (% energy); mean ± SD |
|
|
|
<0.0001 |
|
|
|
<0.0001 |
Protein (% energy); mean ± SD |
|
|
|
<0.0001 |
|
|
|
0.0005 |
Carbohydrate (% energy); mean ± SD |
|
|
|
<0.0001 |
|
|
|
<0.0001 |
Fiber (g/d); mean ± SD |
|
|
|
<0.0001 |
|
|
|
<0.0001 |
There was no significant difference in the mean weight change across the Aus-DQI tertiles (Table
Weight change data (2003 to 2009) and baseline macronutrient intakes (2003) for young women in the Australian longitudinal study on women’s health (ALSWH) by tertile of Australian diet quality index (Aus-DQI).
Aus-DQI tertiles (total sample |
Aus-DQI tertiles (valid TEI subsample |
|||||||
---|---|---|---|---|---|---|---|---|
Number |
1 |
2 |
3 |
|
1 |
2 |
3 |
|
Mean ± SD |
|
|
|
|
|
|
||
ΔWeight (kg); mean ± SD |
|
|
|
0.3 |
|
|
|
0.48 |
Baseline weight (kg); mean ± SD |
|
|
|
<0.0001 |
|
|
|
<0.0001 |
Follow-up weight (kg); mean ± SD |
|
|
|
<0.0001 |
|
|
|
<0.0001 |
Energy intake (kJ/d); mean ± SD |
|
|
|
<0.0001 |
|
|
|
<0.0001 |
|
||||||||
Total fat (% energy); mean ± SD |
|
|
|
<0.0001 |
|
|
|
<0.0001 |
Saturated fat (% energy); mean ± SD |
|
|
|
<0.0001 |
|
|
|
<0.0001 |
Protein (% energy); mean ± SD |
|
|
|
<0.0001 |
|
|
|
<0.0001 |
Carbohydrate (% energy); mean ± SD |
|
|
|
<0.0001 |
|
|
|
<0.0001 |
Fiber (g/d); mean ± SD |
|
|
|
<0.0001 |
|
|
|
<0.0001 |
There was a significant difference in mean weight change across the FAVI tertiles (
Weight change data (2003 to 2009) and baseline macronutrient intakes (2003) for young women in the Australian longitudinal study on women’s health (ALSWH) by tertile of fruit and vegetable index (FAVI).
FAVI tertiles (total sample |
FAVI tertiles (valid TEI subsample |
|||||||
---|---|---|---|---|---|---|---|---|
Number |
1 |
2 |
3 |
|
1 |
2 |
3 |
|
Mean ± SD |
|
|
|
|
|
|
||
ΔWeight (kg); mean ± SD |
|
|
|
0.002 |
|
|
|
0.003 |
Baseline weight (kg); mean ± SD |
|
|
|
0.1 |
|
|
|
0.98 |
Follow-up weight (kg); mean ± SD |
|
|
|
0.3 |
|
|
|
0.17 |
Energy intake (kj/d); mean ± SD |
|
|
|
<0.0001 |
|
|
|
0.1 |
|
||||||||
Total fat (%); mean ± SD |
|
|
|
<0.0001 |
|
|
|
<0.0001 |
Saturated fat (%); mean ± SD |
|
|
|
<0.0001 |
|
|
|
<0.0001 |
Protein (%); mean ± SD |
|
|
|
<0.0001 |
|
|
|
0.0491 |
Carbohydrate (%); mean ± SD |
|
|
|
<0.0001 |
|
|
|
<0.0001 |
Fiber (g/d); mean ± SD |
|
|
|
<0.0001 |
|
|
|
<0.0001 |
In the plausible TEI sub-sample, only those in the top tertile of the ARFS had significantly less weight gain (1.6 kg) compared with those in the lower tertile of the ARFS. In the fully adjusted model, those who were in the top tertile of the ARFS had significantly lower weight gain compared with the lower tertile for the plausible TEI sub-sample, (
Baseline FAVI was a statistically significant negative predictor of weight gain in this group of young women, while ARFS and Aus-DQI were not statistically significant predictors of weight change (Table
Multiple linear regression models to predict six-year weight change in young women from the Australian longitudinal study on women’s health.
Predictor: |
Model* | Tertile |
Total sample: |
|
Valid TEI subsample |
|
---|---|---|---|---|---|---|
ARFS | Crude | 2 | −0.32 (−0.99, 0.28) | 0.29 | −0.60 (−1.7, 0.46) | 0.27 |
3 | −0.69 (−1.3, 0.08) |
|
−1.14 (−2.16, −0.12) |
|
||
Adjusted | 2 | −0.16 (−0.79, 0.47) | 0.63 | −0.93 (−1.96, 0.09) | 0.07 | |
3 | −0.34 (−0.97, 0.30) | 0.29 | −1.59 (−2.63, −0.53) |
|
||
Final | 2 | −0.18 (−0.81, 0.46) | 0.58 | −0.95 (−2.0, 0.7) | 0.07 | |
3 | −0.38 (−1.03, 0.27) | 0.25 | −1.6 (−2.67, −0.56) |
|
||
| ||||||
Aus-DQI | Crude | 2 | −0.05 (−0.71, 0.60) | 0.876 | −0.68 (−1.79, 0.42) | 0.2 |
3 | −0.51 (−1.14, 0.12) | 0.112 | −0.10 (−1.12, 0.92) | 0.8 | ||
Adjusted | 2 | −0.04 (−0.71, 0.63) | 0.905 | −0.80 (−1.93, 0.31) | 0.2 | |
3 | −0.60 (−1.25, 0.06) | 0.073 | −0.46 (−1.52, 0.61) | 0.4 | ||
Final | 2 | −0.05 (−0.73, 0.63) | 0.885 | −0.81 (−1.95, 0.33) | 0.2 | |
3 | −0.62 (−1.32, 0.07) | 0.078 | −0.45 (1.53, 0.62) | 0.4 | ||
| ||||||
FAVI | Crude | 2 | −0.96 (−1.62, −0.31) |
|
−1.60 (−2.67, −0.54) |
|
3 | −1.09 (−1.75, −0.44) |
|
−1.61 (−2.62, −0.58) |
|
||
Adjusted | 2 | −0.61 (−1.28, 0.07) | 0.079 | −1.4 (−2.53, −0.43) |
|
|
3 | −0.68 (−1.36, 0.00) | 0.051 | −1.5 (−2.59, −0.42) |
|
||
Final | 2 | −0.63 (−1.30, 0.05) | 0.070 | −1.5 (−2.4, −0.2) |
|
|
3 | −0.72 (−1.42, −0.03) |
|
|
|
In the plausible TEI sub-sample, we found that those in the second and third tertiles of FAVI had significantly less weight gain compared with the first tertile. More specifically, we found that, in the fully adjustmed model, those who were in the top tertile of FAVI gained the lowest weight compared with other tertiles (
The current study tested three different diet quality indices as predictors of weight change over the subsequent six-year period in a cohort of young women participating in the ALSWH. it demonstrated that higher scores on either a food variety and frequency index (ARFS) or an index based on fruit and vegetable variety and frequency alone (FAVI), predicted lower six-year weight gain in this group of women. In the whole sample the ARFS showed no relationship with prospective weight gain, while the Aus-DQI showed no relationship in either the whole or the plausible TEI sub-sample.
The main findings of this study support the role of increased fruit and vegetable consumption as a key strategy to prevent weight gain, particularly for young women. This is consistent with a recent prospective study by Vioque et al. (2008) [
The Aus-DQI failed to predict weight gain during the followup period in this sample of young women, even though it incorporates sub-scales for the percentage energy from total fat, saturated fat and carbohydrate intakes, and total sodium intake. The limited scoring scale and that it had not been previously validated limit the interpretation of this result. This also may be due to the limited list of energy-dense, nutrient-poor foods, particularly soda, and other sweetened beverages within the DQES which is to be expected given that it was developed more than 20 years ago. Thus, an assumption and limitation are that TEI may be partly underestimated due to the items in the FFQ. In the whole population,we found that the lowest intake of fiber across the Aus-DQI tertiles was for the top tertile, or highest diet quality scores. Among those women with plausible TEI however, we found that the highest Aus-DQI tertile was associated with higher intakes of fiber. This difference is likely due to misreporting of TEI and we expect that the results in the plausible TEI sub-sample are more likely to be more accurate.
The ALSWH cohort is a representative sample of the population of Australian women, and the weight change data from the current study indicate that weight gain is common among young women. In addition, very few young women achieved a high diet quality score. The mean diet quality score in the highest tertile of each index was not high, indicating that interventions seeking to optimise diet quality in this age group are warranted as has been suggested previously [
There are a number of major limitations that need to be addressed. This includes that there are a large number of women with missing data on weight or dietary intake at baseline and follow-up. Furthermore, a limitation that needs to be acknowledged is loss to follow up. In the ALSWH study, attrition is the most common in participants with a lower education, those not born in Australia and those with poorer health or who smoke [
Furthermore, all data were self-reported including weight which introduces a potential reporting bias. A previous validation study of self-reported weight on mid-aged women from the ALSWH demonstrated that there was no clinical difference between self-reported weight and measured body weight [
The strengths of this study include the use of a healthy representative sub-sample derived from ALSWH population, with an adequate followup period. In addition, we used appropriate and rigorous statistical analyses and three different approaches to the measurement of diet quality to reflect the National Dietary Guidelines for Australia, including two based on established methods and one new index based only on fruit and vegetable intakes. This new tool provides a simple approach to diet quality assessment and successfully predicted weight change in this cohort of young women. Further research evaluating and validating the performance of FAVI in other age and gender groups is warranted.
Frequency and variety of fruit and vegetable intakes, and overall diet quality predicted weight gain over six years in this healthy population group of young women. Strategies to encourage young women to more frequently consume a greater variety of fruit, and vegetables are required and may assist to prevent weight gain in this age group.