You are here

  • Home
  • Archive
  • Volume 13, Issue 6
  • Dietary patterns in rural and metropolitan Australia: a cross-sectional study exploring dietary patterns, inflammation and association with cardiovascular disease risk factors

Article Text

Article menu
Nutrition and metabolism
Original research
Dietary patterns in rural and metropolitan Australia: a cross-sectional study exploring dietary patterns, inflammation and association with cardiovascular disease risk factors
  1. Laura Alston1,2,
  2. Melanie Nichols3,
  3. Steven Allender3,
  4. Vincent Versace1,
  5. Leanne J Brown4,5,
  6. Tracy Schumacher4,5,
  7. George Howard6,7,
  8. James M Shikany7,
  9. Kristy A Bolton3,8,
  10. Katherine Livingstone8,
  11. Christina Zorbas3,
  12. Suzanne E Judd7
  1. 1Deakin Rural Health, Deakin University, Warnambool, Victoria, Australia
  2. 2Research Unit, Colac Area Health, Colac, Victoria, Australia
  3. 3Global Centre for Preventative Health and Nutriton, Deakin University, Geelong, Victoria, Australia
  4. 4Department of Rural Health, The University of Newcastle, Tamworth, New South Wales, Australia
  5. 5Hunter Medical Research Institute, New Lambton Heights, New South Wales, Australia
  6. 6Biostatistics, University of Alabama at Birmingham School of Public Health, Birmingham, UK
  7. 7Division of Preventive Medicine, Heersink School of Medicine, University of Alabama at Birmingham, Birmingham, UK
  8. 8Institute for Physical Activity and Nutrition, Deakin University, Geelong, Victoria, Australia
  1. Correspondence to Dr Laura Alston; laura.alston{at}deakin.edu.au

Abstract

Objectives This study sought first to empirically define dietary patterns and to apply the novel Dietary Inflammation Score (DIS) in data from rural and metropolitan populations in Australia, and second to investigate associations with cardiovascular disease (CVD) risk factors.

Design Cross-sectional study.

Setting Rural and metropolitan Australia.

Participants Adults over the age of 18 years living in rural or metropolitan Australia who participated in the Australian Health survey.

Primary outcomes A posteriori dietary patterns for participants separated into rural and metropolitan populations using principal component analysis. Secondary outcomes: association of each dietary pattern and DIS with CVD risk factors was explored using logistic regression.

Results The sample included 713 rural and 1185 metropolitan participants. The rural sample was significantly older (mean age 52.7 compared with 48.6 years) and had a higher prevalence of CVD risk factors. Two primary dietary patterns were derived from each population (four in total), and dietary patterns were different between the rural and metropolitan areas. None of the identified patterns were associated with CVD risk factors in metropolitan or rural areas, aside diet pattern 2 being strongly associated with from self-reported ischaemic heart disease (OR 13.90 95% CI 2.29 to 84.3) in rural areas. There were no significant differences between the DIS and CVD risk factors across the two populations, except for a higher DIS being associated with overweight/obesity in rural areas.

Conclusion Exploration of dietary patterns between rural and metropolitan Australia shows differences between the two populations, possibly reflective of distinct cultures, socioeconomic factors, geography, food access and/or food environments in the different areas. Our study provides evidence that action targeting healthier dietary intakes needs to be tailored to rurality in the Australian context.

  • nutrition & dietetics
  • public health
  • preventive medicine

Data availability statement

Data are available in a public, open access repository. The data are available from the Australian Bureau of Statistics.

http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

https://doi.org/10.1136/bmjopen-2022-069475

Statistics from Altmetric.com

    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.

    STRENGTHS AND LIMITATIONS OF THIS STUDY

    • This study used representative national data sets, which represent the most recent, highest quality and comprehensive diet, disease and biomedical measures population data available currently for Australia.

    • This study provided the first exploration of dietary patterns by rurality in Australia.

    • Due to the nature of the data, remoteness was dichotomised and may not fully consider the heterogeneity of rural areas. Further, the AHS did not sample from very remote populations in Australia.

    • A further strength is that the researchers involved in the data analysis have specific nutrition and dietetics expertise (LA and SEJ), and used eigenvalues and scree plots to guide the determination of the best number of components to extract from the dietary pattern analysis.

    Introduction

    For most populations residing in rural, remote or non-urban areas around the world, the experience of suboptimal health relative to urban (or metropolitan) counterparts is well documented.1–4 In high-income countries including Australia and the USA, diet-related risk factors, such as obesity, hypertension, diabetes and dyslipidaemia, are among the largest contributors to preventable disease and mortality.5 Acknowledging that all levels of remoteness, and the populations within these, are highly heterogeneous, we refer to rural, regional, remote and non-urban areas, as ‘rural’ here-in and metropolitan as ‘metro’. In Australia, residents of rural areas experience higher rates of cardiovascular disease (CVD), diabetes, some cancers and obesity when compared with metro populations.1 4 Furthermore, residents outside metro areas in Australia live almost exclusively in areas of lower socioeconomic status when compared using the nationally consistent Index of Relative Socio-economic Advantage and Disadvantage.6

    A lack of action and attention towards prevention in rural settings has been further exacerbated by a lack of research and evidence in these communities.7–10 Barclay et al demonstrated that there has been minimal research investment in rural areas of Australia, relative to need.10 More recently, research by Alston et al outlined the absolute scarcity of research funding allocated to understanding or improving dietary behaviours in rural, regional and remote Australia.11 This translates to many gaps in our knowledge around the role of diet in driving health disparities in rural areas and extends to how we can advance prevention activity that enables better health in future rural-dwelling generations.

    Despite the lack of research on rural dietary intakes in Australia,11 there is evidence to show diet plays a role in health inequities between rural and metro areas12 13 and that understanding dietsin rural populations are important to inform future evidence-based health promotion initiatives.14 Analysis of the most recent national-level dietary intake and modifiable risk factor data showed that if people living in rural Australia were able to achieve the same behaviour and risk factor profiles as their metro counterparts, the absolute gap in mortality between the two geographies would be reduced by 38%.12 Other modelling has shown that meeting fruit and vegetable recommendations would achieve the highest CVD mortality reductions (~40%) in CVD in rural areas out of all modifiable behaviours.15 Further evidence has shown that public health nutrition priorities are different between rural and metro Australia.15 A recent review that sought to synthesise all dietary data collected from rural Australians highlighted multiple gaps in dietary data collection outside of major cities in Australia, finding only 21 studies in the past 20 years.11 Of the data available on rural dietary intakes, over 50% of the studies collected data using non-validated tools and none looked at dietary pattern analysis specifically for rural populations.11 The majority of studies also did not compare dietary intakes with public health recommendations, meaning it was difficult for the authors to make meaningful conclusions and identify areas for improving dietary intakes in rural areas.11 Alston and Partridge recently highlighted that despite knowledge on the role of diet in the mortality gaps between rural and metro areas, few interventions have been conducted in these settings,16 and further research into dietary intakes across different geography is needed to understand areas of focus.16

    Exploration of dietary patterns in rural areas of Australia and comparison with metro areas has been minimal to date.11 Population-level dietary patterns reflect different dietary practices and culture, and may be linked to inflammation and chronic disease patterns, such as CVD.17 18 Dietary pattern analysis is useful in providing easy to translate information for the general public and also informs public health nutrition campaigns.19 Usually, remoteness is and included in dietary analyses as a confounder rather than analysed separately to assess the possibility of differing diet patterns in rural and metro regions.2 Alongside a lack of dietary data collection and analysis, there has been limited application of dietary inflammatory scoring methods in Australian populations.

    The REasons for Geographic and Racial Differences in Stroke (REGARDS) study has provided comprehensive regional evidence of the association between particular dietary patterns and the risk of stroke in the USA.20 From this study, REGARDS researchers have developed and validated a ‘Dietary Inflammation Score’ (DIS) based on data from intake of food groups and association with inflammatory biomarkers.21 The DIS allows for an estimation of the influence of dietary patterns on systemic inflammation, which increases the risk of chronic disease.21 Exploration of the how the DIS differs between rural and metro areas may provide evidence to further understand areas for nutrition intervention and research that will address health disparities for rural Australians. Using Australia’s most recent comprehensive national survey of biomedical measures and dietary intakes,22 we set out to:

    1. Empirically define dietary patterns and to apply the novel DIS, in data from rural and metropolitan populations in Australia.

    2. Investigate associations of both dietary patterns and the DIS with CVD risk factors.

    Methods

    Setting

    Metro and rural Australia. Very remote areas of Australia were not included in the sampling.23

    Patient and public involvement

    None.

    Study design

    The AHS consists of two separate surveys: the National Health Survey (NHS), which includes the National Nutrition and Physical Activity Survey (NNPAS) and the National Health Measures Survey (NHMS).23 These behavioural and biomedical survey data are the most recently available nationally representative data that include comprehensive dietary data coupled with disease and biomedical measures data.23 The survey was conducted using a stratified multistage area sample of private dwellings, which generated detailed estimates across remoteness areas of Australia.23 Dwellings were selected at random using a multistage area sample of private dwellings for the NHS (25 080 households and 31 837 respondents aged 2 years and over in the core sample), and for the subset NNPAS component, initially included 12 400 dwellings.23 Within each dwelling, a random subsample was selected of one adult over the age of 18 years (and one child aged 2–17 years, data not analysed here). Of those dwellings, 9519 dwellings included people who adequately responded to the survey, yielding a total sample for the survey of 12 153. Dietary intakes were collected in the NNPAS using two, 24-hour dietary recalls (n=12 153, 77.0% response rate).24 A total of 11 246 people participated in the NHMS. For the present analysis, individuals were split into two data sets: ‘rural’ and ‘metro’ population. The ‘rural’ population was defined as all included participants who were not classified by the Australian Bureau of Statistics Australian Geographical Standard Remoteness Areas25 as living in major cities (ie, those living in Inner Regional Australia, Outer Regional Australia and Remote Australia). Self-report measures (such as whether or not participants had ever been told they have high blood pressure, high cholesterol, diabetes mellitus or ischaemic heart disease (IHD)) were collected via the NNPAS survey.23 Biomedical measures were collected via the NHMS and participants were asked to provide both a blood and urine sample, which were then analysed for specific biomarkers.23 Detailed specific sampling and data collection methods are outlined elsewhere.23

    Individuals were excluded if they:

    • Reported had implausible energy intakes, defined as an energy intake to basal metabolic rate ratio of less than 0.9, as per recommendation from the ABS23 from both day 1 or day 2 (removed 4528 participants).

    • Were <18 years of age (removed 3418 participants).

    • Did not participate in the biomedical measure component of the study (removed 1898).

    This resulted in the current analysis samples of 1185 and 713 participants from metro and rural areas, respectively (figure 1). Data were analysed using a complete analysis approach.26

    Figure 1
    Figure 1

    Analytical sample from n=12 153. NNPAS, National Nutrition and Physical Activity Survey.

    Dietary and person-level data

    Food items in the NNPAS basic Confidentialised Unit Record File were classified into 56 food groups based on previous research20 using a combination of 3-digit, 5-digit and 8-digit food codes as required (online supplemental table 1), with reference to nutrient composition and traditional food groupings derived from the AUSNUT 2011–2013 food nutrient database.27 The mean consumption of each group over days 1 and 2 of the recall was calculated.22 If data were from day 1 or 2 of the recall, a single day of data was used. Food consumption data were then matched and merged with person-level demographic and self-report data from the NNPAS and biomedical level data from the NHMS using the ABS unique person identifiers. Data from the NNPAS and NHMS also included both self-report and measured cholesterol, blood pressure, diabetes and the presence of IHD. The mean intakes of key nutrients were calculated for each population.27

    Dietary pattern analysis

    The 56 food groups were then used to derive dietary patterns using principal component analysis (PCA), replicating methods previously used in the REGARDS study20 and consistent with recommendations of a recent systematic review of dietary pattern analysis methodologies.28 PCA is commonly used in dietary patterns research and replaces a set of potentially correlated, predefined food groups with a new set of principal components that are uncorrelated and retain as much of the food variance as possible.29 Patterns were derived separately for rural and metro adult populations. PCA was used for extraction of factors and varimax rotation was used to derive non-correlated factors.29 Following factor analysis in the two populations, we used the Kaiser criterion and scree plots to determine how many patterns to select in each population.29 The Kaiser criterion is defined as an Eigenvalue of >1.0 and widely used for the choice of the number of factors in factor analysis.29 For everyone in the sample, the factor loading of each food group was multiplied by the mean consumption of each food group (in grams per day) to calculate factor scores for each dietary pattern. Adherence to dietary pattern was determined by splitting individual factor scores into quartiles, consistent with previous studies.20 30 31 For example, for a given dietary pattern, participants with a factor score in the top 25% were categorised as high adherers, whereas individuals with a factor score in the lowest 25% were categorised as low adherers.20 30 31 The adherence of a participant to any given dietary pattern did not preclude the participant from being a high or low adherer to any other dietary pattern.

    Dietary Inflammation Score

    The validated DIS was developed by Byrd et al, using REGARDS data from >30 000 participants, and has been applied to multiple other studies exploring dietary data.21 31 32 To apply the DIS to the data in this study, food groups and supplements were categorised according to the DIS 19 groupings (consisting of 18 food groups and 1 multivitamin/mineral supplement group).21 The DIS weights were developed in the REGARDS study data, by assessing the strengths of the multivariable-adjusted associations of each food group component with measured circulating biomarkers of inflammation, including high-sensitivity C reactive protein, interleukin-6 (IL-6), IL-8 and IL-10.21 The DIS was then calculated by multiplying the food group components (in g/day) by the DIS weightings for each group. The population was then standardised by sex, to standardised to have a mean of 0 z-scores with a mean of 0 and SD of 1. Detailed scoring methods for the DIS are documented elsewhere.21

    Statistical analysis

    Data were analysed using STATA (V.17.0, StataCorp) using weightings to account for the complex survey design. T-tests of means and proportions were used to understand differences between rural and metro samples’ demographic characteristics and overall nutrient intakes. Logistic regression was used to assess the association of each dietary pattern with dichotomous outcome variables, including: self-reported presence of IHD (yes/no), self-reported diabetes (yes/no), self-reported high cholesterol (yes/no) and self-reported hypertension (yes/no), measured cholesterol (normal/high), measured blood pressure (normal/high), diabetes (estimated from average HbA1c (high/low) and weight status category (normal/overweight and obesity, calculated from measured body mass index (BMI). The DIS s were compared with the four measured biomedical outcomes due to the objective measure of these biomedical data, against the validated DIS, as opposed to self-reported outcomes. Confounders were selected based on the dietary patterns literature and included sex, age (65 years and over vs below 65 years), socioeconomic status (using the Socio-Economic Index for Areas (SEIFA) 2011, Index of Relative Socio-Economic Disadvantage quintiles,33 mean energy intake (kilojoules per day), physical activity (meeting vs not meeting physical activity recommendations of 150 min/week,34 categories for highest level of education attained (bachelor degree, diploma, certificate, certificate not otherwise classified) and smoking status (current weekly, current less than weekly, ex-smoker or non-smoker).20 22 35

    Results

    Demographics

    There were differences in demographic characteristics between the rural and metro sample (outlined in online supplemental table 2). The rural sample was significantly older than the metro sample, with a mean age of 52.7 compared with 48.6 years in metro areas, and a significantly higher proportion of the population disadvantaged (SEIFA), compared with the metro population. Significantly higher proportions of the rural population reported having hypertension, high cholesterol and IHD than the metro population. The rural population also had significantly higher saturated fat and sodium intakes and measured overweight/obesity compared with the metro population.

    Dietary patterns

    Based on the Kaiser criterion and scree plots,28 two primary dietary patterns emerged in each of the rural and metro populations (ie, a total of four dietary patterns). The top 10 factor loadings for each dietary pattern are outlined in figure 2 (full factor loadings presented in online supplemental table 3). In rural areas, pattern 1 ‘veg, red meat and fruit diet’ (2A) had high factor loadings for other vegetables, tomatoes, red meat, potatoes, cruciferous vegetables, fruit, leafy green vegetables, dark vegetables, yoghurt. Pattern 2, ‘high fat, sweet biscuits and coffee diet’ in rural areas had high loadings for added fat, butter, added sugars, white bread, sweet biscuits, soup, processed meat, high fat milk, coffee and tea (2B). In metro areas, pattern 1 ‘high fat, sugar-sweetened beverages (SSB), beer and chips diet’ (2C) had high factor loadings for high fat milk, high fat dairy, SSBs, white bread, beer, chips, salty snacks, butter, processed meat and added fats. Pattern 2 includes, ‘vegetable, butter and red meat diet’ other vegetables, tomatoes, potatoes, added fats, butter, red meat, eggs, dark vegetables, wine and leafy green vegetables (2D). Table 1 shows the mean intake of nutrients by dietary pattern among the highest adherers by either rural or metro population. The table shows differences between the mean intakes among high adherers of either pattern.

    Figure 2
    Figure 2

    Retained factors for rural and metro populations with top 10 weighted food groups.

    View this table:
    Table 1

    Mean intakes of nutrients by dietary pattern among the highest adherers and rurality

    Table 2 shows the ORs for the association of each quartile of adherence to either pattern 1 or 2 in rural and metro populations with self-reported IHD and with CVD risk factors. In rural areas, the highest and second quartiles of adherence to dietary pattern 1 were both strongly associated with having self-reported hypertension (Q1 OR 2.54, 95% CI 0.97 to 6.63; Q2 OR 2.90, 95% CI 1.16 to 4.36). The metro population also had significant results for the hypertension for the highest and Q2 of diet pattern 1. Self-reported diabetes or high cholesterol was not significantly associated with adherence to either dietary pattern in either population. People in rural areas with self-reported IHD were significantly more likely to be high adherers to diet pattern 2 (OR 13.90, 95% CI 2.29 to84.3).

    View this table:
    Table 2

    Associations between dietary pattern adherence and self-reported risk factors in rural and metro populations

    Table 3 shows dietary pattern adherence and associations with measured CVD risk factors. The odds of having high measured cholesterol increased with adhering to dietary pattern 1 across rural and metro populations but was not significant. Neither dietary pattern was significantly associated with having diabetes across the two populations. In the metro population high adherers to diet pattern 1 were more likely to have high blood pressure (OR 1.78, 95% CI 1.00 to 31.6) and overweight/obese if they were high adherers to diet pattern 2.

    View this table:
    Table 3

    Association between dietary pattern adherence and measured risk factors across rural, metro populations

    Dietary inflammatory score

    Table 4 shows the association of each quartile of the DIS with measured CVD risk factors. Across the rural and metro populations, there were mixed results, with a higher DIS trending toward having suboptimal measured risk factors compared with the lowest DIS for hypertension and obesity. The associations were not statistically significant except in rural populations, where Q2 and the highest quartile of the DIS were associated with being overweight/obese

    View this table:
    Table 4

    Dietary inflammation score and association with measured cardiovascular disease risk factors

    Discussion

    This is the first to explore differences in diet through dietary pattern analyses separately for rural and metro populations using AHS data and the first to apply the novel DIS developed by REGARDS researchers21 to Australian data. Our study provides evidence that dietary patterns are different between rural and metro populations in Australia. Research seeking to improve nutrition and prevent CVD needs to consider the influence of remoteness as a factor impacting on dietary intakes and risk factor profiles. Further, approaches that are effective and appropriate for preventing CVD and improving dietary intake patterns may differ between rural and metro areas.

    Unsurprisingly, the dietary patterns identified in both rural and metro areas across the Australian population were not in accordance with public health recommendations (ie, no pattern was consistently high in vegetables, fruit, wholegrains and whole grains lean meats or included minimal processed foods), consistent with the wider literature on diet patterns.4 36 There were few strong associations between dietary patterns and self-reported or measured CVD risk factors. The exceptions included that high adherence to dietary pattern 2 (‘high fat, biscuits and coffee diet’) in rural areas was significantly associated with self-reported IHD, and high adherers to diet pattern 2 (‘vegetable, butter and red meat diet’) in metro areas being more likely to have a higher measured BMI. Differences in results between measured and self-reported factors may be explained by differences in the proportion of people reporting a diagnosis of hypertension, compared with those who were measured to have higher blood pressure. A previous analysis found that 16% of a national sample of Australians under-reported hypertension in comparison to their measured blood pressure.37

    Reflective of the broader literature on differences in health and health behaviours by rurality,4 12 13 15 our study showed differences between dietary patterns by location. The dietary pattern 2 in rural areas showed a higher factor loading for added fats, compared with metro dietary patterns, which is consistent with previous research.15 A study by Alston et al found that there were differences in the relative contribution of different diet components to CVD deaths in rural compared with metro areas, concluding that addressing high fat intakes needs to be a priority in rural Australia.15 Potential reasons for the different dietary patterns in rural compared with metro areas may include different food culture, socioeconomic factors environments and food access in the different areas.38 39 For example, food environments in rural areas have been shown to be relatively unhealthy,39 with rural communities tending to experience high food insecurity,40 which may encourage consumption of higher fat and energy dense processed foods.

    Although there is evidence for the role of diet and lifestyle in rural health inequities, this analysis of dietary patterns also showed that the higher prevalence of CVD risk factors in rural areas is likely to be due to more than just dietary factors. Previous studies suggest that rural–metro health inequities are likely to arise from complex interactions between factors that include both health behaviours and structural challenges such as access to services and resources.1 Broadly, the literature investigating the link between dietary patterns and CVD and risk factors is mixed, with the use of multiple different methods making it difficult to draw comparisons between studies. For example, another Australian study using the AHS dataset, but different categorisations of foods to define patterns and different statistical analysis methods, identified a healthier dietary pattern that was associated with a healthier metabolic profile across the entire population (but did not separate the population by rurality).35 Another study, that produced results more similar to our findings, looked at adults aged 45 years and older across the whole Australian population (ie, not split by rurality), and identified three dietary patterns using factor analysis, yet found that adhering to a healthier dietary pattern was not associated with a healthier metabolic profile.41

    It is difficult to make comparisons with the wider literature using the DIS, as our study examined its cross-sectional association with measured risk factors at a single point in time rather than its longitudinal association with CVD risk factors or mortality, as has more commonly been analysed.21 31 32 Most studies that have examined the association of the DIS and mortality in cohort data have shown associations between a more inflammatory diet, and increased risk of mortality for CVD and cancers. Our findings are similar to another study in a healthy population that found no association between the DIS and cardiorespiratory fitness.30 A major difference may also be that the DIS components were derived and validated with data from a food frequency questionnaire, that assesses diet over an extended period, whereas the AHS used a 24-hour dietary recall to assess diet, which collects information on dietary intake over the previous 24 hours.2222 The lack of statistically significant results (aside from Q2 and Q4 of the DIS and overweight/obesity in the rural population) may also be due to the small sample size used in this study.

    Although the dietary patterns identified in this study did not show strong associations with CVD risk factors overall, this may reflect limitations of the dietary data (including limited number of foods on the instrument). In addition, dietary intakes at a single point in time do not capture the cumulative impact of lifetime diet, and there is a potential for reverse causality in cross-sectional data (where, eg, those with known CVD risk factors may have altered their dietary intakes). Also, we were only able to analyse metro compared with rural populations as a dichotomous indicator of rurality, due to the limited nature of these data. This meant that our study could not examine the heterogeneity of health and environments across different areas of remoteness. More dietary intake data are needed to understand dietary patterns across all areas of remoteness, considering updated remoteness measures in Australia, such as the Modified Monash Model.6 11 42 This is of particular need, following years of low investment in diet research in rural Australia despite well-documented health inequities in which diet plays an important role.2 Future surveys that capture remoteness, dietary patterns, and also clinical variables that allow the calculation of CVD risk (eg, Framingham Risk Equation43) would not only allow for investigations into remoteness with more granularity than undertaken here,43–45 but also an examination of the interaction between these variables.

    This study used representative national data sets, which represent the most recent, highest quality and comprehensive diet, disease and biomedical measures population data available currently for Australia. This study provided the first exploration of dietary intakes by rurality; however, due to the nature of the data, remoteness was dichotomised and does not consider the heterogeneity of rural areas. Further, the AHS did not sample from very remote populations in Australia,23 and due to missing data and implausible intakes, our sample size was greatly reduced from the original sample which may further reduce generalisability of the results. The nature of the data removed was also not at random and based on recommendation from the AHS and previous literature. As there is evidence that there has been a lack of nutrition research specific to rural areas,2 11 we did not explore rurality as a subgroup analysis and analysed the two groups separately to ensure a detailed understanding of different dietary patterns. This may reduce the generalisability of the results. Another limitation is that the PCA method that we used does require some subjectivity to extract dietary components. Decisions that are influenced by subjectivity include those around the number of factors to extract and description of the components for each of the dietary patterns identified. However, the researchers involved in the data analysis have specific nutrition and dietetics expertise (LA and SEJ), and the use of eigenvalues and scree plots guided determination of the best number of components to extract. Further, the cross-sectional study design used here cannot infer causal relationships or the influence of unmeasured confounding factors such as access to healthy food, local food environment differences or other social norms that may influence diet, that are not assessed in the AHS. The identification of confounders was selected based on previous literature.35

    Conclusion

    Two primary dietary patterns emerged each in rural and in metro areas, with differences between the two populations. Neither pattern was strongly associated with self-reported or measured CVD risk factors, aside from IHD in rural areas and overweight/obesity in metro areas. Our study provides evidence that action targeting healthier dietary intakes and CVD risk factors needs to be tailored to rurality. Further dietary analyses comparing rural and metro areas is required to build knowledge on different dietary intervention priorities between the two populations.

    Data availability statement

    Data are available in a public, open access repository. The data are available from the Australian Bureau of Statistics.

    Ethics statements

    Patient consent for publication

    Ethics approval

    Due to the nature of the data, being previously collected and deidentified, an exemption from ethics review was obtained from Deakin University (ref: 2021-382).

    Acknowledgments

    We would like to acknowledge the contributions made by members of the survey advisory group who made invaluable contributions to the design of the AHS. We would also like to acknowledge the expert advice of the Deakin University Biostatistics team.

    References

    1. ↵
      1. Alston L,
      2. Allender S,
      3. Peterson K, et al
      . Rural inequalities in the Australian burden of ischaemic heart disease: a systematic review. Heart Lung Circ 2017;26:122–33. doi:10.1016/j.hlc.2016.06.1213
      OpenUrl
    2. ↵
      1. Alston L,
      2. Raeside R,
      3. Jia SS, et al
      . Underinvestment in nutrition research for at-risk populations: an analysis of research funding awarded in Australia from 2014 to 2021. Nutr Diet 2022;79:438–46. doi:10.1111/1747-0080.12740
      OpenUrl
    3. ↵
      1. Logan RI,
      2. Castañeda H
      . Addressing health disparities in the rural United States: advocacy as caregiving among community health workers and promotores de Salud Int J Environ Res Public Health 2020;17:9223. doi:10.3390/ijerph17249223
    4. ↵
      1. Australian Institute of Health and Welfare
      . Australia’s health 2018 Canberra, Australia. 2018. Available: https://www.aihw.gov.au/reports/australias-health/australias-health-2018/contents/indicators-of-australias-health/fruit-and-vegetable-intake
    5. ↵
      1. Melaku YA,
      2. Renzaho A,
      3. Gill TK, et al
      . Burden and trend of diet-related non-communicable diseases in Australia and comparison with 34 OECD countries, 1990-2015: findings from the Global Burden of Disease Study 2015. Eur J Nutr 2019;58:1299–313. doi:10.1007/s00394-018-1656-7
      OpenUrlPubMed
    6. ↵
      1. Versace VL,
      2. Skinner TC,
      3. Bourke L, et al
      . National analysis of the modified Monash model, population distribution and a socio-economic index to inform rural health workforce planning. Aust J Rural Health 2021;29:801–10. doi:10.1111/ajr.12805 Available: https://onlinelibrary.wiley.com/toc/14401584/29/5
      OpenUrl
    7. ↵
      1. Alston L,
      2. Bourke L,
      3. Nichols M, et al
      . Responsibility for evidence-based policy in cardiovascular disease in rural communities: implications for persistent rural health inequalities. Aust Health Rev 2020;44:527–34. doi:10.1071/AH19189
      OpenUrl
    8. ↵
      1. Alston L,
      2. Field M,
      3. Brew F, et al
      . Addressing the lack of research in rural communities through building rural health service research: establishment of a research unit in Colac, a medium rural town. Aust J Rural Health 2022;30:536–9. doi:10.1111/ajr.12860
      OpenUrl
    9. ↵
      1. Alston L,
      2. Nichols M,
      3. Allender S
      . Policy makers' perceptions of the high burden of heart disease in rural Australia: implications for the implementation of evidence-based rural health policy. PLoS One 2019;14:e0215358. doi:10.1371/journal.pone.0215358
    10. ↵
      1. Barclay L,
      2. Phillips A,
      3. Lyle D
      . Rural and remote health research: does the investment match the need Aust J Rural Health 2018;26:74–9. doi:10.1111/ajr.12429
      OpenUrlPubMed
    11. ↵
      1. Alston L,
      2. Walker T,
      3. Kent K
      . Characterizing dietary intakes in rural Australian adults: a systematic literature review. Nutrients 2020;12:3515. doi:10.3390/nu12113515
    12. ↵
      1. Alston L,
      2. Peterson KL,
      3. Jacobs JP, et al
      . Quantifying the role of modifiable risk factors in the differences in cardiovascular disease mortality rates between metropolitan and rural populations in Australia: a macrosimulation Modelling study. BMJ Open 2017;7:e018307. doi:10.1136/bmjopen-2017-018307
    13. ↵
      1. Pullen R,
      2. Kent K,
      3. Sharman MJ, et al
      . A comparison of diet quality in a sample of rural and urban Australian adults. Nutrients 2021;13:4130. doi:10.3390/nu13114130
    14. ↵
      1. Schumacher TL,
      2. Alston L,
      3. Wakely L, et al
      . Characterizing the health of older rural Australians attending rural events: implications for future health promotion opportunities. Int J Environ Res Public Health 2022;19:3011. doi:10.3390/ijerph19053011
    15. ↵
      1. Alston L,
      2. Jacobs J,
      3. Allender S, et al
      . A comparison of the modelled impacts on CVD mortality if attainment of public health recommendations was achieved in metropolitan and rural Australia. Public Health Nutr 2020;23:339–47. doi:10.1017/S136898001900199X
      OpenUrl
    16. ↵
      1. Alston L,
      2. Partridge SR
      . Limited dietary interventions in rural Australian communities: a systematic review. Nutr Diet 2021;78:57–68. doi:10.1111/1747-0080.12612
      OpenUrl
    17. ↵
      1. Willett WC
      . Diet and health: what should we eat Science 1994;264:532–7. doi:10.1126/science.8160011
      OpenUrlAbstract/FREE Full Text
    18. ↵
      1. Hu FB,
      2. Rimm EB,
      3. Stampfer MJ, et al
      . Prospective study of major dietary patterns and risk of coronary heart disease in men. Am J Clin Nutr 2000;72:912–21. doi:10.1093/ajcn/72.4.912
      OpenUrlAbstract/FREE Full Text
    19. ↵
      1. Willett WC
      . Diet and health-finding a path to veritas. Eur J Epidemiol 2018;33:127–35. doi:10.1007/s10654-018-0378-8
      OpenUrl
    20. ↵
      1. Judd SE,
      2. Gutiérrez OM,
      3. Newby PK, et al
      . Dietary patterns are associated with incident stroke and contribute to excess risk of stroke in black Americans. Stroke 2013;44:3305–11. doi:10.1161/STROKEAHA.113.002636
      OpenUrlAbstract/FREE Full Text
    21. ↵
      1. Byrd DA,
      2. Judd SE,
      3. Flanders WD, et al
      . Development and validation of novel dietary and lifestyle inflammation scores. J Nutr 2019;149:2206–18. doi:10.1093/jn/nxz165
      OpenUrl
    22. ↵
      1. Australian Bureau of Statistics
      . Australian health survey: usual nutrient intakes. 2015. Available: https://www.abs.gov.au/statistics/health/health-conditions-and-risks/australian-health-survey-usual-nutrient-intakes/latest-release
    23. ↵
      1. Australian Bureau of Statistics
      . 4363.0.55.001 - Australian health survey: users' guide, 2011-13 2013. n.d. Available: https://www.abs.gov.au/ausstats/abs@.nsf/Latestproducts/F87826FDF716EB94CA257B8D00229E83?opendocument
    24. ↵
    25. ↵
      1. Australian Bureau of Statistics
      . Australian statistical geography standard (ASGS). 2022. Available: https://www.abs.gov.au/statistics/statistical-geography/australian-statistical-geography-standard-asgs
    26. ↵
      1. Hughes RA,
      2. Heron J,
      3. Sterne JAC, et al
      . Accounting for missing data in statistical analyses: multiple imputation is not always the answer. Int J Epidemiol 2019;48:1294–304. doi:10.1093/ije/dyz032
      OpenUrlCrossRefPubMed
    27. ↵
      1. Food Standards Australia and New Zealand
      . AUSNUT 2011-13 food nutrient database. 2013. Available: https://www.foodstandards.gov.au/science/monitoringnutrients/ausnut/ausnutdatafiles/Pages/foodnutrient.aspx
    28. ↵
      1. Zhao J,
      2. Li Z,
      3. Gao Q, et al
      . A review of statistical methods for dietary pattern analysis. Nutr J 2021;20:37. doi:10.1186/s12937-021-00692-7
    29. ↵
      1. Hair JFR,
      2. Tatham RL,
      3. Black WC
      . Multivariate data analysis with reading4th ed. New Jersy: Prentice Hall, 1995: 500.
    30. ↵
      1. Farazi M,
      2. Jayedi A,
      3. Noruzi Z, et al
      . Association of dietary and lifestyle inflammation score with cardiorespiratory fitness. Front Nutr 2022;9:730841. doi:10.3389/fnut.2022.730841
    31. ↵
      1. Farhadnejad H,
      2. Tehrani AN,
      3. Jahromi MK, et al
      . The association between dietary inflammation scores and non-alcoholic fatty liver diseases in Iranian adults. BMC Gastroenterol 2022;22:267. doi:10.1186/s12876-022-02353-3
    32. ↵
      1. Sohouli MH,
      2. Hadizadeh M,
      3. Mardali F, et al
      . Association between novel dietary and Lifestyle inflammation indices with risk of breast cancer (Brca): a case-control study. Nutr J 2022;21:14. doi:10.1186/s12937-022-00766-0
    33. ↵
      1. Australian Bureau of Statistics
      . Socio-economic indexes for areas. 2022. Available: https://www.abs.gov.au/websitedbs/censushome.nsf/home/seifa
    34. ↵
      1. Australian Government Department of Health and Aged Care
      . Physical activity and exercise guidelines for all Australians. Canberra, Australia: Australian Federal Government, 2021. Available: https://www.health.gov.au/topics/physical-activity-and-exercise/physical-activity-and-exercise-guidelines-for-all-australians
    35. ↵
      1. Livingstone KM,
      2. McNaughton SA
      . Association between diet quality, dietary patterns and cardiometabolic health in Australian adults: a cross-sectional study. Nutr J 2018;17:19. doi:10.1186/s12937-018-0326-1
    36. ↵
      1. Council NHaMR
      . Australian guide to healthy eating Australia. 2017. Available: https://www.eatforhealth.gov.au/guidelines/australian-guide-healthy-eating
    37. ↵
      1. Peterson KL,
      2. Jacobs JP,
      3. Allender S, et al
      . Characterising the extent of misreporting of high blood pressure, high cholesterol, and diabetes using the Australian health survey. BMC Public Health 2016;16:695. doi:10.1186/s12889-016-3389-y
    38. ↵
      1. Whelan J,
      2. Millar L,
      3. Bell C, et al
      . You can't find healthy food in the bush: poor accessibility, availability and adequacy of food in rural Australia. Int J Environ Res Public Health 2018;15:2316. doi:10.3390/ijerph15102316
    39. ↵
      1. Alston L,
      2. Versace V,
      3. Brown E, et al
      . Understanding the healthfulness of outlets providing lunch and dinner meals: a census of a rural food retail environment in Victoria, Australia. Aust N Z J Public Health 2021;45:65–70. doi:10.1111/1753-6405.13057
      OpenUrl
    40. ↵
      1. Kent K,
      2. Alston L,
      3. Murray S, et al
      . The impact of the COVID-19 pandemic on rural food security in high income countries: a systematic literature review. Int J Environ Res Public Health 2022;19:3235. doi:10.3390/ijerph19063235
    41. ↵
      1. Bell LK,
      2. Edwards S,
      3. Grieger JA
      . The relationship between dietary patterns and metabolic health in a representative sample of adult Australians. Nutrients 2015;7:6491–505. doi:10.3390/nu7085295
      OpenUrl
    42. ↵
      1. Australian Government
      . Modified Monash model. Canberra, Australia: Australian Federal Government, 2022. Available: https://www.health.gov.au/health-topics/rural-health-workforce/classifications/mmm
    43. ↵
      1. Davis-Lameloise N,
      2. Philpot B,
      3. Janus ED, et al
      . Occupational differences, cardiovascular risk factors and Lifestyle habits in South Eastern rural Australia. BMC Public Health 2013;13:1090. doi:10.1186/1471-2458-13-1090
    44. ↵
      1. Versace VL,
      2. Beks H,
      3. Charles J
      . Towards consistent geographic reporting of Australian health research. Med J Aust 2021;215:525. doi:10.5694/mja2.51344
    45. ↵
      1. Beks H,
      2. Walsh S,
      3. Alston L, et al
      . Approaches used to describe, measure, and analyze place of practice in dentistry, medical, nursing, and allied health rural graduate workforce research in Australia: a systematic Scoping review. Int J Environ Res Public Health 2022;19:1438. doi:10.3390/ijerph19031438

    Supplementary materials

    Footnotes

    • Twitter @katmlivingstone

    • Contributors LA, SEJ, VV, MN, KAB, SA, GH and JMS: formulated the research question(s). LA and SEJ: designing the study. LA, MN and SEJ: analysed the data and LA, SEJ, VV, MN, KAB, SA, GH, JMS, KL, LJB, CZ and TS: assisted with interpreting the findings and all authors contributed to the writing and reviewing of the article. LA is the overall guarantor for this work.

    • Funding LA is supported by a National Heart Foundation fellowship (ID: Grant/Award Number: 102530 (LA)). Two authors (TS and LJB) are employed through the University of Newcastle Department of Rural Health, and VLV through Deakin Rural Health, under the Australian Government funded Rural Health Multidisciplinary Programme. MN is supported by an NHMRC Ideas Grant GNT2002234. KL is supported by a National Health and Medical Research Council Emerging Leadership Fellowship (APP1173803).

    • Competing interests None declared.

    • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.