Ultra-processed food consumption and obesity in the Australian adult population

The study investigates whether higher consumption of ultra-processed foods is associated with greater adiposity (BMI, waist circumference) and higher prevalence of obesity and abdominal obesity among Australian adults. The context is Australia’s high burden of non-communicable diseases and rising obesity prevalence over recent decades. Ultra-processed foods, as defined by the NOVA classification, are industrial formulations with cosmetic additives and poor nutrient profiles, which are increasingly prevalent in global and Australian diets. Prior evidence, including randomized and observational studies, suggests links between ultra-processed foods and weight gain/obesity. The study aims to quantify these associations in a nationally representative Australian adult sample and assess consistency across age, sex, and physical activity strata.

The paper references a growing body of cross-sectional and longitudinal studies across multiple countries linking ultra-processed food intake with weight gain, incident overweight/obesity, and higher BMI. A randomized controlled trial demonstrated increased energy intake (~500 kcal/day) and weight gain on an ultra-processed diet versus a non-ultra-processed diet, despite matched macronutrients and sodium. Ecological analyses have associated national availability/sales of ultra-processed products with obesity prevalence and BMI trajectories. Proposed mechanisms include obesogenic nutrient profiles (high free sugars, unhealthy fats, energy density; low fiber and micronutrients), processing-related factors (altered food structure, enhanced palatability, increased eating rate, impaired satiety signaling), beverage-specific effects, and potential gut microbiota alterations. Ultra-processed foods’ convenience, affordability, aggressive marketing, and larger portion sizes may displace minimally processed foods and increase energy intake. Ultra-processed consumption has also been linked with wider health risks (mortality, CVD, metabolic syndrome, cancer, diabetes, depression, gastrointestinal disorders).

Design and data source: Cross-sectional analysis using the Australian National Nutrition and Physical Activity Survey (NNPAS) 2011–2012, a nationally representative household survey employing complex, stratified, multistage probability sampling. Participants: Adults aged 20–85 years with complete BMI and waist circumference (WC) data. Exclusions: Pregnant/lactating women; implausible energy intakes (<1st or >99th percentile); missing outcome data. Final analytical sample: 7411 adults. A sensitivity analysis (to consider reverse causality) excluded individuals with extreme BMI (<p1, >p99), on special diets (weight loss/health reasons), or with diagnoses of diabetes, heart or kidney disease, yielding n = 4610. Dietary assessment: Two non-consecutive 24-h recalls collected (first in-person, second by telephone ~28 days later); analyses used the first recall to avoid bias from lower response on day 2. Food coding based on AUSNUT 2011–2013; mixed dishes disaggregated using the AUSNUT Food Recipe File. Foods classified by NOVA into: unprocessed/minimally processed; processed culinary ingredients; processed; ultra-processed (e.g., packaged breads, soft drinks, confectionery, cookies, breakfast cereals, flavored yoghurts, reconstituted meats, ready meals, instant soups/noodles). Exposure: Dietary share of ultra-processed foods as percent of total energy intake, categorized into quintiles (Q1 lowest to Q5 highest). Outcomes: BMI (kg/m²) and WC (cm) as continuous; obesity (BMI ≥30 kg/m²) and abdominal obesity (WC ≥88 cm women; ≥102 cm men) as binary outcomes. Covariates: Sex; age (continuous); education (≤9 years; 10–12 years; 12 years with graduate degree); income (SEIFA quintiles); zone (major cities; inner regional; other); country of birth (Australia/English-speaking country vs other); physical activity level (active if meeting ≥150 min/week in the last week; inactive otherwise); smoking status (never, former, current). Statistical analysis: Participants grouped into quintiles of ultra-processed energy share; participant characteristics compared across quintiles using chi-square (categorical) and unadjusted linear regression treating quintile as ordinal (continuous). Associations between ultra-processed share and outcomes assessed with linear (BMI, WC) and logistic regression (obesity, abdominal obesity), with unadjusted and multivariable-adjusted models including all covariates. Trend tests reported (P-trend). Subgroup analyses (multivariable adjusted, exposure as continuous per 10% increase) by age group (20–39, 40–59, ≥60 years), sex, and physical activity level (active/inactive). Sensitivity analysis excluding individuals with potential reverse causality factors as above; results compared to full sample. Survey weights and Stata survey procedures (v14) used to account for complex sampling and enable population inference. Model fit assessed via residual distribution plots.

- Ultra-processed foods contributed on average 38.9% of total energy among Australian adults; Q1 mean 12.7% (0–21.7%), Q5 mean 74.2% (62.1–100%). - Compared to Q1, individuals in Q5 were younger, more socioeconomically disadvantaged, more likely to be Australian/English-country born, inactive, current smokers, had higher energy intake, and were less likely higher educated or living in major cities (all P < 0.01). - Mean BMI and WC in the sample were 27.4 kg/m² and 92.8 cm; prevalence of obesity 26.5% and abdominal obesity 40.2%. - Significant dose-response associations (P-trend ≤ 0.001) between higher quintiles of ultra-processed food share and higher BMI, higher WC, and greater odds of obesity and abdominal obesity. Adjusted Q5 vs Q1 differences: • BMI: +0.97 kg/m² (95% CI 0.42, 1.51). • WC: +1.92 cm (95% CI 0.57, 3.27). • Obesity: OR 1.61 (95% CI 1.27, 2.04). • Abdominal obesity: OR 1.38 (95% CI 1.10, 1.72). - Subgroup analyses (per 10% increase in ultra-processed energy share) showed positive associations across age, sex, and physical activity strata; associations tended to be stronger among older adults (≥40 years), females (for BMI/WC) and inactive individuals; some WC/abdominal obesity associations in the youngest group were not statistically significant. - Sensitivity analysis excluding participants with potential reverse causality increased the magnitude of associations in Q5 vs Q1; significant dose-response associations persisted, including for WC and abdominal obesity.

The study demonstrates that a greater dietary share of ultra-processed foods is associated with higher BMI and WC and increased odds of obesity and abdominal obesity in a representative sample of Australian adults, even after adjusting for socio-demographic factors, physical activity, and smoking. These findings address the research question by showing consistent dose-response relationships and robustness across multiple subgroups and sensitivity analyses. The results align with RCT evidence indicating that ultra-processed diets increase ad libitum energy intake and cause short-term weight gain, and with longitudinal and cross-sectional studies across diverse populations linking ultra-processed consumption to weight gain and obesity. The discussion outlines plausible mechanisms: unfavorable nutrient profiles (high sugars, fats, energy density; low fiber/micronutrients), processing-related impacts on oral processing, eating rate, satiety signaling, and gut microbiota, as well as contextual factors (convenience, marketing, affordability, portion sizes) that may displace minimally processed foods and drive overconsumption. Given high and rising ultra-processed consumption in high-income countries, and associations with multiple adverse health outcomes, the findings underscore the relevance of considering food processing level in dietary guidance and obesity policy in Australia.

Higher consumption of ultra-processed foods is associated with greater adiposity and higher prevalence of obesity and abdominal obesity among Australian adults. This is the first nationally representative Australian analysis to document these associations, adding to growing international evidence on the role of food processing in adiposity. Despite the cross-sectional design, the associations are biologically plausible and consistent with experimental and longitudinal data. The results support dietary advice and policy actions to reduce ultra-processed food consumption and increase availability and affordability of unprocessed/minimally processed foods. Future research should extend to varied populations to assess context-specific magnitudes and determinants of ultra-processed intake and obesity, and conduct mechanistic studies to clarify causal pathways between processing and adiposity.

- Cross-sectional design precludes causal inference and temporality; reverse causality cannot be ruled out. Sensitivity analyses excluding individuals with special diets, extreme BMI, or diet-related diseases showed stronger associations but cannot fully eliminate reverse causality. - Dietary assessment based on a single 24-h recall may not reflect usual intake; the decision to use only day 1 avoided potential sampling bias from lower day-2 response but may bias associations toward the null. - Potential misreporting in 24-h recalls (particularly underreporting of foods perceived as unhealthy) could attenuate associations; implausible energy reporters were excluded to mitigate this. - Possible residual confounding from unmeasured variables (e.g., parity, menopause). - Potential misclassification in assigning foods to NOVA categories due to database design not tailored to processing classification, though standardized criteria and conservative approaches were used with independent review.