Short-term improvements in diet quality in people newly diagnosed with type 2 diabetes are associated with smoking status, physical activity and body mass index: the 3D case series study

Lifestyle modification is the first-line treatment following a new diagnosis of type 2 diabetes (T2D). Diet affects glycaemic control via effects on body weight and glucose–insulin homeostasis, and early, tight glycaemic control is associated with improved outcomes and reduced mortality. However, evidence on diet quality among people with T2D has largely been cross-sectional, limiting understanding of how diet quality changes immediately after diagnosis. This study asks: to what extent do people newly diagnosed with T2D improve their diet quality in the short term, and which demographic, behavioural, or health factors are associated with such improvements? Understanding these early changes can inform interventions to achieve early glycaemic control.

Multiple dietary patterns (Mediterranean, DASH, low-GI, moderately low-carbohydrate, vegetarian/vegan, and general healthy eating) improve glycaemic and cardiometabolic outcomes in T2D. Guidelines recommend tailoring dietary patterns to preferences and metabolic goals, with no single pattern suitable for all. The DASH pattern benefits glycaemic control, blood pressure, body weight, waist circumference and lipids in T2D and was used here to assess diet quality. A prior systematic review by the authors showed people with T2D typically have low-quality diets (insufficient fruit, vegetables, whole grains, low-fat dairy; higher red/processed meats), but available studies were cross-sectional. Prospective studies are needed to understand immediate post-diagnosis dietary changes and associated factors to inform early interventions.

Design and participants: The 3D study is a national prospective case-series of Australian adults (>18 years) diagnosed with T2D within the previous 6 months, followed at baseline, 3, 6, 9, and 12 months. This paper reports baseline and 3-month data. Informed consent obtained; study registered (ANZCTR ACTRN12618000375257) and ethics approved (Griffith University 2017/951). Recruitment via Diabetes Australia email invitations (May–Aug 2018). Measures: Interviewer-administered telephone surveys (~30 min) conducted by Accredited Practising Dietitians. Dietary intake: single 24-hour dietary recall per time point collected via ASA-24 (Australian version) and analyzed in FoodWorks to compute DASH scores (8–40) using standard scoring across seven components: fruits, vegetables, nuts/legumes, whole grains, low-fat dairy, red/processed meats, sodium, sugar-sweetened beverages. Health/behavioural measures at each timepoint: smoking status; weight, waist (tape provided); height at baseline; BMI categorized per WHO; physical activity via IPAQ short form; medication use; psychological distress via Kessler K10. Demographics (baseline): age, sex, education, living arrangement, social class, language, Indigenous status, income; remoteness via ARIA; socioeconomic status via SEIFA quintiles. Change definition: Change in diet quality was computed as 3-month DASH minus baseline. Participants classified as ‘improvers’ (≥3-point increase, based on prior evidence linking ≥3-point increases to HbA1c improvements) vs ‘maintainers’ (<3-point increase, no change, or decrease). Data management and analysis: Data entry double-checked on 10% sample; plausibility checks and outlier verification. Representativeness assessed against national datasets (LWD, MILES-2, ABS National Health Survey) by χ² goodness-of-fit. Longitudinal analyses examined crude associations between baseline demographics, health, and diet-related behaviours and being an improver vs maintainer. Categorical variables compared by Pearson’s χ² or Fisher’s Exact test; continuous variables by independent t-test/ANOVA or Mann–Whitney U/Kruskal–Wallis as appropriate. Flow and retention: 14,108 invitations; 415 responses (2.9%); 225 completed baseline; 203 completed 3-months (90.2% retention). Non-completers at 3 months were younger (p=0.006), had less time since diagnosis (p=0.018), and higher baseline DASH (p=0.029).

- Sample characteristics: 225 baseline participants (56% male); 203 completed 3 months. Cohort comparable to Australian diabetes cohorts by gender and BMI but younger and differed by remoteness and SES. Mean time from diagnosis to baseline: 114.5 ± 41.1 days. Over half had obesity (59.4%); >90% exceeded recommended waist circumference. - Baseline diet quality: Mean DASH 24.4 ± 4.7 (range 12–37), midrange of possible 8–40; no sex difference in DASH (p=0.623). Males had higher energy intake (p<0.001), higher vegetable (p=0.004) and whole-grain serves (p=0.034) than females. - Short-term change: 31.0% (63/203) were ‘improvers’ (≥3-point DASH increase) by 3 months; 69.0% (140/203) were ‘maintainers’. - Baseline predictors of improvement: Improvers had lower baseline DASH (21.1 ± 4.7 vs 25.4 ± 4.0; p<0.001), driven by lower fruit (p<0.001), vegetables (p=0.033), low-fat dairy (p=0.004), whole-grains (p=0.014), and higher SSBs (p<0.001), red/processed meats (p=0.053), and nuts/legumes (p=0.037). Improvers had higher baseline physical activity (high IPAQ: 19.7% vs 7.1%; p=0.028), were less likely to be current smokers and more likely ex-smokers (smoking status p=0.018). Mean BMI was lower among improvers (males: 29.5 ± 5.3 vs 32.0 ± 6.5; p=0.048; overall: 30.5 ± 6.0 vs 32.5 ± 6.7; p=0.045). No differences in demographic variables (sex, age, education, living situation, income, social class, remoteness, SES). - Additional observations: Medication use at baseline trended higher among improvers (73.8% vs 60.6%; p=0.071). Energy intake differences between groups were not significant.

The study addressed whether individuals newly diagnosed with T2D improve diet quality shortly after diagnosis and which factors relate to such improvements. About one-third achieved a clinically meaningful improvement (≥3 DASH points) over 3 months. Improvements were linked to lifestyle behaviours rather than demographics: non-smoking status, higher physical activity, and lower BMI (especially among men) at baseline were associated with greater likelihood of improvement. Individuals with poorer initial diet quality showed the largest gains, suggesting either greater room for improvement or changes initiated prior to baseline among maintainers who had higher initial DASH scores. The inverse association between BMI and diet quality change in men echoes prior cross-sectional evidence linking higher BMI with poorer diet quality, indicating that men with higher BMI at diagnosis may require additional support to enact dietary changes. The positive association between physical activity and diet quality supports integrated interventions targeting both behaviours early after diagnosis. Current smokers were less likely to improve diet quality, underscoring the need for targeted support for smoking cessation alongside dietary counseling in newly diagnosed T2D. Demographic invariance suggests that behavioural and clinical factors outweigh socio-demographic determinants in predicting early dietary change within this cohort.

This is among the first prospective investigations of immediate post-diagnosis diet quality change in T2D. Approximately one-third of participants achieved clinically meaningful short-term improvements in diet quality. Improvements were associated with lifestyle factors—non-smoking, higher physical activity, and lower BMI (in men)—rather than demographics. Clinical practice should prioritize integrated behavioural support early after diagnosis, particularly for smokers, individuals with low physical activity, and men with higher BMI. Future research should evaluate how early changes in diet quality relate to long-term metabolic outcomes and test targeted, combined diet–physical activity–smoking cessation interventions.

- Low response rate (minimum estimated 2.9%) may introduce non-response bias; email recruitment reach uncertain. - Lag between diagnosis and baseline (mean ~114 days) may have allowed dietary changes prior to baseline, potentially attenuating observed change. - Under-representation of Aboriginal and Torres Strait Islander peoples and non-English speakers, limiting generalizability. - Single 24-hour dietary recall per time point cannot capture day-to-day variability and may be subject to recall and social desirability bias. - DASH score summarizes diet quality but may mask differences in specific food patterns yielding similar scores. - Sample distribution differed by state, and cohort differed from national data on age, remoteness, and SES, affecting representativeness.