Aligning patient and planetary health: planetary health diet adherence and inflammatory bowel disease prognosis across Western and Eastern cohorts

Abstract

This study aims to evaluate whether planetary health diet (PHD) could be a potential tool to achieve sustainability in IBD management. We included 1455 IBD participants from the UK Biobank (UKB) and 205 from the ONE-IBD cohort. Dietary intake was captured through validated questionnaires and seven PHD adherence algorithms were evaluated. Environmental impact was estimated using dietary carbon footprints. Outcomes included IBD-related surgery, colorectal cancer, and all-cause mortality in the UKB, while readmission and recurrence in ONE-IBD. Cox models estimated hazard ratios (HRs) with 95% confidence intervals (CIs). During a median 10.5-year follow-up, 191 surgeries, 78 CRC, and 232 deaths occurred. Adherence to PHD were inversely associated with surgery risk (HR_{per SD} 0.67–0.80). All algorithms were linked to lower dietary carbon footprints (r −0.061 to −0.468). A modified-Knuppel algorithm provided similar protection (HR_{per SD} = 0.73, 95% CI 0.60-0.89) and environmental correlation (r = −0.244). In ONE-IBD, the modified algorithm was associated with lower readmission risk (HR_{per SD} = 0.80, 95% CI 0.65–1.00). Proteomics data suggest potential benefits through improved intestinal barrier integrity and reduced fibrosis. To conclude, higher PHD adherence is associated with better IBD outcomes and lower environmental burden, supporting that the PHD may represent a promising and sustainable dietary pattern in IBD management.

Inflammatory bowel disease (IBD), including Crohn’s disease (CD) and ulcerative colitis (UC), is a chronic immune-mediated inflammatory gastrointestinal tract disorder. Global epidemiological data indicate a rising prevalence of IBD, imposing a substantial and growing burden on healthcare systems¹. The varied and currently unpredictable treatment response and recurrent disease flares contribute to a poor prognosis, including a need for readmission, IBD-related surgery, colorectal cancer(CRC) incidence, and even premature death. These adverse clinical outcomes significantly impair patients’ quality of life and increase the economic burden to the society². Given these patient and public health implications associated with IBD flare, the current treatment target is sustained remission³. Notably, the theme “Sustainability in IBD and beyond” for the European Crohn’s and Colitis Organization (ECCO) 2025 focuses on holistic sustainable approaches in IBD aiming to minimize the disease impact in the patient, the society, and even to the environment. This encompasses disease prevention, treatment optimization, and efficient resource use in disease management³. Aligned with this focus, dietary management has been proposed as an accessible and appropriate option to achieve prognostic sustainability in IBD, by promoting short-, medium-, and long-term remission and reducing healthcare costs associated with adverse clinical outcomes^4,5.

A variety of foods have the potential to influence IBD prognosis^5,6. Studies have linked to a diet is low in sugar or artificially sweetened beverage⁷, high fiber intake⁸ or plant-based ingredients⁹ to improved IBD prognosis. The Planetary Health Diet (PHD), also known as the EAT-Lancet Diet, advocates a plant-based dietary regimen, with only moderate or small amounts of fish, dairy, and meat recommended¹⁰. Considering the different economic/agricultural systems and cultural traditions around the world, foods were categorized into 13 groups. In this dietary pattern, less than 60% of total energy intake is contributed by grains, while protein is sourced mainly from legumes and nuts¹⁰. Although animal-derived foods had an inverse impact on the planet, low to moderate amounts of fish, poultry, and dairy products were still included in this frame, considering the health benefits of fish-derived omega-3 fatty acids and its accessibility to the deprived population. Similarly, potatoes were also included for its global accessibility, although it may impede chronic disease management due to its impact on the glycemic load^9,10. Featuring great accessibility across geographical regions and economic strata, the PHD aims to harmonize human health, economic viability, and environmental sustainability, promoting dual benefits for both humans and the planet¹¹. While previous studies have linked PHD adherence to a reduced incidence of IBD^12,13, evidence supporting its beneficial effect on the prognosis of IBD remains scarce. Therefore, it is important to determine whether adherence to PHD improves outcomes in patients with IBD and evaluate its potential to advance both prognostic and environmental sustainability.

Leveraging data from the UK Biobank (UKB) cohort and validation Eastern cohort, we conducted a prospective cohort study with three aims: (1) to systematically investigate the association between PHD adherence and long-term prognosis for IBD using different algorithms of PHD; (2) to clarify the effective components of PHD that are associated with prognosis; (3) to examine whether the IBD prognostic sustainability achieved by PHD aligns with dietary environmental sustainability in the study population; In addition, we used proteomics data to further explore underlying mechanism of how PHD influence intestinal health.

Results

Baseline characteristics

The baseline characteristics of the UKB participants who were categorized into three groups as tertiles of Knuppel’s score (0–14) are shown in Table S1. A total of 1455 participants were included in the study, among whom the mean age (SD) was 60.1 years (7.8 years) at baseline, and 757 (52%) were female. Participants with high adherence to the PHD were more likely to be female, possess a higher educational level, engage in sufficient physical activity, and exhibit lower BMI and CCI scores. Details for the characteristics of micronutrient intake are shown in Table S2.

In the Omics, Nutrition, and Environment Synthesis for Inflammatory Bowel Disease (ONE-IBD) cohort, a total of 205 participants were included in this study, among whom the mean age (SD) was 34.1 years (13.1 years) at baseline, and 54 (26.3%) were female. Participants with higher PHD adherence were more likely to be older and have higher fiber intake. The specific characteristics of the ONE-IBD population in this study are shown in Table S3.

Association between adherence to PHD and the prognostic sustainability of IBD

To evaluate the relationship between adherence to PHD and the prognostic sustainability of IBD, we analyzed the association between scores by seven algorithms of PHD and the risk of IBD-related surgery in participants with IBD, CD, and UC, using three models of different multivariable adjustment levels (Fig. 1 and Table S4).

Fig. 1. Association between scores by different algorithms of PHD and IBD-related surgery in participants from UKB with IBD, CD, and UC in a fully adjusted model.

IBD inflammatory bowel disease, CD Crohn’s disease, UC ulcerative disease, CI confidence interval, HR hazard ratio. a Fully adjusted model is adjusting for age, sex, physical activity, TDI, BMI, smoking status, alcohol drinking, total energy intake, carbohydrate intake, protein intake, fat intake, and added sugars intake.

Results from the fully adjusted Cox regression models demonstrated that a higher PHD score was significantly associated with reduced IBD-related surgery risk. Specifically, four algorithms including Knuppel’s (HR _{per SD}: 0.74, 95% CI: 0.61–0.89, P = 0.002), Kesse-Guyot’s (HR _{per SD}: 0.67, 95% CI: 0.56–0.82, P < 0.001), Cacau’s (HR _{per SD}: 0.79, 95% CI: 0.65–0.96, P = 0.019), and Hanley-Cook’s (HR _{per SD}: 0.80, 95% CI: 0.66–0.98, P = 0.031) demonstrated inverse associations with a reduced risk of surgery in participants with IBD, and showed significant inverse linear trends (all P for trend < 0.05). Among all algorithms, Knuppel’s score exhibited a significant inverse association with the risk of IBD-related surgery (HR _{T3 vs T1}: 0.47, 95% CI: 0.29–0.77, P for trend = 0.003), which was consistent with mortality (HR _{T3 vs T1}: 0.58, 95% CI: 0.37–0.90, P for trend = 0.016, shown in Table S5) in the further prognostic exploration. For risk of CRC, Trijsburg’s score revealed a significant inverse association (HR _{T3 vs T1}: 0.35, 95% CI: 0.12–0.98, P for trend = 0.026, shown in Table S6).

In participants with CD, scores by four out of seven PHD algorithms similarly showed significant inverse association with the risk of IBD-related surgery, in which Knuppel’s (HR _{T3 vs T1}: 0.46, 95% CI: 0.22–0.98, P for trend = 0.037) and Cacau’s (HR _{T3 vs T1}: 0.46, 95% CI: 0.22–0.95, P for trend = 0.021) showed a clear linear inverse trend (Table S7). In participants with UC, Knuppel’s (HR _{per SD}: 0.76, 95% CI: 0.59–0.98, P = 0.032) and Kesse-Guyot’s score (HR _{per SD}: 0.65, 95% CI: 0.50–0.85, P = 0.002) also demonstrated an inverse association with the risk of IBD-related surgery (Fig. 1 and Table S8).

Association between adherence to PHD and dual sustainability in IBD and environment in Western population

Since the PHD was designed to promote the dual sustainability for humans and the environment, we analyzed the correlation between scores by PHD algorithms and dietary carbon footprint to evaluate the association between adherence to PHD and environmental sustainability. Pearson correlation analysis revealed significant inverse correlations between all scores and individuals’ dietary carbon footprint. The inverse correlations were observed in Knuppel’s (Pearson r = −0.218, 95% CI: −0.266, −0.168, P < 0.001), Kesse-Guyot’s (r = −0.468, 95% CI: −0.507, −0.427, P < 0.001) and Stubbendorff’s (r = −0.296, 95% CI: −0.342, −0.248, P < 0.001) and other algorithms (Fig. 2A and Table S9).

Fig. 2. Associations of adherence to the PHD with dietary carbon footprint, IBD prognosis, and IBD-related proteins.

A association between dietary carbon footprint and adherence of PHD in UKB population; (B) association between dietary carbon footprint and adherence of PHD in ONE-IBD population; (C) forest plot of association of PHD adherence evaluated by Knuppel’s score and IBD-PHDI with prognosis of IBD in ONE-IBD in fully adjusted model; (D) Heat map of association between PHD adherence and IBD-related protein. (*P < 0.05, **P < 0.01, ***P < 0.001).

To investigate the independent effect of specific PHD food components on IBD prognosis, we evaluated associations between food components in PHD and IBD-related surgery risk in participants with IBD (as well as CD and UC). The analysis revealed an inverse association between most components and IBD-related surgery, except for potatoes and legumes. Specifically, vegetable intake demonstrated significant protective effects across most scoring systems except Hanley-Cook (Table S10).

Based on the observed positive association of potato and legume intake with IBD-related surgery, we developed a modified PHD algorithm from Knuppel’s tailored for individuals with IBD, called IBD-PHDI. Cox regression analysis demonstrated that the score by IBD-PHDI (HR _{per SD}: 0.73, 95% CI: 0.60–0.89, P = 0.002) exhibited a comparable inverse association with IBD-related surgery compared with Knuppel’s. Our further analysis confirmed that the score by IBD-PHDI exhibited significant inverse correlations with dietary carbon footprint (P < 0.001). Specifically, the Pearson correlation coefficient strengthened from −0.218 to −0.244 (Table S11 and Fig. 2C).

Association between PHD adherence and dual sustainability of IBD and environment in Eastern population

To explore the universality of the PHD in Eastern population, we evaluated the association between adherence to PHD and IBD prognosis in ONE-IBD participants. IBD-PHDI demonstrated a significant association with outcome of readmission (HR _{per SD}: 0.80, 95% CI: 0.65–1.00, P = 0.049; HR _{T3 vs T1}: 0.56, 95% CI: 0.33–0.94, P for trend = 0.016) and clinical recurrence (HR _{T3 vs T1}: 0.02, 95% CI: 0.00–0.72, P for trend=0.031). Meanwhile, other algorithm such as Knuppel’s, Hanley-Cook’s, and Colizzi’s also revealed inverse associations between the PHD adherence and clinical recurrence (Fig. 2B and Tables S12, 13). Our further analysis in dietary carbon footprint in ONE-IBD cohort also demonstrated significantly inverse correlation with adherence of PHD in Knuppel’s, Kesse-Guyot’s, Cacau’s, as well as IBD-PHDI algorithms by Pearson correlation coefficient (from −0.837 to −0.116)(Fig. 2C and Table S14).

Association between PHD adherence and IBD-related proteins in participants

To further explore the relationship between PHD and IBD prognosis in the molecular level, we evaluated the association between PHD and 17 IBD-related protein in serum of participants. Significantly inverse correlations were observed in canonical inflammatory pathways (CRP), IL-12/23 inhibitor-related pathways (IL-12B), IL-17 inhibitor-related pathways (IL-17A), biomarkers of intestinal barrier integrity (Occludin (OCLN), Epithelial Cadherin (CDH1), and LY6/PLAUR Domain Containing 8 (LYPD8)), and core protein biomarkers of intestinal fibrosis (Collagen Type III Alpha 1 Chain (COL3A1) and TIMP Metallopeptidase Inhibitor 1 (TIMP1)) (all p < 0.05). Specifically, the intestinal barrier integrity-related protein such as OCLN, showed significantly inverse association with most of algorithms. The details of the association between PHD and IBD-related protein are shown in Fig. 2D and Table S15.

Subgroup and sensitivity analysis

The association between adherence to PHD and prognostic sustainability was consistent across subgroups by age, sex, BMI, physical activity, smoking status, CRP levels, and pre-baseline IBD-related surgery with all P-interaction > 0.05 (Tables S16–S19). The inverse relationship observed in our primary analysis (i.e., the inverse association between PHD algorithm scores by Knuppel’s, Kesse-Guyot’s, Cacau’s, and Hanley-Cook’s and IBD-related surgery) remained after excluding participants who underwent IBD-related surgeries within 1 or 2 years post-baseline, and after adjusting for covariates including medication use and CCI (Tables S20, S21) .

Discussion

In this prospective cohort study across Western and Eastern cohorts, we found that adherence to PHD is associated not only with reduced risk of adverse short-, medium-, and long-term IBD prognosis, but also with better environmental sustainability by reducing individuals’ dietary carbon footprint. Among the components of PHD, vegetable intake contributed most substantially to the risk reduction of adverse IBD prognosis. Meanwhile, we demonstrated that adherence to PHD inversely associated with biomarkers of intestinal barrier integrity (OCLN, CDH1, and LYPD8) and core protein biomarkers of intestinal fibrosis (COL3A1 and TIMP1), which provided potential pathway for how PHD improving IBD prognosis.

The core concept of “sustainability in IBD and beyond” proposed by ECCO 2025 aligns with the United Nations’ Sustainable Development Goals (SDGs) framework, particularly SDG3 (Good Health and Well-being)^3,14. This goal emphasizes that lifelong global management of chronic diseases must integrate environmental, social, and economic sustainability. To achieve this goal in IBD management, diet could play an important role¹⁵. More specifically, the PHD has shown potential in chronic disease management and environmental sustainability. In comparison with other conventional dietary pattern, which served as beneficial factors in IBD, such as the Mediterranean diet, MIND diet, or the Anti-Inflammatory diet that focus on human health only, the PHD offers additional benefits for the planet^16–18. It was, for example, shown that the PHD was associated with a reduced risk of diabetes mellitus and might therefore represent an effective dietary intervention¹⁹. For IBD incidence, inverse associations with PHD adherence were demonstrated in cohort studies^12,13; building on this, our study further explored the link with IBD prognosis. Echoing established research demonstrating an association of higher PHD adherence with decreased mortality and reduced environmental impact²⁰, our study revealed this synergestic association that higher PHD adherence has a positive impact on the well-being of both individuals with IBD and the environment.

However, admittedly, the PHD is not a holy grail in the nutritional management of IBD. Ty Beal et al. have suggested that modifications to the original planetary health diet were needed to achieve micronutrient adequacy for adults, which included increasing foods of animal source and reducing those high in phytate²¹. In this study, participants with higher PHD adherence scores showed a significant inverse trend for intake of heme iron, vitamin B12, vitamin B6, vitamin D, and zinc, which may be attributed to lower animal-derived foods intake. In contrast, intake of vitamin C and magnesium increased with higher adherence. Overall, this suggested risk for inadequate micronutrient intake among those who had greater adherence to PHD, which may warrant dieticians to evaluate and construct a more precise PHD recommendation²².

Meanwhile, it is crucial to formulate a more accurate adjustment component range from the PHD for patients with IBD, who have specific nutritional requirements⁶ and severe nutritional problems²³. In this study, while higher adherence to the PHD had an overall tendency to reduce IBD-related surgery risk, the associations of individual food components with prognosis diverged. Specifically, higher intake of potatoes and legumes (including beans and soy) was associated with increased risk of adverse prognosis, whereas greater consumption of vegetables and eggs showed inverse associations. These divergent effects are likely stem from both the distinct biological impacts of specific foods on intestinal pathophysiology and variations in sustainability priorities embedded within dietary algorithms. Our modified algorithm, IBD-PHDI, which substitutes potatoes and legumes with energy-equivalent alternatives, offered a “win-win” alternative for both the individual and the environment.

The differential associations seen for individual food components also highlight the need to establish dynamic and counterbalanced dietary optimization protocols that ensures balanced integration of individualized health objectives with planetary sustainability imperatives. Take dairy consumption as an example. The PHD diet advocates flexible intake, recommending two daily servings of animal-sourced foods consisting of one serving of dairy products and one serving of non-dairy animal products¹⁰. Although our study did not identify a significant association between moderate dairy intake and IBD prognosis, previous investigations have indicated that whey protein derived from dairy may alleviate intestinal inflammation²³. In addition, dairy serves as a valuable source of micronutrients and high-quality protein to sustain adequate nutritional status, which is particularly critical for patients with IBD. These findings further reflect the balanced nutritional philosophy underlying the development of IBD-specific PHD diet.

We also found a paradoxical trend of positive associations between specific dietary sub-scores and IBD surgical risk, including the grain and fish sub-scores in Kesse-Guyot’s, the poultry sub-score in Colizzi’s, and the nut sub-score in Trijsburg’s. Further examination of scoring criteria demonstrated that Kesse-Guyot’s algorithm employs a dynamic energy-adjusted model with compensatory mechanisms, lacking an upper intake limit for grains and permitting fish intake scores to be offset by plant-based foods. This design may inadvertently promote high-grain/carbohydrate consumption or imbalanced fish intake, potentially adversely influencing IBD prognosis²⁴. Notably, Colizzi’s poultry intake limits are lower than other scoring systems, and such stringent protein restrictions may influence the process of recovery²⁵, thereby negatively impacting IBD prognosis. The Trijsburg is the only algorithm with an upper limit for nut intake. This sustainability-driven design that prioritized environmental goals may conflict with patients’ nutritional needs, such as the micronutrients from nuts, potentially elevating surgical risk²⁶.

In this study, we found that PHD may improve intestinal function through maintaining tight junctions and mitigating intestinal fibrosis to achieve better IBD prognosis; while previous evidences showed that dietary pattern or certain food could impact on IBD through intestinal barrier²⁷ or fibrosis²⁸. We observed a negative correlation between CRP and PHD scores, with serum CRP acting as a marker of acute inflammatory response, indicating that PHD may have the potential to improve the short-term prognosis of IBD. Furthermore, we found that higher PHD scores were associated with lower levels of COL3A1. As a key component of interstitial collagen, COL3A1 is implicated in fibrosis when excessively deposited, contributing to intestinal sclerosis and stricture, which can disrupt normal intestinal function. PHD may enhance the long-term prognosis of IBD by reducing interstitial collagen deposition and slowing the progression of fibrosis.

However, the specific mechanisms underlying this dietary pattern still stay unknown. We speculate that, firstly, certain components of the diet may have biological influence on intestinal status. As is recommended by the PHD, plant-sourced proteins were proven to be protective factors in IBD^29,30, while restricting red meat and sugar effectively lowered the risk of IBD flare-up¹⁸. Likewise, our previous multinational retrospective cohort study also confirmed that plant-based diets were protective factors against the adverse prognosis of IBD⁹, which is consistent with the observation in this study. Secondly, the PHD may influence the gastrointestinal microbiome through precise recommendations of foods³¹. Dysbiosis in IBD often results in dysfunction of the intestinal barrier and persistent inflammation³², which could be improved by healthy diet patterns^17,33. Evidence positioned precision nutrition as a new modulator of gastrointestinal microbiome and metabolism in IBD³⁴. The effect of fine-toning nutritional component in PHD on IBD prognosis could be analogous to a Mediterranean-like diet which, according to a cohort study, was strongly associated with a defined microbial composition and a lower level of subclinical gut inflammation³⁵.

Our study has several strengths, including being the first to propose and validate a dual-dimensional dietary framework integrating IBD prognostic sustainability with planetary environmental sustainability. We also assessed the concordance of these two sustainability goals with seven different algorithms and several rounds of WebQ to calculate PHD adherence. The results showed the same direction of effect, suggesting the reliability of our findings. However, our study has limitations. First, in this observational study, our findings cannot conclude a causal association between PHD and prognosis of IBD, since we cannot control the reverse causation bias and exclude the impact of all confounders. Second, the evaluation of diet depended on self-reporting by participants, which may have been influenced by recall bias or social desirability bias. Moreover, the food categories being not granular enough caused limitations to PHD adherence algorithms. The more precise subtype of the components had varied effects on IBD due to different inflammatory properties. For example, tomatoes and leafy green vegetables had controversial impacts on inflammation, but were assessed into the same category in our study³⁶. Meanwhile, due to the limitation of WebQ and SQFFQ, the algorithms of PHD did not take cooking methods such as grilling or boiling into consideration, which may vary in each category and influence the prognosis of IBD distinctly³⁷. Therefore, the large and simplified categories in the PHD algorithms may cause misclassification bias.

In this study, adherence to the PHD, as recommended by the EAT-Lancet Commission, is associated with benefits for both lowering adverse IBD outcome risks and reducing environmental impact of the diet in global population. Our study suggests potential of PHD to serve as a green and accessible means for achieving dual sustainability in universal IBD management and provides inspiration for chronic immune-mediated disease care systems not only for Western but Eastern people. Application of PHD diet, a gram-quantified recommendation system, may help address the vagueness of qualitative recommendations in conventional dietary guidelines. Furthermore, future research is needed to systematically evaluate strategies to balance clinical benefits and environmental objectives. Considering limitations inherent to observational research, randomized controlled trials that include different populations and more granular dietary considerations are needed to further develop the recommendations of PHD before recommending it to patients with IBD.

Methods

Study population

This study is a prospective cohort study based on the UKB, a large-scale cohort included over 500,000 participants at 22 assessment centers in the United Kingdom (UK). During 2009–2012, participants completed up to 5 rounds of online 24 h dietary recall questionnaires (WebQ), which included 206 foods and 32 beverages commonly consumed. To validate robustness of our results in Western population from UKB, we included Eastern participants from the Omics, Nutrition, and Environment Synthesis for Inflammatory Bowel Disease (ONE-IBD) Study in Asian (ISRCTN38557825) included 699 participants from 2023 to 2024 to evaluate short-term prognostic outcomes with a six-month follow-up period³⁸.

In this study we first included individuals who completed WebQ in UKB and Semi-Quantitative Food Frequency Questionnaire (SQFFQ) in ONE-IBD. We further included patients with baseline IBD, the diagnostic criteria for which are described elsewhere^7–9. Then we excluded participants with implausible energy intakes³⁹, self-reported atypical diets and participants from UKB who completed the WebQ only on one occasion. The detailed inclusion and exclusion criteria and study design of these two cohorts were presented in Fig. 3.

Fig. 3.

Flowchart of the study.

The ONE-IBD study has been approved by the Institutional Review Board of Third Xiangya Hospital, Central South University (ref No: R19057 and No: E25069) and is registered at the ISRCTN registry (No: ISRCTN38557825, https://www.isrctn.com). This UKB project received ethical approval by the North West-Haydock Research Ethics Committee (REC ref. ²¹:/NW/0157). This research was also carried out using the UK Biobank study under application number 232231.

Assessment of exposure

Dietary intake was assessed by WebQ from eligible participants in the UKB and the SQFFQ in ONE-IBD cohort, which had been converted to amount of daily intake⁴⁰. Food intake in WebQ was multiplied by portion sizes from McCance and Widdowson’s tables to estimate the total amount of intake. When validated against interviewer-administered 24 h recalls completed on the same day, the WebQ achieved a mean Spearman’s correlation coefficient of 0.6 across 21 calculated nutrients 10, indicating a moderate correlation. For participants in ONE-IBD, dietary intake in SQFFQ were evaluated by experienced doctors and nutritionist. Intake frequencies reported on a weekly or monthly basis were converted to daily frequencies and then to intake amounts.

To evaluate adherence to PHD, we adopted seven algorithms of PHD as exposure to ensure a comprehensive and robust assessment, which was validated in the previous systematic review²⁰. Knuppel’s score⁴¹ was considered as the primary algorithm to assess PHD adherence given the reported association between Knuppel’s score and incident of IBD^12,13, while the other six were incorporated as secondary analyses. These algorithms are distinct in their scoring systems and their associations with individual health and environmental sustainability²⁰ (Table S22). A higher score in the algorithm represents a better adherence to PHD, meaning that the individual’s dietary intake of each component is closer to the PHD recommendation. All dietary intake data were categorized according to the 14 food components defined by the PHD, which are shown in Tables S23, S24. Detailed scoring criteria and operational definitions are presented in Suppl. Tables A1–A8.

Based on the positive association of IBD-related surgery with potatoes/legumes sub-scores (HR _{per SD} > 1) and the strong inverse association of Knuppel’s score with both adverse IBD prognosis and dietary carbon footprint, we developed the IBD-PHDI, a modified PHD index adapted from Knuppel’s algorithm for individuals with IBD. In the analysis of associations between PHD components and IBD prognosis, higher sub-scores for potatoes and legumes (per SD increase) were associated with increased hazard ratios (HR > 1), suggesting these foods may adversely impact IBD outcomes. Thus, considering the significant inverse association of the integral Knuppel’s algorithm in T3 vs. T1 with both adverse prognosis and dietary carbon footprint in our study, as well as the previous studies that revealed an inverse association of PHD with the risk of IBD via Knuppel’s algorithm, we conducted a modified algorithm called IBD-PHDI based on Knuppel’s, substituting potatoes and legumes with energy-equivalent and similar nutritional components. As legumes were divided into beans and soy in Knuppel’s algorithm, we substituted beans with fish, soy with egg, and potatoes with grains. We substituted beans with fish due to their comparable fatty acid and protein profiles, and replaced soy (tofu) with eggs, given their shared high protein content. Grains were chosen to replace potatoes as both are primary carbohydrate sources.

For instance, while Knuppel’s algorithm recommends daily intakes of 100 g potatoes (78 kcal) and 464 g grains (1622 kcal), our modification added 78 kcal to the baseline grain energy (1622 kcal), yielding a revised total grain energy intake of 1700 kcal. Converted to mass, this corresponds to 486 g of grains. Via this substitution criteria, in the IBD-PHDI, potato, bean, and soy intake were limited to 0 g/d, and the sub-scores would be 0 points when upper than the limit; while grain, fish, and egg intake increased from 464 g/d to 486 g/d, from 100 g/d to 220 g/d, and from 25 g/d to 99 g/d, respectively. The algorithms for other components were consistent with Knuppel’s. The total score is the sum of sub-scores, ranging from 0 to 14. The detailed criteria were shown in Table B1.

Ascertainment of outcome

According to the SPIRIT consensus from the International Organization for the Study of Inflammatory Bowel Diseases (IOIBD), recurrence is considered a short-term impact on the patient’s quality of life, while readmission and IBD-related surgery are classified as medium-term endpoints. In addition, CRC and mortality are regarded as long-term endpoints of IBD⁴. For participants from the UKB, IBD-related surgery was selected as the primary outcome given its critical and sustained impact on long-term prognosis⁴. Secondary outcomes included CRC and mortality. The outcomes were obtained by external linkage to national hospital inpatient records, primary care data, cancer registry, and the death registry. To avoid confounding by pre-existing disease, participants with CRC at baseline were excluded from the analysis of CRC risk. For each outcome, follow-up time was calculated from date of the last WebQ participants completed to date of the incidence of the outcomes, loss to follow-up, death, or the end of follow-up (defined as the most recent update of health data: October 2022 for participants in England, August 2022 for Scotland, and May 2022 for Wales).

For participants in the ONE-IBD study, the short-to-medium-term prognosis over six months was defined by readmission and clinical recurrence, with recurrence indicated by a Crohn’s Disease Activity Index (CDAI) > 150 for CD, or a Simple Clinical Colitis Activity Index (SCCAI) ≥ 2 or Mayo Clinic Score (Mayo) ≥ 2 for UC. These events were recorded using data from the electronic health record system and face-to-face interviews by doctors. For recurrence outcome, we excluded whom had experienced the outcome case at baseline respectively. The follow-up time was calculated in the same methods to UKB cohort.

Algorithm of dietary carbon footprint

We used life cycle assessment-derived dietary carbon footprint metrics to evaluate the environmental impact of the diet. This metrics calculates carbon dioxide equivalents in the entire procedure of food production (e.g., processing, transportation, cooking)⁴². Daily dietary carbon footprint (g/day) per individual was calculated by multiplying food-specific emission metrics (kg CO₂e/kg of food) by individuals’ daily intake quantities (grams), based on the carbon accounting methodology established by Hallström et al.⁴² and implemented in the RISE Food and Climate Database⁴³ for participants from UKB. For participants from ONE-IBD, we used the Chinese Food Life Cycle Assessment Database (CFLCAD) to evaluate the dietary carbon footprint in same calculating methods⁴⁴.

IBD-related protein data

The protein data were obtained from the UK Biobank Pharma Proteomics Project (UKB-PPP), which characterized the plasma proteomic profiles of UKB participants⁴⁵. To investigate the potential molecular mechanisms by which PHD influences the prognosis of IBD, we selected 17 proteins that have been reported to have a close relationship with IBD. Details of the protein function are provided in the Table S25.

Covariates

We chose covariates associated with both diet and IBD based on prior knowledge^46,47, including sex, body mass index (BMI), age, education level, family income, Townsend deprivation index (TDI), smoking and drinking status (never or ever), alcohol intake, physical activity, comorbidities by the Charlson comorbidity index (CCI), and baseline medication use. Dietary factors included total energy, fat, carbohydrate, protein, and added sugars intake. Details regarding covariate processing were presented in Table S26 and Table S27.

Statistical analysis

Based on previous studies showing an inverse association between PHD adherence and IBD risk via Knuppel’s algorithm^12,13, baseline characteristics were stratified by tertiles of Knuppel’s score⁴¹. Continuous variables were shown as means (standard deviation, SD), while categorical variables were shown as numbers (percentages).

In the primary analysis, after passing the proportional hazards (PH) assumption test (P > 0.05), we applied Cox proportional hazards model to estimate the hazard ratio (HR) and 95% confidence intervals (CIs) of the association between seven algorithms of PHD and the risk of IBD-related surgery in participants with IBD. Each score was treated by per-SD change or by tertile (the lowest category as the reference group). We constructed three Cox models to perform a stepwise regression: (1) crude model; (2) minimally adjusted model: adjusting for age, sex, and total energy intake; (3) fully adjusted model: adjusting for age, sex, physical activity, TDI, BMI, smoking status, alcohol, total energy intake, carbohydrate intake, protein intake, fat intake and added sugars intake in UKB cohort, while adjusting for age, sex, total energy intake, education level, income level, BMI, drinking status, and smoking status in ONE-IBD cohort. Variance inflation factors indicated minimal multicollinearity among the included variables (maximum VIF = 1.265).

The following were conducted in the secondary analysis: (1) to explore whether adherence to PHD is beneficial to each subtype of IBD, we assessed the association between scores by seven PHD algorithm and IBD-related surgery in participants with CD and UC; (2) to further evaluate the long-term prognostic sustainability of PHD, we assessed the association of PHD adherence scores with new-onset CRC and mortality in participants with IBD; (3) to evaluate which components are beneficial or harmful for individuals with IBD, we assessed the association between component sub-scores of the seven PHD algorithms and IBD-related surgery in participants with IBD, as well as its subtypes; (4) to evaluate the environmental sustainability of PHD, we assessed the association between PHD adherence scores and dietary carbon footprint per participant via Pearson correlation coefficients; (5) to evaluate the dual sustainability of IBD-PHDI on individuals and environment, we assessed the association of score by modified algorithm IBD-PHDI and components with IBD-related surgery and dietary carbon footprint, and compared it with that by Knuppel’s; (6) to further explore the practicability of the PHD in Chinese IBD population, we assessed the association of PHD adherence scores including IBD-PHDI with short-term prognosis and dietary carbon emission; (7) to explore the potential molecular mechanism of how PHD contacting with better prognosis of IBD, we assessed the association between PHD adherence scores and IBD-related protein in serum via Pearson correlation coefficients.

In our subgroup analysis, we assessed the association between seven algorithm of PHD in per SD and the risk of IBD-related surgery, stratified by age ( < 60, ≥ 60 years), sex (male, female), smoking status (never or ever), BMI ( < 30, ≥ 30 kg/m²), physical activity (inadequate or adequate, identified by median value), serum C-reactive protein level (CRP, < 10 mg/L, ≥ 10 mg/L) and history IBD-related surgery (with or without history of the surgery) in UKB cohort. Additionally, we tested multiplicative interactions. To test the robustness of our results, we performed several sensitivity analyses based on the fully adjusted model. We further (1) excluded outcomes that occurred in the first 1- and 2-year follow-up to mitigate the impact of reverse causation; (2) additionally adjusted for use of medication and CCI to reduce the impact of residual confounding.

The P value of less than 0.05 was considered statistically significant. Statistical analyses were performed using R (version 4.3.1).

Author contributions

Lintao Dan (Formal analysis, Methodology, Writing-original draft), Haiting Shu (Formal analysis, Methodology, Writing-original draft), Yuhao Sun (Methodology, Writing-review and editing), Judith Wellens (Methodology, Writing-original draft), Yang Chen (Writing-review and editing), Minzi Deng (Writing-review and editing), Jialu Yao (Writing-review and editing), Xixian Ruan (Formal analysis, Writing-review and editing), Henit Yanai (Methodology, Writing-review and editing), Xue Li (Writing-review and editing), Britta Siegmund (Methodology, Writing-review and editing), Xiaoyan Wang (Methodology, Writing-review and editing), Fernando Magro (Conceptualization, Methodology, Writing-review and editing), Jie Chen (Conceptualization, Methodology, Formal analysis, Writing-original draft; Writing-review and editing). All authors reviewed the manuscript.

Data availability

UK Biobank data can be publicly available from the UK Biobank (www.ukbiobank.ac.uk/). ONE-IBD is an ongoing prospective IBD multi-omics cohort that includes a wide range of phenotypes, such as dietary data. ONE-IBD holding a positive attitude for data sharing and can be accessible with a case-by-case collaboration issues under guidance of local regulations. For inquiries regarding data sharing and collaboration, please contact PI Jie Chen (med_chenjie@csu.edu.cn).

Code availability

No custom code or novel algorithm was developed for this study. All analyses were performed using standard statistical software. The analytic code used in this study will be made available upon reasonable request.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41538-026-00959-5.