Cohort studyindependent cross-checkfull text checked★ key study I-GRN-04 in-depth review

Potatoes, fries and diabetes

Total and specific potato intake and risk of type 2 diabetes: results from three US cohort studies and a substitution meta-analysis of prospective cohorts.

Overview

Mousavi SM et al.. BMJ 2025 · PMID 40769531 · DOI 10.1136/bmj-2024-082121

Eating fries 3 more times a week was linked to 20% higher diabetes risk; boiled, baked or mashed potatoes and potato/corn chips showed no clear link. Swapping potatoes for whole grains was linked to lower risk, swapping for white rice to higher risk.

Does the study support the specific claim? No · Methodological quality: high

Original publication

Plain-language summary

Study explained simply

What did the researchers want to find out?: Potatoes and diabetes
How was the study done?: Cohorts + meta-analysis
How many people, animals or samples were studied?: 205,107
What was actually found?: Fries linked, boiled not
How large was the observed effect?: HR 1.20
What do the results mean in everyday life?: Prefer boiled potatoes; limit fries
What are the study's limitations?: Observational; US health professionals; boiled/baked/mashed and potato/corn chips each measured together
What does it say about Elawady's specific claim?: Limits the allowance of fried potatoes; does not show harm from boiled potatoes; chips data inconclusive (combined with corn chips)
Do the results apply to Germany and Europe?: Moderate
What can explicitly not be concluded from this study?: Cause

Scientific in-depth review

1. Full reference

Authors: Mousavi SM et al.
Journal: BMJ
Year: 2025
Identifier: PMID 40769531 DOI 10.1136/bmj-2024-082121
Verification: verified – PubMed efetch 2026-10-25
Retracted: no
Corrections: none
Extent checked: full text
Conflicts of interest and funding: NIH grants (UM1 CA186107, U01 CA176726, U01 CA167552) and Friends of FACES/Kids Connect; MRC and NIHR support for UK co-authors; Kuwait Heart Foundation for one author; funders had no role. ICMJE: no support from any organisation for the submitted work, no relevant financial relationships.

2. Research question

Potato types and T2D; substitutions

3. Study design and population

Study design: Cohort study – Three prospective cohorts (NHS, NHS II, HPFS; two-stage individual participant data analysis with repeated diet measures) plus PROSPERO-registered (CRD42023448736) dose-response and substitution meta-analysis of 13 prospective cohorts
Population: US female nurses (NHS n=72,712; NHS II n=90,232) and male health professionals (HPFS n=42,163), free of diabetes, CVD and cancer at baseline; >90% of European ancestry. Meta-analysis: 13 cohorts (5 US, 4 Europe, 2 Asia, 2 Australia), 587,081 participants
Country: USA (cohorts); meta-analysis adds Europe, Asia, Australia
Period: NHS 1984-2020, NHS II 1991-2021, HPFS 1986-2018; meta-analysis search to July 2024
Sample size: 205,107; 22,299 cases; meta 587,081 / 43,471 cases
Inclusion and exclusion criteria: Cohorts: completed baseline FFQ, no prior cancer, MI, angina, stroke, CABG or T2D; excluded incomplete age/potato data, implausible energy (<500/>3500 kcal women, <800/>4200 kcal men), death at/before baseline, baseline-only responders. Meta-analysis: prospective cohorts of potato intake and T2D in populations free of CVD, cancer and T2D; PubMed/Medline, Web of Science, Embase, no restrictions.

4. Methods and endpoints

Exposure / intervention: Cumulative average intake (FFQ every 2-4 years) of total potatoes (baked/boiled/mashed + French fries), baked/boiled/mashed potatoes, French fries, and potato or corn chips (one combined item, analysed separately)
Comparator: <1 serving/week (total) or almost never (specific types); per 3 servings/week; modelled substitution with whole grains, refined grains, white/brown rice, legumes, starchy and non-starchy vegetables
Endpoints: Incident T2D confirmed by supplementary questionnaire using ADA criteria (NDDG criteria before 1998)
Statistical methods: Cohort-specific Cox models stratified by age and calendar time, 3 nested models (energy; + lifestyle, BMI, SES, medications; + food groups and mutual potato adjustment), pooled by fixed-effect IPD meta-analysis (random effects secondary); restricted cubic splines; latency, reverse-causation, surveillance-bias, mediation (BMI) and multiple dietary-modelling sensitivity analyses; serving-based substitution models. Meta-analysis: inverse-variance linear and one-stage non-linear dose-response, ROBINS-E and NOS, Egger test, GRADE and NutriGrade.

5. Actual results

Results: Cohorts (fully adjusted, per 3 servings/week): total potato HR 1.05 (1.02-1.08); >=7 vs <1 serving/week 1.12 (1.02-1.24); French fries 1.20 (1.12-1.28), >=5/week vs almost never 1.27 (1.08-1.49); baked/boiled/mashed 1.01 (0.98-1.05), >=5/week 0.99 (0.93-1.05); potato or corn chips 1.02 (0.98-1.06), >=5/week 0.97 (0.89-1.06). BMI mediated about 50% of the French fries association (HR without BMI 1.60, 1.36-1.88). Substitution of 3 servings/week with whole grains: total potato -8% (5-11%), baked/boiled/mashed -4% (1-8%), French fries -19% (14-25%); replacing total or baked/boiled/mashed potatoes with white rice associated with higher T2D risk; brown rice substitution showed no inverse association. Meta-analysis per 3 servings/week: total potato 1.03 (1.02-1.05; I2 57.3%, n=11); fried 1.16 (1.09-1.23; I2 50.9%, n=9); non-fried 1.01 (0.99-1.03; n=6); boiled 1.01 (0.99-1.03); mashed 1.06 (1.02-1.09; n=2). GRADE: fried high, total moderate, non-fried/mashed low, boiled very low.
Absolute risks: 22,299 T2D cases over 5,175,501 person-years. Crude counts: total potato <1 serving/week 788 cases/270,016 person-years vs >=7/week 1,176/245,130; French fries almost never 2,952/985,067 vs >=5/week 180/26,919. No adjusted absolute risk differences reported.
Authors' conclusion: Higher intake of French fries, but not baked, boiled or mashed potatoes, was associated with higher T2D risk; potato-related risk depends on the replacement food: whole grains lower, white rice higher.

6. Strengths

Very large cohorts with repeated diet assessments over about 30 years and high follow-up
Extensive sensitivity analyses (latency, reverse causation, surveillance bias, baseline-only vs cumulative diet, polygenic risk score)
Substitution analyses addressing the replacement food
Pre-registered meta-analysis with ROBINS-E, GRADE and NutriGrade; analysis code public

7. Weaknesses

Observational: residual confounding (high potato/fries consumers had lower diet quality and activity)
Health professionals, >90% European ancestry; effect modification by race and BMI
FFQ groups baked, boiled and mashed together and potato with corn chips; frying fat changed over time (tallow, partially hydrogenated oils)
Substitution estimates are statistical models, not observed dietary changes; rice intake low in these cohorts
Meta-analysis: 38% of included studies at high risk of bias (ROBINS-E), moderate heterogeneity, regional differences
Weak associations for non-fried potatoes graded low/very low certainty

8. Risk of bias

Orienting tool: ROBINS-E (cohort component); meta-analysis component assessed by the authors with ROBINS-E/GRADE
Type of assessment: structured assessment based on the full text by a single reviewer (no formal double review)
Full formal assessment: no
Overall risk of bias: some concerns
Rationale: Overall unchanged; residual confounding is the main concern. Simplified single-reviewer judgement from the full text; a formal ROBINS-E assessment would need two independent reviewers.

8. Risk of bias
Domain	Judgement	Rationale
Confounding	some concerns	Adjusted for lifestyle, SES, BMI, medications and many food groups; residual confounding by diet quality possible (AHEI 31 vs 42 between extreme potato groups); results robust to AHEI adjustment.
Measurement of exposure	low	Validated FFQ (de-attenuated r 0.61-0.63 vs diet records) repeated every 2-4 years, cumulative averages; portion sizes updated.
Selection of participants	low	Prospective enrolment, clear exclusions unrelated to outcome.
Post-exposure interventions	low	Diet updates stopped after cardiac events in sensitivity analysis; latency analyses address diet change near diagnosis.
Missing data	low	Low missingness (~10% for dietary variables), last observation carried forward, missing indicators; previously validated against multiple imputation.
Measurement of outcome	low	Self-report confirmed by validated supplementary questionnaire with ADA criteria; symptomatic-case sensitivity analysis.
Selection of reported result	low	All potato types and models reported, including null results for boiled potatoes and chips; meta-analysis pre-registered.

Full-text review (25 October 2026): Added category-specific HRs, potato/corn chips null result (1.02, 0.98-1.06), BMI mediation (50% for fries), brown-rice substitution, crude case/person-year counts, meta-analysis heterogeneity, ROBINS-E of included studies and GRADE per potato type, PROSPERO registration, funding details. Bias overall unchanged (some_concerns) with full domains.

9. Assessment of the statistical interpretation

Modest relative associations: French fries HR 1.20 per 3 servings/week (half mediated by BMI); total potato 1.05; boiled/baked/mashed and chips null. Substitution estimates are modelled contrasts. Meta-analysis confirms fried potato association (high GRADE) and null non-fried association.

10. Relevance to the specific claim

Contradicts unrestricted fried potatoes (C-GRN-07); also shows white rice substitution worse (C-GRN-05).

11. Independent contrary findings

12. Transferability to Germany and Europe

Moderate.

13. Overall methodological judgement

High-quality cohort analysis plus pre-registered meta-analysis: French fries are associated with higher T2D risk, non-fried potatoes are not; replacing potatoes with whole grains is associated with lower risk, replacing them with white rice with higher risk.

Multidimensional assessment
Multidimensional assessment
Authenticity of the source	verified
Methodological quality	high
Risk of bias	some concerns
Precision of results	high
Directness	direct
Reproducibility	replicated
Consistency with the overall evidence	consistent
Regional transferability	moderate
Adequacy of the interpretation (by the archive)	not applicable

14. Open questions

Generalisability to Egyptian frying practices and oils; potato chips/crisps analysed separately from corn chips; populations with high rice intake.

15. Linked fact checks

Potatoes in all forms are allowed, including fried potatoes and packaged potato chips (crisps). – Studies with mixed or unclear results
Rice in all forms (including white rice) is allowed without restriction; 'rice is sugar' and two spoons of rice a day sustain life. – Studies contradicting or limiting the claim