Risk assessmentindependent cross-checkfull text checked I-CAN-03 in-depth review
Smoking, alcohol and excess weight lead global avoidable cancer deaths
The global burden of cancer attributable to risk factors, 2010-19: a systematic analysis for the Global Burden of Disease Study 2019.
Overview
GBD 2019 Cancer Risk Factors Collaborators. Lancet 2022 · PMID 35988567 · DOI 10.1016/S0140-6736(22)01438-6
44% of cancer deaths worldwide in 2019 were attributable to known risk factors, smoking first; diet accounted for about 5-6% of cancer burden.
Does the study support the specific claim? Partly · Methodological quality: moderate
Plain-language summary
Study explained simply
- What did the researchers want to find out?
Global attributable cancer burden.
- How was the study done?
GBD modelling.
- How many people, animals or samples were studied?
204 countries.
- What was actually found?
4.45 million cancer deaths (44.4%) attributable to modelled risks in 2019; tobacco 33.9% of male and 10.7% of female cancer DALYs; dietary risks 5.9% and 5.1%.
- How large was the observed effect?
4.45 million deaths.
- What do the results mean in everyday life?
Avoid tobacco and alcohol, keep healthy weight.
- What are the study's limitations?
Model-based estimates; sparse data in many countries; infections and UV not included.
- What does it say about Elawady's specific claim?
Diet is one factor among several.
- Do the results apply to Germany and Europe?
High.
- What can explicitly not be concluded from this study?
That the remaining 56% of cancer deaths have no external cause (infections and UV were not modelled); individual risk for a given person.
Scientific in-depth review
1. Full reference
- Authors
- GBD 2019 Cancer Risk Factors Collaborators
- Journal
- Lancet
- Year
- 2022
- Identifier
- PMID 35988567 DOI 10.1016/S0140-6736(22)01438-6
- Verification
- verified – PubMed esummary + efetch 2026-10-25
- Retracted
- no
- Corrections
- none
- Extent checked
- full text
- Conflicts of interest and funding
Funding: Bill & Melinda Gates Foundation; funder had no role in design, data collection, analysis, interpretation or writing. Many individual author disclosures (e.g. consulting fees, honoraria, stock options), all stated as outside the submitted work.
2. Research question
Global cancer burden attributable to risk factors.
3. Study design and population
- Study design
- Risk assessment – Comparative risk assessment within the Global Burden of Disease Study 2019 (modelling study; population attributable fractions from meta-analysed relative risks and modelled exposures)
- Population
Global population; 204 countries and territories, by age, sex, region and Socio-demographic Index (SDI)
- Country
Global
- Period
2010-2019
- Sample size
Global estimates; 82 risk-outcome pairs (23 cancer types, 34 risk factors); 1,000 draws for uncertainty
- Inclusion and exclusion criteria
Risk-outcome pairs with convincing or probable causal evidence by World Cancer Research Fund criteria; ICD-10 chapter 2 neoplasms except Kaposi sarcoma. Not modelled as risk factors: ultraviolet radiation, infectious agents such as Helicobacter pylori, HPV (only 'unsafe sex'), hepatitis B/C.
4. Methods and endpoints
- Exposure / intervention
Behavioural (tobacco, alcohol, dietary risks, unsafe sex, low physical activity etc.), environmental and occupational (air pollution, radon, 13 occupational carcinogens), metabolic (high BMI, high fasting plasma glucose) risks
- Comparator
Theoretical minimum risk exposure level (counterfactual optimal exposure)
- Endpoints
Risk-attributable cancer deaths and DALYs in 2019, and change 2010-2019 (counts and age-standardised rates)
- Statistical methods
Six-step comparative risk assessment: WCRF causal criteria; updated systematic reviews and meta-analyses of relative risks (incl. MR-BRT); exposure modelling (DisMod-MR 2.1, ST-GPR); TMREL; PAF by age/sex/location/year; mediation matrix for joint effects; cancer deaths from CODEm with MIR-transformed registry data; 95% uncertainty intervals from 1,000 draws (GATHER-compliant).
5. Actual results
- Results
2019: 4.45 million (95% UI 4.01-4.94) risk-attributable cancer deaths = 44.4% (41.3-48.4) of cancer deaths; 105 million (95.0-116) DALYs = 42.0% (39.1-45.6). Males 2.88 million deaths (50.6%, 47.8-54.1); females 1.58 million (36.3%, 32.5-41.3). Leading Level 2 risk for DALYs: tobacco, 33.9% (32.3-35.4) of male and 10.7% (9.9-11.5) of female cancer DALYs; then in males alcohol 7.4% (6.7-8.2), dietary risks 5.9% (4.4-8.3), air pollution 4.4% (3.4-5.5); in females unsafe sex 8.2% (7.0-8.8), dietary risks 5.1% (4.0-6.7), high BMI 4.7% (2.8-7.0), high fasting plasma glucose 3.6% (1.0-7.5). Most detailed level: smoking, alcohol, high BMI lead. Lung cancer 36.9% (34.2-39.3) of attributable deaths. 2010-2019: attributable deaths +20.4% (12.6-28.4), DALYs +16.8%; age-standardised attributable death rate -6.9% (0.9-12.8) and DALY rate -7.8% (1.4-14.0); metabolic risks rose most (+34.7% deaths). Alcohol 4.9% and high BMI 4.6% of cancer deaths.
- Absolute risks
4.45 million risk-attributable cancer deaths and 105 million DALYs in 2019; leading regional age-standardised attributable death rates e.g. central Europe 82.0 (71.0-94.9) and western Europe 63.8 (58.4-69.7) per 100,000 person-years
- Authors' conclusion
Leading risk factors were behavioural, while metabolic risks increased most 2010-2019; reducing exposure would decrease cancer mortality and DALYs, with context-specific policies. A large proportion of cancer burden might not be avoidable through the estimated risk factors; cancers remain fundamentally linked to genetics and ageing; individuals with cancer should not be blamed; prevention must be coupled with early diagnosis, screening and effective treatment.
6. Strengths
- Comprehensive, standardised framework across 204 countries and 82 risk-outcome pairs
- Causal criteria (WCRF) and updated meta-analyses for relative risks
- Uncertainty propagated through all estimation steps
- Mediation adjustment to avoid double counting of overlapping risks
- Public data and GATHER compliance
7. Weaknesses
- Model-based; sparse registry and exposure data in many low-SDI countries, with greater uncertainty (authors)
- Residual exposure measurement bias likely despite adjustment; exposure model performance not formally assessed (authors)
- Infections (H. pylori, HPV, hepatitis B/C) and UV radiation not included as risk factors, so the attributable share is underestimated (about 8.9% of cases attributable to infections per Global Cancer Observatory)
- PAFs assume causal relative risks and transferable risk functions; TMREL choices influence estimates
- Pre-COVID estimates; reliance on historical registry data
8. Risk of bias
- Orienting tool
- none applicable (modelling/comparative risk assessment; no standard RoB tool fits)
- 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
Structured judgement of main validity threats for a modelling study; overall unchanged. Single-reviewer structured judgement from the full text; a formal assessment requires two independent reviewers and a protocol.
| Domain | Judgement | Rationale |
|---|---|---|
| Causal basis of risk-outcome pairs | low | Only pairs with convincing/probable evidence (WCRF criteria) with updated systematic reviews. |
| Input data (cancer burden) | some concerns | Registry, vital registration and verbal autopsy data of variable quality; MIR transformation where mortality data are absent; lagged data. |
| Exposure estimation | some concerns | Sparse exposure data in many settings; crosswalk adjustments; residual measurement bias likely (authors). |
| Model assumptions | some concerns | PAF depends on TMREL, risk curves and mediation matrix; relative risks largely from observational cohorts in high-income settings. |
| Completeness of risk factors | some concerns | Infections and UV excluded; attributable fraction underestimates preventable burden. |
| Reporting | low | Uncertainty intervals for all estimates; full appendix and public data tools. |
Full-text review (25 October 2026): Added risk-specific shares (tobacco 33.9%/10.7%, dietary risks 5.9%/5.1%, alcohol, unsafe sex, high BMI, FPG), DALY figures, 2010-2019 changes in counts vs age-standardised rates, methods detail, excluded risk factors (infections, UV, H. pylori, HPV, hepatitis B/C), authors' statements on genetics/ageing and non-blame; full funding/COI. Bias moved from a bare 'none applicable' to a structured domain judgement (overall unchanged: some concerns).
9. Assessment of the statistical interpretation
About 56% of cancer deaths are not attributed to the modelled risks; this residual includes ageing, genetics and chance but also unmodelled preventable causes (infections, UV). Dietary risks account for a modest share (5.9% of male and 5.1% of female cancer DALYs) compared with tobacco (33.9% and 10.7%). Rising absolute counts with falling age-standardised rates reflect population growth and ageing.
10. Relevance to the specific claim
Shows smoking first, diet-related factors as part; over half of cancer deaths not explained by modifiable risks.
11. Independent contrary findings
12. Transferability to Germany and Europe
High (central and western Europe have among the highest age-standardised attributable death rates).
13. Overall methodological judgement
Authoritative context on modifiable cancer risks: tobacco dominates; diet contributes a modest share; a large proportion of cancer is not attributable to known modifiable risks.
| Multidimensional assessment | |
|---|---|
| Authenticity of the source | verified |
| Methodological quality | moderate |
| Risk of bias | some concerns |
| Precision of results | moderate |
| Directness | direct |
| Reproducibility | replicated |
| Consistency with the overall evidence | consistent |
| Regional transferability | high |
| Adequacy of the interpretation (by the archive) | not applicable |
14. Open questions
Burden attributable to infections and UV within the same framework; validation of exposure models in low-SDI settings; post-COVID trends.
15. Linked fact checks
- Cancer is the body's own plan ('a self plan'): body cells are ordered to transform because this was necessary for the person's life; it is not caused by external agents. – Studies contradicting or limiting the claim
- Smoking (and similar exposures) should not be blamed for cancer; only massive exposures such as Hiroshima cause cancer, and there is 'no such thing as early death' from smoking. – Studies contradicting or limiting the claim
- Third-party site: the Tayebat system can serve as a preventive lifestyle against cancer (less processed meat, refined sugar, fried food, soft drinks), not as treatment; cancer patients must follow their oncology team and keep screening. – Studies supporting the claim
- Cross-system summary: avoiding the system's forbidden foods (industrial eggs, chicken, dairy, white flour, UPF, soft drinks, leafy greens, legumes, acid blockers, supplements) prevents cancer. – Studies contradicting or limiting the claim
- Food is the first (leading) cause of modern (non-communicable) diseases. – Studies contradicting or limiting the claim
- Cigarettes are beneficial or at least not harmful: asked whether cigarettes are beneficial he answered 'Yes, beneficial'; elsewhere he said smoking has nothing to do with atherosclerosis and is 'not harmful at all' organically. – Studies contradicting or limiting the claim