Meta-analysisindependent cross-checkfull text checked★ key study I-INTB-21 in-depth review
Even one cigarette a day harms the heart
Low cigarette consumption and risk of coronary heart disease and stroke: meta-analysis of 141 cohort studies in 55 study reports.
Overview
Hackshaw A et al.. BMJ 2018 · PMID 29367388 · DOI 10.1136/bmj.j5855
One cigarette a day carries about half the heart-disease risk of 20 a day.
Does the study support the specific claim? No · Methodological quality: moderate
Plain-language summary
Study explained simply
- What did the researchers want to find out?
Quantify light-smoking risk
- How was the study done?
Meta-analysis
- How many people, animals or samples were studied?
141 cohort studies in 55 reports (millions of participants)
- What was actually found?
Large excess risk even at 1/day
- How large was the observed effect?
CHD RR 1.48-2.19 at 1/day
- What do the results mean in everyday life?
There is no safe amount of smoking for heart and brain blood vessels.
- What are the study's limitations?
Observational; one database, English only; no protocol; modelled from summary data
- What does it say about Elawady's specific claim?
Contradicts 'cigarettes are beneficial/not harmful'.
- Do the results apply to Germany and Europe?
High
Scientific in-depth review
1. Full reference
- Authors
- Hackshaw A et al.
- Journal
- BMJ
- Year
- 2018
- Identifier
- PMID 29367388 DOI 10.1136/bmj.j5855
- Verification
- verified – PubMed efetch 2026-10-25
- Retracted
- no
- Corrections
Errata published (PubMed ErratumIn PMIDs 29643098, 30487298)
- Extent checked
- full text
- Conflicts of interest and funding
ICMJE: no support from any organisation for the submitted work; no relevant financial relationships (full text).
2. Research question
What is the risk of coronary heart disease and stroke from light smoking?
3. Study design and population
- Study design
- Meta-analysis – Systematic review and meta-analysis of prospective cohorts
- Population
Generally healthy adults in prospective cohorts (high-risk-only cohorts excluded); for CHD in men about 3.07 million participants and >75,000 cases; stroke in men about 3.53 million and >73,000 cases; CHD in women about 2.56 million and >36,000 cases; stroke in women about 3.78 million and 62,000 cases
- Country
Multinational
- Period
Medline, English-language articles 1946 to May 2015
- Sample size
55 publications containing 141 cohort studies (e.g. 26 reports for CHD in men, 18 for CHD in women, 17 for stroke in men, 10 for stroke in women in within-study analyses)
- Inclusion and exclusion criteria
Prospective cohorts with >=50 CVD events, HRs/RRs vs never smokers adjusted at least for age (or age-specific incidence), >=3 smoking categories, generally healthy populations, sex-specific or age- and sex-adjusted results. Six studies showing decreasing risk with increasing consumption were excluded (justification in supplement).
4. Methods and endpoints
- Exposure / intervention
1, 5 or 20 cigarettes/day
- Comparator
Never smokers
- Endpoints
Coronary heart disease, stroke
- Statistical methods
MOOSE; per-study log-linear variance-weighted regression of log RR on cigarettes/day (category midpoints adjusted for carboxyhaemoglobin/cotinine), predicting RR at 1, 5 and 20 cigarettes/day; excess-RR ratio (RR1-1)/(RR20-1) vs expected ~5%; random-effects meta-regression and random-effects meta-analysis (RevMan); subgroup by multiple-confounder adjustment, age, sex; Newcastle-Ottawa for one subset; sensitivity for follow-up to 1995 and possible former smokers in reference group.
5. Actual results
- Results
CHD men: RR 1.48 (1.30-1.69) at 1 cigarette/day, 1.58 (1.39-1.80) at 5/day, 2.04 (1.86-2.24) at 20/day; 1 cigarette/day carried 46% (IQR 24-56%) of the excess risk of 20/day; multiply adjusted studies 1.74 vs 2.27 (53%). CHD women: 1.57 and 2.84 (31%); multiply adjusted 2.19 and 3.95 (38%). Stroke men: 1.25 (1.13-1.38) and 1.64 (1.48-1.82) (41%); adjusted 1.30 and 1.56 (64%). Stroke women: 1.31 (1.13-1.52) and 2.16 (1.69-2.75) (34%); adjusted 1.46 and 2.42 (36%). 'Good quality' (NOS) CHD studies in men: RR 1.62 (1.45-1.82) at 1/day. Expected under linearity: ~5%.
- Absolute risks
not reported in full text (relative risks only; authors cite that 15-20% of CVD events may be attributable to smoking)
- Authors' conclusion
Smoking only about one cigarette per day carries around half the excess risk of CHD and stroke of smoking 20 per day; no safe level of smoking exists for cardiovascular disease; smokers should quit rather than cut down.
6. Strengths
- Very large evidence base (141 cohorts, millions of participants)
- Prospective designs only; three complementary analytic approaches with consistent results
- Sensitivity analyses for confounding adjustment, reference-group contamination, follow-up period and study quality
- Consistent with external dose-response model (CPS-II) and second-hand smoke data
7. Weaknesses
- Single database (Medline), English only; no registered protocol
- Aggregate data only; log-linear modelling of category midpoints; within-study correlation not fully modelled
- Exclusion of six studies with inverse dose trends could bias towards steeper low-dose risk
- Study quality formally assessed only for one subset (NOS)
- Observational: residual confounding, smoking misclassification (authors argue bias would mainly attenuate); baseline smoking only
- No absolute risks
8. Risk of bias
- Orienting tool
- AMSTAR 2
- 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
- high
- Rationale
Raised from some_concerns (ROBINS-E label replaced by AMSTAR 2, the appropriate tool for a meta-analysis): strict AMSTAR 2 confidence is critically low (no protocol, Medline-only search, no systematic RoB assessment). The core finding (non-linear, high excess CVD risk at low consumption) is nevertheless robust across methods, consistent with an established causal relation, CPS-II modelling and second-hand smoke evidence. Simplified single-reviewer judgement from the full text; a formal AMSTAR 2/ROBIS assessment needs two independent reviewers.
| Domain | Judgement | Rationale |
|---|---|---|
| Item 2 (critical) protocol | high | No protocol or registration reported. |
| Item 4 (critical) comprehensive search | high | Medline only, English-language only; reference lists checked. |
| Items 5-6 duplicate selection and extraction | some concerns | Abstracts reviewed by two authors; independence of extraction not described. |
| Item 7 (critical) excluded studies | some concerns | Flowchart and justification for six excluded studies in supplement. |
| Item 9 (critical) RoB of individual studies | high | No systematic RoB assessment; NOS applied only to 26 CHD-in-men studies post hoc. |
| Item 11 (critical) meta-analytic methods | some concerns | Pragmatic log-linear modelling of summary data; three approaches consistent; within-study correlation not modelled. |
| Item 13 (critical) RoB in interpretation | low | Confounding, misclassification and reference-group contamination discussed with sensitivity analyses. |
| Item 14 heterogeneity | low | Explored by adjustment level, age and sex. |
| Item 15 (critical) publication bias | some concerns | No formal small-study test; minimum 50 events used to limit small-study bias. |
| Item 16 review COI | low | No funding or conflicts declared. |
Full-text review (25 October 2026): Added inclusion criteria, participant/case totals, CIs for main RRs, excess-risk proportions with IQRs, adjusted-study and NOS subgroup results, modelling method and author-stated limitations, COI. Tool corrected from ROBINS-E to AMSTAR 2; overall bias raised from some_concerns to high (critically low review confidence), with note that the conclusion is robust.
9. Assessment of the statistical interpretation
Relative risks vs never smokers; about 40-50% of the excess CHD/stroke risk of 20 cigarettes/day is already present at 1 cigarette/day, far above the ~5% expected under linearity. Estimates are higher in studies adjusted for multiple confounders.
10. Relevance to the specific claim
Contradicts the claim that smoking is harmless or unrelated to atherosclerosis.
11. Independent contrary findings
12. Transferability to Germany and Europe
High.
13. Overall methodological judgement
Methodologically limited review conduct, but the main conclusion is robust and consistent with the wider causal evidence: even light smoking substantially raises cardiovascular risk; there is no safe level.
| Multidimensional assessment | |
|---|---|
| Authenticity of the source | verified |
| Methodological quality | moderate |
| Risk of bias | high |
| Precision of results | high |
| Directness | direct |
| Reproducibility | replicated |
| Consistency with the overall evidence | consistent |
| Regional transferability | high |
| Adequacy of the interpretation (by the archive) | not applicable |