Lack of an association or an inverse association between low-density-lipoprotein cholesterol and mortality in the elderly: a systematic review
Uffe Ravnskov, David M Diamond, Rokura Hama, Tomohito Hamazaki, Bjorn Hammarskjold, Niamh Hynes, Malcolm Kendrick, Peter H Langsjoen, Aseem Malhotra, Luca Mascitelli, Kilmer S McCully, Yoichi Ogushi, Harumi Okuyama, Paul J Rosch, Tore Schersten, Sherif Sultan, Ralf Sundberg - BMJ Open, 2016
This systematic review reports that, in people over 60, higher LDL cholesterol was not associated with higher mortality and was usually inversely associated with all-cause mortality, with no cohort showing the harm predicted by the cholesterol hypothesis. It is, however, a contested, skeptic-authored review: several authors (including Ravnskov, Kendrick, and Malhotra) have written books criticising the cholesterol hypothesis, it was funded by the Western Vascular Institute, and it relied on crude vote-counting with no pooled effect estimate, drawing heavy post-publication criticism (CEBM Oxford, Science Media Centre).
Key findings
- In a vote-counting systematic review of 19 cohort studies (30 cohorts, 68,094 community-dwelling people aged 60+), high LDL cholesterol was inversely associated with all-cause mortality in most of the elderly.
- An inverse association between all-cause mortality and LDL-C appeared in 16 cohorts (statistically significant in 14), representing 92% of the participants in whom the association was recorded; in the remainder, no association was found.
- For cardiovascular mortality, 2 of 9 cohorts showed mortality highest in the lowest LDL-C quartile (statistically significant) and 7 showed no association.
- No cohort showed the higher mortality with higher LDL-C predicted by the cholesterol hypothesis; the authors argue this questions LDL-lowering guidelines in the elderly. This is a contested, skeptic-authored review using crude vote-counting with no pooled effect estimate.
Effect measures
- Other: Inverse association with all-cause mortality in 16/30 cohorts (14 statistically significant), representing 92% of participants in whom the association was recorded; no association in the rest; no cohort showed a positive (harmful) association
- Other: CV mortality highest in the LOWEST LDL-C quartile (statistically significant) in 2/9 cohorts; no association in 7/9 cohorts
Why this evidence tier (Low)
- Risk of bias:
- Crude vote-counting of cohort studies with no pooled meta-analytic effect estimate, no formal risk-of-bias appraisal of the included cohorts, and selection/inclusion choices that were heavily criticised. Residual confounding and reverse causation (serious illness lowers LDL and raises mortality) are not addressed.
- Precision:
- No pooled effect size and no confidence intervals are reported; the synthesis counts the direction of associations rather than quantifying them.
- Directness:
- The endpoints (all-cause and cardiovascular mortality in the elderly) are direct and important, but the observational design, unadjusted for reverse causation, limits causal directness.
- Consistency:
- The conclusion conflicts with the randomized-trial and Mendelian-randomization evidence that LDL is causal for atherosclerotic disease; it is an outlier, contested finding.
- Funding / COI:
- This is a SKEPTIC-authored review: several authors (including Ravnskov, Kendrick, and Malhotra) have written books criticising the cholesterol hypothesis, and the study was supported by a grant from the Western Vascular Institute. It was heavily criticised after publication (e.g. CEBM Oxford post-publication peer review, the UK Science Media Centre) for its crude vote-counting and absence of pooled estimates. It has not been retracted and carries no erratum.
Low certainty. The review honestly reports that no included cohort showed higher mortality with higher LDL-C in the elderly, but it is a contested, skeptic-authored, Western Vascular Institute-funded vote-counting review with no pooled effect estimate, and it was widely criticised on methodological grounds.
- Population:
- Community-dwelling individuals aged >=60 years from the general population; 19 cohort studies comprising 30 cohorts with 68,094 elderly people (all-cause mortality recorded in 28 cohorts, cardiovascular mortality in 9 cohorts).
- Conflicts of interest:
- SKEPTIC-authored review. Competing interests (verbatim): "TH has received speaker fees from Nissui Pharmaceutical and Nippon Suisan Kaisha. KSM has a US patent for a homocysteine-lowering protocol. RH, HO, RS and UR have written books with criticism of the cholesterol hypothesis." Several lead authors (Ravnskov, Kendrick, Malhotra, and others) are prominent cholesterol-hypothesis critics. The study was supported by a grant from the Western Vascular Institute. No retraction or erratum is attached, but the review was substantively criticised post-publication (CEBM Oxford, UK Science Media Centre) for its vote-counting methodology.
- Funding:
- Supported by a grant from the Western Vascular Institute.
Limitations
- Crude vote-counting design with no pooled meta-analytic effect estimate and no confidence intervals.
- Does not address reverse causation - serious illness lowers LDL cholesterol and raises mortality - which can manufacture an apparent inverse association.
- SKEPTIC-authored review: several authors (including Ravnskov, Kendrick, and Malhotra) wrote books criticising the cholesterol hypothesis, and it was funded by the Western Vascular Institute.
- Heavily criticised after publication (e.g. CEBM Oxford, UK Science Media Centre) for its methodology, though it has not been retracted.
- Conclusions conflict with the randomized-trial and Mendelian-randomization evidence that LDL is causal.