Published a year before TRAVERSE reported, the Hudson meta-analysis was the most rigorous synthesis of testosterone cardiovascular safety then available. It identified 35 randomized trials covering 5,601 men, obtained the underlying individual patient records from 17 of them, and found cardiovascular events in 7.5% of testosterone-treated participants versus 7.2% on placebo — odds ratio 1.07, 95% CI 0.81 to 1.42.[1]
Its interpretation was carefully bounded: no evidence of increased short- to medium-term cardiovascular risk, alongside an explicit statement that long-term safety data were lacking.
Why individual patient data changes the argument
A conventional meta-analysis pools what trials published — the summary numbers in each paper. That inherits every choice the original authors made about which outcomes to report and how to categorise them, and it cannot recover anything they left out.
An individual patient data (IPD) meta-analysis asks investigators for the underlying participant records and re-analyses them under one consistent definition. It is slower and depends on cooperation, but it can classify every event the same way across every trial and examine whether an effect differs by patient characteristics.
That distinction is the whole reason this paper carried weight. The 2013–2015 cardiovascular scare rested substantially on a meta-analysis whose event definitions were later contested. Hudson answered with the design that removes exactly that failure mode.
What it did
- Search: MEDLINE, Embase, Science Citation Index, the Cochrane registers and DARE, covering literature from 1992 onwards, searched 27 August 2018. Registered with PROSPERO (CRD42018111005).
- Inclusion: men aged 18+ with screening testosterone at or below 12 nmol/L (350 ng/dL); any testosterone formulation, dose or route for at least three months; placebo comparator.
- Screening: 9,871 citations → 225 full-text reports → 35 primary studies (5,601 participants, mean age 65) reported across 109 peer-reviewed publications.
- IPD obtained: 17 studies (49%), 3,431 participants, from nine countries, mean duration 9.5 months. 18 studies did not supply IPD.
- Analysis: one-stage meta-analysis on the IPD, plus a two-stage analysis integrating the non-IPD studies.
- Risk of bias: judged low in most IPD studies (71%).
What it found
Cardiovascular events: 120 of 1,601 (7.5%) on testosterone versus 110 of 1,519 (7.2%) on placebo. Odds ratio 1.07 (95% CI, 0.81–1.42), P = .62.[1]
Deaths: 6 of 1,621 (0.4%) on testosterone versus 12 of 1,537 (0.8%) on placebo. Odds ratio 0.46 (95% CI, 0.17–1.24), P = .13 — fewer deaths on treatment, but not significantly so.[1]
Breaking the cardiovascular events down, the frequent categories were arrhythmia (52 versus 47), coronary heart disease (33 versus 33), heart failure (22 versus 28) and myocardial infarction (10 versus 16).
The subgroup analyses found no association between cardiovascular risk and patient age, baseline testosterone, or smoking status.
The limitation the authors put in their own conclusion
The mean IPD trial duration was 9.5 months. That is the ceiling on what this analysis could see.
The authors' interpretation says it directly: no evidence that testosterone increased short- to medium-term cardiovascular risk, but a paucity of data evaluating long-term safety, and long-term data are needed.
This is the honest shape of the pre-TRAVERSE evidence base — reassuring as far as it reached, and openly silent past a year. TRAVERSE was designed to fill precisely that gap, running a mean 33 months in a population selected for elevated cardiovascular risk, and reached a compatible conclusion on a much longer horizon.
Where it sits now
Hudson has not been superseded so much as joined. It answered the short-to-medium-term question with the strongest available method; TRAVERSE answered the longer-term question with a single adequately powered trial. Two different designs, two different windows, pointing the same way — which is a stronger position than either alone.
The historical arc that led here, from the 2010 trial stopped early through the meta-analysis behind the FDA warning, is covered in the cardiovascular season of The T Files.
Bottom line
Hudson pooled the raw participant data from 17 trials and the published data from 18 more, and found cardiovascular event rates essentially identical between testosterone and placebo across 5,601 men. Its own stated limit was duration: a mean of 9.5 months in the IPD set, with long-term safety explicitly unresolved. Read together with TRAVERSE, it forms the evidentiary basis for the current position that properly dosed testosterone in hypogonadal men does not raise cardiovascular event rates.
References
1 source
- Hudson J, Cruickshank M, Quinton R, et al. Adverse cardiovascular events and mortality in men during testosterone treatment: an individual patient and aggregate data meta-analysis. Lancet Healthy Longev. 2022;3(6):e381–e393. doi:10.1016/S2666-7568(22)00096-4 · PMID 35711614
Practical questions about membership, labs and using the spray are answered in the Learn Hub FAQ.