Professor Rory Collins declines to release raw datasets for statin trials
Executive summary
Professor Rory Collins and the Cholesterol Treatment Trialists’ (CTT) collaboration have long been central custodians of individual patient data from many major statin trials, but independent calls for release of the full raw datasets have repeatedly met resistance or have not resulted in open access; Collins and colleagues say they have pooled and analysed patient-level data for specified endpoints and have flagged practical barriers to immediate public release [1] [2]. The BMJ, Lancet commentators, and outside critics have all urged independent scrutiny of adverse‑event data and fuller transparency, while Collins’ group has defended the reliability of randomized trial evidence and published large-scale meta-analyses based on the datasets to which they have access [3] [4] [5].
1. The custody of the trials: who actually holds the patient‑level data
Collins and the CTT collaboration are the only investigators known to have seen full participant‑level data for some—but not all—major statin trials, because trial sponsors, trialists and CTT pooled data under agreements that limited wider sharing, a fact acknowledged by Collins’ team in correspondence with The BMJ [1] [2]. The CTT analyses cited by Lancet and other major reviews used these pooled individual‑level data for cause‑specific mortality, major vascular events and selected cancers, rather than an unfiltered dump of every collected variable across trials [2] [5].
2. The BMJ’s demand for independent scrutiny and Collins’ response
The BMJ launched a campaign calling for anonymised individual patient data to be made available for independent analysis after high‑profile critiques in 2013 questioned trial reporting of harms and benefits, and an expert panel explicitly recommended wider access to raw datasets [3] [2]. Collins pushed back vigorously—lodging complaints to the Committee on Publication Ethics and demanding corrections or retractions of papers he considered misleading—and maintained the primacy of blinded randomized evidence while arguing that some criticisms misread or misrepresented trial data [3] [6] [7].
3. What Collins’ group has offered instead of raw datasets
Collins and colleagues acknowledge that, to date, their meta‑analyses have been restricted to specific patient‑level outcomes (mortality, major vascular events, site‑specific cancers) and that assembling harmonised adverse‑event datasets across trials is “non‑trivial”; they indicated plans to compile and make tabulated adverse‑event data available once analyses are complete rather than release all raw source files immediately [2]. Critics call this insufficient, arguing that narrative reviews and meta‑analyses confined to selected endpoints cannot settle questions about minor but common side effects or trial run‑in exclusions [4] [6].
4. The core of the disagreement: harms, run‑in periods and representativeness
A central point of contention is whether the randomized trials under‑report symptomatic side‑effects and whether trial designs (including run‑in periods that excluded statin‑intolerant individuals) biased harm estimates downward; BMJ commentators and external authors have pressed for access to the raw adverse‑event data to study these issues, while Collins’ team acknowledges heterogeneity and complexity in harms reporting and has resisted immediate blanket data release [8] [2] [6]. Lancet and other defenders of Collins warn that the randomized evidence still offers the most reliable estimate of net benefit, and that misreporting could harm public health if it deters patients from effective therapy [9] [10].
5. Motives, conflicts and the politics of data access
Observers on both sides point to potential motives: critics stress that many contributing trials were industry‑funded and that trial sponsors and trial leaders—including CTT members—have gatekeeping power over data [6] [11], while Collins and allies emphasize government and charity funding of meta‑analyses and the risk that selective or flawed secondary analyses could mislead clinicians and patients [6] [10]. The BMJ’s campaign and editorial stance have themselves been accused of institutional bias by some supporters of Collins, illustrating that disagreements over data access sit atop professional reputations, patient safety claims, and editorial incentives [3] [4].
6. What the public record shows — and what it does not
Public documents show Collins’ group has not accepted wholesale public release of every raw dataset, has limited published meta‑analyses to certain endpoints, and has committed to providing harmonised adverse‑event summaries when feasible; independent panels have repeatedly called for broader access, but the practical mechanics and legal/contractual constraints of releasing anonymised raw trial files remain unresolved in the public record [2] [3] [4]. Reporting does not, in the sources provided, yield a single documented instance of Collins voluntarily releasing full raw trial datasets for unrestricted third‑party reanalysis, nor do the sources detail the precise contractual or privacy barriers that would have to be cleared to do so [1] [2].