Two global collaborations use data in very different ways. Both raise the same question for Scotland: are we organised well enough to make the most of what we already have?

By Professor David Cameron

I came to clinical research through numbers. I studied mathematics before medicine and have since spent years designing and leading breast cancer trials, as well as serving on the teams of many others. Randomised trials remain the gold standard because randomisation balances the hidden biases that we cannot see or measure. Yet trials cannot answer every question, not least because many patients are never represented in them.

That limitation drew me towards other forms of evidence and, in particular, two collaborations that could hardly be more different. One seeks to include entire populations; the other considers only patients enrolled in randomised trials. What connects them is a question I find genuinely interesting: what can, and cannot, be done with data held by others? Each also offers a useful perspective on Scotland’s current position and its priorities for the future.

The sick man of Europe

For many years, UK politicians had a ready response to concerns about cancer outcomes: survival was improving with each decade, the figures were historical, and there was therefore no pressing need to act. Until about twenty years ago, that argument arguably contributed to a lack of sustained momentum within the health service to address the cancer problem.

The flaw in the argument was that outcomes elsewhere were also improving, often more quickly. The UK remained near the bottom of international rankings. The International Cancer Benchmarking Partnership (ICBP) was established around fifteen years ago by Mike Richards and colleagues to examine that pattern clearly and fairly, with Michael Coleman among the epidemiologists involved. It is now funded partly by Cancer Research UK (CRUK) and partly by its partner jurisdictions. John Butler, a gynaecological oncologist in London, wrote the foundational paper setting out its purpose. The partnership brings together clinicians, researchers, cancer epidemiologists and cancer registries from around the world. It operates at the level of jurisdictions: geographical territories that organise cancer care in a particular way, usually through state funding. Participation requires high-quality population-based cancer registration, universal access to healthcare, broadly comparable healthcare spending, and a willingness both to fund membership and to share and learn. These conditions help to explain the variation in participation. Not every Australian state is involved, whereas all four UK nations and every Canadian province take part.

Unlike international programmes that simply monitor cancer outcomes, the ICBP investigates why variation occurs. It benchmarks each jurisdiction against its partners, then examines possible explanations across the whole pathway, from public awareness of cancer signs and symptoms to diagnosis, treatment and national cancer strategies. This work has already produced evidence that informs clinical practice and policy across participating countries.

Scotland’s long-established population-based cancer registration and nationally linked routine records make it a quietly strong participant in this work. Many jurisdictions would envy these resources, and participation depends on a willingness to share reliable data. Much of my own non-trial work, through the Scottish Government’s Innovative Healthcare Delivery Programme and the national cancer medicines data groups I chair, has focused on making better use of them within NHS Scotland. We therefore bring a genuine asset to the partnership, although the uncomfortable question is what those data reveal.

Consider survival between 1950 and 2010, with the four UK nations ranked separately alongside sixteen other Western European countries. From 1950 to 2000, Scotland had the lowest survival for lung cancer in men, with a similar pattern among women; only recently have we moved from the bottom of the table. Peaks in incidence can be linked to historical changes in adult smoking, but smoking rates have since fallen sharply. Why, then, do our outcomes remain so poor? Breast cancer presents a slightly less stark picture, although Scotland has generally ranked third or fourth from the bottom. This is hardly an accolade. Some aspect of our system may be failing, but Scotland also occupies a potentially productive position: we have both a recognised problem and the data needed to investigate it.

The working hypothesis is that comparisons of outcomes, treatments, time to treatment, care pathways, routes to diagnosis and workforce may show where poorer-performing jurisdictions can improve. The answer may not be the newest drug or surgical technique, but an established practice that already works well elsewhere. Phase III is now beginning with three research teams. The International Agency for Research on Cancer in Lyon is studying survival, including the striking concept of estimating ‘missed lives saved’: how many more people would be alive if poorer-performing jurisdictions matched the best? Showing that Scotland might save hundreds, perhaps thousands, of lives by matching outcomes in parts of Australia sends a powerful political message. University College London is undertaking detailed analyses of specific cancer pathways, building on earlier findings that some treatments begin particularly slowly in the UK. A third team, funded internally by CRUK, is examining workforce and the structure and function of multidisciplinary teams. I serve as Scottish vice-chair, working alongside a Canadian chair and John Butler as English clinical adviser, and report to a programme board representing every jurisdiction.

It all started with a letter

The second collaboration is the Early Breast Cancer Trialists’ Collaborative Group (EBCTCG), a statistical team in Oxford established by Sir Richard Peto. In many respects, it is the mirror image of the ICBP. The ICBP studies whole populations and seeks to exclude no one; the EBCTCG pools individual patient data from randomised trials that address a shared question.

The group began with a letter in 1984, following a meeting of trialists held, famously, not in Oxford but at Heathrow. Tamoxifen was already known to be an effective, though non-curative, treatment for advanced breast cancer, and trials around the world were testing its use in patients with curable disease. Each trial observed fewer recurrences, but none was large enough to determine whether tamoxifen cured patients or merely delayed recurrence. By sharing individual patient data with the Oxford team, the trialists created an overview that showed clearly that tamoxifen did cure some patients. Early chemotherapy was subsequently shown to do the same. This was the beginning of a partnership that has now lasted more than forty years.

Its operating model is important. When a randomised trial has addressed a question that Oxford wishes to investigate, the team asks the trialists to share their individual patient data. Ownership remains with the trialists, who join the steering committee and review outputs before publication. Oxford checks the data for imbalances in randomisation and other inconsistencies, although it does not verify them against source records. The process is slow, especially for commercially sponsored trials, from which extracting data has sometimes taken five years. Its reward is scale: the group now holds records for almost three-quarters of a million patients across hundreds of trials. Scotland diagnoses about 5,000 breast cancers each year. Even if every patient entered a trial, accumulating 700,000 cases ourselves would take well over a century.

Far from being a spectator, Scotland has contributed through adjuvant trials led by or recruiting from Edinburgh and other Scottish centres. I was chief investigator of the UK adjuvant trial TACT2 and of BEATRICE, the global trial of adjuvant bevacizumab in triple-negative disease. I also co-led the international lapatinib and capecitabine trial that secured the treatment’s initial worldwide marketing authorisation, and worked on the HER2 trials HERA, APHINITY and ALTTO through the Breast International Group, the Brussels-based network of 57 academic breast cancer trials groups that I now chair. In time, every Scottish patient enrolled in one of these studies contributes to an Oxford overview capable of changing practice worldwide. This is the understated value of consistent recruitment and registry-quality follow-up, and it depends on Scotland choosing to participate in the trials in the first place.

The scale of the EBCTCG’s output is evident in the publication of around thirty overview papers between 1995 and 2019, followed by a steady series since then, mostly in The Lancet and The Lancet Oncology. Topics have included ovarian suppression, extended endocrine therapy, regional nodal radiotherapy, anthracyclines and taxanes, trastuzumab, and a 2024 pooled analysis of recurrence reductions among more than 155,000 women. Because most randomised trials in curable breast cancer enrol between 1,000 and 5,000 patients, these overviews provide evidence on an entirely different scale. The full list appears on the EBCTCG publications page. Current work includes lobular breast cancer, which accounts for roughly one in ten cases, behaves differently clinically, and continues to prompt debate over its response to treatment. Other questions concern how BMI influences risk and outcome, which patients are affected, and whether menopausal status matters. Researchers are also examining whether Ki67, a marker of cell proliferation, is prognostic, predictive or both, while updating earlier analyses such as those of bisphosphonate trials.

Edinburgh’s contribution extends to the operation of the group itself. Leadership of trials included in the meta-analyses led to my appointment to the steering committee and, subsequently, to the vice-chair role. Karen Taylor now leads our scientific contribution, following John Bartlett, and we hold Breast Cancer Research Foundation funding to create a digital slide archive. In collaboration with the Oxford statistical team, this resource will allow artificial intelligence methods to be applied to diagnostic slides.

Where this leaves Scotland

Together, these collaborations clarify Scotland’s opportunity. Through the ICBP, our unusually strong population data can be used to investigate persistently poor outcomes and learn from systems that perform better. Through the EBCTCG, Scottish trial participants contribute to evidence that changes practice internationally, provided that recruitment remains strong and records remain complete. Neither model works well in a fragmented system.

With around 5,000 breast cancer cases each year, Scotland cannot out-recruit larger countries by working one centre at a time, and those cases are too valuable to disperse across uncoordinated efforts. Acting nationally offers a better approach. The case for the Scottish Breast Studies Group is practical rather than sentimental: coordination would allow centres to recruit collaboratively to the trials that inform global overviews, rather than compete with one another. Linking trial information to Scotland’s registry and routine records could also address real-world questions about patients who never enter trials. A national group would offer trainees and early-career researchers a route into wider collaborations, while giving Scotland a coherent voice within the EBCTCG, the Breast International Group and the ICBP. My experience as a small cog in these large organisations, in Oxford, Brussels and the benchmarking partnership, has taught me that influence comes from arriving organised, with good data and reliable delivery. Scotland already has those ingredients; a studies group provides the means to bring them together.

What it adds up to

Both stories demonstrate what becomes possible when data are treated as a shared resource. Population records can expose weaknesses that averages and assurances obscure; pooled trial data can answer questions beyond the reach of any individual study. Neither achievement is effortless. Governance, permissions, collection and alignment all require sustained work, as does the inclusive stewardship needed when data are held across academic and commercial organisations. Yet there is genuine satisfaction in being a small part of a collaboration that changes care. Greater satisfaction may come when a country the size of Scotland learns to contribute as a coordinated whole.