The Oxford Overviews

How cumulative evidence changed the field

By the early 1980s, randomised trials in breast cancer had multiplied across Europe and North America, but each was sized to answer its own question and most were underpowered to detect moderate treatment effects with confidence. The survival benefit from adjuvant chemotherapy or tamoxifen, if real, would be real but small — a reduction in the annual death rate of perhaps fifteen or twenty percent. Detecting that reliably required far more events than any single trial had accumulated.

The solution was systematic. In 1984, Richard Peto and colleagues at Oxford convened what became the Early Breast Cancer Trialists' Collaborative Group, assembling not published summaries but the underlying patient-level records from every eligible randomised trial worldwide. Individual patient data meta-analysis is technically more demanding than combining published results, but it is also more trustworthy: it allows uniform analysis, catches errors in reported figures, and avoids the publication bias that plagues literature-based pooling.

The first overview was published in the New England Journal of Medicine in 1988 and covered roughly 30,000 women across trials of chemotherapy, tamoxifen, ovarian ablation and radiotherapy. Its conclusions were precise enough to be acted on. Polychemotherapy reduced annual mortality by around fifteen to sixteen percent in women under fifty. Tamoxifen, particularly when given for two years or more, reduced annual mortality in oestrogen-receptor-positive disease by comparable margins. Neither finding had been demonstrable from any single trial; both emerged clearly from the pool.

The overviews have been repeated at roughly five-year intervals since, each iteration incorporating new trials and longer follow-up from older ones. The 1992 update confirmed that longer tamoxifen duration improved outcomes further. The 2005 cycle showed that the absolute survival gains from adjuvant therapy were larger than earlier follow-up had suggested, because breast cancer deaths accumulate over fifteen years. The 2011 radiotherapy overview established that post-mastectomy chest-wall irradiation reduced not just local recurrence but breast-cancer mortality — a finding that required pooling because the absolute effect, once again, was modest per trial.

The method itself became a template. The Cochrane Collaboration, founded in 1993 partly in the intellectual orbit of these overviews, institutionalised systematic review across medicine. Within breast cancer, the Oxford overviews remain the benchmark: periodically renewed, rigorously maintained, and capable of settling arguments that decades of individual trials had left open.