One of the central tenets of modern scientific education is this simple idea: association does not equal causation.
Which is to say: when a given scientific study finds that one factor (say, obesity), is associated with a specific outcome (say, an increase in colorectal cancer risk), that’s all we should take away from it – that obesity is associated with colorectal cancer.
But here at AICR, we go further. We say, for example, that obesity is a cause of colorectal cancer – and nine other kinds of cancer as well. We say alcoholic drinks are a cause of cancers of the breast, esophagus, liver, mouth and colorectum. We say diets high in red and processed meat are a cause of colorectal cancer.
We know that talking about cancer causality in this way sets us apart from many others who talk about the science of health. Some critics believe that when we cite any factor as a cause of cancer, we’re going “beyond the data.”
Because we’re not going beyond the data, we’re following them closely. In our Second Expert Report and Continuous Update Project (CUP) Reports, we have developed a strict set of criteria for judging scientific evidence in a transparent and objective way.
Only when all of these criteria are met – when data from multiple well-designed epidemiological studies lines up with evidence from mechanistic studies and these findings are judged to be strong, graded, and biologically plausible – do we say that a causal relationship exists.
To be clear, when we say obesity is a cause of colorectal cancer, we of course don’t mean it’s the only one. There’s an crucial difference between calling something a cause of cancer and calling it the cause of cancer.
We know that many factors come together over the course of a lifetime to contribute to the cancer process, and simply being at increased risk doesn’t mean cancer is inevitable. But the causal link between obesity and colorectal cancer exists, and the amount it increases a person’s risk can be quantified using our rigorous and comprehensive systematic review and analysis.
Yet there are those who continue to believe that causation can only truly be demonstrated using randomized controlled trials, or RCTs. In RCTs, subjects are assigned to either a control group or a group that receives the specific intervention being studied. At the end of the study, the two groups’ outcomes are statistically compared to each other.
RCTs are how cancer drugs are tested. One group of patients receives the drug being tested, the other receives a standard treatment. RCTs are ideally suited to studying the effects of individual agents – drugs, supplements, etc. – over relatively short time periods.
But when studying something as complex and ever-changing as the human diet, and its role in the development of a disease that can take decades to manifest, RCTs alone are simply not designed to provide the answers we seek. They’re a useful tool, but we need others; we need to broaden and diversify our search.
That’s why we created our systematic “portfolio approach” to analyzing the evidence. RCTs are included when they’re available, and their findings are one part of the totality of evidence we consider.
The strongest evidence for a given causal link comes when consistent findings are observed across different types of studies, ideally in different populations. There’s a lot more about the precise criteria we use, what kind of studies we consult, and how and why we grade the strength of the evidence in our reports.
But here’s the bottom line: at AICR, we use the term “a cause of cancer” mindfully. We use it when and only when we subject a possible link between a lifestyle factor and cancer risk to our robust, systematic and rigorous analysis, and judge the total body of evidence to be clear, consistent and strong (technically, either probable or convincing).
That’s a very high bar to clear, and it should be. But whenever you see us call something a cause of cancer, you can know that it’s firmly grounded in a careful and objective review of the available scientific evidence.
Published on July 17, 2015