Observational studies will become more important in deciding future food policy, a leading cancer researcher has suggested.
Professor Kay-Tee Khaw, principle investigator in the European Prospective Investigation of Cancer (EPIC-Norfolk), from the University of Cambridge, said there was a lot of controversy surrounding cancer and diet.
Khaw said that because some theories of diet related disease cannot be tested out by closely controlled experiments, it has become necessary to make use of observational studies.
The traditional view, however, is that because observational studies lack the probabilistic equivalency between groups that controlled experiments provide, their results are much less convincing.
While making the case for observational studies, Khaw urged caution: “Correlations are not causations, but they are the beginning of our hypotheses.” She was speaking following the recent inaugural meeting of the Food Standards Agency’s (FSA’s) General Advisory Committee on Science.
Khaw said there were key lessons to be learned from observational studies in which associations had not been subsequently substantiated.
Notably, most misleading findings result from small scale, poor quality studies, often not subjected to peer review, not published in scientific journals and not replicated. She added that non-experimental studies were especially susceptible to bias when there were selection effects stemming from personal choices, such as lifestyle and supplements.
She urged particular caution where major interests were concerned - such as commercial considerations or strongly held clinical views. Lastly, she said causal claims, even from good studies, needed to be based on similar findings from multiple studies using varying designs.
Consultant and former chief executive of the FSA Jon Bell asked how the uncertainty that existed in science was to be measured and explained in such a way that others might understand it. “Where do you set the [certainty] line above which you decide policy?” he added.
Khaw replied: “You are never going to be able to demonstrate absolute safety, so we have to continue surveillance. That has to be more integrated, as in the US. Implement policies and monitor the outcomes.”