What is risk?
Researchers use the term risk to describe the probability of a disease occurring given exposure to a particular factor. For example, what is the risk of developing dementia for people who exercise 3 times a week or more, compared to people who don’t exercise? Risk in the population is said to be increased or decreased by exposure to the factor, with decreased risk referred to as a protective effect.
Risk describes the probability of the disease at the population level, rather than for any individual. So, if the evidence shows that people who exercise at least 3 times a week have half the risk of developing dementia, that means that half as many people in the population who exercise regularly will get dementia, compared to the population of people who don’t exercise. It doesn’t mean that if I exercise regularly my personal dementia risk will be halved. But it would put me in the lower risk group.
How are risk levels determined?
Risk is estimated from two basic types of research – observational studies and intervention studies. Observational studies collect real-life information about exposure to a risk or protective factor and subsequent development of disease. For example, an observational study might measure how much exercise people do, then follow them up a number of years later to see who has developed dementia, and how that is related to their earlier exercise habits.
A single observational study is usually not enough to be sure that a factor has a real effect on disease. Findings need to be replicated in other studies of other groups of people. Sometimes, not all studies agree. A meta-analysis takes the data from all similar studies and combines them, providing an overall result from all the relevant research. This is a higher level of evidence that is more convincing than a finding from any single study.
Intervention trials are usually conducted after there is sufficient evidence from observational studies. Randomised controlled trials (RCTs) are considered the gold standard in determining whether disease risk can be reduced by intervening to alter a risk factor. They involve randomising individuals to a treatment (e.g. drug or behavioural intervention) or a no-treatment (placebo) condition.
Problems with interpreting the evidence can arise when the findings from observational studies are not replicated when tested in RCTs, and this has often been the case in the field of dementia risk research. However, the intervention trial may have been undertaken with the wrong timing and/or duration of exposure to show an effect.
Many of the risk and protective factors for dementia are believed to exert their influence during midlife or perhaps across the lifespan. If a short duration RCT involving older adults suggests an intervention does not work, this is not enough evidence to ignore observational findings that the factor does influence dementia risk, perhaps at other life stages or over a longer duration. Because dementia typically occurs in late life and develops gradually, we will likely never have RCT evidence for some risk factors. Long term RCTs are impractical and for some risk factors (such as smoking, alcohol, diabetes or cholesterol) would be unethical as we cannot expose people to health risks or deny them necessary treatment simply to see how this might affect dementia risk.