4: How can someone know that the research evidence applies to them?
All decisions rely on previous experience of some kind – individual or collective. Fair tests of treatments such as randomized trials are simply well organized versions of that experience designed to minimize biases. Well organized or not, there will always be some uncertainty about how well previous experience can shape our advice for the next person.
So if the patients who had been studied in the fair tests had a similar condition, at a similar stage or severity, to the individual in question, the most reasonable assumption is that the individual would get a similar response, unless there was a good reason to think they or their condition were substantially different.
Of course, even if the evidence is applicable, a patient might reasonably ask: ‘people are all different so surely they may respond differently?’ The ‘fair test’ of a treatment will only tell us what works on average, but rarely guarantees it will work equally well in everyone; and it cannot usually predict who will suffer unwanted side-effects. Research evidence can be used to guide what treatment is likely to be best, and then tried in an individual.
With some skin rashes, for example, evidence-based treatment could be applied to one area of the body, using another area as a control. By comparing responses in the two areas, both doctor and patient can tell whether it works, or whether there is an adverse effect. Indeed it’s common to try a ‘test patch’ when first using some skin treatments, such as acne treatments on the face.
Mostly, however, we don’t have the convenience of such a straightforward comparison. For some chronic and non-lifethreatening problems, such as pain or itch, it is possible to try repeated periods on and off a drug in the same patient. This approach is also called an n-of-1 trial, meaning that the number (n) of participants in the trial is one – a single patient. With such tests in individual patients, the principles for a fair comparison still apply, including an unbiased or blinded assessment of outcome, etc. Ideally, then, we would use placebo controls of skin treatments or pills, but this is often difficult to organize.
For many conditions, however, we cannot ‘try it and see’: the outcome is too remote or too uncertain. For example, it is impossible to know whether aspirin will prevent a patient’s stroke until it is too late. This is a problem in most cases of preventive medicine, and also with treatments for many acute conditions, such as meningitis, pneumonia or snake bite, where we don’t have the opportunity to test it in each individual patient and see. So we then have to rely on whether and how to apply the evidence from the experience of studying others.
In practice, if we are happy the evidence applies, it is then important to ask how the severity of the condition in the patient (or the predicted level of risk in those who are still well) compares with that of the people in the studies. In general, patients with more severe illness have more to gain from treatment. So if severity is equal to or greater than those in studies that showed a treatment to be beneficial, we can generally be confident about the applicability of the evidence. If their illness is less severe (or if still well, they are at relatively low predicted risk) the key issue is whether a smaller benefit than that seen in the studies might still be considered worthwhile.
GET-IT Jargon Buster
GET-IT provides plain language definitions of health research terms