Epidemiology is the science of studying health-related events that affect populations. Like all science, it is built on the fundamental belief that precise observations and measurements, combined with careful reasoning in the light of existing knowledge, is the most effective way to proceed with that study. But epidemiologists must be very careful in interpreting the results of their study: they have to recognize the potential for error, being on the lookout for possible artefacts. One significant class of errors is bias, defined in epidemiology as an error in design or execution of a study, which causes results consistently distorted in one direction because of nonrandom factors. Bias can occur in randomized controlled trials but tends to be a much greater problem in observational studies.
In this short article, author considers an observational study that investigate the association between sun exposure and risk of multiple sclerosis where a population based case-control design was used. A case-control study is retrospective in design,data about potential risk factors are collected retrospectively. Information about past sun exposure was collected by questionnaire. The outcome of the study is higher sun exposure during childhood and early adolescence is associated with a reduced risk of multiple sclerosis. But a question raises from the author: might the above case-control study and its results have been prone to any kind of bias?
To read the article you may sign up for a free trial on BMJ.