[Approach to epidemiological cohort designs].

In neurological cohort studies participants are those who are "at risk" of developing the outcome (i.e., stroke, dementia, parkinsonism) and should be free of such event when their follow-up is begun. They are classified and controlled at the outset by their risk or "exposition" factors. There are several alternatives of selecting the study population, although the most known are those of the large population bases (for example, Framingham Heart Study, the Rochester MN, etc.). There should be standardized procedures of classification and follow-up and the investigator who makes the diagnosis should be blinded to the exposure status, when it is possible. In cohort studies, we will consider several frequency measures, such as incidence and risk, and the most frequent association measures, such as relative risk, risk difference, and rate ratio. The advantages of a cohort study in neurology are: a) the exposure is known to occur prior to the onset of the outcome and is measured prior to the occurrence of the outcome, b) multiple exposures and outcomes (end-points) can be studied simultaneously, and they usually have good control, c) it is good for rare exposures (if there was adequate sampling), and it is good for common outcomes. The disadvantages of a cohort studies are: a) they are expensive and time-consuming; b) possibility of change in investigators for very long studies, or attrition of participants; c) not good for rare outcomes; d) exposures can change, exposures are not independent of other potential causes and prognostic factors, and e) knowledge of exposure may alter surveillance for the disease or alter the likelihood that a particular diagnostic label will be applied. It is useful for the clinical neurologist to know and to understand these terminology and concepts.