January 10th, 2018
Measures of Association
Measures of association are used to identify the relationship existing among variables thus help to make conclusions about the sample from which data was collected. They provide either the likelihood or odds at which one variable predicts the other, thus help in making conclusions on aspects such as group membership. There are various measures employed to determine the relationship between and among variables, but this paper focuses on the relative risk and odds ratio and then reviews the use of logistic regressions in case-control studies and cohort studies.
Relative Risk. Relative risk is a ratio expressing the likelihood that a member of a group will develop a specific outcome to the likelihood of a member of an opposing group developing such an outcome. For instance, the relative risk (RR) of an individual exposed to a risk factor for a disease contracting such a disease, is the probability of developing the disease in the exposed group dividend the probability of developing the disease in the unexposed group (Kahn & Sempos, 1989). The RR ratio thus measures the strength of the association between the outcome and the risk factor following exposure or non-exposure to such a risk (Fletcher & Fletcher, 2005, p 85).
Use of RR to assess the strength of the association depends on the study design. Since RR calculation requires the calculation of cumulative incidence, i.e. the number of new incidences divided by the population at risk during a specified time period (Kahn & Sempos, 1989, p.45), only study designs that allow for such data can anable calculation of RR. Thus, RR can be calculated using data from cohort studies and clinical trials but not with data from case-control studies, which do not evaluate the incident rates following exposure to the risk factor under evaluation (Fletcher & Fletcher, 2005, p. 84). For such latter studies, the odds ratio is more appropriate.
Go to part two here.