Seuri M, Granfors K
Department of Community Health and General Practice, University of Kuopio, Finland.
Int J Epidemiol. 1992 Dec;21(6):1185-92. doi: 10.1093/ije/21.6.1185.
In seroepidemiology when a single sample is taken from exposed and non-exposed subjects, the results of serological tests are usually reduced to a dichotomy: normal (low or no antibodies) or elevated (high or antibodies found). The choice of the cutoff point between these values has been based on various vaguely defined methods and the results depend, at least partly, on the method chosen. An alternative method is introduced here which is not based on the absolute value of a serological test but on the rank order of the values of the test. In this ordered series the risk ratio (RR) and the upper and lower confidence limits (CL) are plotted for each exposed subject as the cutoff value is moved from the highest value to the lowest one. The method is applicable in quantitative immunoassays. If a priori defined RR and confidence limit or limits are reached at any point of the plot, the data support the original hypothesis. The advantage of this alternative method is that no single cutoff value is needed.
