A statistically defined endpoint titer determination method for Immunoassays

Results of immunoassays for which no positive standards are available are o ften expressed as endpoint titers. The endpoint titer is defined as the rec iprocal of the highest analyte dilution that gives a reading above the cuto ff. Unfortunately, there is no generally accepted rule for the determinatio n of these cutoff values. In enzyme-linked immunosorbent assays (ELISA) a v alue two or three times the mean background or negative control reading is sometimes used. Other investigators set the cutoff arbitrarily at a certain absorbance value. These procedures do not provide statistically meaningful information about the risk of overtitration or false low titers. We have s olved this problem by devising a practical method for establishing a statis tically valid cutoff. The procedure involves calculating the upper predicti on limit using the Student t-distribution. The mathematical formula which d efines the upper prediction limit is expressed as the standard deviation mu ltiplied by a factor which is based on the number of negative controls and the confidence level (1 - alpha). Appropriate factors are provided for 2 to 30 negative controls and for confidence levels ranging from 95% to 99.9%. Our new method is more reliable than other nonstatistical procedures yet do es not require sophisticated computation. It can be applied to a variety of immunoassays provided that negative controls are available. (C) 1998 Elsev ier Science B.V. All rights reserved.