Reproducibility of an indirect immunofluorescence test for varicella-zoster antibodies

The need has arisen for a rapid test for detection of antibody production in the central nervous system (CNS) in patients with varicella-zoster (VZV) encephalitis. We have developed an immunofluorescence test based on semiquantitative determination of antibody concentration. In this work we have systematically examined some of the most important errors that may influence the quantification of the system. In the reported experiments six samples with known antibody dilutions and one control sample were tested for fluorescence "in the blind" on 716 slides. On each slide two samples were tested in six two-fold dilutions. The reproducibility of the differences between two titres was best when the fluorescence of each test sample in each pair had been directly compared under the microscope. Pairs of samples which had been tested in random order with very small chance of being directly compared, showed a higher number of errors. Of 224 pairs having four-fold or greater concentration difference, only one was not detected by the improved test. However, with only two-fold difference, the failure increased to 35%. Only three out of 62 pairs with the same concentration of antibodies falsely showed two-fold or greater difference of titres. No paired titres showed a difference opposite to what was expected. Based on the results reported in this paper we have started to use the test in routine diagnosis of encephalitis.