Parallel studies of HLA antigens in patients with rheumatic heart disease and scleritis: comparisons with three control populations.

Parallel studies were carried out on HLA antigens in patients with rheumatic heart disease and scleritis. Comparison with one control population showed a significant excess of BW15 in both disease samples, while a comparison with two other control populations, showed the excess not to be significant. Possible reasons for this discrepancy are discussed, together with the effect on statistical significance of a small percentage of false antigen assignments in one of the samples. A small systematic serological false assignment of an antigen can, by itself, produce a significant result more easily if the frequency of the antigen being detected is low than if it is high. It is suggested that this effect may contribute to the discrepant significant results obtained by different workers in some HLA and disease studies.