Severity of chest disease in CF patients in relation to their genotypes.

in our second study (25%) was markedly lower than that sampled in our first study (75%). It is possible that this may have contributed to the reduction of effect in the second study. Nothen et al found that family history of schizophrenia had no effect on genotype distributions. However, the possibility remains that their patients differed from ours with respect to treatment response or to some other as yet undetermined variable. Nothen et al point out correctly that the controls in the French sample were not in Hardy-Weinberg equilibrium and showed an excess of heterozygotes (though X2 is in fact 5 47 and not 8-7). However, the Cardiff controls did not show significant departure from Hardy-Weinberg equilibrium and nor did the further set of 98 controls typed in Car-diff. Moreover, we should remember that the significant excess of homozygotes seen in the patients as compared with Hardy-Weinberg expectations is not dependent upon the distribution of genotypes seen in the controls. Furthermore, the difference between patients and controls remains significant when the French controls are removed from the analysis (X2=6 8, p=0009). In view of the fact that the French controls were all members of hospital staff, whereas the controls in the first Cardiff study were married in members of families seeking DNA diagnosis for non-psychiatric diseases, we too wondered whether these differences might reflect heterozygote advantage. However, when we compared a sample of members of staff from our Institute, all of whom are graduates (n = 37), with our other controls (n= 129), no significant homozygosity effect (p = 066) or other differences were found and no significant departures from Hardy-Weinberg equilibrium were observed (both p values = 0 9). Morell is correct to point out that our findings could have resulted from allelic dropout. We have also considered this possibility and have retyped our sample with a second set of primers placed internally to the first. The results we obtained were identical to those obtained from the first set of primers. assessed severity of lung disease by calculating % predicted values for FEV, and FVC and comparing single values in subjects of different ages, as did Johansen et al' in a similar study. Although in widespread use, this method has deficiencies, especially when analysing lung function data over time. The widely held assumption that a given % predicted value means the same for subjects of different ages is untrue; …