EDITOR, - Ward F M Posthuma and colleagues suggest that preventive hormone replacement therapy is unjustified because the observed protection from stroke and ischaemic heart disease could be accounted for by selection of healthy women in uncontrolled studies.1 Their own arguments and citations, however, are also biased.

Central to their thesis is the apparent correlation of the degree of reduction of risk with lowered incidence of cancer across these studies. They reason that this proves that “healthy cohorts” were recruited rather than that cardiovascular events were genuinely prevented by hormone replacement therapy. Women seeking hormone replacement therapy, however, tend to have higher socioeconomic status. This in turn is associated with a raised incidence of the commonest malignancy in women, which tends to bias the studies in the opposite direction. There is also limited support for the authors' sweeping statement that “cancers are if anything increased” by oestrogens. For combined oestrogen and progestagen replacement to physiological concentrations, as commonly used in Britain, the evidence is non-existent. The authors reference an American study that is hard to interpret and suggests a small increase in breast cancer in current users of hormone replacement therapy (possibly an artefact arising from increased surveillance),2 while data elsewhere in their paper from a large number of such studies show an overall reduction in total malignancies. There is no unequivocal evidence for any increase in cancer other than endometrial carcinoma, and that only with unopposed oestrogen treatment.3

The authors also ignore the fact that reduction of postmenopausal bone loss is a principal indication for blanket hormone replacement therapy. This is less controversial, though it is still to be the subject of a definitive trial showing a fall in fractures. Such a study is indeed desirable. For treatment of asymptomatic subjects there is no substitute for …