Objective: To reexamine the mortality experience of a cohort of long-term users of hormone replacement therapy (HRT) in comparison with that reported previously for the same cohort of women, paying particular attention to cardiovascular mortality, deaths from breast and endometrial cancer, and deaths attributed to suicide or suspected suicide.
Design: Longitudinal cohort of 4544 long-term users of HRT amongst whom mortality is being monitored prospectively in comparison with expected rates in the female population of England and Wales (taking account of age and calendar period).
Subjects: 4544 women, all of whom were recruited from specialist menopause clinics around Britain and had taken at least one year's continuous HRT at the time of recruitment to the study.
Main outcome measures: All cause mortality, cardiovascular mortality, deaths from female cancers, deaths attributed to suicide or suspected suicide.
Results: Overall mortality (based on 236 deaths over the entire study period from recruitment to December 1988) remained significantly lower than expected on the basis of national rates (relative risk (RR) 0.56, 95% confidence limits (CL) 0.47-0.66). When specific causes were considered, the only mortality ratios greater than unity were for injury, poisoning and violence (1.54, 95% CL 1.02-2.06), and for suicide and suspected suicide ('suicide') (2.40, 95% CL 1.68-3.11). Comparison of the ratios for the 112 additional deaths with those obtained in our previous analysis revealed that one of the few ratios to show any increase was that for breast cancer mortality. This rose from a significant deficit of 0.55 (95% CL 0.28-0.96) in the earlier period to 1.00 (95% CL 0.55-1.45) in the later period. There was also a suggestion of an increase in breast cancer risk with increasing duration since first use of HRT. Most of the other cause-specific ratios were very similar over the two periods. The ratio of death from all circulatory diseases was notably lower in the later analysis (RR 0.37, 95% CL 0.15-0.58) than in the earlier analysis (0.51, 95% CL 0.36-0.69), as were all of the subcategories of cardiovascular death. The mortality ratio for cancer of the ovary and uterine adnexa fell from 1.12 in the previous analysis to 0.63 (95% CL 0-1.41). The mortality ratio for 'suicide' also decreased, but was only slightly lower in the later period. As before, however, there was evidence of a relatively high prevalence of prior psychiatric problems amongst the recent 'suicide' deaths, suggesting that the excess of deaths from 'suicide' may be a manifestation of selection.
Conclusion: These data are consistent with a beneficial effect of HRT on cardiovascular diseases, although updated information comparing progestogen-opposed and -unopposed treatment is not available. The increase in breast cancer mortality contrasts with the pattern for all other specific causes examined; taken together with the suggestion of an increase in breast cancer mortality with increasing interval since first exposure to HRT, this finding is somewhat worrying.