HRT in a general practice population: further observations on the ‘healthy cohort’ effect

Abstract

We recently highlighted the current uncertainty surrounding the use of hormone replacement therapy (HRT) in relation to coronary heart disease (CHD).1 As we pointed out in our paper, recent large prospective randomised studies have failed to confirm observations from several large epidemiological surveys suggesting a beneficial effect of HRT in preventing CHD.3 Few arguments had been raised to try to explain the discrepancy, but one was the possible bias in patient selection in that women on HRT might have been healthier, received more health monitoring and taken a greater interest in modifying cardiovascular risks. However, the observations we recently reported did not support such a ‘healthy cohort’ hypothesis, because HRT users appeared to be less likely to undergo investigations such as cervical smears and mammograms.1 As most of the HRT given to patients in epidemiological studies were mainly prescribed and managed by their general practitioners (GP), we hypothesised that this ‘healthy cohort’ effect could perhaps relate more to GPs who are actively involved in HRT/menopause care. To test this hypothesis, we extended our previous survey of HRT use to another general practice in the west of Birmingham, which had a regular menopause clinic. Using the practice computer system, we reviewed notes from 143 HRT users and 131 age-matched non-users (as controls) who were randomly selected from the 619 women aged between 40 and 60 years in the practice population of 6900. There were no significant differences in age, body mass index and the presence of each of the cardiovascular risk factors between the two groups. However, there were significantly fewer Indo-Asian women on HRT, although none of them had any known contraindications for the therapy (Table 1). The main indication for HRT use was the presence of menopausal symptoms, such as hot flushes and genitourinary discomfort (Table 2). Only 3 (2.1%) patients in our study were taking HRT specifically to prevent osteoporosis and none for the prevention of cardiovascular disease. In addition, patients with a known history of cardiovascular disease were not excluded from being prescribed HRT. On the other hand, there were no known contraindications for HRT in 90% of the non-users, although up to 45% had documented menopausal symptoms; the same number of patients had osteoporosis as had HRT users. Interestingly, none of the patients who were offered HRT by their GP but who subsequently refused had a history of CHD. Thus, as in our previous report,1 it appears that cardiovascular risk or cardiovascular prevention was not a factor to be considered by either patients or GPs on whether or not to prescribe HRT. Unlike in our previous report, HRT users in this practice made significantly more visits to their GPs and had regular blood pressure checks, mammograms and lipid profile tests, as well as more frequent cervical smear tests (apart from those who had had a hysterectomy) than non-users (Table 1). HRT users also made more visits to their GP within the previous 12 months (p<0.001). In conclusion, our results are broadly in accordance with our previous report and support the fact that women taking HRT were no healthier in terms of cardiovascular risk than non-users, nor did they seek more cardiovascular preventive care than non-users. However, proponents of the ‘healthy cohort’ hypothesis have suggested that patients on HRT were monitored more closely, so may have biased the results of previous epidemiological case controlled studies. Our extended observations in this general practice, which actively managed postmenopausal women in a menopause clinic, suggest that the ‘healthy cohort’ effect may actually exist, and be GP-initiated. Nevertheless, this does not necessarily mean that women who are monitored more closely are healthier. Pending further data from more prospective randomised controlled trials, the jury is still out regarding the use of HRT for cardiovascular prevention.2