18 March, 2019:
Publication: Use of postmenopausal hormone therapy and risk of Alzheimer’s disease in Finland: nationwide case-control study by Hanna Savolainen-Peltonen, Paivi Rahkola-Soisalo, Fabian Hoti, Pia Vattulainen, Mika Gissler, Olavi Ylikorkala, Tomi S Mikkola BMJ 2019;364:l665.
Should this warrant a change in prescribing practices for MHT?
The simple answer is no.
This recent publication in the BMJ reports a case-control study from Finland comparing the use of menopausal hormone therapy (MHT) among Finnish women with and without Alzheimer’s disease (AD). The researchers report that systemic MHT ever-use was higher among women with AD (18.6%) compared with those without AD (17.0%). They conclude, “Use of postmenopausal systemic hormone therapy is accompanied with an increase in the risk of Alzheimer’s disease in postmenopausal women” and that, “[this] data should be implemented into information for present and future users of hormone therapy”.
The IMS does not agree with either concluding statements. This is because an association between MHT and AD is not evidence for cause and effect. There are many instances in medicine where research observations have not stood the test of subsequent randomized clinical trials [1,2]. Unfortunately, in this case the chance of an appropriately powered randomised trial ever being done is vanishingly small.
The study was based on the Finnish drug registries, so the sample is large. However, the study has a number of important limitations, acknowledged by the authors, that necessitate caution in interpreting these findings. Like all registry studies there was lack of information about key confounding factors, including other established dementia risk factors, and the timing of initiation of MHT. As an observational study, it is limited by ascertainment bias. Cases of AD were identified via a national reimbursement register. Whereas the Finnish Drug Reimbursement Register has a high positive predictive value for AD (most people identified will actually have AD), the sensitivity is in the order of 65%. This means that up to 35% of people with AD may not be identifiable by this process, and potentially some included as ‘controls’. Such an ascertainment bias may have influenced the study findings either way.
The study collected data on the use of MHT from 1994. Through the nineties, and until the first outcomes of the Women’s Health Initiative Trials, MHT was thought to prevent cardiovascular disease and cognitive decline, such that women at increased risk for both vascular dementia and AD may have been preferentially prescribed MHT. Hence, the finding of a small excess in the present study could simply reflect this bias. As suggested by the authors, women may have been prescribed MHT once they started to develop early signs of cognitive dysfunction, so that at least part of the association could be “reverse causality”. Also, in this study, some women classified as having AD may have had vascular dementia or mixed AD/vascular dementia, both of which may have been worsened by MHT.