select search filters
briefings
roundups & rapid reactions
before the headlines
Fiona fox's blog
An observational study published in The BMJ looks at menopausal hormone therapy and dementia.
Dr Anita McGrogan, Senior Lecturer in the Department of Life Sciences at the University of Bath, said:
“This study evaluates the association between HRT use and diagnosis with dementia finding a small increase in risk. While the study looks in detail at different HRT preparations, length of prescribing and some patient characteristics, more information is needed before trying to generalise these findings to other populations. For example there are other risk factors for dementia such as smoking, physical activity, levels of obesity that have not been addressed.
“The prescribing and availability of HRT in Denmark may be different to other countries such as the UK where we know that prescribing and access to HRT differs and may be related to ability to seek this out and pay for the treatment. In the paper, those taking HRT for a long time may have different reasons for this than those taking HRT for a short time (and a different risk of dementia); similarly those with more health conditions may have greater access to healthcare providers and could result in dementia symptoms being more likely to be identified and investigated.”
Dr Susan Kohlhaas, Executive Director of Research & Partnerships at Alzheimer’s Research UK, said:
“Dementia is the number one cause of death amongst women in the UK. It affects about twice as many women compared with men, and this difference can’t be explained by differences in longevity alone. It’s therefore important to understand what contributes to dementia risk, as this could help develop prevention strategies for those most at risk in the future.
“Understanding whether hormone replacement therapy (HRT) plays a role in dementia risk has been a priority for the research community in recent years. Such studies, however, have so far been inconclusive and contradictory, with some suggesting that HRT reduces risk of dementia, and others suggesting it increases risk. Reasons for this may be because of differences in age when HRT is given, differences in how HRT is used, and different types of HRT used over the years.
“This large, nationwide Danish study has found an association between both short- and long-term use of HRT and an increased risk of dementia. Unfortunately, it doesn’t provide a conclusive answer on whether HRT contributes to dementia risk, because the study did not consider some of the other factors known to be involved, such as social isolation, smoking or dietary factors like alcohol.
“Women need to understand the implications of deciding to take HRT in terms of benefits and risks and, at the moment, as far as dementia is concerned, we’ll need to wait for more research to give clearer answers. In the meantime, people should speak to a qualified healthcare professional if they would like to know more about the known benefits and risks of taking HRT to manage symptoms of the menopause.”
Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:
“This new piece of research uses data from a large number of women, and the statistical methods used are appropriate. But it can’t avoid some important limitations of this type of research. The researchers did find evidence that the chance of having a dementia diagnosis was greater in women who had previously used menopausal hormone therapy, and indeed found more detail on how the chance of dementia varied depending on how long women took the therapy, which type of dementia was involved, and other things. But, because the study was observational, it cannot establish that the higher chance of dementia was actually caused by using the hormone therapy.
“Indeed, I’d say, it’s important to read not just the research paper but also the linked editorial by Kantarci and Manson, which goes into some detail on why the new research cannot give evidence that taking hormone therapy definitely causes the risk of dementia to increase. The editorial authors write, “Although Pourhadi and colleagues’ study was done carefully […], the observed associations could be artefactual and should not be used to infer a causal relationship between hormone therapy and dementia risk. These findings cannot inform shared decision making about use of hormone therapy for menopausal symptoms.” In other words, though the new study has increased knowledge, it hasn’t done so in a way that is relevant yet to the decisions that a woman and her doctor should make on whether and when to use menopausal hormone therapy.
“Indeed the researchers on the new study seem to agree. They write in their research paper [in the Abstract], “Further studies are warranted to determine whether these findings represent an actual effect of menopausal hormone therapy on dementia risk, or whether they reflect an underlying predisposition in women in need of these treatments.” They are saying that yes, one possibility is that the hormone therapy causes an increase in dementia risk later in life. But another possibility is that there is something pre-existing in some women’s physiology that, separately, makes it more likely that they will have menopausal symptoms that need hormone treatment, and makes it more likely that they will (later) have a dementia. And their study couldn’t distinguish between these two possibilities. If it’s the second possibility, the pre-existing disposition, then avoiding menopausal hormone treatment won’t reduce the risk of dementia , because it’s not what’s causing the increased dementia risk. And that’s why it’s sensible to ask for more research to try to distinguish between these possibilities. (Indeed there may be other possibilities for what causes what). That future research would have to use different methods, because an observational study like this one just can’t, on its own, make the distinction.
“So why can’t this study establish cause and effect? The researchers used data from the comprehensive national registries of health data that exist in Denmark. They obtained data on over 5,500 women in Denmark who had a dementia diagnosis between 2000 and 2018. They matched each of these women with ten others of the same age, who had not had a dementia diagnosis, and obtained data on those so-called ‘controls’. Then they compared the past histories of all these women, in terms of what menopausal hormone therapy they had taken, and other aspects of their health, education, income and several other quantities. They were particularly interested in whether the women who had had a dementia diagnosis were more likely to have taken hormone therapy for menopausal symptoms.
“The snag is that the women who had dementia, and the controls, differed in other ways apart from their use of menopausal hormone therapies. On average, for example, the women with a dementia had rather lower levels of education, lower household incomes, and were more likely to have had certain diseases such as diabetes. Maybe a difference on one or more of these other attributes was the cause of increased dementia risk, and not the hormone therapy at all. Of course the researchers were aware of this possibility, and they carried out statistical adjustments to try to allow for these other attributes. After doing that, there was still an association between hormone therapy and dementia. But that still doesn’t establish that the hormone therapy is the cause of the increased dementia risk. There may be other factors that are important, that the researchers could not make adjustments for, because they had no data for the adjustment. One possibility is the existence of some predisposition that independently affects the chance of needing hormone therapy and the chance of having dementia. It’s not even clear exactly what this predisposition might be, so there’s no clearly relevant data on it in the registries, and the researchers couldn’t possibly make an adjustment for it.
“The writers of the linked editorial point out that the research findings themselves indicate that there are probably other relevant factors that weren’t taken into account. The research found indications of increased dementia risk for women who had less than one year of hormone therapy, and the editorial authors say that it is “not biologically plausible” that such a short term use could cause an increased dementia risk. So the increased dementia risk in these women with very short-term hormone use must be caused by something other than their hormone use, and the adjustments that the researchers did must not have been able to take that into account.
“In situations like this, a single observational study can’t establish what causes what. It remains possible that there is a cause-and-effect association between menopausal hormone therapy and dementia, but there are other important possible explanations, and that’s why clinical decisions can’t yet be informed by these findings.
“Usually in this kind of situation, the position on cause and effect has to be established by looking at data from many different studies. You might get the impression from the press release that previous studies have also provided strong indications that it’s the hormone therapy that is at least part of the cause of the increased dementia risk. But that’s not the overall impression I get from the research paper.
“The press release says, “These findings align with the largest clinical trial carried out on this topic”, the Women’s Health Initiative Memory Study. But the researchers themselves (in their paper) point out that that trial only included women aged 65 and over, whereas the focus of the new research is on women who would typically use hormone therapy at younger ages. They mention some other issues with that trial too, which, I’d say, make it clear that that trial was really answering a rather different question about hormone therapy and dementia risk. And they, rightly, mention that two smaller randomised clinical trials did not report an association between some menopausal hormone use and later dementias. So the picture from these previous trials isn’t actually all that clear.
“The press release also says that other large observational studies have shown and association between menopausal hormone therapy and dementia. That’s true, but because those studies are observational, there has to be doubt about what’s causing what. And again the researchers on the new study point out deficiencies in what those studies were able to do. There are two of them, both produced findings that differ from those of the new study in some respects anyway, and both point out that the increase in risk, whatever is causing it, is ‘small’ or ‘slight’.
“So I don’t think that all these different studies, together, have yet got as far as indicating that taking menopausal hormone therapy does cause increased dementia risk. It’s important to try to establish whether this is true or not, but that will indeed require substantially more research, as the researchers on the new study recommend.”
Dr Sarah-Naomi James, Senior Research Fellow at the MRC Unit for Lifelong Health and Ageing at University College London, said:
“The editorial seems very fair – the study has strengths in utilising nationwide data it has available and their effort to try to differentiate between different types and duration of HRT use is admirable, well-needed and seems fairly robust.
“However, the study has fundamental limitations in its ability to interpret and understand the true underlying causal pathways of the observed association, as both the exposure (why you would be prescribed HRT in the first place, and why you would be prescribed certain types and duration of medication use) and the outcome (dementia diagnosis) have many things in common that influence them, and so this association may be artificial.
“For example, changes in sleep or mood are very common symptoms of menopause and reasons to seek out HRT; meanwhile we are starting to understand that sleep and mood may play an important role in in the expression and progression of dementia.
“The best way to understand whether HRT medication itself causes dementia comes from clinical trials, and to date, there is not enough evidence to support a direct link from the medication itself, and this new study alone should not change practice.
“What is clear from this study is that more research is warranted to understand the exacerbated risk of AD in women, including the role and patterns of HRT on diseases that cause dementia, but also to understand the wider context, encompassing menopausal symptoms, as well as life course and sociocultural influences that impact women in this transitional period of their life.”
Prof David Curtis, Honorary Professor, UCL Genetics Institute, University College London (UCL), said:
“The study shows that women who were prescribed HRT are somewhat more likely to be subsequently given a dementia diagnosis. The effect size is moderate: if the background risk of getting a dementia diagnosis in women who did not receive HRT was 10% then the risk in women who did get HRT would be around 12.5%.
“The accompanying editorial is really quite dismissive about the possibility that this is showing any real causal effect. It describes the idea that a brief course of HRT could increase dementia risk as ‘biologically implausible’, and I agree. This provides strong grounds for concluding that the association is due to confounding factors rather than being because taking HRT actually increases dementia risk.
“The authors of the editorial suggest a number of reasons for the observed association. One of these is that women taking HRT would see doctors more often and hence be more likely to be diagnosed with dementia earlier. An even more convincing suggestion is that some women were actually prescribed HRT because they had memory problems. (There was an idea that women going through the menopause who had memory problems and similar symptoms might have these alleviated by HRT.) It is typical that people who will eventually develop dementia will have mild memory problems years before they receive a formal diagnosis. So the editorial is really saying that women who eventually develop dementia may be more likely to be prescribed HRT in the years before they are diagnosed.
“Because the risk that the association is due to confounding is so great, and because formal clinical trials do not show that HRT causes dementia, the experts who wrote the editorial strongly state that the findings of this study should not have any influence on decisions about whether or not somebody should receive HRT. I agree with this position.”
Dr Amanda Heslegrave, Senior Research Fellow at the UK Dementia Research Institute, said:
“This is a large observational study and the data from a health registry should be of a good quality. It is important to note that the data was not collected specifically for this study.
“I don’t believe you can suggest a causal link from this data, it is known that many women who seek HRT at or around menopause do so because of concerns around memory and cognition, potentially confounding data.
“There is research that suggests HRT can be protective with respect to dementia, also other research that the paper cites, that HRT is associated with dementia – this suggests to me that we really don’t know the whole story and targeted research is required.
“We cannot tell from this study whether or not different routes of administration have the same apparent effect – this too needs to be investigated.
“The prevalence of all cause dementia over Alzheimer’s disease tells us that diagnosis continues to be poor and needs to be improved urgently.
“This research may cause alarm for women taking HRT, but it highlights just how much we still don’t know about the effects of hormones on women’s brain health and with promising treatments on the horizon it should be a call to action to make this a priority area of research.”
Prof Gill Livingston, Professor of Psychiatry of Older People, University College London, said:
“The press release accurately reflects the science. This study is good quality large research with excellent detailed data from the whole female population of Denmark from aged 45 in the beginning of the study. It finds that there is an overall increase in risk of dementia by 24% over the life of the study in women taking combined oestrogen and progestin HRT treatment. It does not find the same risk for other groups.
“It is observational so does not prove a causal link. However the possibility of a causal link is increased by the new finding (as better records were available) that the longer people took this type of HRT then the more likely they were to develop dementia. In addition, the HRT prescription preceded the dementia by many years. Also the researchers took into account in the analysis other risk factors for dementia.
“It is possible that women developing dementia are more likely to use HRT in the first place (reverse causation) as they have, for example, more low mood which may precede dementia. It might also be that dementia is diagnosed more rapidly if people are in touch with the health care system because they are getting HRT.
“This fits with other good quality evidence. The existing evidence of other studies have similar but not as detailed findings as populations are smaller with less detailed health records. E.g. another observational study1.
“Also there is better evidence from a meta-analysis of trials where HRT was given, also suggesting HRT may increases the risk of dementia.2
“I think this is in line with what we know already. However I do not think it is widely known that all medical treatment has risks and benefits. This new study gives us more information. I think it is clear that HRT should not be given to prevent dementia as it does not and may increase the risks. Women should discuss with their doctors and weigh up the benefits compared to the risks.”
1 Vinogradova Y, Dening T, Hippisley-Cox J, et al. Use of menopausal hormone therapy and risk of dementia: nested case-control studies using QResearch and CPRD databases. BMJ 2021;374:n2182. doi: 10.1136/bmj.n2182.
2 Zhou HH, Yu Z, Luo L, et al. The effect of hormone replacement therapy on cognitive function in healthy postmenopausal women: a meta-analysis of 23 randomized controlled trials. Psychogeriatrics 2021;21(6):926-38. doi: 10.1111/psyg.12768 [published Online First: 20211007]
Comment supplied by Fatti per Capire, an SMC-like organisation providing the views of scientists on controversial topics in Italy:
Prof Andrea Lenzi, Professor in Endocrinology at the University of Rome La Sapienza, and President of the Biosafety, Biotechnology and Life Science Committee of the Prime Minister Office of Italy, said:
“This new study is extremely interesting in terms of the number of people examined, but undoubtedly needs to be further investigated in order to understand the effect of any pre-existing vascular conditions or other factors that were not examined.
“This need for further investigation is typical of any observational study, as it can only highlight correlations and cannot establish cause and effect. The correlation could theoretically be linked to other reasons, or be random or not so strong, and all this can only be ruled out by careful verification.
“In any case, what the researchers observed is certainly further confirmation of the fact that HRT must be customised to avoid individual risks related to vascular conditions or even cancer risks (which for HRT in recent years have been greatly reduced in any case). When it comes to deciding on HRT, gynaecologists, endocrinologists and cardiologists must work side by side.
“In addition, it must be considered that the study looked at women who had received HRT many years ago, as they were of menopausal age around 2000. Over the years, however, the therapies have changed a lot, both in terms of type of drugs and dosages, and therefore the side effects of today’s therapy may be quite different from those of HRT from back then.
“It is also necessary to look carefully at the statistical data, because dementia, although worryingly widespread, is not a common condition in women, and it is necessary to understand in how many real cases a possible percentage increase in risk translates.
“What is certain is that today’s average life span has lengthened a lot and we cannot accept that a woman should live for 30 or 40 years without the hormones on which her well-being depends. The challenge then is to find the best therapies and minimise the side effects (after having understood them well, of course) – certainly not to give up therapy.”
Comments sent out by our colleagues at SMC Australia:
Prof Martha Hickey, Professor of Obstetrics and Gynaecology and Head of Menopause Services at the University of Melbourne and the Royal Women’s Hospital, said:
“Around 14% of Australian women take menopausal hormone therapy (MHT) for the treatment of troublesome menopausal symptoms, and most will take combined (estrogen and progestogen) preparations. In the past, MHT was recommended for the prevention of chronic diseases including dementia.
“This nationwide study from Denmark provides important new information showing an increased risk of all cause dementia with combined MHT. This was observed with both continuous combined and sequential MHT preparations and risk increased with longer duration of use. These findings are similar to those reported in a large randomised controlled trials of older postmenopausal women taking MHT.
“However, in this new study even women using MHT in the years immediately after menopause had an increased dementia risk. MHT is an effective treatment for hot flushes and night sweats but is not indicated for other reasons.
“Strengths of this study include the large sample size and capture of all prescriptions for MHT from 1995. Limitations include the exact type of dementia and that women taking estrogen-only MHT (following hysterectomy) were not included.
“This is an observational study so cannot prove that MHT causes dementia. However, the findings are concerning and reinforce recent guidance from the United States Preventive Task Force (2022) that HRT should not be used for the prevention of chronic conditions including dementia.”
Scientia Professor Kaarin Anstey, Director of the Ageing Futures Institute at the University of New South Wales and a Senior Principal Research Scientist at Neuroscience Research Australia, said:
“This paper reports on a large observational study of national registry data from Denmark that examined the link between HRT and incident dementia. Focusing on combined oral estrogen and progesterone, the authors found an increased risk of dementia in women who had ever taken HRT. When this was broken down by duration of exposure, there was an increase risk of all cause dementia of about 20% when taking HRT for one or more years which appeared to increase to around 40% risk when taking HRT for eight to 10 years.
“Previous research on this association has reported mixed findings (both increased and decreased risk of dementia associated with HRT).
“There are some important factors to consider when interpreting the results of this study. The sample included women who were aged 50-60 in the year 2000. Therefore the types of HRT that the study are based on were those prescribed in Denmark during this period and there may be factors associated with the patterns of prescribing at that time which do not apply currently. A strength was that the database captured nearly the entire population.
“The study only examined orally administered HRT and not dermal so results cannot be generalised to dermal administered HRT. Overall, while the study findings are of significant concern, there are some caveats that need to be considered.
“The follow-up interval meant that the study identified dementia cases occurring with a median age of 70 which is far younger than would usually be seen if the cohort had been followed up for longer. The usual median age for dementia diagnosis is in the 80s. The results may indicate earlier onset of dementia in women who took HRT in Denmark. The authors also report low rates of Alzheimer’s disease of 26% whereas in the population one would expect 60-70% of dementia cases to be diagnosed with Alzheimer’s disease if the sample were follow-up up into the 80s. This may reflect the way diagnoses were recorded in the registry. The authors note that the findings may also be explained by indication bias. This means the symptoms that lead to women taking HRT may also be linked to increased risk of dementia. The overall differences in the expected rates of dementia suggest that there are some significant biases in this study which may be due to the methodology. Further work is needed to clarify this important issue.”
Prof Susan Davis, Director of the Women’s Health Research Program in the School of Public Health and Preventive Medicine at Monash University, said:
“The observational study of Pourhardi et al published in the BMJ reports the associations between the use of combined menopausal hormone therapy (MHT: estrogen with a synthetic progestin) and the likelihood of subsequent dementia.
“The average use of MHT was just under four years, virtually all use was oral therapy (90%) and the majority took a formulation of oral estrogen with oral norethisterone (83.5% of those who developed dementia and 80.6% of the control group). So, the findings are fundamentally driven by this specific combination of therapy that included a specific synthetic progestin therapy.
“The authors say they excluded women who had had ever had a hysterectomy. Women who have had a hysterectomy only require estrogen therapy without a progestin. They say that in their analysis ‘Compared with never users of menopausal oestrogen-progestin therapy, systemic or vaginal oestrogen only treatment, and perimenopausal progestin only therapy, ever users of menopausal oestrogen-progestin treatment were more likely to develop all cause dementia…’. So this gives rise to the question as to whether estrogen-only users were also included in the comparator group as stated, and if so, what proportion of the comparator group were using estrogen-only therapy is unclear.
“Around the year 2000 in Denmark estrogen-only therapy comprised about 38% of MHT use (excluding vaginal estrogen only therapy). So why did the researchers not report on the association between estrogen only therapy and dementia risk? This might indicate whether the observed risk might be attributable to the progestin component when prescribed with estrogen.
“As for all observational studies, despite statistical adjustments for variables such as weight or education etc, women who choose to take MHT are different from women who do not. The main difference is symptoms. So, while multiple adjustments have been made in the analysis for dementia risk factors such as living alone and education, no consideration was given to the reasons why women take MHT.
“So, the real elephant in the room for this analysis is: is the observed risk of dementia in MHT users is due to the use of oral synthetic MHT or due to the reasons why women ever took MHT? This has been overlooked.
“As the study included women aged 50-60 years in 2000, just before the publication of the Women’s Health Initiative (WHI) Study that raised the concern of MHT use and breast cancer – specifically estrogen plus progestin use and breast cancer, many of the women in this study are likely to have stopped MHT because of this perceived risk, but still had symptoms.
“The conclusion that even short term use increased dementia risk may be because women either stopped MHT for fear of breast cancer after the WHI data was published, or were told by their doctor to stop, and had untreated ongoing symptoms of flushes, low mood and poor sleep that increased their subsequent risk of dementia.
“Longer term users would have most likely continued to use MHT, despite warning of breast cancer risk because of the severity of their symptoms.
“So before drawing the conclusion that oral oestrogen with norethisterone causes a greater risk of dementia, one must consider whether MHT use is simply a surrogate marker for the postmenopausal symptoms that might directly contribute to dementia – hot flushes and night sweats, and poor sleep.”
‘Menopausal hormone therapy and dementia: nationwide, nested case-control study’ by Nelsan Pourhadi et al. was published in The BMJ at 23.30 UK time on Wednesday 28 June.
DOI: 10.1136/bmj-2022-072770
Declared interests
Prof Kevin McConway: ‘I am a Trustee of the SMC and a member of its Advisory Committee. My quote above is in my capacity as an independent professional statistician.’
Prof David Curtis: ‘I declare no conflict of interest.’
Prof Gill Livingston: ‘I have no interests to declare.’
Dr Amanda Heslegrave: ‘I don’t have any disclosures apart from a real desire to push research in this area forward!’
Prof Martha Hickey declares she received funding from Que oncology for a clinical trial of a non-hormonal treatment for vasomotor symptoms (now finished)
Prof Susan Davis reports honoraria from Besins Healthcare, Mayne Pharma, Health Ed, BioSyent, Lawley Pharmaceuticals, and Que Oncology. She has served on Advisory Boards for Mayne Pharma, Astellas Pharmaceuticals, Theramex, and Gedeon Richter and has been an institutional investigator for Que Oncology and Ovoca Bio. Research support from Australian NHMRC, MRFF, Heart Foundation
Kaarin Anstey has not declared any conflicts of interest.
For all other experts, no reply to our request for DOIs was received.