How Long Can You Safely Take Menopause Hormone Therapy

Looking at that new study (Part 1)

We have very little data on the long-term use of menopause hormone therapy (MHT), given the absence of long-term clinical trials. The 2022 Menopause Society Guidelines state, “Of note, the continued use of hormone therapy in healthy women aged older than 65 years at low risk for breast cancer and CVD (cardiovascular disease) is limited by insufficient evidence regarding safety, risks, and benefits.” This conclusion is based largely on expert opinion because of the lack of data. The gist of the recommendation is we think people who are at low risk for complications can continue hormones past age 65 if they still feel they need them, but honestly, we don’t really know if there are risks or not. 

Many people have asked me here about the long-term use of MHT, and I have avoided writing about it so far because there was so little data. How do you write about something when it’s almost all opinion? I really had nothing else to add beyond the 2022 Menopause Society Position Statement. Basically, my opinion was, sure, if you are low risk and it’s helping, it seems reasonable, especially if you are at risk for osteoporosis. However, I am pleased to say that the meandering course of science generally rewards procrastinators. Now, we have a new study published in the journal Menopause that specifically looks at menopause hormone therapy for women aged 65 and older. 

The Study: The Basics

The researchers used electronic medical records from the US Centers for Medicare and Medicaid Services (CMS). This massive database includes billing for prescription medications and encounters, like office and hospital visits. This means researchers could look at who received hormones and who didn’t, as well as evaluate a variety of important health outcomes: death, five different types of cancer (breast, colon, lung, endometrium, and ovary), cardiovascular disease (six different types, including stroke and blood clots), and dementia. The sheer volume of patients in this database meant that there was a lot of data. 

One exciting thing about this paper is that given the wealth of data, they did not just look at estrogen vs. estrogen and a progestogen (either progesterone or progestin); they were able to get data on oral, transdermal and vaginal estrogen as well information about dose. They divided estrogen into low, medium (or standard), and high doses. The medium dose for oral was 0.625 of conjugated equine estrogens (CEE or Premarin), 1 mg of oral estradiol, and 5 μg of oral ethinyl estradiol. A 50 μg transdermal dose was considered the medium transdermal dose. This is also the dosing that I consider medium (you can read more about dosing here). They also had data on the different types of progestogens as well as data on various combinations of estrogen plus progestogens. 

In the main analysis, the investigators grouped low-dose transdermal and low-dose vaginal therapies together, meaning a 25 mcg estradiol patch was in the low-dose category, but so were vaginal estrogen tablets and rings. However, one of these delivers estrogen into the blood (patch), and the other does not. There is more data on these two groups in the supplement, so I have also used that data to help clarify some of the findings. Honestly, lumping low-dose transdermal and vaginal therapies together seemed like an odd choice. 

This study is observational, which means women were not randomly assigned to hormones or a placebo and then followed. Instead, they likely chose hormone therapy in the same way most women do: either they asked for it because of symptoms, or they felt it would improve their health, or their doctor recommended it for one of those same two reasons. We also don’t know how long the women were taking estrogen before enrolling in Medicare. The researchers chose women when they first became available for Medicare and had Part D benefits and followed them until they developed one of the outcomes in the study, died, or left the health plan. They collected data from 14 years of billing records, and people needed to be included in the Medicare plans for a minimum of 6 months to be included in the study. They selected a large number of control women who were not on hormones for comparison. 

Assigning causation with observational studies isn’t typically possible, so it’s important to understand the study's limitations. For example, we don’t know if people who were taking hormone therapy after the age of 65 were more likely to have a healthy diet, were more likely to exercise, or were more likely to be screened for heart disease. We don’t know if the group of women who received hormones were overall at lower risk for cardiovascular disease before they started hormone therapy. We don’t know if doctors who prescribe hormone therapy are more up to date, and so their patients receive more meticulous health screening and care. As there is data emerging on improved health outcomes with women physicians over male physicians, might women doctors be more likely to prescribe MHT?

My general rule of thumb with observational studies is to think of them as puzzle pieces. They are helpful but cannot tell the whole story. And we rarely (ever?) change medical practice based on a single observational study.

The Findings

The next few sections will be pretty in-depth, and I certainly felt like this meme while reviewing all the tables in the supplemental data. If you want it all, carry on. If you just want the conclusions, head to the end and the “Putting it All Together” section.

There was a lot of data: 1,522,914 women who used menopause hormone therapy at some point and 9,421,414 who did not. This is impressive. 

There was a significant difference in the results for estrogen alone versus estrogen and a progestogen. Remember, for those with a uterus, a progestogen (or bazedoxifene) is required to prevent endometrial cancer (this study did not look at bazedoxifene). 

As the findings are somewhat complex, given the permutations and combinations of doses as well as routes of therapy (oral vs. transdermal vs. vaginal vs. injection), for simplicity’s sake, I am going to divide this study into two posts. This post will focus on the estrogen-only arm, and the next post (hopefully will be out in the next day or two) will discuss the estrogen and progestogen arm. 

Overall Mortality

Using estrogen therapy (ET) without progesterone or progestin reduced the risk of mortality by 19% over the study period compared with no estrogen. 

All the estrogens reduced mortality, with estradiol at 21% better than Premarin at 13%. Here are the decreased risks of mortality broken down by estrogen delivery method:

• Oral: decrease by 11%

• Transdermal: decrease by 20%

• Vaginal: decrease by 30% 

The benefit of transdermal versus oral fits with other observational data; however, the vaginal route showing the lowest rate of mortality needs more analysis because this bucket includes therapies that don’t enter the blood and those that do. Let’s look at the supplemental data and reduction in mortality:

• Low dose vaginal (tablets and the ring): 28% reduction 

• Medium dose vaginal: 38% reduction

• Low dose transdermal: 22% reduction 

• Medium dose transdermal: 25% reduction

How does vaginal estrogen that doesn’t enter the blood (vaginal tablets and ring) perform better than estrogen that enters the blood from a patch? This doesn’t make sense biologically based on what we know or what we think we know. 

Could vaginal estrogen result in that many fewer urinary tract infections such that the mortality rate is so reduced (urinary tract infections can lead to falls and bloodstream infections) that the benefit is as good as giving estrogen that enters the blood, or is this a spurious finding? This is why it’s important to be cautious about interpreting observational data, as here, we have some difficult results to explain. 

Overall, the data for mortality generally favored transdermal, estradiol, and medium to lower doses. This supports what we currently believe about hormone therapy for younger women. 

Cancer 

The study examined five cancers: breast, colon, lung, endometrial, and ovarian. We assume that most people who are taking estrogen alone have had a hysterectomy, so, of course, endometrial cancer rates will be lower. One can hardly attribute that to the medication. Also, we know from clinical trials that estrogen by itself is associated with an increased risk of endometrial cancer (read this to learn more), so the idea that giving estrogen without progesterone or progestin lowers the rate of endometrial cancer is absurd. Also, many people who have had a hysterectomy will have had their ovaries removed, which would lower the rate of ovarian cancer. Therefore, I am only going to focus on breast, colon, and lung cancers. 

Overall, the rate of breast cancer was 16% lower for people taking estrogen therapy versus those taking no hormones. In the long-term follow-up of the WHI, there was a 21% risk reduction of breast cancer risk associated with 0.625 mg of Premarin, and in this study, the risk of reducing breast cancer with 0.625 mg of Premarin was 26%. So, similar to the mortality data discussed above, the outcomes of the observational data looked better than those of the randomized clinical trial. One explanation is that the study may have a small healthy user bias (or other bias). 

The greatest benefit for breast cancer reduction was seen with Premarin, a 23% reduction, versus 12% for estradiol. By route, the reduction in breast cancer risk was 23% for oral, 14% for transdermal, and 5% for vaginal. If these results are correct, then estradiol offers a benefit for breast cancer prevention but is not as robust as Premarin.

Premarin is definitely the winner for breast cancer prevention, and I guess that this is likely because it isn’t just estrogen; some of the compounds are selective estrogen receptor modulators. Still, the most common regimen that we prescribe, transdermal estradiol, was associated with a 14% reduced risk of breast cancer, and the risk was pretty similar whether a low, medium, or high dose was used. 

The conclusion here is that long-term use of estrogen does not appear to increase the risk of breast cancer. (Note: this applies to estrogen-only preparations, not combinations that include progesterone or progestin). 

Cardiovascular Disease

The researchers evaluated six different kinds of heart disease: ischemic heart disease, heart failure, atrial fibrillation, heart attack, blood clots (also known as thromboembolism), and stroke. 

Regarding ischemic heart disease (IHD), there was a 4% increased risk associated with estrogen therapy. The authors point out that IHD is the most common cause of heart disease, and in the study, 1.6 million women experienced IHD, so if a 4% increase were a true effect, it could result in 64,000 additional cases a year. The increased risk was essentially the same for both Premarin and estradiol. The risk was very much related to the delivery method, with injections associated with a 17% increased risk, oral 4%, vaginal 3%, and transdermal 1%. 

I have some thoughts here. These results really raise alarm bells about estrogen injections, which are never recommended. Also, low and medium doses had a minimal increased risk of IHD, but high-dose estrogen resulted in a 9% increased risk. Again, this largely supports what we believe: the lowest dose is likely the safest for the heart. 

Other heart conditions were reduced with estrogen: congestive heart failure was reduced by 5%, atrial fibrillation by 4%, and heart attack by 11%. Transdermal therapy was associated with the lowest risk. 

There was a 3% reduction in blood clots, which was essentially the same for transdermal vs oral therapy. This is a perplexing finding as there is quite a lot of data showing a lower risk of clots with transdermal therapy versus oral. Another finding that really raises eyebrows is that low-dose vaginal estrogens were associated with a 6% lower risk of blood clots. This doesn’t seem biologically plausible and does raise concerns that many of those using estrogens may be healthier in other ways that can’t be accounted for within the study. Again, this points out the limits of an observational study. 

Stroke and Dementia

With estrogen therapy, there was a 1% increase in stroke, which is within the margin of error (which means it is not possible to say if it is a true effect or not). The increased risk of stroke with Premarin was 5%, but estradiol showed a 1% decrease (which was within the margin of error). High-dose estrogen therapy increased the risk of stroke by 8%; the standard dose was neutral, and for the lowest dose, there was a 6% reduction. Oral and injection had a 6% increased risk, and both transdermal and vaginal therapies were in the 5-6% reduced risk range. 

However, looking closer at the vaginal therapies and reduction in stroke risk raises some questions:

• Low dose vaginal: 8% reduction

• Medium dose vaginal: 4% reduction (not statistically significant)

• Low dose transdermal: 10% reduction

• Medium dose transdermal: 7% reduction

The medium-dose vaginal therapy should theoretically be delivering the same dose of estradiol to the blood as medium-dose transdermal therapy, and yet the results are different. And the mechanism by which low dose vaginal estrogen reduces stroke is a mystery, unless, of course, strokes related to bladder infections are far more common than we think. 

There was an overall 2% reduced risk of dementia with estrogen, with Premarin giving a 1% increased risk and estradiol a 3% reduced risk. The lowest risks for dementia were seen with transdermal and lower-dose therapies. Looking at transdermal therapy, a low dose was associated with an 11% reduced rate of dementia, a standard dose 8%, and a high dose a 6% reduction. Low-dose oral estradiol was associated with a 4% reduction in risk of dementia, a medium dose with no change, and a high dose with a 7% increased risk.

When we evaluate the vaginal versus the transdermal therapies for dementia, there are, again, some issues. Here are the rates of dementia by therapy:

• Low dose vaginal: 12% reduction

• Medium dose vaginal: 1% reduction (within the margin of error, so neutral)

• Low dose transdermal: 11% reduction

• Medium dose transdermal: 8% reduction

How can estrogen entering the blood from medium-dose transvaginal therapy provide no benefit for dementia, while estrogen entering the blood from transdermal therapy offers a benefit? And how can low-dose transvaginal therapy offer as much benefit as low-dose transdermal? This doesn’t seem biologically plausible; unfortunately, the authors offer no explanation. 

Limitations

There are some findings that match clinical trial data, such as the benefits of Premarin. Still, there are also findings with the low-dose vaginal estrogens that are difficult to reconcile. Another example that I haven’t touched on already is how low-dose vaginal therapy reduces the risk of lung cancer by 19%?

None of these limitations means the study is good or bad; it's just that it’s important to be aware of the limitations and understand that drawing firm conclusions about many of the health outcomes is challenging. Overall, the greater the benefit seen, the more likely it is to be true, but given findings like low-dose vaginal therapy reduced the risk of lung cancer by 19%, it does suggest there is probably some healthy user bias or another confounder that isn’t accounted for. Or maybe we have no idea about what low-dose vaginal estrogen can achieve!

Also, we don’t have data on years of use. We know the average follow-up for those who were followed to death was 4.1 years, but the authors don’t break down the data by years of use, so the risk increase/reduction is for the study period. 

Putting it All Together

In this study, estrogen therapy after age 65 was not associated with increased health risks and helps us reassure people taking estrogen, whether it is transdermal, vaginal (low dose or higher doses), or oral, that continuing beyond age 65 seems reasonable when desired.

For several health outcomes, the higher the dose of estrogen, the greater the risk, so this supports the overall belief that people should aim for the lowest dose that controls symptoms. While many people may need doses of 2 mg oral estradiol or 100 mcg transdermal (meaning high doses) to control symptoms early on, it seems the philosophy of trying to reduce people to at least a medium dose over time seems prudent. My practice for people who want to continue estrogen over age 65 is to choose dose equivalents that are 50 mcg transdermal/1 mg oral estradiol/Premarin 0,625 mg or lower. 

Estrogen therapy alone is associated with a lower risk of breast cancer, with Premarin showing the greatest benefit with a 23% reduction (Premarin results were close to the WHI findings). However, transdermal estradiol was associated with a 14% reduction in breast cancer risk. As expected, low-dose vaginal estrogen, which should not be absorbed, was neutral from a breast cancer reduction perspective. Colon and lung cancer reduction was fairly similar across all the estrogens. The reduction in lung cancer with low-dose vaginal estrogen doesn’t seem biologically plausible. 

Improvement in heart disease was less certain. Estrogen therapy raised the risk of the most common cardiovascular condition, ischemic heart disease, but lower doses of oral and transdermal estrogen were essentially neutral. The other cardiac outcomes were either slightly beneficial with estrogen therapy or had no effect. This study largely supports the hypothesis that low and medium-dose estrogen therapy is neutral for the heart of women who started taking estrogen before age 65. There wasn’t a consistent, strong signal for benefit.

Some of the data for estrogen to prevent stroke or dementia was a little conflicting. If there is a benefit, it is limited to the low and standard doses of transdermal therapy, but given some of the findings, I would caution any great conclusion about the benefit here. However, it’s very reassuring that the only increase in risk was seen with the higher dose oral therapies.  

People taking long-term estrogen alone can be fairly reassured that as long as they are on a medium or low dose, they are not likely to incur any significant risk. Transdermal may offer the most benefit/least harm, except for Premarin being superior for breast cancer prevention. Having this information should reassure women and allow them to consider doses and routes that may be the best for them based on their individual risk/benefit ratio.

In my opinion, this study doesn’t really provide evidence to support estrogen therapy to protect the heart or prevent dementia. It also doesn’t raise major concerns about risks, especially for those using low and medium doses.  

Some of the results raise questions, so I would exercise caution in making big claims about health benefits that aren’t supported robustly elsewhere. To be clear, this doesn’t make it a bad study; it’s just important to understand that some unknown factors may affect some of the outcomes. We need more studies to answer some of the other questions raised by the data. This is why observational studies can be kludgy.

My big takeaway? This study did not find a big safety signal for concern for women who continue estrogen after age 65, which is good. And the best outcomes were associated with low and medium doses of estrogen and transdermal and vaginal therapy.

References

Baik, SH, Baye F, McDonald CJ. Use of menopausal hormone therapy beyond age 65 years and its effects on women's health outcomes by types, routes, and dosesMenopause ():10.1097/GME.0000000000002335, April 9, 2024. | DOI: 10.1097/GME.0000000000002335

NAMS Position Statement. The 2022 Hormone Therapy Position Statement of The North American Menopause Society. Menopause. 2022;29:767-794.

Cynthia A. Stuenkel et al. Treatment of Symptoms of the Menopause: An Endocrine Society Clinical Practice Guideline, The Journal of Clinical Endocrinology & Metabolism, Volume 100, Issue 11, 1 November 2015, Pages 3975–4011, https://doi.org/10.1210/jc.2015-2236.