Postmenopausal hormone therapy and mortality before and after the Women’s Health Initiative study
We observed a diminished systemic HT prevalence and an elevated mortality danger for systemic HT end users in the two study populations in the wake of the 2002 WHI publication with the most pronounced inclination offered in the singleton populace. The conclusions suggest an alteration in the HT user profile just after 2002 with a unique pattern of HT use, most likely owing to the healthiest users determining to either fall systemic HT or switching to local HT, as tips improved subsequent the WHI publication. It highlights the value of satisfactory baseline characteristics when inspecting HT use, as confounders may possibly differ markedly with altering HT person profile over time.
In line with earlier reports, we noticed a drop in systemic HT prevalence concerning 2000 and 2005 for systematic HT people in the menopausal or postmenopausal age9,10,11,12. The WHI publication in 2002, and subsequent media consideration and alteration in HT prescription guidelines, is generally deemed the principal explanation for the observed decrease, as systemic HT was recommended kept in the lowest probable dosage for the shortest amount of money of time and was not to be utilised by asymptomatic girls15,23,24.
Scientific trials evaluating HT users to non-people uncovered no association in between systemic HT and all-induce mortality3. This is supported by a Danish observational study of HT initiated ahead of 2002, which observed no association concerning systemic HT and general mortality in a large cohort of just about 30,000 girls25. But, meta-analyses of both scientific and observational studies found a lessened danger of all-result in mortality, if systemic HT was initiated at age < 60 years26,27. This finding aligns with our study, as we observed a lower risk of all-cause mortality for systemic HT users in the age group 56–60 in 2000, before the WHI publication. However, a change occurred in 2005, after the WHI study, as the mortality risk among systemic HT users was like that of non-users, supporting a selection hypothesis rather than causality.
A smaller German observational study examined differences in the pre- and post-WHI HT user profile shortly after the 2002 WHI publication and found a decline in prevalence especially amongst women with higher social status, lower body mass index (BMI), and healthier lifestyle28.
This aligns with a Canadian study also performed shortly after the 2002 WHI publication, which indicated a shift in the profile of HT users. A decline in prescriptions was observed and gynaecologists now preferred to prescribe lower doses of systemic HT if necessary. Women who were prescribed HT in the year after the WHI publication had fewer medical visits on average (7.47 vs. 6.36), yet a higher number of different classes of drugs per month (1.07 vs 1.18)29.
Our study, designed to examine long term outcomes and HT user differences in a large study population, supports these previous findings, as we observed a decreased prevalence, changing mortality risk from lower than to similar to that of the background population, and increased years of education for systemic HT users between 2000 and 2005, which altogether suggests an alteration in the systemic HT user profile following the 2002 WHI publication.
Danish twins have previously shown to have a mortality rate similar to that of the background population30. While the association between HT and mortality in the twin study population showed no clear trend across all age groups, likely due to the small sample size, it did show an increasing mortality for systemic HT users aged 56–60 between 2000 and 2005.
Alignment of the findings from the singleton and twin study populations suggests that the increased mortality after 2002 is due to a selection rather than causal effects. The selection may be a result of the abrupt discontinuation of systemic HT following the WHI publication, indicating a paradigm shift away from the otherwise generalised perception of systemic HT users being healthier than non-users13,28,31,32. Our study supports this hypothesis, as we observed an increased mortality risk after the 2002 WHI publication, but it must also be mentioned that the mortality risk is still close to one. So even if the post-WHI HT user is considered unhealthier, the overall mortality risk is not increased compared to the background population.
The strengths of our study include the linkage of nationwide registries enabled by the unique personal identification number, which allows us to study data from the universal healthcare system on a large, random sample from the general Danish population and Danish twins. This, combined with the access to information on education, provided a favourable setting for investigation of the association between HT and all-cause mortality as the registries provided a long and full follow-up with minimal room for selection bias. A major strength is the access to a unique study population of Danish twins provided by the DTR enabling the exposure-discordant twin design, which controls for potential genetic confounding and shared environmental confounding17.
Some limitations must also be mentioned. Our findings of systemic HT users being better educated than non-users and being slightly better educated after the 2002 WHI publication indicates a shift in the HT user profile. Our study could have benefitted from additional information on lifestyle and menopause e.g., age of menopause, smoking, and BMI, to illuminate other potential differences in HT user profiles. Another limitation was lack of statistical power in the twin study population, which hindered a further examination of HT user category differences. Lack of statistical power in both study populations further prevented us from dividing HT exposure into dosage and regimen. There is also the possibility of left truncation bias, as the DNPR only contains information on HT exposure from 1995 onwards. Although the DNPR provides excellent assessment of prescriptions and complete national coverage21, we only have information on redeemed prescriptions and thus have no indication of the HT user’s compliance. This could potentially overestimate the use of HT.
Following the 2002 WHI publication, we found a decreased systemic HT prevalence and a change from lower to similar mortality risk compared to the background population for systemic HT users within the singleton and twin study populations. These findings suggest an altogether different HT user profile with a different pattern of HT use, perhaps driven by the healthiest users deciding to either drop systemic HT or switching to local HT as recommendations changed following the WHI publication. Mortality is a crude outcome and is, in this study, used as a summary measure of health and could in future studies be complemented by suggested HT user alteration in other measures of health and lifestyle. Our study highlights the importance of examining, in future studies, not only the differences between HT users and non-user, but also the differences between HT users before and after 2002, as they may vary on subtle and not easily assessed health and lifestyle risk factors, which may be related to initiation, regimen and dose of HT28,33. This emphasises that confounder control is required when investigating the influence of HT, but also that confounders may alter with the altered HT user profile before and after 2002.