Postmenopausal hormone therapy and dementia, cognition, and mortality

Laura Løkkegaard Johansen

Research output: ThesisPh.D. thesis

40 Downloads (Pure)


As life expectancy increases, so does the concerns for the health and well-being at advanced ages. Especially, loss of cognitive function and risk of dementia are two concerns that pose great personal and societal consequences. So, prevention of such diseases is of increasing relevance, and one major focus hasbeen health in mid-life, as this may play an essential role in aging.A pivotal mid-life event for women is menopause, which is a phase potentially influenced by e.g., vasomotor, genitourinary, and musculoskeletal symptoms that can last well into the postmenopausal years. Away to relieve these symptoms is by using hormone therapy (HT) consisting of oestrogen with or withouta progestogen – a treatment that has been known and used for decades. HT was, due to findings fromobservational studies, generally believed beneficial in the relief of menopausal and postmenopausalsymptoms and disease prevention. However, this perception changed when findings from a large clinical,randomized trial, the Women’s Health Initiative (WHI) study, were published in 2002 concluding that therisk of using systemic HT outweighed the benefits of disease prevention. These findings prompted widespread media attention, and they led to alterations in HT guidelines, which caused a sharp decline in theprevalence of systemic HT.Several explanations for the discrepant findings on HT and disease prevention between observationalstudies and clinical trials have subsequently been posed. Both an introduction of a healthy user bias inobservational studies prior to the 2002 WHI publication and the timing of HT in relation to menopausehave been suggested as possible reasons. Environmental and genetic factors, not controlled for in theobservational studies, may also play a part.

Twin studies could help illuminate especially the latter hypothesis as twins present a unique study population due to shared genetic factors and early childhood environment. This thesis aimed to investigate the association between HT and the outcomes dementia, cognition, and mortality in a study population of
Danish female twins from the Danish Twin Registry and a 5% random sample of female singletons from the Danish background population. Both study populations had to meet the age criteria of being born before 1950 and being alive by 1995 to ensure a minimum age of 45 years and thus being close to or past the average menopausal age when information on HT became available from a Danish nationwide register on medicine. Information on dementia diagnoses and mortality was also obtained through Danish nationwide health registries with observation time beginning in 1995. Cognitive function was registered in the Danish Twin Registry and assessed as a part of The Longitudinal Study of Aging Danish Twins beginning in 1995 and the Middle-Aged Danish Twin Study beginning in 1998 using a cognitive composite score.

Paper 1 examined the association between systemic HT and dementia and found an overall borderline statistically significant tendency towards an increased risk of dementia for systemic HT users in both then study samples of twins and singletons. A statistically significant increase in risk of dementia for systemic
HT users was observed before the 2002 WHI publication in both study populations. This finding could indicate confounding by indication as alterations in guidelines after 2002 may have led to an exclusion of the most fragile women. It indicates a potential change in the HT user profile before and after the 2002 WHI publication.
As persistent mild cognitive impairment has shown a high risk of progressing to dementia, Paper 2 investigated the association between HT and cognitive function in Danish female twins and found that systemic HT users aged 70+ had a lower cognitive function than non-users in analyses adjusted for age, education, social class, and unobserved familial confounding. Longitudinal data on systemic HT users aged 70+ showed that the lower cognitive function was most explicit before 2002, whereas after 2002 the cognitive function was closer to that of non-users. This finding aligned with the tendency observed in Paper 1. However, longitudinal data in younger twins (aged 50-69) showed a lower cognitive function in systemic HT users after 2002 compared to non-users, and a change in the HT user profile seemed to have occurred as those who changed from systemic HT to local HT after 2002, or dropped it altogether, performed cognitively better within this age group. A likely explanation for the observed tendencies in this study is the introduction of a selection bias in the wake of the 2002 WHI publication. To further elaborate on these findings, Paper 3 investigated the association between HT and mortality in different age groups before and after the 2002 WHI publication in both a study population of twins and singletons. Analyses were adjusted for education and unobserved familial confounding, respectively. In both study populations, the prevalence of systemic HT decreased markedly following the 2002 WHI publication, while local HT use increased. Among twins aged 56-60, the mortality risk changed from lower before 2002 to similar to that of the background population for systemic HT users following the 2002 WHI publication – a tendency also observed in singleton females aged 56-75. These findings suggest an altogether different HT user profile after 2002, perhaps driven by the healthiest users deciding to either drop
systemic HT or switch to local HT as recommendations changed following the WHI publication.

Overall, we found it likely that selection rather than causality underlies the observed associations between HT and dementia, cognition, and mortality in Danish twins and singletons, based on studies spanning across year 2002, as the risk of the various outcomes for HT users appeared to differ before and after 2002. This implied that systemic HT users had undergone a selection following the 2002 WHI publication, potentially due to the alterations made in the guidelines on HT prescription after 2002. Our findings highlight the importance of examining not only the differences between HT users and non-user, but also the differences in HT users before and after 2002, and it emphasizes the importance of adequate confounder control in future observational studies when examining HT use, as confounders may vary markedly over time and may be related to initiation, regimen, and dose of HT. Alternatively, as we are now 20 years from the 2002 WHI publication, observational studies could focus solely on HT users after 2002, as they may provide a more stable base for generalisation of results to current HT users.
Original languageEnglish
Awarding Institution
  • University of Southern Denmark
  • Christensen, Kaare, Principal supervisor
  • Thinggaard, Mikael, Co-supervisor
  • Hallas, Jesper, Co-supervisor
  • Osler, Merete, Co-supervisor
Date of defence24. Oct 2022
Publication statusPublished - 20. Sept 2022

Note re. dissertation

Print copy of the full thesis is restricted to reference use in the Library.


Dive into the research topics of 'Postmenopausal hormone therapy and dementia, cognition, and mortality'. Together they form a unique fingerprint.

Cite this