In this population study, use of MHT showed no association with sleep problems in adjusted analyses, while using sleep medication, reporting anxiety/depression, smoking and alcohol use, doing daily exercise and lower levels of life-satisfaction were associated with more sleep problems. Reporting good health indicated less sleep disturbance.
The relatively low rate of MHT use evident in the present data from 2007 (7.6%) may be influenced by the findings from the Women Health Initiative (WHI) trial [28] and the Heart and Estrogen/progestin Replacement Study (HERS) II [29], which raised concerns about the risks and benefits of MHT. These studies reported a significant increase in breast cancer for users of MHT, without the expected primary preventive effect on cardiovascular disease and other chronic conditions. The studies resulted in a more cautious prescribing practice among doctors [30]. In the years following the WHI and HERS II studies, use of MHT declined from 44% of US women using or having used MHT in 1988–1994 to 4.7% of women in 2010 [31]. Norwegian gynaecologists’ attitudes toward MHT also shifted from being quite liberal to prescribe MHT, to becoming rather reluctant, after the aforementioned studies were published [32, 33]. Clinicians applied more strict indications for MHT, prescribed lower doses, and shortened the treatment duration of MHT [30]. The uncertainty regarding the actual cost vs. benefit to the user probably left many women reluctant to take MHT, or desisted from this type of treatment [33]. In more recent years, however, opinions regarding use of MHT have shifted away from the scepticism that came with the first WHI publications, as a result of new research, including a recent study on the long-term effects of the WHI study [34]. The latest recommendations from the North American Menopause Society [35] support use of MHT for treating bothersome VMS, but not for disease prevention. Consequently, our interpretations of the results bear these two perspectives, both historic and current, in mind.
There are some possible explanations for the lack of association between use of MHT and sleep in the present study. Our findings may confirm previous indications that MHT does not influence sleep quality: The Wisconsin Sleep Cohort Study (2003) of 589 premenopausal, perimenopausal and postmenopausal women found that postmenopausal women had the best sleep and this was somewhat worsened in women who were taking MHT [36]. In that study, the authors did not find evidence that hot flashes caused sleep disturbances. Other researchers also claim that the relationship between VMS and sleep disturbance during menopause is not well defined, and sleep problems are not necessarily due to VMS [37]. On the other hand, several studies report that sleep problems related to vasomotor symptoms can be improved with MHT [38]. Most trials comparing MHT with placebo have shown improvement in perceived sleep quality and self-reported sleep problems in women with VMS at baseline [18]. Other studies, such as an RCT comparing a low-dose antidepressant and low-dose MHT with placebo in menopausal women with hot flashes, showed that both medications modestly reduced symptoms of sleep problems compared to placebo [39]. The strong association between use of prescribed sleep medication and sleep problems in the present study probably reflects that women seek other treatments for their sleep difficulties, such as hypnotic drugs. This is in line with an earlier research that identified drug use (hypnotics and hormone therapy) as risk factors for sleep disturbance. Such use is suggested being caused by women with most menopausal symptoms, who despite a proven effect of the medication, still have more symptoms than observed in the general population [7].
In the present study, the reports of VMS in MHT users were high, with more than half of the women having major symptoms. The survey question in the HUNT3 was posed as the following; “Do/did you perceive hot flashes during menopause?” Positive answers could indicate that the women had a history of VMS without medication, giving a justification for their present prescription. The ambiguity of referencing either the past or the present time in the question gives support to this interpretation. A more likely explanation is that many women remain untreated, or do not receive adequate medication despite climacteric symptoms.
Of interest is that use of systemic hormone therapy was highest in the age group 55 to 64 years (10.4%), and was still used by women in the 65–75 years age group. Some may find it difficult or unnecessary to quit, as most women experience a reappearance of symptoms after treatment cessation [40]. When hormone therapy is recommended for women younger than 60 for the treatment of VAS and bone loss, for women aged 60 or older the risks of heart disease, stroke, venous thromboembolism, stroke and dementia become greater [35]. The 2017 hormone therapy position statement from The North American Menopause Society [35] recommends that clinicians individualize their decision-making, determine the proper dosage, and actively include the women in the shared decision-making process. Menopausal symptoms that interrupt sleep may be more troublesome than daytime symptoms and this should be considered when targeting therapy [41].
Anxiety and depression were significantly associated with poor sleep quality. Previous studies have identified mood disturbances as a strong predictor of poor sleep [42], and the relationship between sleep disturbance and depressed mood is likely bidirectional [43]. Although it is postulated that periods with hormone fluctuations across the female reproductive lifecycle represent ‘a window of vulnerability’ for depression [44], the low anxiety and depression scores reported in the present study are in line with the research literature that most women do not experience depression during menopause or in their midlife years [45].
In the adjusted analyses, our data showed no association between moderate frequency of exercise and sleep, except that doing daily exercise was associated with more sleep disturbance. Experience has also shown that daily exercise not necessarily remove vasomotor symptoms or provide good sleep. Earlier research about the effect of exercise on vasomotor symptoms and sleep is also inconclusive, and in a Cochrane review investigating the effectiveness of exercise in the treatment of VAS, the interventions differs in the different studies, due to e.g., content, intensity, frequency, and length [46]. According to the same Cochrane review, only a single, small study suggested that MHT is more effective than exercise to moderate VAS symptoms [46]. More research is needed within this area.
Strengths and weaknesses
A strength of the study is the population-based approach with the inclusion of 13,060 women with self-reported data on sleep. Further, the study included important socio-demographic information, lifestyle, self-perceived health, and psychological factors, allowing for thorough adjustment of potentially confounding variables. Another strength is the use of a complete national prescription database (NorPD), which is not influenced by recall-bias and non-response.
However, our study also has some limitations. First, we were unable to exclude women for which MHT is contra-indicated, such as those with breast cancer or thrombosis, because these medical conditions are not specified in the HUNT database. Even though we lack this information, the numbers would be quite low, and we consider the risk of bias by this factor to be small. Another limitation is that we cannot be certain that the medication was actually taken as prescribed, but we have ensured that the prescriptions were filled at the pharmacy. Low adherence to MHT arises mainly from concerns about possible adverse effects [47]. Moreover, we do not have data on prescribed daily doses. However, women who experience a good therapeutic effect would most likely take their prescribed medication. Lastly, we don’t actually know whether the women were pre-, peri-, or post-menopausal, and whether MHT might have different associations with sleep in these different groups. The age groups were a proxy for these stages, but exact determination was not feasible for this type of population-based study.
