*Correspondence: Dr E. Banks, Cancer Research UK Epidemiology Unit, Radcliffe Infirmary, Gibson Building, Oxford OX2 6HE, UK.
Objective To describe national patterns of use of hormone replacement therapy (HRT) in 1996–2000.
Design Population-based study of women aged 50–64.
Setting England and Scotland.
Population A total of 1,091,250 women were recruited between May 1996 and December 2000.
Methods Women invited for screening at 66 NHS breast screening units were sent a questionnaire to complete before they were screened; 71% of women screened participated.
Main outcome measures Prevalence of use of HRT.
Results Overall, 33% of women reported that they were currently using HRT, the average duration of use being 5.8 years; 50% were ever-users. Current use of HRT was about twice as common at age 50–54 than 60–64, but varied little by time or region, the prevalences being 33%–34% in each year from 1996 to 2000; 30% in Scotland and 35% in southeast England. The prevalence of current use of HRT varied substantially depending on the woman's history of gynaecological surgery and past health, and was as follows in women with a history of: bilateral oophorectomy (66%); hysterectomy without bilateral oophorectomy (48%); neither hysterectomy nor bilateral oophorectomy (27%); breast cancer (6%); other cancer (25%); stroke (24%); venous thromboembolism (24%); diabetes (25%); heart disease (31%); or hypertension (31%). There was considerably less variation in the prevalence of use of HRT according to deprivation index, education, parity, body mass index, exercise, smoking or alcohol consumption.
Conclusions HRT is currently used by around one-third of women aged 50–64 in Britain and appears to be influenced considerably more strongly by a woman's medical and surgical history than by socio-economic or lifestyle factors.
Use of hormone replacement therapy (HRT) in the UK increased rapidly during the early 1990s, from about 10% in 1990 to 30% in 1995, among women aged 50–641. However, there are few national data on the use of HRT, on recent trends in use, or on the characteristics of users. This article describes the patterns of use of HRT among 1,091,250 women recruited into a population-based study of women aged 50–64 between May 1996 and December 2000.
The Million Women Study is a population-based multicentre prospective study in the UK, the main aim of which is to investigate the effects of HRT on health. The methods are described in detail elsewhere2. In brief, all 50- to 64-year-old women in the UK who are registered with a general practitioner are invited to attend the National Health Service Breast Screening Programme (NHSBSP) once every three years, at one of about 100 breast screening units throughout the country. Since 1996, women invited to screening at 66 participating breast screening centres received a questionnaire for the Million Women Study accompanying their invitation. If they wished to join the study, they were asked to return a completed questionnaire when they attended for screening. Collaborating screening centres are listed at the end of this article and their location is shown in Fig. 1.
The self-administered recruitment questionnaire includes questions on lifestyle and socio-demographic factors, reproductive factors, past health and use of HRT. Signed consent for follow up is requested. The questionnaire can be viewed at the study website 〈http://www.millionwomenstudy.org〉. The confidential completed questionnaires are sent from the breast screening units to the Million Women Study co-ordinating centre in Oxford, where they are checked and data are entered by scanning, using dedicated operator-verified data capture software (Eyes and Hands, Readsoft). An estimated 71% of those attending routine breast screening at participating centres returned a completed questionnaire2.
The data presented here relate to the participants' status at the time of recruitment. Women are selected for screening according to their year of birth and a small proportion (4.9%) are actually aged 49 or 65 at the time of routine screening. Such women are usually close to their 50th birthday or have just turned 65 at recruitment and so they are included here as being aged 50 and 64, respectively.
Current users of HRT are defined as those reporting that they were using HRT when they completed the questionnaire and ever-users are women reporting that they had ever used HRT. Women were asked which specific proprietary preparation of HRT they had used most recently, and for these analyses, the preparations are grouped as: oestrogen alone, oestrogen and progestogen or ‘other’ type of HRT (including tibolone, progestogen only, androgens and vaginal preparations), according to the hormonal constituents of each preparation listed in the British National Formulary3.
Women are classified as having had a ‘bilateral oophorectomy’ if they reported a bilateral oophorectomy, with or without hysterectomy. They are classified as having had a ‘hysterectomy without oophorectomy’ if they reported having had a hysterectomy but not a bilateral oophorectomy. Women who did not report having had either a bilateral oophorectomy or a hysterectomy are classified as having ‘neither hysterectomy nor bilateral oophorectomy’. Participants' medical histories were classified according to their reports of ever having had breast cancer, stroke, thromboembolism, diabetes, heart disease and hypertension (except hypertension during pregnancy). Women reporting that they had had cancer (other than breast cancer or non-melanoma skin cancer) are classified as having had ‘other cancer’. These categories are not exclusive as women can report more than one condition.
Two indices relating to social class were available. One variable is the reported highest level of attained education and the second is the Townsend deprivation score. The Townsend score is based on each participant's postcode and this score was classified into tertiles for these analyses4.
Initial analyses relate to the patterns of use of HRT in the study population as a whole and according to age and history of hysterectomy and/or oophorectomy. The prevalence of current use of HRT is then presented in relation to participants' medical history, markers of use of health services and lifestyle factors, adjusting for age (50–54, 55–59, 60–64), past hysterectomy/oophorectomy (bilateral oophorectomy, hysterectomy without oophorectomy, neither) and deprivation index (Townsend score, in tertiles).
Logistic regression models are used to estimate adjusted prevalences by setting each adjustment covariate to its mean value based on the observations used in the analysis. Observations with missing values are not included in the analyses except in logistic regression models where they are categorised as ‘unknown’. Differences between prevalences are assessed using log-likelihood χ2 tests. Standard errors (SE) are reported since, due to the large sample size, most confidence intervals are extremely narrow. All analyses are carried out using version 5.0 of the STATA statistical package for Windows and the adjusted prevalences were calculated using the ‘adjprop’ function of this package5.
These analyses are based on the reports from a total of 1,091,250 women, aged 50–64, who were recruited into the Million Women Study between May 1996 and December 2000.
Overall, 50% (SE 0.1%) of the women reported that they had ever used HRT and 33% (SE 0.1%) reported that they were current users (Table 1). The mean duration of use of HRT was 4.9 years (SE 0.01) in ever-users, 5.8 years (SE 0.01) among current users and 2.9 years (SE 0.01) among past users. The prevalence of current use of HRT remained relatively constant, at 34% (SE 0.3%) among women recruited in 1996, 33% (SE 0.1%) among those recruited in 1997 and 1998 and 34% (SE 0.1%) among those recruited in 1999 and 2000. The proportion of past users increased from 14% (SE 0.2%) in 1996 to 19% (SE 0.1%) in 2000. The proportion of women who were currently using HRT showed a slight north–south gradient across the country, with the prevalence of use being highest in southeast England (35%, SE 0.1%) and lowest in Scotland (30%, SE 0.2%) (Fig. 1).
Table 1. Distribution of characteristics related to use of HRT in study participants. Values are given as n (%).
Hysterectomy without oophorectomy
* 98% of women reporting bilateral oophorectomy also reported having had a hysterectomy.
** Includes tibolone, progestogen only, androgens, vaginal preparations.
† Numbers do not always add up to totals due to missing values.
Age at starting HRT use (ever-users)
Mean age (SE)
Duration of HRT use (ever-users)
Mean duration in years (SE)
Time since cessation of HRT use (past users)
<5 years ago
5–9 years ago
Mean years since stopping (SE)
Type of HRT use (current users)
Oestrogen and progestogen
The prevalence of HRT use was strongly related to whether or not a woman had had a hysterectomy and/or bilateral oophorectomy (Fig. 2) and to her age.
Overall, 66% (SE 0.2%) of women who reported having had a bilateral oophorectomy were currently using HRT, compared with 48% (SE 0.1%) of those who reported hysterectomy without oophorectomy and 27% (SE 0.1%) of those who did not report having had either of these operations (Table 1). The duration of use of HRT was also related to gynaecological surgery, the average in ever-users being 6.7 years for women reporting a bilateral oophorectomy, 5.7 years for women reporting a hysterectomy without oophorectomy and 4.1 years in women who had neither operation. Furthermore, the age pattern of use differed significantly according to the type of surgery the participant had experienced (Fig. 2). On average, women with a bilateral oophorectomy began use at an earlier age and reached a higher prevalence of use than those with intact ovaries and uterus; women with a hysterectomy without oophorectomy had intermediate values. Overall, women aged 50–54 reported about twice the prevalence of current use of HRT than women aged 60–64.
The onset of use of HRT was closely linked to when a woman ceased menstruating, as well as to the type of gynaecological surgery she had had. Among women reporting that they had ever used HRT, 40% (SE 0.1%) reported commencing use of HRT prior to ceasing menstruation. Among women who had started HRT use after ceasing menstruation, the proportion of ever-users whose use began within five years of their periods ceasing was 82% (SE 0.2%) for women who had had a bilateral oophorectomy, 34% (SE 0.2%) for women with a hysterectomy without oophorectomy and 77% (SE 0.1%) for women with a natural menopause.
Among women who reported current use of HRT, 41% (SE 0.1%) specified use of preparations composed of oestrogen(s) without progestogen, 53% (SE 0.1%) specified use of combined oestrogen and progestogen and 6% (SE 0.04%) reported use of preparations containing other types of hormones [including 0.4% (SE 0.01%) who reported use of vaginal preparations] (Table 1). However, the type of preparation in current use was closely related to whether or not the respondent reported having had a hysterectomy. The vast majority (98%) of women who had a bilateral oophorectomy also had a hysterectomy and 93% (SE 0.1%) of current users who had had a bilateral oophorectomy reported use of preparations containing oestrogen alone. Similarly, among women who had had a hysterectomy without bilateral oophorectomy, 94% (SE 0.1%) of current users specified use of preparations containing oestrogen(s) alone. By contrast, among women who did not report having had a hysterectomy or bilateral oophorectomy, 96% (SE 0.04%) of current users reported use of combined oestrogen and progestogen or ‘other’ preparations.
The prevalence of current use of HRT was significantly related to deprivation score; it was 34% (SE 0.1%), 33% (SE 0.1%) and 30% (SE 0.1%) among women from the least deprived, medium deprived and most deprived areas, respectively (χ23 for heterogeneity adjusted for age and previous hysterectomy/oophorectomy = 565.7, P < 0.00001). The prevalence of current use of HRT according to a woman's medical history, markers of use of health services and lifestyle factors is shown in Fig. 3. The results are adjusted for age, previous hysterectomy/oophorectomy and deprivation score. Figures 4, 5 and 6 show results separately according to whether or not a woman had had a bilateral oophorectomy, hysterectomy without oophorectomy or neither of these operations, respectively.
Current use of HRT was strongly related to a reported history of cancer and/or conditions related to cardiovascular disease (Fig. 3). The prevalence of current use of HRT was only 6% among women who had had breast cancer in the past and 27% (SE 0.1%) among women reporting that their mother or sister had a history of breast cancer. The prevalence of current use of HRT was around 25% in women reporting a history of cancer (other than breast cancer and non-melanoma skin cancer), diabetes, stroke or thromboembolic disease and around 31% in women reporting a history of heart disease or hypertension.
Among women reporting a history of cardiovascular disease or related conditions, the only indicator of disease severity available was whether or not they were currently receiving treatment for this condition. Among women with a history of diabetes, 24% (SE 0.3%) of those currently receiving treatment for diabetes reported current use of HRT, compared with 29% (SE 0.8%) of those not receiving treatment (χ21 for heterogeneity = 46.6, P < 0.00001). Among women with a history of thromboembolic disease, the prevalence of current use of HRT was 18% (SE 0.5%) among women reporting current treatment for it, compared with 24% (SE 0.2%) among those who were not being treated (χ21 for heterogeneity = 121.5, P < 0.00001). For women reporting heart disease, the prevalence of current use of HRT was 30% (SE 0.3%) among women reporting current treatment and 31% (SE 0.4%) among those not reporting current treatment (χ21 for heterogeneity = 3.6, P= 0.06). For women reporting a history of hypertension, the prevalence of current use of HRT was 31% (SE 0.2%) among those reporting current treatment and 30% (SE 0.1%) among those not reporting current treatment (χ21 for heterogeneity = 9.8, P= 0.002).
The prevalence of current use of HRT also varied significantly according to use of various medications or interventions, which can be taken to be markers of use of health services (Fig. 3). For example, women reporting that they had had a tubal ligation, had used oral contraceptives in the past or had attended breast cancer screening previously were more likely to be using HRT than women who did not report these interventions.
Current use of HRT was significantly more common in women who were more educated, less deprived, less overweight, drank more alcohol, smoked more and exercised more, than in women who did not report these levels of exposure. However, the use of HRT varied only slightly according to these socio-economic and lifestyle factors—with the most extreme variation being for alcohol consumption where the prevalence ranged from 27% to 38% across divisions of this factor (Fig. 3).
The pattern of variation in the prevalence of current use of HRT according to a woman's medical history, markers of use of health services and lifestyle was broadly similar for women who had had a bilateral oophorectomy, hysterectomy without oophorectomy or neither of these operations (Figs 4–6).
These results indicate that use of HRT is common among British women aged 50–64, that use has remained fairly constant from 1996 to 2000 and that there is little variation in the prevalence of use across the country. Furthermore, the prevalence of use of HRT is strongly related to a woman's age, previous gynaecological surgery and to her history of cancer or cardiovascular disease, but is less affected by socio-economic and lifestyle factors.
The data presented here are derived from self-administered questionnaires. Validation studies within this population indicate that the reported use of HRT and current treatment for conditions such as heart disease, diabetes, clotting disorders and hypertension is reliable; compared with general practice prescriptions, agreement for current use of HRT, including type of HRT used, is at least 95% (kappa statistic for agreement ranged from 0.90 to 0.96) and for current illnesses, ranges from 89% to 99% (kappa 0.53–0.92)6. Current use of HRT in women who participate in the Million Women Study is slightly higher than in all women aged 50–64, largely due to the higher prevalence of use of HRT in women of this age who attend for screening7 and, using these data, we estimate that the prevalence of current use of HRT is of the order of 25–30% among all women in Britain aged 50–64. Women who participate in the Million Women Study are similar to the general population from which they are derived in terms of their age and use of prescription medications for conditions such as hypertension, heart disease, diabetes, thyroid disease, asthma and depression/anxiety7. This similarity, along with the fact that internal comparisons within a cohort are less prone to distortion by selection than external ones, suggests that the main findings regarding the characteristics of users of HRT are generalisable to the broader population.
Although use of HRT in the UK increased rapidly during the early 1990s1, our data show that the prevalence of current use of HRT among women aged 50–64 has remained fairly constant from 1996 to 2000. There was slight regional variation only in the use of HRT, with the prevalence of current use being 30% in Scotland and 35% in southeast England. This apparent lack of substantial change since 1996 is supported by data from the UK General Practice Research Database, which showed that the prevalence of prescriptions for ‘sex hormones’ among women aged 45–64 (which would mostly be HRT) was 27% in 1996 and 28% in both 1997 and 19988.
Large variations in the prevalence of current use of HRT were seen according to age and previous gynaecological surgery. Use of HRT was more common, began earlier and was of longer duration among women reporting bilateral oophorectomy and/or hysterectomy compared with those who did not report these operations. Increased use of HRT among women with hysterectomy and/or oophorectomy has been noted before in studies from the UK and elsewhere9–13. In keeping with current clinical guidelines and the findings of previous studies11,14,15, whether or not the woman had had a hysterectomy is strongly related to the type of HRT preparation used.
Current use of HRT also varied considerably according to women's history of cancer or cardiovascular disease or related disorders. Current use was less common in women with a history of cancer or cardiovascular disease, or related disorders, such as diabetes or hypertension. Furthermore, among women with diabetes and thromboembolic disease, use of HRT was less common still if they reported currently being treated for the condition. A number of studies, mostly from the United States, have found that women who use HRT have a more favourable cardiovascular risk profile than non-users16–19. Many of these differences are present prior to commencement of use of HRT and are therefore likely to reflect selection to use, rather than a direct effect of HRT18,19. Most previous studies have tended to focus on factors such as cholesterol level, glucose tolerance, body mass index and lifestyle. The small number of studies that have examined use in relation to previous illness have had limited statistical power although their overall findings are in keeping with those shown here13,20–22.
Somewhat higher prevalences of current use of HRT were seen in relation to markers of increased use of health services, such as previous tubal ligation, oral contraceptive use and breast cancer screening. Use of HRT was also more common among women of higher socio-economic status, measured by both educational attainment and deprivation index. Following adjustment for deprivation, leaner women and women who exercised regularly were more likely to use HRT than women of higher BMI and those who did not report regular exercise. While some of these findings suggest that women who use HRT have a healthier lifestyle than women who do not, this does not hold true for all factors. Indeed, current and ex-smokers had higher adjusted levels of use of HRT than women who never smoked. The variation in current use of HRT seen according to previous use of health services, demographic and lifestyle factors was, however, of a lesser magnitude than that seen for factors such as age, gynaecological surgery and history of illness.
While the importance of lifestyle factors in determining the use of HRT and subsequent health is often stressed, the role of pre-existing disease has received far less attention. Yet, the presence or absence of conditions such as diabetes and hypertension exerts a stronger influence on the subsequent risk of disease and death than many lifestyle factors. This study shows that women with such conditions are less likely to be prescribed HRT, and that variations in the prevalence of use of HRT associated with medical and surgical history are considerably greater than the variations associated with different lifestyles. Studies investigating the relationship between the use of HRT and the subsequent risk of disease or death need to account properly for the pre-existing differences between women who do and do not use HRT. Biases resulting from the selective prescribing of HRT according to past health are potentially more serious than those resulting from differences in lifestyle.
NHS Breast Screening Centres collaborating in the Million Women Study (in alphabetical order) are as follows: Avon, Aylesbury, Barnsley, Basingstoke, Bedfordshire and Hertfordshire, Cambridge and Huntingdon, Chelmsford & Colchester, Chester, Cornwall, Crewe, Cumbria, Doncaster, Dorset, East Berkshire, East Cheshire, East Devon, East of Scotland, East Suffolk, East Sussex, Gateshead, Gloucestershire, Great Yarmouth, Hereford & Worcester, Kent (Canterbury, Rochester, Maidstone), Kings Lynn, Leicestershire, Liverpool, Manchester, Milton Keynes, Newcastle, North Birmingham, North East Scotland, North Lancashire, North Middlesex, North Nottingham, North of Scotland, North Tees, North Yorkshire, Nottingham, Oxford, Portsmouth, Rotherham, Sheffield, Shropshire, Somerset, South Birmingham, South East Scotland, South East Staffordshire, South Derbyshire, South Essex, South Lancashire, South West Scotland, Surrey, Warrington Halton St Helens and Knowsley, Warwickshire Solihull and Coventry, West Berkshire, West Devon, West London, West Suffolk, West Sussex, Wiltshire, Winchester, Wirral and Wycombe.
The Million Women Study Co-ordinating Centre staff are as follows: Simon Abbott, Emma Bailey, Krys Baker, Emily Banks, Isobel Barnes, Angela Balkwill, Valerie Beral, Judith Black, Kate Bonner, Anna Brown, Diana Bull, Andrea Burrile, Becky Cameron, Judith Clarke, Andrea Cliff, Barbara Crossley, Jane Dee, David Eastwood, Dave Ewart, Laura Gerrard, Jane Green, Elizabeth Hilton, Ann Hogg, Joy Hooley, Anna Hurst, Carol Keene, Caroline Kennedy, Nicky Langston, Christine Lewis, Joanne Mathie, Karen Pennington, Gillian Reeves, Simon Reid, Brian Roden, Emma Sherman, Moya Simmonds, Elizabeth Spencer, Alison Timadjer, Steve Warby.
The Steering Committee members are: Joan Austoker, Emily Banks, Valerie Beral, Ruth English, Julietta Patnick, Richard Peto, Gillian Reeves, Martin Vessey and Matthew Wallis.
The Writing Committee members are: Emily Banks, Isobel Barnes, Valerie Beral and Gillian Reeves.
The authors would like to thank the many women who are participating in the study. The contribution from many individuals at each of the collaborating NHS Breast Screening Centres (listed above) is also gratefully acknowledged. The Million Women Study is supported by Cancer Research UK, The Medical Research Council and the National Health Service Breast Screening Programme.