1The Million Women Study Collaborators are given in the Appendix.
Valerie Beral, Cancer Epidemiology Unit, University of Oxford, Richard Doll Building, Roosevelt Drive, Oxford OX3 7LF, UK. Tel.: +44 1865 289658; fax: +44 1865 289610. E-mail: firstname.lastname@example.org
Summary. Background: Current use of menopausal hormone therapy (HT) increases the risk of venous thromboembolism (VTE) and the formulations used may affect risk.
Methods: A total of 1 058 259 postmenopausal UK women were followed by record linkage to routinely collected National Health Service hospital admission and death records. HT use and risk of VTE was examined using Cox regression to estimate relative risks (RRs) and 95% confidence intervals (CIs).
Results: During 3.3 million years of follow-up, 2200 women had an incident VTE, diagnosed, on average, 1.5 years after last reporting HT use. RRs in current vs. never users at last reporting varied by HT formulation: the risk was significantly greater for oral estrogen-progestin than oral estrogen-only therapy (RR = 2.07 [95%CI, 1.86–2.31] vs. 1.42 [1.21–1.66]), with no increased risk with transdermal estrogen-only therapy (0.82 [0.64–1.06]). Among users of oral estrogen-progestin, the risk from HT varied by progestin type, with significantly greater risks for preparations containing medroxyprogesterone acetate than other progestins (2.67 [2.25–3.17] vs. 1.91 [1.69–2.17]; Pheterogeneity = 0.0007). Current users of oral HT at last reporting had twice the risk of VTE in the first 2 years after starting HT than later (Pheterogeneity = 0.0006). Associations were similar for deep vein thrombosis with and without pulmonary embolism. Over 5 years, 1 in 660 who had never used HT were admitted to hospital for (or died from) pulmonary embolism, compared with 1 in 475 current users of oral estrogen-only HT,1 in 390 users of estrogen-progestin HT containing norethisterone/norgestrel, and 1 in 250 users of estrogen-progestin HT containing medroxyprogesterone acetate.
Conclusions: The risk of VTE varied considerably by HT formulation, being greatest in users of oral estrogen-progestin HT, especially formulations containing medroxyprogesterone acetate.
Randomized controlled trials and observational studies have demonstrated a clear increase in the risk of venous thromboembolism (VTE) (pulmonary embolism and/or deep vein thrombosis) with current use of hormone therapy for the menopause [1–5], hereafter referred to as hormone therapy (HT). Greater risks have been observed in users of combined estrogen-progestin therapy than in users of estrogen-only therapy [6–10] and with use of oral compared with transdermal therapy [11–13], but most of the previous studies included too few cases to allow a detailed investigation of specific HT formulations [2,3,14,15]. Little has been published on the risks associated with different types of progestins. We report here on the relationship between type of HT used and the incidence of VTE in a large UK cohort of postmenopausal women.
Data collection, follow-up and outcomes
The Million Women Study is a population-based prospective study that recruited 1.3 million women through National Health Service (NHS) breast screening clinics in England and Scotland in 1996–2001, representing one in four UK women aged 50–64 at this time. Women provided information on their use of HT, socio-demographic and anthropometric factors, and medical and reproductive history at recruitment, and a second questionnaire was sent to study participants 3 years later to update this information on HT use and other factors, with a 65% response rate. Study questionnaires can be viewed at http://www.millionwomenstudy.org.uk. The study has been approved by the Oxford and Anglia Multi-centre Research Ethics Committee and all study participants provided written consent.
All UK residents registered with the NHS are issued an NHS number, a unique personal identifier for all NHS health records. All study participants are followed by record linkage using their NHS number and other personal details for deaths , cancer registrations , emigration  and NHS hospital admissions [17,18], thus providing close to complete follow-up throughout the study period. The hospital admission data include details of admissions to NHS-funded hospitals, including inpatient (i.e. overnight) and day-case (i.e. not overnight) admissions, dating from 1981 in Scotland and 1997 in England. For every hospital record clinical diagnoses are coded using the World Health Organisation International Classification of Diseases 10th revision (ICD-10) , and all procedures are coded using the Office of Population Censuses and Surveys Classification of Surgical Operations and Procedures, 4th revision (OPCS-4) . For these analyses the main outcome, VTE, is defined as the first diagnosis following recruitment into the study of pulmonary embolism (ICD-10 I26) or deep vein thrombosis (I80-I82) as an inpatient/day-case hospital admission or as the underlying cause of death. To examine the reliability and completeness of hospital diagnoses we wrote to general practitioners (who hold a summary of all NHS medical records for their patients) of a random sample of about 1000 women with and 1000 women without a hospital record of VTE, asking whether they had a record of a VTE event over an 8-year period: 93% of hospital diagnoses were confirmed by the general practitioners and only three women with no hospital record of VTE were reported by their general practitioner to have had a diagnosis of VTE during the 8 years.
Women were excluded from analyses if: they were premenopausal or perimenopausal (as previously defined ) throughout follow-up; they had a history of cancer (except non-melanoma skin cancer, ICD-10 C44) at recruitment; they had reported a history of blood clots or treatment for clotting problems at recruitment; they had a hospital record for VTE before recruitment or had surgery in the 12 weeks prior to recruitment ; or HT use was unknown. Person-years were calculated from the date of recruitment (or for women who were not postmenopausal at recruitment, from the date when they were first defined as postmenopausal ) to whichever came first of: the date of first hospital admission for VTE; date of first cancer registration; date of death; date of emigration; or the end of follow-up – the last date of follow-up in these analyses was 31 December 2002 or 48 months after the last recording of HT use, whichever came first, because many women in the UK ceased HT use after publication of the first report of results from the Women’s Health Initiative randomized trial in 2002 [23,24]. Women were also censored at the date of their first inpatient/day-case operation because of the significant association between surgery and VTE, which may also result in a change in use of HT around the time of their operation. Person-years were calculated from 1 April 1997 for the small proportion of women (5%) recruited in England before that date, as hospital admission data were not available before then.
Cox regression was used to estimate the relative risk (RR) and 95% confidence interval (CI) of hospital admission or death for VTE in relation to use of HT, with attained age as the underlying time variable. Comparisons across categories of HT exposure were made using standard chi-squared tests for heterogeneity, calculated from the change in log likelihood on adding extra terms. All analyses were routinely stratified by geographical region of residence (10 regions corresponding to the areas covered by the cancer registries at recruitment) and socioeconomic status in quintiles (based on the Townsend method ), and adjusted for body mass index (< 25, 25–29.9, ≥ 30 kg m−2). The effect of adjusting for other potential confounders including smoking status (never, past, current), alcohol intake (none, < 7, 7–13, ≥ 14 drinks per week), strenuous exercise (never, < once, once, ≥ twice per week), oral contraceptive use (never, past, current), hysterectomy, acute myocardial infarction, heart failure, inflammatory bowel disease, respiratory failure, stroke or varicose veins was also examined, but there was no material change in the findings. Women with missing values for any of the adjustment variables were assigned to a separate category for that variable and sensitivity analyses were performed, firstly excluding women with missing data on potential confounders (< 6%) and secondly using multiple imputation for these values, with little change to the relative risk estimates for any type of HT use.
Women were classified as being current, past or never users of HT, as reported at the time of last contact. This was done initially using information from recruitment and women contributed person-years within the appropriate category up to the date that they completed the second study questionnaire, when they were reclassified whenever possible using updated HT information provided at resurvey. Current users of HT at last contact were further classified by type of HT last used, time since first use, age at first use, time between menopause and first use, and duration of use at last contact. Proprietary HT preparations were classified based on the method of administration and their hormonal constituents using the British National Formulary . Women were defined as users of oral preparations if they were taking a tablet formulation of HT, and users of transdermal preparations if they were using a patch or gel formulation of estrogen with or without a progestin (in patch or tablet form). Where the type of HT for current users at last contact could not be categorized, women were classified as unknown/other and were included in all analyses in a separate category. For current users of HT at last contact, the time since first use of HT was assumed to increase by 1 year for each year of follow-up.
Standardized incidence rates for hospital admissions or deaths for VTE were calculated using the rates in never users of HT as the standard and adjusting for region of residence, socioeconomic quintile and body mass index. In the results RRs are reported with 95% CIs in square brackets. In the plots RRs are represented by squares, whose sizes are inversely proportional to the variance of the logarithm of the relative risk; 95% CIs are indicated by lines. Analyses were conducted using stata 11.1 statistical software (Stata Corp., College Station, TX, USA).
These analyses included 1 058 259 postmenopausal women, of whom 36% were current users and 19% were past users of HT at the time of last contact. Few current users became past users (and vice versa) during follow-up; based on a sample of women completing the follow-up questionnaire prior to 1 January 2003, only 7% of current users of estrogen-only and 9% of current users of estrogen-progestin ceased use each year , whilst 1% of never users and 4% of past users became current users . Among current users at the time of last contact, 23% last used an oral estrogen-only formulation, 14% last used a transdermal estrogen-only formulation, and 52% last used a combined estrogen-progestin formulation. Table 1 shows the baseline characteristics of participants, by use of HT at last contact. Past and current users at last contact differed primarily from never users of HT in that they were younger and thus more likely to have used oral contraceptives. Current and past users of HT were broadly similar in the characteristics shown in Table 1; however, there was an obvious difference in the proportions of women who had had a hysterectomy among the different types of HT users, due to the increased risk of endometrial cancer with estrogen-only preparations [29,30].
Table 1. Characteristics of the study population by use of hormone therapy recorded at the time of last contact
Use of hormone therapy at the time of last contact
Estimates calculated in women with no missing values.
Mean age, years (SD)
Mean body mass index, kg m−2 (SD)
Upper third socio-economic group (%)
Current smokers (%)
Ever used oral contraceptives (%)
Mean alcohol intake among drinkers, g day−1 (SD)
Mean number of full-term pregnancies (SD)
Woman-years of follow-up (1000s)
Number with incident venous thromboembolism
Study participants were followed for a total of 3.3 million person-years, with a mean of 3.1 years per woman. During follow-up of the 2200 women first diagnosed with VTE, 2083 women were hospitalized (95% of all VTE cases), and for 117 cases (5%) their first VTE diagnosis was at death. Of the 2083 hospitalized VTE cases, 901 had pulmonary embolism and 1182 had deep vein thrombosis without pulmonary embolism. The mean time between reporting HT use and VTE diagnosis was 1.5 years.
Compared with never users, past users of HT at last contact had no increased risk of VTE (RR 0.95 [95%CI, 0.84–1.08], Figure 1). Among past users there was no significant change in risk with increasing time since last use (RRs 0.96 [0.72–1.28] and 0.76 [0.61–0.95] for < 2 years and 2+ years since last use of oral HT; Pheterogeneity = 0.2), and there was no difference between the relative risks of pulmonary embolism and deep vein thrombosis without pulmonary embolism (0.97 [0.81–1.16] vs. 0.94 [0.79–1.12]).
Current users at last contact had a highly significant increased risk of VTE overall (1.59 [1.45–1.75]), but risks varied by both type and method of administration of HT (Figure 1): current users of oral estrogen-progestin HT had the highest relative risk of VTE, with twice that in never users (2.07 [1.86–2.32], P < 0.0001); current users of oral estrogen-only HT had a lower but still significant increased risk (1.42 [1.22–1.66], P < 0.0001); but there was no evidence of an increased risk of VTE for current users of transdermal estrogen-only HT (0.82 [0.64–1.06], P = 0.1). There were few users of transdermal estrogen-progestin HT and of tibolone at last contact, with only 20 and 42 users, respectively, developing VTE (1.06 [0.68–1.65], P = 0.8 and 1.09 [0.80–1.48], P = 0.6), but confidence intervals for both risk estimates are wide.
Figure 2 shows the relative risks of VTE in current users of the three main groupings of HT use at the time of last contact (transdermal estrogen-only, oral estrogen-only and oral combined estrogen-progestin) by various factors. There was little variation in risk of VTE for most of the factors investigated here, apart from time since first use, where there was a significant difference between those who started < 2 years ago and those who started two or more years ago for both oral estrogen-only and oral estrogen-progestin users (Pheterogeneity = 0.01 and 0.05, respectively). The relationship of time since first use of oral HT and VTE is illustrated in Figure 3 where the 3-fold relative risk of VTE during the first 2 years of use (3.30 [2.35–4.63]) was almost twice that in later time periods (1.65 [1.36–2.00], 1.69 [1.48–1.93] and 1.79 [1.57–2.05] for 2–5 years, 5–10 years and 10+ years since first use, respectively; RRs for < 2 years vs. 2+ years were 3.32 [2.36–4.66] vs. 1.72 [1.56–1.91], Pheterogeneity = 0.0006).
There was some heterogeneity across the relative risks of VTE by body mass index for current oral estrogen-progestin users at last contact (Pheterogeneity = 0.01) and suggestive evidence of a decreasing trend with increasing body mass index for current oral estrogen-only users (Ptrend = 0.08). There was no significant difference between the relative risks of pulmonary embolism and deep vein thrombosis without pulmonary embolism for any of the three types of current HT use shown in Figure 2 (0.74 [0.51–1.09] vs. 0.90 [0.64–1.25] for transdermal estrogen-only; 1.39 [1.10–1.75] vs. 1.45 [1.17–1.80] for oral estrogen-only; 1.90 [1.61–2.23] vs. 2.24 [1.93–2.59] for oral estrogen-progestin; Pheterogeneity ≥ 0.4 for each comparison). Neither were there any differences in risk of VTE by duration of use, age at first use or time between menopause and first use.
Figure 4 shows the relative risks of VTE in current users of oral HT at last contact, by constituent, dose and regimen. There was no significant difference in risk of VTE among oral estrogen-only users by type of estrogen (Pheterogeneity = 0.9) or by dose of estradiol (Pheterogeneity = 0.3), although there was borderline evidence of a greater risk with higher compared with lower doses of oral equine estrogen (Pheterogeneity = 0.06). There was also little difference in the relative risks of VTE among transdermal estrogen-only users by dose of estradiol (0.60 [0.31–1.17] vs. 0.89 [0.67–1.17] for higher (> 50 mcg) vs. lower (≤ 50 mcg) doses, respectively; Pheterogeneity = 0.3).
The risk of VTE varied significantly by the type of progestin among current oral estrogen-progestin users at last contact, with higher relative risks (compared with never users) found for preparations containing medroxyprogesterone acetate than those containing norethisterone/norgestrel (2.67 [2.25–3.17] vs. 1.91 [1.69–2.17]; Pheterogeneity = 0.0007). There was a slight difference between the RRs associated with current use of continuous vs. sequential oral combined HT (Pheterogeneity = 0.05) and this was mainly due to the larger proportion of users of combined HT containing medroxyprogesterone acetate using continuous preparations (79% continuous, 21% sequential) – there was little difference between the risks for use of continuous and sequential preparations within each progestin group (for norethisterone, 1.91 [1.56–2.35] vs. 1.58 [1.13–2.20], Pheterogeneity = 0.3, and for medroxyprogesterone acetate, 2.78 [2.31–3.35] vs. 2.18 [1.45–3.28], Pheterogeneity = 0.3; there were no women taking continuous norgestrel). There were no significant differences between equine estrogen and estradiol, or between higher and lower doses of these estrogens, once the type of progestagen had been taken into account (data not shown).
The baseline characteristics of current users of oral HT at last contact according to type of estrogen (for estrogen-only users) and type of progestin (for estrogen-progestin users) are shown in the Appendix. There was little difference between users of different estrogen types or progestin types for the characteristics shown, although among users of oral estrogen-progestin preparations those taking norethisterone were slightly more likely to have had a hysterectomy than users of other progestin types. Adjustment for these factors had no material effect on the overall findings.
In this population, the standardized rate for hospital admission or death for VTE in never users of HT was 3.1 (95%CI, 2.9–3.3) per 1000 women over 5 years. The corresponding rates for current oral HT users at last contact per 1000 women over 5 years were 4.5 (3.8–5.2) for estrogen-only preparations, 6.0 (5.3–6.8) for combined HT containing norethisterone/norgestrel, and 8.4 (7.1–9.9) for combined HT containing medroxyprogesterone acetate. The rates for current transdermal users at last contact were not different to those of never users. The standardized rates for hospital admission or death for pulmonary embolism were 1.5 (1.4–1.7) per 1000 women over 5 years for never HT users, 2.1 (1.7–2.7) for current users of oral estrogen-only HT, 2.6 (2.1–3.1) for current users of combined HT containing norethisterone/norgestrel, and 4.0 (3.1–5.2) for current users of combined HT containing medroxyprogesterone acetate.
Our findings provide additional evidence that current use of oral HT in postmenopausal women increases the risk of VTE and that the risk varies by HT formulation. Among current users of oral HT at last contact in this study, the risk of VTE for combined estrogen-progestin users was substantially greater than that for estrogen-only users (RRs 2.07 vs. 1.42; Pheterogeneity < 0.0001), with a relative risk estimate for the direct comparison of combined HT vs. estrogen-only HT of 1.46 (1.23–1.72). Among users of oral estrogen-progestin HT, use of preparations containing medroxyprogesterone acetate (e.g. Provera®) was associated with a significantly higher risk of VTE than preparations containing norethisterone/norgestrel (RRs 2.67 vs. 1.91; Pheterogeneity = 0.0007). Use of transdermal estrogen HT was not associated with an increased risk (RR = 0.82). Among current users of oral HT at last contact the risk of VTE was higher during the first 2 years of use than later (RRs 3.32 vs. 1.72). Duration of use of HT, time between menopause and first HT use, and age at first use were not significantly associated with the risk of VTE in current users of HT. Risk decreased rapidly after use stopped, with no increased risk in past users of HT.
Our findings are consistent with those from previous studies [5,7,8,10,13,31,32], including a meta-analysis of results from randomized controlled trials  and a recent large record-linkage study in the UK , in showing that the risk of VTE was greater with oral estrogen-progestin than oral estrogen-only HT. These differences were also seen in the Women’s Health Initiative (WHI) trials where estrogen-only HT was randomized in women with a hysterectomy and estrogen-progestin HT was randomized in women with a uterus [9,10,33]. Other differences have been found in the Million Women Study between the effects of estrogen-only and combined HT for endometrial cancer, where the excess risk is greater for estrogen-only than combined HT , and for breast cancer, where the excess risk is greater for combined estrogen-progestin than estrogen-only HT ; the total excess risk of cancer is greater in users of estrogen-progestin than estrogen-only HT .
The lack of power in many studies has led to conflicting results for transdermal HT use and risk of VTE. Early small studies suggested that transdermal users had a similar risk of VTE to oral HT users [3,14], but a recent meta-analysis concluded that use of transdermal HT was associated with a lower risk of VTE than oral HT , consistent with our findings. We also reported that transdermal HT use was associated with a lower risk of gallbladder disease than use of other types of HT .
We found a significantly higher relative risk of VTE during the first 2 years of use of HT, in keeping with results from randomized trials [10,31,36]. Previous studies have investigated duration of HT use and found higher risks for the first year or two, decreasing thereafter [3,4,13,14,37,38], but we did not find a significant association between duration of HT use and risk of VTE.
Among oral estrogen-progestin users at last contact, we found a significantly greater risk of VTE in users of formulations containing medroxyprogesterone acetate than the other progestins (Pheterogeneity = 0.0007). To our knowledge, no other study has directly compared risk of VTE by specific types of progestins. The large randomized controlled trials of combined HT all used medroxyprogesterone acetate as their progestin constituent [5,9,32], and so no direct comparison with other progestin types can be made across trials. One UK study grouped oral combined HT into pregnane derivatives (which include medroxyprogesterone acetate) and nortestosterone derivatives (which include norethisterone) and found higher risks associated with the pregnane derivatives  (RRs 1.72 vs. 1.48), consistent with our findings. The higher risk of VTE associated with use of preparations containing medroxyprogesterone acetate than other progestins appears to be specific for VTE in this cohort, as we found no such differences for breast, ovarian or endometrial cancer [21,28,29].
We found no significant difference in risk of VTE by estrogen type, and some evidence for an increased risk with higher compared with lower doses of oral equine estrogen (Pheterogeneity = 0.06). Other smaller studies with comparisons for different types of estrogen and their doses have shown mixed findings [2–4,8,14,37], and a recent meta-analysis found a significantly higher risk of VTE with high doses of estrogen than with low doses . Few studies have examined the relationship between HT and VTE subdivided by body mass index (BMI), and there are mixed findings in both our study and studies by others .
This study has a number of strengths. VTE is relatively uncommon and this large prospective study of over one million women with excellent follow-up and over 2000 incident VTE events allowed a detailed examination of the association between different types of HT and VTE. We thus had sufficient power to investigate how risks varied between types of HT, by progestin type, and by method of administration. Prospectively collected information on HT use was linked with routinely collected NHS hospital admission and death records. Both the recording of pulmonary embolism diagnoses and the accuracy of the linkage process for NHS datasets have been shown to be good [17,41]. This linked dataset allowed potential confounding to be minimized by adjusting for age, region of residence, socioeconomic quintile and BMI (as well as allowing the investigation of other potential confounders).
Pulmonary embolism is a serious condition that will almost always be treated as an inpatient or day-case hospital admission or result in death, so almost all cases should be included in this study. Deep vein thrombosis is increasingly treated in a community (general practice) or outpatient setting [42,43]. Our investigations confirmed that these data are reliable for hospital admissions with VTE, especially for pulmonary embolism, because out of 864 women without a hospital record for VTE, none had a general practitioner diagnosis of pulmonary embolism, and relatively few women with a general practitioner diagnosis of deep vein thrombosis were not admitted to hospital; furthermore, general practitioners confirmed the hospital diagnosis in 93% of VTE cases. Thus, our absolute risks for the more serious condition of pulmonary embolism are likely to be correct. Any misclassification of deep vein thrombosis arising from VTE diagnoses not recorded in the hospital data would lead to a slight underestimation of the rate of VTE, whereas there might be a slight overestimation of the rate due to the small proportion of hospital cases not confirmed by the patients’ general practitioners. There was little difference between the relative risks of deep vein thrombosis with and without pulmonary embolism when examined for each of the three HT types here (Figure 2).
Prospectively collected information on HT use was self-reported and the agreement between self-reported use of HT and general practitioner prescriptions has been shown to be excellent in this cohort . Updating HT information using information collected 3 years after recruitment and censoring HT use 48 months after last contact should have minimized misclassification of HT use. The average time between last reported HT use and VTE was only 1.5 years, and over this short period comparatively few women would have changed from one category of HT use to another: an estimated 1% of never users became current users, 4% of past users became current users, and 8% of current users ceased use, each year . Any residual random misclassification of HT use would slightly dilute the estimates of relative risk but would not produce spurious associations . Correcting for such misclassification, by use of the regression dilution approach , would increase the logarithm of the relative risk among current users by a factor of approximately 1.2. Non-random misclassification of exposure, for example if women change HT use because of preclinical symptoms, is likely to be small in this study as the exposure was recorded on average 1.5 years before diagnosis, and the onset of VTE is usually acute.
Estrogen-only preparations are not usually given to women with a uterus because of the increased risk of endometrial cancer (96% of women here reporting current use of oral estrogen-only HT had had a hysterectomy –Table 1), and it is unclear whether there are other major differences in prescribing. Our results for estrogen-only HT are similar to those found in the WHI trial for women who had had a hysterectomy, and the difference between estrogen-only and estrogen-progestin is similar to that reported by WHI . However, the main new finding here is the variation in risk of VTE between types of progestin amongst estrogen-progestin HT users, almost all of whom have not had a hysterectomy.
Although many VTE risk factors were examined as potential confounders here, other known risk factors, such as family history and thrombophilia, were not recorded in this study and so could not be taken into account in these analyses; thus there is some potential for residual confounding here if the prescribing of HT does take these factors into account. This may lead to a slight underestimation of the relative risks for current users, but thrombophilia is rare and so this is unlikely to materially alter our findings.
Our findings suggest that both route of administration and the specific constituents of HT have a significant impact on the risk of VTE in middle-aged women. The risk of VTE was highest for current users of oral combined estrogen-progestin HT at the time of last contact and in particular for users of estrogen-progestin preparations containing medroxyprogesterone acetate. The risk was lower but significantly increased in users of oral estrogen-only HT, but there was no increased risk for current users of transdermal estrogen HT. Among current oral HT users, over a 5-year period one in 475 users of estrogen-only HT, one in 390 users of estrogen-progestin HT containing norethisterone/norgestrel and one in 250 users of estrogen-progestin HT containing medroxyprogesterone acetate would be admitted to hospital for (or die from) pulmonary embolism, compared with one in 660 who never used HT. Thus, one case of pulmonary embolism could be avoided for every 1295 current users of oral HT at last contact who used estrogen-only rather than estrogen-progestin, and among women using oral combined HT one pulmonary embolism in 700 women could be avoided by use of norethisterone or norgestrel rather than medroxyprogesterone acetate as the progestin constituent.
We thank all the women who participated in the Million Women Study, the study steering committee as well as ISD Scotland, the Information Centre for Health and Social Care and Northgate Solutions for the linkage of hospital records. We would also like to thank A. Goodill for his assistance in the preparation of the figures.
The Million Women Study is supported by Cancer Research UK, the UK Medical Research Council and the UK National Health Service Breast Screening Programme.
The sponsors did not have any input into study design, study conduct, data collection, management, analysis or interpretation, nor did they influence the preparation, review or approval of the manuscript.
Disclosure of Conflict of Interests
The authors state that they have no conflict of interest.
Million Women Study Steering Committee: Emily Banks, Valerie Beral, Ruth English, Jane Green, Julietta Patnick, Richard Peto, Gillian Reeves, Martin Vessey and Matthew Wallis.
Million Women Study Coordinating Centre Staff: Simon Abbott, Naomi Allen, Miranda Armstrong, Angela Balkwill, Emily Banks, Vicky Benson, Valerie Beral, Judith Black, Anna Brown, Diana Bull, Benjamin Cairns, Kathy Callaghan, Karen Canfell, Dexter Canoy, James Chivenga, Barbara Crossley, Francesca Crowe, Dave Ewart, Sarah Ewart, Lee Fletcher, Toral Gathani, Laura Gerrard, Adrian Goodill, Jane Green, Lynden Guiver, Elizabeth Hilton, Sau Wan Kan, Carol Keene, Oksana Kirichek, Mary Kroll, Nicky Langston, Isobel Lingard, Bette Liu, Maria-Jose Luque, Lynn Pank, Kirstin Pirie, Gillian Reeves, Andrew Roddam, Keith Shaw, Emma Sherman, Evie Sherry-Starmer, Helena Strange, Siân Sweetland, Alison Timadjer, Sarah Tipper, Ruth Travis, Xiaosi Wang, Joanna Watson, Lucy Wright, Tienyu Yang, Heather Young.
Collaborating UK NHS Breast Screening Centres: Avon, Aylesbury, Barnsley, Basingstoke, Bedfordshire and Hertfordshire, Cambridge and Huntingdon, Chelmsford and 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 and Worcester, Kent, 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, Wycombe.
Characteristics of the study population by current use of known type of oral hormone therapy recorded at the time of last contact.
Current users of known types of oral hormone therapy at the time of last contact
Oral estrogen plus progestin
Equine estrogen (n = 60 438)
Estradiol (n = 19 818)
Medroxyprogesterone acetate (n = 57 049)
Norethisterone (n = 78 293)
Norgestrel (n = 41 240)
Estimates calculated in women with no missing values.