EE indicates esterified estrogen; P, progestin; CEE, conjugated equine
Smith NL, Heckbert SR, Lemaitre RN, Reiner AP, Lumley T, Weiss NS, Larson EB, Rosendaal FR, Psaty BM. Esterified Estrogens and Conjugated Equine Estrogens and the Risk of Venous Thrombosis. JAMA. 2004;292(13):1581-1587. doi:10.1001/jama.292.13.1581
Author Affiliations: Departments of Epidemiology
(Drs Smith, Heckbert, Reiner, Weiss, and Psaty), Medicine (Drs Lemaitre and
Psaty), Biostatistics (Dr Lumley), and Health Services (Dr Psaty), University
of Washington, Seattle; Center for Health Studies, Group Health Cooperative,
Seattle, Wash (Drs Heckbert, Larson, and Psaty); and Leiden University Medical
Center, Leiden, the Netherlands (Dr Rosendaal).
Context Clinical trial evidence indicates that estrogen therapy with or without
progestins increases venous thrombotic risk. The findings from these trials,
which used oral conjugated equine estrogens, may not be generalizable to other
Objective To compare risk of venous thrombosis among esterified estrogen users,
conjugated equine estrogen users, and nonusers.
Design, Setting, and Participants This population-based, case-control study was conducted at a large health
maintenance organization in Washington State. Cases were perimenopausal and
postmenopausal women aged 30 to 89 years who sustained a first venous thrombosis
between January 1995 and December 2001 and controls were matched on age, hypertension
status, and calendar year.
Main Outcome Measure Risk of first venous thrombosis in relation to current use of esterified
or conjugated equine estrogens, with or without concomitant progestin. Current
use was defined as use at thrombotic event for cases and a comparable reference
date for controls.
Results Five hundred eighty-six incident venous thrombosis cases and 2268 controls
were identified. Compared with women not currently using hormones, current
users of esterified estrogen had no increase in venous thrombotic risk (odds
ratio [OR], 0.92; 95% confidence interval [CI], 0.69-1.22). In contrast, women
currently taking conjugated equine estrogen had an elevated risk (OR, 1.65;
95% CI, 1.24-2.19). When analyses were restricted to estrogen users, current
users of conjugated equine estrogen had a higher risk than current users of
esterified estrogen (OR, 1.78; 95% CI, 1.11-2.84). Among conjugated equine
estrogen users, increasing daily dose was associated with increased risk (trend P value = .02). Among all estrogen users, concomitant
progestin use was associated with increased risk compared with use of estrogen
alone (OR, 1.60; 95% CI, 1.13-2.26).
Conclusion Our finding that conjugated equine estrogen but not esterified estrogen
was associated with venous thrombotic risk needs to be replicated and may
have implications for the choice of hormones in perimenopausal and postmenopausal
Clinical trial evidence from the Heart and Estrogen/progestin Replacement
Study (HERS) and the Women’s Health Initiative (WHI) indicates that
estrogen therapy, with or without progestins, is associated with an increased
risk of venous thrombosis (VT) in postmenopausal women.1- 3 The
HERS investigators observed a relative risk of 2.7 (95% confidence interval
[CI], 1.4-5.0) for estrogen plus progestin therapy, and the WHI investigators
observed a relative risk of 2.1 (95% CI, 1.6-2.8) and 1.3 (95% CI, 1.0-1.8)
for estrogen with and without concomitant progestin use, respectively. Each
of these trials evaluated oral conjugated equine estrogens (CEEs) and medroxyprogesterone
Various estrogen compounds and their modes of administration are known
to differ in estrogen constituents, metabolism, and their affinity for estrogen
receptors.4,5 Clinical comparison
studies of these compounds have focused mainly on mode of delivery, and outcomes
have generally been restricted to the relief of menopausal symptoms and the
occurrence of vaginal bleeding.6- 8 Nearly
all comparisons have evaluated the most commonly used estrogens, CEE and micronized
estradiol. Esterified estrogens (EEs) have received less attention. All these
products, nonetheless, continue to be used to treat menopause-related vasomotor
symptoms in perimenopausal and postmenopausal women.
In October 1999, the Group Health Cooperative (GHC) pharmacies switched
the standard postmenopausal estrogen therapy from EE to CEE for current and
new users of hormone therapy. Formulary switches such as these occur in health
maintenance organization (HMO) settings when medications are thought to be
therapeutically interchangeable.9- 11 The
formulary change occurred during data collection for a case-control study
of cardiovascular outcomes that included VT, which presented us with the opportunity
to examine the association of oral EE and CEE with VT risk in perimenopausal
and postmenopausal women.
The setting for this observational study was GHC, a large HMO in western
Washington State. Health care delivery and medication prescribing at GHC are
based primarily on GHC treatment guidelines. This case-control study of VT
was part of a larger, ongoing, population-based, case-control study of VT,
myocardial infarction, and stroke and shares with it a single control group.12,13 The study was approved by the GHC
Human Subjects Review Committee, and informed consent was obtained from individuals
who could be contacted and waived by the committee for those who could not
Study participants were perimenopausal and postmenopausal female GHC
members aged 30 to 89 years. Cases were all GHC members who experienced a
first deep VT (DVT) or pulmonary embolism (PE) diagnosed between January 1,
1995, and December 31, 2001, the most recent year of complete and cleaned
data in this ongoing study. The date of the VT served as an index date before
which information on hormone use and other exposures was ascertained. Controls
were a random sample of GHC members who comprised a pool of individuals shared
by several case-control studies. The control group was frequency matched by
age (within decade), sex, treated hypertension status, and calendar year of
identification to myocardial infarction cases, the largest case group. All
controls for this analysis met the same inclusion criteria as VT cases and
had no history of DVT or PE. For controls, the index date was a randomly chosen
date within the calendar year from which they were selected as a control.
Women with VT were identified from inpatient and outpatient care settings.
In the inpatient setting, International Classification of
Diseases, Ninth Revision codes were abstracted from GHC hospitalization
records, which included hospital stays at GHC and non-GHC facilities. In the
outpatient setting, GHC pharmacy records were used to identify women who were
dispensed a prescription for a low-molecular-weight heparin for nonhospitalized
DVT treatment. Additionally, women were identified from 3 GHC clinics where
a pharmacy-based outpatient treatment protocol for DVT was implemented in
Trained medical record abstractors reviewed the medical records of all
potential cases to verify the diagnosis of VT and to determine how the diagnosis
was made. Events were classified as study eligible if they were diagnosed
by an imaging modality (Doppler or duplex ultrasound, computed tomography,
pulmonary angiography, or ventilation-perfusion scan) or by physician judgment
in the presence of symptoms or according to treatment strategies. Ninety-two
percent of the eligible cases had positive diagnostic imaging test results.
Menopausal status at the index date was defined by the cessation of
ovarian function that occurred naturally or through a bilateral oophorectomy
for cases and controls and was based on information collected from the GHC
medical record. A woman was considered perimenopausal at the onset of menopausal
symptoms. If menopausal status was not explicitly stated in the record, women
aged 55 years and older were considered postmenopausal.
Hormone Use. Use of hormones was determined
by using the GHC computerized pharmacy database that contains records of all
prescriptions filled through GHC since 1977. More than 95% of GHC members
in this age group fill almost all prescriptions through GHC pharmacies.14 Pharmacy data contain detailed information that includes
GHC member identification number, drug name, date the prescription was filled,
medication strength, quantity of medication prescribed, and dosing instructions
or the number of days the supply of medication would last.
Oral estrogen was classified into 3 subgroups: (1) CEE, such as Premarin;
(2) EE, such as Estratab andMenest; and (3) other estrogens, primarily micronized
estradiol, which accounted for less than 1% of all estrogen prescriptions
during the 7 years of the study. The progestin prescribed was almost exclusively
medroxyprogesterone acetate and was dispensed as a separate pill from estrogen
for virtually all subjects.
A woman was considered a current user of a hormone if she received enough
medication with her last prescription to last until her index date according
to an assumption of 80% compliance. An 80% compliance adjustment was made
by increasing the number of days that a prescription would last by 25%. We
excluded women for whom there was no record of a GHC pharmacy prescription
being filled in the 5 years before the index date (n = 39), women
using progestin without estrogen at the index date (n = 38), and
women who were current users of creams or patches and not using estrogen pills
(n = 54).
Daily oral estrogen dose was calculated from computerized pharmacy data
by using pill strength and dosing instructions or number of supply days. For
estrogens other than CEE and EE, estrogen dose was based on CEE equivalents.15 Recency of therapy initiation for current hormone
users was calculated for CEE and EE separately and was defined as the number
of days between the index date and the date of the first prescription fill
for the drug that was continually used through the index date. Continual use
was defined as consecutive prescription refills, assuming 80% compliance,
while also allowing for a 90-day gap between run-out dates and refills. For
subjects who were switched from EE to CEE, recency of use pertained only to
the duration of use of CEE and did not include time accumulated using EE.
Clinical and Demographic Information. Demographic
and health-status information was obtained by review of the entire GHC ambulatory
medical record up to the index date. Medical conditions included treated hypertension,
congestive heart failure, a history of stroke, coagulation disorders (lupus
anticoagulant; protein C, protein S, and antithrombin deficiencies; and polycythemia
vera), hysterectomy, previous oral contraceptive use, and recent inpatient
surgical procedures. Information was collected on clinical measures that included
most recent weight and height. Demographic information included birth date
and race, which was based on clinician notes.
Cancer history information was collected from a GHC cancer registry
file, which included all cancers except nonmelanoma skin cancers. Previous
hospitalization data and fracture data were collected from GHC administrative
files that include diagnoses from inpatient and outpatient care delivered
at GHC and non-GHC facilities. Surviving subjects were invited to participate
in a telephone interview in which information such as race and smoking status
is more reliably collected than from a medical record. In the interview, subjects
had the choice of 4 race categories or an option to specify a fifth.
Missing values for demographic and clinical characteristics collected
from the medical record or telephone interview were imputed with IVEware software.16 Missing data were uncommon, and no more than 2% of
the data for any variable was imputed.
Multivariate logistic regression was used to model the association between
current use of estrogen hormones and the risk of VT. Models estimated relative
risks with odds ratios (ORs) and produced 95% CIs. All multivariate models
were adjusted for matching variables that included age (continuous and indicator
variables by decade), index year (indicator variables by year), and treated
hypertension status (absent vs present). We examined multivariate models that
included additional adjustments for suspected confounding variables such as
race (white vs other), body mass index (weight in kilograms divided by height
in meters squared), cancer history (cancer diagnosed within 5 years of index),
hysterectomy status, previous oral contraceptive use, prevalent congestive
heart failure, history of stroke, smoking status, recent inpatient surgery,
hospitalizations that lasted at least 2 nights in the 30 days before the index
date, and fracture of the pelvis or lower limb in the 30 days before the index
date. Covariates that confounded risk estimates were retained in final multivariate
models using SAS statistical software, version 8.2 (SAS Institute Inc, Cary,
Primary analyses compared current CEE and current EE use with nonuse
of hormones and current CEE use with current EE use. The latter comparison
minimized issues of confounding by hormone-therapy indication or contraindication.
Further analyses stratified the use of CEE and EE by concomitant progestin
use. Sensitivity analyses of the primary findings were conducted assuming
100% compliance with hormone therapy instead of 80%, excluding women who had
a predisposing VT risk factor (cancer, recent hospitalization, or fracture)
and excluding VT events that were not confirmed with an imaging test.
Estrogen dose analyses, adjusted for progestin use, were restricted
to current users of CEE and EE. The daily, modal dose of CEE and EE was 0.625
mg. The low dose mean was 0.3 mg for CEE and EE; high dose mean was 1.67 mg
for EE (range, 1.25-2.5 mg) and 1.16 mg for CEE (range, 0.9-2.5 mg). On average,
81% of estrogen users used the modal dose. The modal dose of EE served as
the reference group for dose analyses. Analyses for recency of starting were
restricted to current users, and subjects were classified as users for less
than 1 year, 1 to 5 years, or more than 5 years, with the latter category
serving as the reference. Sensitivity analyses additionally restricted subjects
to those who started using hormone therapy while enrolled at GHC. This restriction
required that, according to GHC pharmacy data, there be at least 6 months
of no hormone use before initiation.
We identified 586 perimenopausal and postmenopausal women who experienced
a first VT: 426 with DVT alone (73%), 68 with DVT and PE (12%), and 92 with
PE (16%). Thirty-three of the PE events were fatal. Characteristics of the
case patients and controls (n = 2268) are presented in Table 1. Compared with controls, cases were more
likely to have risk factors for VT, including a history of cancer, prevalent
heart failure, and a recent hospitalization or major fracture. A similar percentage
of cases (37%) and controls (37%) were current users of an oral estrogen with
or without concomitant progestin use.
The percentage of controls using any estrogen therapy remained fairly
constant during the 7 years of observation (32%, 36%, 39%, 37%, 36%, 38%,
and 39% for 1995-2001, respectively) and represents the use of hormones in
the larger GHC population of postmenopausal women in the age range included
in the study. The Figure presents the
use of EE and CEE estrogen types during this period. The GHC formulary switch
from EE to CEE in October 1999 resulted in a substantial proportion of women
switching estrogen type. Characteristics of the study population according
to hormone use are presented in Table 2.
Compared with nonusers, users were younger, more likely to be white, less
likely to have a history of cancer, and more likely to have had a hysterectomy.
Among users of hormone therapy, age, hysterectomy status, and daily dose varied
by estrogen type.
The risk of VT associated with hormone therapy is presented in Table 3. Compared with women not currently using
hormones, current users of EE had no increase in VT risk (OR, 0.92; 95% CI,
0.69-1.22), whereas current users of CEE had an elevated risk (OR, 1.65; 95%
CI, 1.24-2.19) in analyses adjusted for matching factors and the confounding
factors of race and cancer history. Other covariates were considered in the
multivariate model but were not included in the final model because they did
not confound the hormone-VT association. When adjusted analyses were restricted
to hormone users to minimize potential indication bias, current use of CEE
was associated with an increase in VT risk compared with current use of EE
(OR, 1.78; 95% CI, 1.11-2.84). There was little evidence of confounding in
hormone-only models (Table 3, models
3 and 4).
Current hormone exposure was further divided to differentiate estrogen
therapy opposed and unopposed with progestin (Table 4). Compared with women not currently using hormone therapy,
current users of EE alone, EE with progestin, and CEE alone had no appreciable
increase in VT risk, whereas current users of CEE with progestin had a 2-fold
increase in risk (OR, 2.17; 95% CI, 1.49-3.14) in adjusted analyses. In adjusted
analyses restricted to hormone users, current users of CEE alone had a borderline
increase in VT risk (OR, 1.74; 95% CI, 0.96-3.16), and current users of CEE
with progestin had nearly a 3-fold increase in risk (OR, 2.94; 95% CI, 1.60-5.40)
compared with current users of EE alone. In analyses restricted to estrogen
and progestin users, current use of CEE with progestin was associated with
a 2-fold increase in VT risk (OR, 2.02; 95% CI, 1.05-3.88) compared with current
use of EE with progestin.
There were 470 controls and cases who used concomitant progestin among
the 1011 users of CEE or EE. The risk of VT associated with the current use
of estrogen plus progestin was increased compared with use of estrogen alone
(OR, 1.60; 95% CI, 1.13-2.26) in analyses that adjusted for estrogen type,
cancer history, race, and the matching variables.
Sensitivity analyses did not meaningfully alter the risk estimates for
the association of EE and CEE with VT. When current use was defined assuming
100% compliance with hormone prescriptions, the OR for VT risk was 0.99 (95%
CI, 0.74-1.33) for EE use and 1.50 (95% CI, 1.12-2.01) for CEE use compared
with nonuse. When controls and cases who had predisposing VT risk factors
were excluded, the OR was 1.11 (95% CI, 0.80-1.56) for EE use and 1.71 (95%
CI, 1.23-2.37) for CEE use compared with nonuse. When VT events not confirmed
by imaging were excluded, the OR was 0.99 (95% CI, 0.74-1.33) for EE use and
1.75 (95% CI, 1.30-2.35) for CEE use vs nonuse.
Compared with current use of the modal dose of EE, low- and high-dose
EE were not associated with an increased or decreased risk of VT in adjusted
analyses (Table 5). When compared with
current use of the modal dose of EE, low-dose CEE was not associated with
an increased risk of VT, whereas modal and high-dose CEE were associated with
increased VT risk in adjusted analyses. When analyses were restricted to women
who used CEE, there was a positive dose-response relationship between CEE
dose and VT risk (P value for Mantel-Haenszel χ2 for trend=.02).
Compared with controls and cases using EE for more than 5 years, starting
EE in the year before the index date (OR, 1.26; 95% CI, 0.56-2.83) or 1 to
5 years before the index date (OR, 1.08; 95% CI, 0.54-2.15) was not associated
with an increased risk in adjusted analyses. Compared with control and case
using CEE more than 5 years, starting CEE in the year before the index date
(OR, 0.89; 95% CI, 0.40-1.97) or 1 to 5 years before the index date (OR, 1.54;
95% CI, 0.72-3.28) was also not associated with an increased risk. Odds ratios
did not differ appreciably in sensitivity analyses.
In this observational study, compared with nonuse of hormone therapy,
CEE use but not EE use was associated with an increased risk of VT in adjusted
analyses. Compared with EE use, CEE use was associated with an increase in
risk. Among CEE users, there was evidence of a dose-response relationship
with VT risk. No association was detected between recency of starting either
CEE or EE therapy and VT risk. Use of estrogen plus progestin was associated
with an increase in risk of VT compared with use of estrogen alone.
Our results are in agreement with and expand on previous experimental
and observational findings of VT risk associated with hormone use in postmenopausal
women. For CEE plus progestin, the HERS relative risk of 2.7 (95% CI, 1.4-5.0)
and the WHI relative risk of 2.1 (95% CI, 1.6-2.8) are similar to the OR of
2.17 (95% CI, 1.49-3.14) reported in this study for CEE plus progestin use
compared with nonuse of hormones.1,2 For
CEE without progestin, the WHI relative risk of 1.3 (95% CI, 1.0-1.8) is again
similar to our findings of an OR of 1.31 (95% CI, 0.91-1.88).3 For
EE, no clinical trial or observational data are available for VT risk.
Comparison of hazard ratios (2.1 vs 1.3) from the CEE plus medroxyprogesterone
acetate and the CEE-alone arms of WHI would suggest that this progestin compound
is associated with a 62% increase in risk. We observed a 60% increase in risk
(OR, 1.60; 95% CI, 1.13-2.26). Unlike findings from the CEE plus progestin
clinical trials that showed larger risks in recent hormone starters compared
with longer-term users,1,2 we
did not find that recent starting—particularly use in the first year—was
associated with an increased risk of VT.
Several issues should be considered when this study’s findings
are evaluated. First, the use of hormone therapy was not randomly assigned.
Women and their physicians chose whether to use hormones according to clinical
indication, which can induce confounding. The type of estrogen received, however,
was dictated primarily by changes over time in the GHC formulary and not by
patient or physician choice: women treated before October 1999 primarily received
EE, whereas women treated after this date received CEE. This unanticipated
estrogen switch was not included as a primary aim of the original research.
Use of hormone therapy was prospectively collected in the GHC pharmacy database
and not subject to recall or information bias. The design of this study is
population-based so that controls are representative of the population from
which the cases arose. Findings are generalizable only to similar populations,
primarily white perimenopausal and postmenopausal women without a previous
Conjugated equine estrogens contain 10 known biologically active estrogen
compounds, as well as others that have yet to be described, which has prevented
the manufacture of a generic version.4,17 The
primary compounds are estrone sulfate, constituting approximately 53% of the
estrogens, and equilin sulfate, constituting about 25% of the product.5,17 Esterified estrogens contain approximately
80% estrone sulfate and approximately 11% equilin sulfate.5 In
vivo, there is active conversion of estrogen components so that plasma concentration
ratios of these hormones tend to be similar across estrogen products other
than transdermal estrogen preparations, which bypass liver metabolism and
produce higher estradiol levels.5,18 Comparative
pharmacologic data for CEE and EE are limited.18
In both experimental and observational settings, prescription estrogens,
primarily micronized estradiol and CEE, are associated with changes in the
plasma levels of several proteins or markers of the coagulation, anticoagulation,
and fibrinolysis pathways that create a prothrombotic environment. These compounds
increase plasma levels of tissue activatable fibrinolysis inhibitor antigen19,20; protein C21;
factors VII, IX, and X22- 25;
and D-dimer20,26 and decrease
the levels of protein S,27,28 soluble
thrombomodulin,29 antithrombin,21,27,28 and
tissue plasminogen activators.24,28 Few
studies have addressed EE’s effects on clotting factors, although its
role in stress response, bone density, lipid levels, and endometrial hyperplasia
has been published with data from placebo-controlled trials.30,31 Clinical
comparison studies of EE and CEE are limited to a single crossover trial that
reported significantly better short-term cognitive function and depression
scores during EE use compared with CEE use in menopausal women.32
Data from this observational study provide a comparison of oral estrogen
products in relation to the risk of first VT. Findings suggest that compared
with nonuse of hormone therapy, oral CEE therapy increases the risk of VT
dose-dependently, EE does not influence VT risk, and the use of any estrogen
in combination with medroxyprogesterone acetate increases risk as well.
Our findings for VT risk need to be replicated, and the association
of EE and CEE with other adverse outcomes of estrogen use should be investigated.
If replicated, these findings for VT may have implications for the choice
of hormone in treating menopause-related vasomotor symptoms in perimenopausal
and postmenopausal women.
Corresponding Author: Nicholas L. Smith,
PhD, Cardiovascular Health Research Unit, 1730 Minor Ave, Suite 1360, Seattle,
WA 98101 (email@example.com).
Author Contributions: Dr Smith had full access
to all of the data in the study and takes responsibility for the integrity
of the data and the accuracy of the data analysis.
Study concept and design: Smith, Heckbert,
Reiner, Rosendaal, Psaty.
Acquisition of data: Smith, Heckbert, Psaty.
Analysis and interpretation of data: Smith,Heckbert,
Lemaitre, Reiner, Lumley, Weiss, Larson, Rosendaal, Psaty.
Drafting of the manuscript:Smith.
Critical revision of the manuscript for important
intellectual content: Heckbert, Lemaitre, Reiner,Lumley, Weiss, Larson,
Statistical analysis:Smith, Heckbert, Lemaitre,Lumley.
Obtained funding: Smith, Heckbert, Psaty.
Administrative, technical, or material support:
Smith, Heckbert, Larson, Rosendaal, Psaty.
Study supervision: Smith, Heckbert, Weiss,
Funding/Support: The research reported in this
article was supported by National Heart, Lung, and Blood Institute grants
HL73410 (Dr Smith), HL60739 (Dr Psaty), HL68639 (Dr Psaty), HL43201 (Dr Psaty),
HL74745 (Dr Psaty), HL68986 (Dr Heckbert), and National Institute on Aging
grant AG09556 (Dr Psaty).
Role of the Sponsor: The funding organizations
had no role in the conduct and design of the study; in the collection, analysis,
and interpretation of the data; or in the preparation, review, or approval
of the manuscript.