COMPARED WITH PLACEBO,

IGINAL INVESTIGATION Esterified Estrogen and Conjugated Equine Estrogen and the Risk of Incident Myocardial Infarction and Stroke Rozenn N. Lemaitre, PhD, MPH; Noel S. Weiss, MD, DrPH; Nicholas L. Smith, PhD; Bruce M. Psaty, MD, PhD; Thomas Lumley, PhD; Eric B. Larson, MD, MPH; Susan R. Heckbert, MD, PhD Background: Clinical trials of conjugated equine estrogen (CEE) or estradiol vs placebo in postmenopausal women have found no effect or an elevated risk of myocardial infarction (MI) and stroke. The association of these end points with the use of esterified estrogen (EE) is unknown. Methods: We examined the risk of MI and stroke associated with current use of CEE, use of EE, or nonuse of hormones in a population-based case-control study in a health maintenance organization. Cases were all postmenopausal women with an incident MI (n=1644) or stroke (n=1080). Controls (n=4205) consisted of a random sample of postmenopausal women without MI or stroke. Current use of postmenopausal hormones was assessed using a computerized pharmacy database. Results: There was no difference in risk of MI or stroke associated with current use of CEE or EE compared with nonuse or for current use of CEE compared with EE. In analyses restricted to hormone users, there was a suggestion of higher ischemic stroke risk associated with CEE alone (without progestin) compared with EE alone (odds ratio, 1.57; 95% confidence interval, 0.98-2.53). There was also a suggestion that when initiated in the previous 6 months, CEE was associated with a higher risk of MI than EE (odds ratio, 2.33; 95% confidence interval, 0.93-5.82). Conclusion: Further study may be warranted of the effects of EE on the risk of cardiovascular end points. Arch Intern Med. 2006;166:399-404 Author Affiliations: Cardiovascular Health Research Unit, Departments of Medicine (Drs Lemaitre, Psaty, and Larson), Epidemiology (Drs Weiss, Smith, Psaty, and Heckbert), Biostatistics (Dr Lumley), and Health Services (Drs Psaty and Larson), University of Washington, and Center for Health Studies, Group Health Cooperative (Drs Psaty, Larson, and Heckbert), Seattle. COMPARED WITH PLACEBO, oral conjugated equine estrogen (CEE) and oral estradiol have been associated in several clinical trials 1-7 with either an increased risk or no difference in risk of coronary disease and stroke in postmenopausal women. However, a third form of postmenopausal estrogen, esterified estrogen (EE), has not been studied in clinical trials or observational studies of coronary disease or stroke. For editorial comment see page 389 Conjugated equine estrogen, harvested from the urine of pregnant mares, 8 includes 10 estrogen compounds, all of which are biologically active, and EE consists primarily of the sulfate esters of estrone and equilin. 9 The different estrogen compounds could have different CME course available at www.archinternmed.com biological effects and different effects on cardiovascular end points, but there are few comparative studies. At Group Health Cooperative (GHC), a large health maintenance organization (HMO), changes were made in 1991 10 and again in 1999 in the standard postmenopausal estrogen on the formulary, first from CEE to EE and then back again to CEE. At the time, the 2 medications were considered therapeutically interchangeable. The first switch was based on purchase cost considerations. The second switch happened when the GHC formulary was altered based on supply issues with EE and the concurrent development of an affiliation with Kaiser Permanente, another HMO. These formulary changes occurred during data collection for a case-control study of myocardial infarction (MI) and stroke in postmenopausal women and provided the opportunity to compare the risk of incident MI and stroke associated with the use of CEE, EE, or no hormone therapy. METHODS STUDY DESIGN AND PARTICIPANTS The setting for this project was GHC, a staffmodel HMO. The case subjects were all postmenopausal female GHC members aged 30 to 79 years with an incident MI between July 1, 399

Table 1. Characteristics of the Cases and Control Subjects Characteristic Myocardial Infarction Cases (n = 1644) Ischemic Stroke Cases (n = 830) Hemorrhagic Stroke Cases (n = 250) Controls (n = 4205) Age, mean (SD), y 68.1 (8.3) 70.0 (7.7) 66.8 (9.9) 67.3 (8.5) White race, % 93.1 88.8 87.8 91.0 Years of enrollment, mean (SD), 18.1 (11.6) 18.8 (12.2) 19.6 (12.6) 19.5 (11.4) Physician visits in past year, mean (SD), 7.5 (6.8) 8.4 (7.5) 6.5 (6.4) 5.9 (5.5) Hypertension, % 51.8 60.1 46.0 44.4 Systolic blood pressure, mean (SD), mm Hg 142.0 (21.6) 146.7 (23.3) 143.9 (23.8) 137.9 (20.3) Cholesterol, mean (SD), mg/dl 250.2 (50.1) 241.6 (45.5) 232.9 (48.8) 234.8 (42.0) Weight, mean (SD), kg 73.6 (18.0) 73.5 (17.9) 69.4 (16.6) 72.2 (16.8) Diabetes mellitus, % 22.1 23.9 8.8 7.5 Angina, % 15.1 9.3 4.8 4.9 Congestive heart failure, % 8.8 10.1 3.6 3.0 Current smoking, % 29.4 19.2 23.7 14.2 Bilateral oophorectomy, % 20.9 23.8 23.3 21.6 Current estrogen use, % 22.8 25.8 30.4 29.3 Current progestin use, % 8.8 8.8 15.2 11.8 SI conversion factor: To convert cholesterol to millimoles per liter, multiply by 0.02586. 1986, and December 31, 2001, or with an incident ischemic or hemorrhagic stroke between July 1, 1989, and December 31, 2001. Cases were identified by hospital discharge diagnosis codes and Washington State death records and were validated by medical record review as previously described. 11,12 Control subjects were postmenopausal women sampled at random from the GHC population and were frequency matched to cases on age, the presence of treated hypertension, and calendar year of identification. Controls met the same eligibility criteria as cases but had not had an MI or a stroke. Each participant was assigned an index date. For cases, the index date was the hospital admission date for the first MI or stroke or the date of death for those with out-of-hospital fatal events. For controls, the index date was a random date within the year for which they were sampled. We excluded case subjects whose MI or stroke was a complication of a procedure or surgery, and we excluded those with fewer than 4 visits before their index date to increase the likelihood that information would be available in the medical record on important clinical characteristics. ASSESSMENT OF CURRENT USE OF HMONE THERAPY AT THE INDEX DATE The GHC computerized pharmacy database was used to assess current hormone use as previously described. 13 To determine current use at the index date, we searched the pharmacy data for the hormone prescription immediately preceding the index date. If a woman received enough pills to last until her index date (assuming 80% compliance), she was considered a current user at the index date. Analyses using a 100% compliance assumption yielded similar results (data not shown). Recency of starting the current hormone regimen at the index date was calculated separately for CEE and EE and was defined as the number of days between the index date and the date of the first prescription fill for the drug continually used before 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. Data from telephone interviews indicated that 96.8% of controls and 94.7% of MI and stroke cases filled all or almost all (90%-100%) of their prescriptions at a GHC pharmacy. For this analysis, we excluded current (at the index date) users of progestin without estrogen (n=69; 1.0% of otherwise eligible cases and controls), current users of estrogen cream or patches without estrogen pills (n=138; 2.0%), women with no record of a prescription for any drug in the 5 years before the index date (n=52; 0.7%), and current users of oral estradiol (n=39; 0.6%). All current users of postmenopausal estrogen included in the analysis used either oral EE or oral CEE. For CEE and EE, a medium daily dose of estrogen was defined as 0.625 mg, low as less than 0.625 mg, and high as greater than 0.625 mg. STATISTICAL ANALYSIS Results of observational studies of hormone use and cardiovascular disease are susceptible to bias to the extent that women who use and do not use hormone therapy have a different underlying risk of cardiovascular disease. 14,15 For this reason, our main analyses were restricted to current users of hormone therapy at the index date, with users of EE serving as the reference group. We used logistic regression analysis to obtain odds ratios (s) for the separate outcomes of MI, ischemic stroke, and hemorrhagic stroke. Data were missing on current smoking, angina, or systolic blood pressure for fewer than 1% of the women; values were imputed for these women using multiple imputation (Stata statistical software; StataCorp, College Station, Tex). RESULTS There were 1644 eligible postmenopausal women who had an incident fatal or nonfatal MI between July 1, 1986, and December 31, 2001, and 1080 women who had an incident fatal or nonfatal stroke between July 1, 1989, and December 31, 2001 (830 with ischemic stroke and 250 with hemorrhagic stroke). There were 4205 eligible control subjects, of whom 3507 were assigned index dates between July 1, 1989, and December 31, 2001, the range of index dates for the stroke cases. The mean age of cases and controls was 67 to 70 years, and the average duration of enrollment in GHC before the index date was 18 to 20 years (Table 1). As expected, risk factors for atherosclerotic disease were more prevalent in MI and ischemic stroke cases than in control subjects, and hemorrhagic stroke cases had 400

Table 2. Risk Factors in Relation to Current Hormone Use and Type in Control Subjects Variable No Current Use (n = 2972) EE Only (n = 366) EE With Progestin (n = 285) CEE Only (n = 378) CEE With Progestin (n = 204) Age, mean, y 68.6 66.2 62.1 65.1 62.2 White race, % 90.0 91.8 93.7 96.0 87.8 Years of enrollment, mean, 19.6 19.8 21.7 16.8 18.3 Physician visits in past year, mean, 5.9 6.4 5.3 6.8 5.0 Hypertension, % 44.2 57.7 36.8 44.7 33.8 Systolic blood pressure, mean, mm Hg 139.0 137.2 131.2 138.0 131.6 Cholesterol, mean, mg/dl 237.0 234.2 222.1 235.4 224.3 Weight, mean, kg 72.0 74.4 72.3 72.1 70.3 Diabetes mellitus, % 8.3 7.1 3.5 5.8 4.4 Angina, % 5.6 4.9 1.4 3.2 2.5 Congestive heart failure, % 3.1 4.6 1.4 2.1 1.5 Current smoking, % 15.1 12.0 11.2 11.5 13.7 Bilateral oophorectomy, % 17.2 49.3 1.4 55.5 5.4 Time since starting current regimen of hormone therapy, mean, y NA 3.7 2.9 2.9 2.1 Abbreviations: CEE, conjugated equine estrogen; EE, esterified estrogen; NA, not applicable. SI conversion factor: To convert cholesterol to millimoles per liter, multiply by 0.02586. 160 140 Conjugated Equine Estrogen Esterified Estrogen 120 Control Subjects, 100 80 60 40 20 0 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Year Figure. Time trends in the use of conjugated equine estrogen and esterified estrogen at Group Health Cooperative, 1986-2001. a higher mean systolic blood pressure and a higher rate of smoking than control subjects. Users of hormone therapy were, in general, healthier and had fewer risk factors for atherosclerotic disease than nonusers (Table 2). Users of either EE or CEE in combination with a progestin were younger and, except for smoking, had fewer risk factors than users of EE or CEE alone or nonusers. In 1991, EE replaced CEE as the standard postmenopausal estrogen on the GHC formulary (Figure). In 1999, a second change was made from EE back to CEE. For both formulary switches, the predominant postmenopausal estrogen prescribed changed substantially within 1 year of implementation. Overall, there was little or no difference in the adjusted risk of MI, ischemic stroke, or hemorrhagic stroke associated with the use of EE or CEE with or without progestin compared with nonuse of hormone therapy (Table 3). Results were similar after further adjustment for congestive heart failure, number of physician visits, and use of statins and when women with congestive heart failure or angina were excluded. Education (as a measure of socioeconomic status) was available for 60% of the women; there was no evidence of confounding by educational achievement in analyses limited to this subset. All further analyses were restricted to current users of hormone therapy, with EE users as the reference group, and were conducted for MI and ischemic stroke cases and controls only. Because of the small number of hemorrhagic stroke cases, this group was not considered in CEE-EE comparisons. Compared with EE use, the use of CEE was not associated with an increased risk of MI (adjusted, 0.93; 95% confidence interval [CI], 0.65-1.31) 401

Table 3. Risk of MI, Ischemic Stroke, and Hemorrhagic Stroke in Relation to the Use and Type of Postmenopausal Estrogen in Users and Nonusers Current Use and Type of Estrogen* MI Cases vs 1986-2001 MI Cases, (95% CI)* Ischemic Stroke Cases, Stroke Cases vs 1989-2001 (95% CI)* Hemorrhagic Stroke Cases, (95% CI)* None 2972 1270 1.00 2408 616 1.00 174 1.00 EE alone 366 117 0.84 (0.66-1.07) 366 67 0.86 (0.64-1.16) 24 0.89 (0.56-1.40) EE plus progestin 285 78 0.87 (0.66-1.15) 285 47 0.98 (0.69-1.38) 24 1.13 (0.71-1.80) CEE alone 378 113 0.75 (0.59-0.95) 286 74 1.16 (0.87-1.56) 16 0.83 (0.48-1.43) CEE plus progestin 204 66 0.87 (0.64-1.19) 162 26 0.86 (0.55-1.34) 12 1.02 (0.54-1.91) Abbreviations: CEE, conjugated equine estrogen; CI, confidence interval; EE, esterified estrogen; MI, myocardial infarction;, odds ratio. *Adjusted for age, year, hypertension, diabetes mellitus, angina, current smoking, and systolic blood pressure. Table 4. Risk of Myocardial Infarction and Ischemic Stroke in Relation to Type of Estrogen in Current Users of Postmenopausal Estrogen Variable Myocardial Infarction Cases, (95% CI) Ischemic Stroke Cases, (95% CI) All current estrogen users* EE 651 195 1.00 651 114 1.00 CEE 582 179 0.93 (0.65-1.31) 448 100 1.31 (0.88-1.97) By progestin* EE alone 366 117 1.00 366 67 1.00 CEE alone 378 113 1.06 (0.70-1.60) 286 74 1.57 (0.98-2.53) EE plus progestin 285 78 1.00 285 47 1.00 CEE plus progestin 204 66 1.12 (0.71-1.77) 162 26 0.96 (0.53-1.72) By daily estrogen dose Low-dose EE 68 13 1.00 68 13 1.00 Low-dose CEE 78 17 1.30 (0.54-3.14) 53 5 0.49 (0.15-1.58) Medium-dose EE 545 175 1.00 545 96 1.00 Medium-dose CEE 419 132 1.00 (0.69-1.45) 324 74 1.28 (0.82-1.98) High-dose EE 38 7 1.00 38 5 1.00 High-dose CEE 85 30 2.22 (0.82-5.97) 71 21 2.59 (0.83-8.07) By recency of starting current estrogen regimen 6 mo, EE 50 8 1.00 50 12 1.00 6 mo, CEE 63 30 2.33 (0.93-5.82) 63 19 1.06 (0.43-2.65) 6 mo, EE 601 187 1.00 601 102 1.00 6 mo, CEE 519 149 1.02 (0.71-1.46) 385 81 1.34 (0.87-2.06) Abbreviations: CEE, conjugated equine estrogen; CI, confidence interval; EE, esterified estrogen;, odds ratio. *The s are adjusted for age, year, hypertension, diabetes mellitus, angina, current smoking, and systolic blood pressure. The s are adjusted for age, year, and hypertension. or ischemic stroke (adjusted, 1.31; 95% CI, 0.88-1.97) (Table 4). The was little changed by additional adjustment for progestin use, heart failure, and number of physician visits in the year before the index date. Among users of estrogen alone (without progestin), there was no difference in MI risk between users of CEE and EE, but for ischemic stroke, there was a suggestion of higher risk associated with use of CEE alone compared with EE alone (, 1.57; 95% CI, 0.98-2.53; P=.06). Among women who used a high daily dose of estrogen ( 0.625 mg), there was a suggestion of higher risk of MI (, 2.22; 95% CI, 0.82-5.97; P=.12) and ischemic stroke (, 2.59; 95% CI, 0.83-8.07; P=.10) associated with CEE use compared with EE use, but the number of subjects in these analyses was small and the CIs were wide. In addition, there was a suggestion of higher risk of MI associated with use of CEE initiated within 6 months of the index date compared with EE use initiated during the same period (, 2.33; 95% CI, 0.93-5.82; P=.07). COMMENT In this observational study, overall there was no difference in risk of MI, ischemic stroke, or hemorrhagic stroke among nonusers of hormone therapy, current users of CEE, and current users of EE. Compared with EE use, there was a suggestion that CEE use was associated with higher risk of ischemic stroke in women who used estrogen without progestin, with higher risk of MI and ischemic stroke in women using more than 0.625 mg of estrogen per day, and with higher risk of MI in women 402

who initiated their current estrogen regimen within 6 months of their index date. The strengths of this study include the large number of MI and stroke cases, the use of population-based cases and controls, the validation of case diagnoses, and the comparable ascertainment of potential confounding factors and hormone use in cases and controls. All subjects were enrollees of an HMO and thus had similar access to health care. The choice of CEE vs EE was largely guided by changes in the formulary rather than by patient characteristics. However, patients and their physicians selfselected hormone therapy and its regimen and dose, and it is possible that this self-selection could have introduced bias. In several of the exposure categories we studied, particularly those relating to estrogen dose and recency of starting the current estrogen regimen, the number of case subjects was small; thus, our findings should be viewed as hypothesis generating. Results from the Women s Health Initiative hormone therapy trials indicated that use of CEE combined with medroxyprogesterone acetate increased the risks of coronary heart disease and stroke 16,17 and that use of CEE alone increased the risk of stroke compared with placebo. 4 In contrast, in the present study, the estimated risks associated with CEE use with and without progestin tended to be lower compared with nonuse, as has often been found inobservationalstudies.recently,prenticeandcolleagues 18 reported that the discrepancy in results between the Women s Health Initiative randomized clinical trial and the Women s Health Initiative observational study could be largely explained by 2 factors: (1) the risks in the trial decreased across time and hormone users in the observational study were largely long-term users and (2) hormone users had fewer risk factors than nonusers in the observational study. The second factor has been called the healthy user bias (hormone users are inherently at lower risk for coronary heart disease and stroke than nonusers). 19 After taking into account time since initiation and confounders, risk estimates for coronary heart disease and venous thromboembolism for the trial and observational study did not differ significantly. However, some discrepancy remained in the risk estimates for stroke. Ray 20 proposed the new-user design as an observational study method to identify and control for risks that may vary across time. In the present study, few women had recently initiated hormone therapy, and our estimates largely reflect longterm use. In an effort to minimize the healthy user bias, we removed nonusers from the main analyses and compared CEE users with EE users. Users of either hormone are more similar to each other than to nonusers (Table 2). Furthermore, the choice of CEE or EE was based on the formulary and not on patient characteristics. For these reasons, the bias that arises from comparisons with nonusers would be minimized in the comparisons between users of 2 types of hormone therapy, CEE and EE. In the Women s Estrogen for Stroke Trial 7 and the Estrogen in the Prevention of Reinfarction Trial, 5 both secondary prevention trials, there was no difference between oral estradiol alone and placebo in the risk of stroke or coronary disease events, respectively. All trials to date have studied either CEE or estradiol; to our knowledge, there are no clinical trial data of the association of EE with cardiovascular event risk compared with either placebo or another estrogen. Few studies have directly compared the effects of CEE and EE in relation to any end point. Small short-term studies showed that achieved serum levels of estrone and estradiol were similar with CEE and EE use, 21 and shortterm EE treatment was associated with improved scores on a depression scale and a cognitive performance test compared with use of CEE. 22 In the first large study to examine the risk of major cardiovascular end points in users of CEE and users of EE, our group 23 suggested that compared with nonusers of estrogen, the risk of venous thromboembolism was higher in users of CEE (, 1.65; 95% CI, 1.24-2.19) but not in users of EE (, 0.92; 95% CI, 0.69-1.22). Furthermore, compared with use of EE, use of CEE was associated with higher risk of venous thromboembolism (, 1.78; 95% CI, 1.11-2.84). 23 Some evidence suggests differences in the effects of various constituents of CEE and EE on the lipid profile, low-density lipoprotein oxidation, insulin action, vasomotor tone, and vascular wall metabolism. 24 Such differences may underlie any differences in the effects of the 2 estrogen preparations on major cardiovascular disease end points. Published studies indicate that EE has been used in 2 HMOs in the United States, 10,25 but outside of the HMO setting, CEE has dominated sales of postmenopausal estrogen. 26 Only a small proportion ( 2%) of prescriptions for postmenopausal estrogens filled at retail pharmacies between August 2003 and August 2004 were for EE, according to data from NDCHealth, a health care information company. Thus, in the United States, further study of differences in the health effects of CEE and EE may require the initiation of clinical trials. In this observational study, overall, there was no difference in risk of MI or stroke associated with use of CEE vs EE, but there was a suggestion of higher ischemic stroke risk associated with use of CEE alone and higher MI risk associated with recent initiation of CEE use. Further study may be warranted of the effects of EE on the risk of cardiovascular end points. Accepted for Publication: November 12, 2005. Correspondence: Rozenn N. Lemaitre, PhD, MPH, Cardiovascular Health Research Unit, University of Washington, 1730 Minor Ave, Suite 1360, Seattle, WA 98101-1448 (rozenl@u.washington.edu). Author Contributions: Drs Lemaitre and Heckbert had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Financial Disclosure: None. Funding/Support: This study was supported by grants HL40628, HL43201, HL53375, and HL68639 from the National Heart, Lung, and Blood Institute, Bethesda, Md, and by grant 9970178N from the American Heart Association, Dallas, Tex. REFERENCES 1. Hulley S, Grady D, Bush T, et al. Randomized trial of estrogen plus progestin for secondary prevention of coronary heart disease in postmenopausal women. JAMA. 1998;280:605-613. 403

2. Simon JA, Hsia J, Cauley JA, et al. Postmenopausal hormone therapy and risk of stroke: the Heart and Estrogen-progestin Replacement Study (HERS). Circulation. 2001;103:638-642. 3. Rossouw JE, Anderson GL, Prentice RL, et al. Risks and benefits of estrogen plus progestin in healthy postmenopausal women: principal results from the Women s Health Initiative randomized controlled trial. JAMA. 2002;288:321-333. 4. Anderson GL, Limacher M, Assaf AR, et al. Effects of conjugated equine estrogen in postmenopausal women with hysterectomy: the Women s Health Initiative randomized controlled trial. JAMA. 2004;291:1701-1712. 5. Cherry N, Gilmour K, Hannaford P, et al. Oestrogen therapy for prevention of reinfarction in postmenopausal women: a randomised placebo controlled trial. Lancet. 2002;360:2001-2008. 6. Clarke SC, Kelleher J, Lloyd-Jones H, Slack M, Schofiel PM. A study of hormone replacement therapy in postmenopausal women with ischaemic heart disease: the Papworth HRT atherosclerosis study. BJOG. 2002;109:1056-1062. 7. Viscoli CM, Brass LM, Kernan WN, Sarrel PM, Suissa S, Horwitz RI. A clinical trial of estrogen-replacement therapy after ischemic stroke. N Engl J Med. 2001; 345:1243-1249. 8. Bhavnani BR. Pharmacokinetics and pharmacodynamics of conjugated equine estrogens: chemistry and metabolism. Proc Soc Exp Biol Med. 1998; 217:6-16. 9. O Connell MB. Pharmacokinetic and pharmacologic variation between different estrogen products. J Clin Pharmacol. 1995;35:18S-24S. 10. Baluch WM, Gardner JS, Krauss RH, Scholes D. Therapeutic interchange of conjugated and esterified estrogens in a managed care organization. Am J Health Syst Pharm. 1999;56:537-542. 11. Lemaitre RN, Heckbert SR, Psaty BM, Smith NL, Kaplan RC, Longstreth WT Jr. Hormone replacement therapy and associated risk of stroke in postmenopausal women. Arch Intern Med. 2002;162:1954-1960. 12. Psaty BM, Heckbert SR, Atkins D, et al. The risk of myocardial infarction associated with the combined use of estrogens and progestins in post-menopausal women. Arch Intern Med. 1994;154:1333-1339. 13. Heckbert SR, Weiss NS, Koepsell TD, et al. Duration of estrogen replacement therapy in relation to the risk of incident myocardial infarction in postmenopausal women. Arch Intern Med. 1997;157:1330-1336. 14. Barrett-Connor E. Commentary: observation versus intervention: what s different? Int J Epidemiol. 2004;33:457-459. 15. Humphrey LL, Chan BK, Sox HC. Postmenopausal hormone replacement therapy and the primary prevention of cardiovascular disease. Ann Intern Med. 2002; 137:273-284. 16. Manson JE, Hsia J, Johnson KC, et al. Estrogen plus progestin and the risk of coronary heart disease. N Engl J Med. 2003;349:523-534. 17. Wassertheil-Smoller S, Hendrix SL, Limacher M, et al. Effect of estrogen plus progestin on stroke in postmenopausal women: the Women s Health Initiative: a randomized trial. JAMA. 2003;289:2673-2684. 18. Prentice RL, Langer R, Stefanick ML, et al. Combined postmenopausal hormone therapy and cardiovascular disease: toward resolving the discrepancy between observational studies and the Women s Health Initiative clinical trial. Am J Epidemiol. 2005;162:404-414. 19. Matthews KA, Kuller LH, Wing RR, Meilahn EN, Plantinga P. Prior to use of estrogen replacement therapy, are users healthier than nonusers? Am J Epidemiol. 1996;143:971-978. 20. Ray WA. Evaluating medication effects outside of clinical trials: new-user designs. Am J Epidemiol. 2003;158:915-920. 21. Jurgens RW Jr, Downey LJ, Abernethy WD, Cutler NR, Conrad J. A comparison of circulating hormone levels in postmenopausal women receiving hormone replacement therapy. Am J Obstet Gynecol. 1992;167:459-460. 22. Friebely JS, Shifren JL, Schiff I, Regestein QR. Preliminary observations on differing psychological effects of conjugated and esterified estrogen treatments. J Womens Health Gend Based Med. 2001;10:181-187. 23. Smith NL, Heckbert SR, Lemaitre RN, et al. Esterified estrogens and conjugated equine estrogens and the risk of venous thrombosis. JAMA. 2004;292: 1581-1587. 24. Ansbacher R. The pharmacokinetics and efficacy of different estrogens are not equivalent. Am J Obstet Gynecol. 2001;184:255-263. 25. Andrade SE, Gurwitz JH, Cernieux J, Fish LS. Evaluation of a formulary switch from conjugated to esterified estrogens in a managed care setting. Med Care. 2000;38:970-975. 26. HershAL,StefanickML,StaffordRS.Nationaluseofpostmenopausalhormonetherapy: annual trends and response to recent evidence. JAMA. 2004;291:47-53. Clinical Trial Registration Announcement In concert with the International Committee of Medical Journal Editors (ICMJE), Archives of Internal Medicine will require, as a condition of consideration for publication, registration of clinical trials in a public trials registry (such as http://clinicaltrials.gov or http://controlled -trials.com). Trials must be registered at or before the onset of patient enrollment. This policy applies to any clinical trial starting enrollment after March 1, 2006. For trials that began enrollment before this date, registration will be required by June 1, 2006. The trial registration number should be supplied at the time of submission. For details about this new policy see the editorials by DeAngelis et al in the September 8, 2004 (2004;292: 1363-1364) and June 15, 2005 (2005;293:2927-2929) issues of JAMA. 404