NONRANDOMIZED STUDIES OF EFFECTS OF HORMONE REPLACEMENT THERAPY

• Nurses' Health Study prospectively studied 59,000 women for 16y. RR of "major coronary events" was 0.60 and 0.39 for users of unopposed estrogen and combined HRT, respectively, compared with never-users of postmenopausal HRT, after adjustment for other risk factors. RR for CVA was 1.27 and 1.09, respectively (NEJM 335:453, 1996-JW)
• Nurses' Health Study followed 70,533 women over 20y; after adjusting for risk factors and age, RR for coronary events was 0.61 for current HRT users c/w never-users; sig. reduction in risk applied to both subgroups using estrogen-only and combined estrogen-progestin HRT (Ann. Int. Med. 133:933, 2000--JW)
• Case-control trial of 438 postmenopausal women hosp'd w/MI vs. 438 age-matched controls found no sig. diff. in current or past use of HRT or duration of HRT and MI, though wide confidence intervals (Ann. Int. Med. 127:501, 1997-JW)
• In a corollary study as part of the Beta-Blocker Evaluation of Survival Trial (BEST), mortality among 493 women > 50yo with Heart Failure who were HRT users vs. non-users, after adjustment for potential confounders, was 0.6%; this advantage was only noted in women with a non-ischemic etiology for heart failure.  Adjustment was made for LVEF, NYHA functional classification, etiology of HF (ischemia = CAD, no ISCHEMIA = no CAD), race, age, diabetes, and treatment group assignment (bucindolol or placebo).  No difference in survival was seen between pts on combination-HRT and unopposed estrogen.
• One case-control study showed no difference in reduction of MI risk with combined vs. unopposed estrogen (Arch Int. Med. 154:1333, 1994-Med. Letter)

Dementia:

• Longitudinal study of 1124 women with no Alzheimer's, Parkinson's, or CVA at baseline; followed 1-5y. RR for developing Alzheimer's over f/u period was 0.4 for estrogen users; remained sig. after adjusting for age, education, ethnicity, and apolipoprotein E genotype. Less risk with longer durations of use. (Lancet 348:429, 1996-JW)
• Cohort study followed at 8877 female residents of a retirement community for 14 years. Compared 248 who died with dx of Alzheimer's to 1000 controls matched for year of birth and death. OR for Alzheimer's in estrogen users was 0.65 regardless of route of administration. Risk decreased with increased dose and duration (Arch. Int. Med. 156:2213, 1996-JW)
• Retrospective study of 514 women who participated in the Baltimore Longitudinal Study of Aging, which involved gathering info on meds etc. and Q2y physical and neurologic testing, found that RR of Alzheimers disease in estrogen users c/w nonusers was 0.46; no change in effect w/increasing duration of use. This cohort was followed from 1978-94 so probably many were on unopposed estrogen but summary doesn't state (Neurology 48:1517, 1997--AFP)
• 2705 postmenopausal women prospectively studied x 6y; estrogen use, controlling for demographic characteristics and health-related variables, was not ass'd with sig. reduced risk for cognitive decline as measured by the Short Portable Mental Status Questionnaire (Am. J. Epidem. 153:137, 2001--JW)

Venous thromboembolism:

• Case-control study of 103 British women aged 45-64 with DVT and 178 matched controles with dx's not related to HRT use; excluded women with h/o Ca, recent surg, or bedrest or h/o thrombotic event inc. prior DVT, MI, CVA; OR for DVT among current HRT users was 3.5; Absolute increase in incidence was from 1/10,000 women/year to 3/10,000 women/yr (Lancet 348:977, 1996)
• Case-control study of 42 women from Seattle with DVT and 168 controls found RR 3.0-4.0 for DVT with HRT use (Lancet 348:981, 1996)
• In a case-control study of postmenopausal women 45-70yo with idiopathic venous thromboembolism, after adjusting for potential confounders, current use of oral estrogen was associated with OR of 3.9 for VTE relative to never-users; transdermal estrogen was ass'd with OR 0.9 (Lancet 362:428, 2003--AFP)
• In a case-control study of postmenopausal women 30-89yo with a venous thromboembolism and pts matched on age, hypertension status, and calendar year, use of esterified estrogen was not ass'd with sig. increased risk but conjugated equine estrogen was ass'd with a sig. increased risk; for the latter, a sig. dose-response effect was noted (OR 1.78) (JAMA 292:1581, 2004--abst)

Breast cancer:

• Prospective study of 23,000 women 35yo and older followed for mean 5.7y.  RR of breast Ca for any estrogen use was 1.1; 1.7 after 9y.  Most women were on estradiol; NO INC. RISK for women on conjugated estrogens! (22% of cohort).  Combination estrogen-progestin associated with RR of 4.0 with 6y or longer of use (NEJM 321:293, 1989)
• Prospective study of 4500 women on HRT, for mean duration 5.5y.  43% had combination with progestins, though "less than now considered adequate"; RR for breast Ca was 1.59; less with combination HRT than with unopposed estrogen (Br. J. Obs. Gyn. 94:620, 1987)
• In a case control study looking at use of estrogen-only and combined HRT, with 537 ALL WHITE pts 50-64yo with breast Ca and 492 ALL WHITE controls aged 50-64y, for most HRT users, estrogen was Premarin, progestin was Provera, after controlling for various known risk factors for breast Ca:
  • RR 0.9 (95% CI 0.6-1.1) with ever use of estrogen-only HRT
  • RR 1.0 (95% CI 0.5-2.0) with >20y use of estrogen-only HRT
  • RR 0.9 (95% Ci 0.7-1.3) with ever use of combined HRT
  • RR 0.4 (95% CI 0.2-1.0) with use for >8y of combined HRT; for this figure the authors note that limited # of subjects were involved (prob. because combined HRT only recently became common), and study had 80% power to detect a 2.5-fold increase of RO of breast Ca in pts using combined HRT > 8y. (JAMA 274:137, 1995)
• In a prospective study of women 30-55yo, there was no increased risk for breast Ca with past use of estrogens; RR with current use was 1.36; not affected by duration of use.  In a followup analysis published in 1995, 
  • RR for breast Ca was 1.32 for current users of unopposed estrogen
  • RR 1.41 for current users of combined HRT (both c/w never-users of postmenopausal estrogen)
  • RR 1.46 for use of any postmenopausal HRT for 5-9y and for 1.46 10y or more
  • RR 1.71 for women 60-64yo taking any HRT for >5y

  ("Nurses' Health Study," JAMA 264:2648, 1990; NEJM 332:1589, 1995)

• In a prospective study of 37,105 postmenopausal women 55-69yo with no h/o non-skin Ca or mastectomy; over 11y of f/u, duration of HRT was sig. ass'd with risk for invasive Ca with a "favorable histology" (RR c/w never-users was 4.4 for HRT use < 5y and 2.6 for HRT use > 5y!) but not ass'd with risk for DCIS or invasive ductal or lobular Ca ("Iowa Women's Health Study," JAMA 281:2091, 1999--JW)
• In a case-control study of 1897 postmenopausal women w/breast Ca and 1637 postmenopausal controls (none who had had hysterectomy), HRT was ass'd with significantly increased risk for breast Ca (RR 1.1 for each 5y period of use of any HRT; 1.24 for combined HRT, 1.06 for estrogen-only HRT (J. Nat. Ca. Inst. 92:328, 2000--JW)
• In a cohort study of 46,355 postmenopausal women being screened for breast Ca, only use of HRT within the previous 4y was ass'd with detectable increase in risk of breast Ca; RR was increased by 0.01 (sig.) with each year of estrogen-only use and 0.08 (sig.) with each year of combined estrogen-progestin use (adjusted for BMI, age at menopause, mammographic screening, education, and age). Diff. between estrogen-only and combined-HRT incidences of brast Ca were sig; no diff. in invasiveness seen in breast Ca's in women on one vs. the other type of HRT. The sig. increased risk of breast Ca for both estrogen-only and combined HRT was limited to women with BMI < 24.4--there was a trend towards increased risk of breast Ca with use of combined HRT (though not with use of estrogen-only HRT) in heavier women (JAMA 283:485, 2000)
• Meta-analyses
  • One concluded that estrogen HRT doesn't incr. risk of breast Ca; another concluded that risk of breast Ca incr. with 15y of estrogen HRT but not before (Arch. Int. Med. 151:67, 1991; JAMA 265:1985, 1991; ref'd in NEJM rvw.)
  • One summing up comparisons of 52,000 women w/invasive breast Ca and 108,000 women w/o it (Lancet 350:1047, 1997-JW) concluded:

  1. RR for breast Ca increased w/increasing duration of HRT
  2. RR increased 2.3% for each year of use (sig.)
  3. No sig. increased risk for women who had not taken HRT in >5y

Overall mortality

• In a case-control study of 3,637 deaths and 34,625 matched controls among a cohort of female RN's aged 30-55 and with no h/o cardiovascular disease or cancer at intake, all among postmenopausal women who had provided info about postmenopausal hormone use, after adjustment for potential confounders, current HRT use was ass'd with RR 0.63 for death vs. never-users (sig.); apparently mostly due to reduction in CAD deaths (RR 0.47; sig.); as well as some reduction in deaths due to Ca (RR 0.71; sig.; authors mention a RR of 0.46 for mortality from colorectal Ca and 0.76 for mortality from breast Ca among current users but don't give CI's.).  Past use of hormones not ass'd w/sig. altered risk of death (RR 1.03; nonsig.)
  • Current users w/ 1 or more coronary risk factors (smoking, high chol., HTN, DM, parental hx of premature MI, BMI 29 or higher) had a particularly sharp reduction in risk for death (RR 0.51 for death; sig.); RR 0.89 for death for those w/o the above risk factors for CAD (nonsig.; 95% CI 0.62-1.28).
  • Current users w/mother or sister with breast Ca had similar RR of death c/w never users as current users w/o such a family hx (RR 0.65 vs; 0.60; both sig.).
  • Current users < 50yo had RR of death c/w never-users of 1.05 (nonsig.).
  • Current users w/BMI > 29 had similar RR of death c/w never-users as women in other weight groups (0.54 vs. 0.63 for women w/BMI < 23; both sig.).
  • For combined HRT, RR of death for current users was 0.46 (sig.); for estrogen-only HRT, RR of death for current users was 0.69 (sig.)

("Nurse's Health Study," NEJM 336:1769, 1997)

• In a prospective study of 8900 women followed prospectively for 7.5y, estrogen users had RR 0.8 for death from any cause.  Mortality decreased with increased duration of use and was lower for current than for past users. RR for mortality was 0.6 for current users for more than 15y.  Most decrease evident in cardiac causes of death.  Abstract didn't mention combined regimens (Arch. Int. Med. 151:75, 1991).
• In a retrospective comparison of 232 postmenopausal estrogen users for >5y (mean length of use 17y) with 222 age-matched controls, current users had RR 0.54 for death from any cause in users; mostly due to reduced cardiovascular deaths; abstract didn't mention combined regimens (Obs. Gyn 87:6, 1996)
• In a case-control of 3,600 postmenopausal women from Nurses' Health Study who had died and 35,000 living, age-matched controls, after adjustment for various risk factos, current hormone users had sig. reduction in overall mortality c/w never-users of hormones (RR 0.63).  There was less risk reduction for women who had taken hormones x >10y (RR 0.8), mostly due to increase in death from breast Ca in that group.  Current users w/ at least 1 cardiovascular risk factor had a 50% reduction in death, while those w/no cardiac risk factors had no sig. reduction.  For past users, protection against death disappeared about 5y after stopping hormones (NEJM 336:1769, 1997-JW)