Current issues of ACP Journal Club are published in Annals of Internal Medicine


Review: Observational studies adjusting for socioeconomic status and lifestyle show no association between HRT and CAD


ACP J Club. 2003 Mar-Apr;138:40. doi:10.7326/ACPJC-2003-138-2-040

Related Content in this Issue
• Companion Abstract and Commentary: Review: Risks and benefits of HRT comparing various sources of evidence

Source Citation

Humphrey LL, Chan BK, Sox HC. Postmenopausal hormone replacement therapy and the primary prevention of cardiovascular disease. Ann Intern Med. 2002;137:273-84. [PubMed ID: 12186518]



What is the effectiveness of postmenopausal hormone replacement therapy (HRT) for primary prevention of cardiovascular disease?

Data sources

Studies were identified by searching MEDLINE and Cochrane databases from 1966 to December 2000 and by reviewing bibliographies of relevant studies and other publications.

Study selection

Studies were selected if they were randomized controlled trials (RCTs), cohort studies, or case–control studies that assessed the use of HRT for primary prevention of cardiovascular disease in postmenopausal women and if English-language abstracts were available.

Data extraction

Data were extracted on study design, type of HRT (unopposed estrogen or estrogen plus progestin), definitions of cardiovascular disease, and potential risk factors included in multivariate models. 2 investigators independently assessed the quality of individual studies. Formal review and meta-analysis were limited to studies of good or fair quality and included RCTs, cohort studies with internal controls, and population-based case–control studies with ≥ 3 years of follow-up. HRT use was classified as current, past, ever, or any (combined current, past, or ever).

Main results

Only the results for coronary artery disease (CAD) incidence are reviewed here (3 cohort studies, 9 case–control studies, and 1 small RCT). Studies that did not adjust for socioeconomic status (SES) found that current and past use of HRT reduced CAD incidence, whereas studies that adjusted for SES found no association between any measure of HRT use and CAD (Table). Similar results were found when analyses were stratified by studies that adjusted for alcohol consumption, exercise, or both.


Meta-analysis of studies that adjust for socioeconomic status or alcohol consumption and exercise shows that current, past, ever, or any use of HRT does not reduce coronary artery disease (CAD). Studies that do not adjust for these factors show a reduced risk for CAD with current and past use.

Source of funding: Agency for Healthcare Research and Quality.

For correspondence: Dr. L.L. Humphrey, Veterans Affairs Medical Center and Oregon Health and Science University, Portland, OR, USA. E-mail

Table. Relative risk for coronary artery disease according to use of hormone replacement therapy (HRT)

Use of HRT Relative risk (95% credible interval)
Unadjusted* Adjusted†
Current 0.71 (0.64 to 0.78) 0.97 (0.82 to 1.16)
Past 0.78 (0.69 to 0.87) 1.07 (0.90 to 1.27)
Ever Not reported 1.11 (0.84 to 1.53)
Any Not reported 1.04 (0.79 to 1.44)

*Includes data from studies that did not adjust for socioeconomic status.
†Includes data from studies that adjusted for socioeconomic status.


The 2 meta-analyses by Humphrey and colleagues and Nelson and colleagues, and much of the medical literature, do not clearly define postmenopausal HRT. Although authors often claim that many observational studies show the benefits of HRT on CAD risk, the data almost completely relate to unopposed estrogen (1). Few epidemiologic studies have evaluated estrogen plus progestin administered as continuous combined therapy. A limited number of studies were included in the meta-analyses by Humphrey and Nelson and their colleagues, and the included studies sometimes failed to specify type of HRT. Studies published in the 1980s and earlier likely refer to unopposed estrogen therapy.

The lack of specificity in type of HRT is problematic because of the potential for different effects of estrogen alone and estrogen plus progestin. It is important to note that the meta-analyses used a common protocol to apply standard criteria for quality (2) and included only studies of higher quality. For studies assessing CAD outcomes, the authors identified a relation between poorer quality studies and greater protection against CAD.

Changing patterns of use (estrogen alone replaced by combined estrogen plus progestin) and indications for use (e.g., starting women on therapy before menopause to reduce bone loss) may account, in part, for differences in results. Also, acute effects of combined therapy (e.g., the prothrombotic and proinflammatory effects of progestins) may be missed in prospective studies because of lack of attention to early events and follow-up.

Data from both epidemiologic studies and RCTs of HRT consistently show other thrombotic effects, such as stroke and pulmonary embolus. This argues, at least in part, against a bias related to this pathway in the observational studies. Although SES appears as one other important variable in this analysis, the authors note that the range of cardiovascular risk factors controlled for in observational studies also varied substantially. The authors, however, did not point out that studies that controlled for SES observed similar results before and after such control, which suggests that this is not an explanation for the discrepancy between RCT and observational study results. Is confounding by indication changing over time concurrent with changing patterns of drug combination? Is SES merely a marker for more recent studies that evaluate estrogen plus progestin? Alternatively, is the timing of exposure in relation to menopause the explanatory factor? Animal studies suggest that estrogens have beneficial effects in the early stages of atherogenesis but reduced beneficial effects in the final stages of plaque complications (3).

How can drug formulation change over time and most associations observed in epidemiologic studies (i.e., breast cancer, pulmonary embolus, stroke, colon cancer, cholecystitis, and osteoporotic fracture) be consistent, and yet CAD outcomes diverge? Consistent evidence across the RCTs argues against chance.

The consistency between non-RCTs and RCTs for noncardiac outcomes is reassuring and supports other evaluations of the contribution of different study designs to evaluation of medical therapies (4−7). Such studies comparing designs may be limited in power, but on average, well-designed observational studies show a similar magnitude of estimated benefits as RCTs. We should not focus solely on study design but must also consider the formulation and timing of use of postmenopausal HRT that is being evaluated.

Graham Colditz, MD
Channing Laboratory
Boston, Massachusetts, USA


1. Stampfer MJ, Colditz GA. Estrogen replacement therapy and coronary heart disease: a quantitative assessment of the epidemiologic evidence. Prev Med. 1991;20:47-63. [PubMed ID: 1826173]

2. Harris RP, Helfand M, Woolf SH, et al. Current methods of the US Preventive Services Task Force: a review of the process. Am J Prev Med. 2001;20:21-35. [PubMed ID: 11306229]

3. Mikkola TS, Clarkson TB. Estrogen replacement therapy, atherosclerosis, and vascular function. Cardiovasc Res. 2002;53:605-19. [PubMed ID: 11861031]

4. Colditz GA, Miller JN, Mosteller F. How study design affects outcomes in comparisons of therapy. I: Medical. Stat Med. 1989;8:441-54. [PubMed ID: 2727468]

5. Ioannidis JP, Haidich AB, Pappa M, et al. Comparison of evidence of treatment effects in randomized and nonrandomized studies. JAMA. 2001;286:821-30. [PubMed ID: 11497536]

6. Benson K, Hartz AJ. A comparison of observational studies and randomized, controlled trials. N Engl J Med. 2000;342:1878-86. [PubMed ID: 10861324]

7. Concato J, Shah N, Horwitz RI. Randomized, controlled trials, observational studies, and the hierarchy of research designs. N Engl J Med. 2000;342:1887-92. [PubMed ID: 10861325]