|Mortality associated with hormone replacement therapy in younger and older women: a meta-analysis
|Salpeter S R, Walsh J M, Greyber E, Ormiston T M, Salpeter E E
The authors concluded that hormone replacement therapy reduced mortality in studies with a mean baseline participant age lower than 60 years, but not in studies with a mean baseline participant age greater than 60 years. Population and treatment differences between the included studies, and low event rates, particularly in the 'younger' group, mean that these conclusions might not be reliable.
To evaluate the effects of hormone replacement therapy (HRT) on mortality in younger and older postmenopausal women.
MEDLINE, EMBASE, CINAHL and the Cochrane CENTRAL Register were searched from 1966 to September 2002 without any language restrictions; the search terms were reported. In addition, selected journals and reference lists of identified studies were screened.
Study designs of evaluations included in the review
Randomised controlled trials (RCTs) of more than 6 months' duration were eligible for inclusion in the review.
Specific interventions included in the review
Studies that compared HRT with placebo or no treatment were eligible for inclusion. The included studies evaluated transdermal or oral oestrogens with or without progestins. The control groups received placebo, no treatment or calcium supplementation. Details of specific drug treatments were reported.
Participants included in the review
Studies of postmenopausal women were eligible for inclusion. In the included studies, the mean ages at baseline were 62.2 and 63.4 years in the treatment and control groups, respectively (age range: 36 to 87 years). The studies had wide-ranging inclusion and exclusion criteria; some included women with narrowing of the carotid arteries, recent stroke, cerebrovascular disease, healthy women post hysterectomy or oophorectomy, and women with recent myocardial infarction, coronary disease, coronary stenosis, elevated cholesterol, ovarian cancer and rheumatoid arthritis, and some were restricted to specific age groups.
Outcomes assessed in the review
Studies that reported at least one death were eligible for inclusion. If this information was not reported in otherwise eligible studies, mortality data were requested from the authors. The review assessed total mortality and mortality due to cardiovascular, cancer and all other causes.
How were decisions on the relevance of primary studies made?
Two reviewers independently selected the studies; interobserver agreement was measured.
Assessment of study quality
Two reviewers independently assessed validity; interobserver agreement was measured. The studies were assessed for method of randomisation and adequate allocation concealment, blinding of the providers and patients, description of withdrawals and drop-outs, and use of intention-to-treat analysis. Studies meeting all quality criteria were classified as A, studies that partially met one or more criteria were classified as B, and studies that did not meet one or more criteria were classified as C.
Two reviewers independently extracted the data and resolved any disagreements by consensus.
For each study, odds ratios (ORs) of mortality outcomes were calculated. Authors of studies with a duration of more than 6 months were contacted for information about deaths during the study. For patients who withdrew from studies due to adverse events, deaths reported after withdrawal were included. For crossover studies, only data from the first phase were used.
Methods of synthesis
How were the studies combined?
Pooled ORs of mortality outcomes (total, cardiovascular and cancer-related and all other causes) were calculated, along with 95% confidence intervals (CIs), using a random-effects model. Funnel plots were used to examine the potential for publication bias.
How were differences between studies investigated?
Statistical heterogeneity was assessed using the chi-squared statistic and the Q-value. The influence on mortality of age at baseline was examined by analysing separately studies with a mean baseline age of less than 60 years and studies with a mean baseline age greater than 60 years. Differences between older and younger age groups were compared using the logarithm of the OR. A sensitivity analysis was used to examine the effects of using different age cut-offs (56 to 63 years). A subgroup analysis was used to examine the effects on mortality of the type of HRT (unopposed oestrogen versus combined oestrogen plus progestin). Linear regression was used to compared mortality in different age groups (younger than 56 years, 56 to 63 years, and older than 63 years). Analyses were repeated after excluding lower quality (C score) studies.
Results of the review
Thirty RCTs were included (n=26,708 providing 119,118 years of patient follow-up).
Thirteen studies were allocated an A quality score, ten a B score, and seven a C score. The mean drop-out rate was 11.5% in the treatment groups and 10.6% in the control groups.
For all ages combined, there was no statistically significant difference between HRT and control in total mortality (OR 0.98, 95% CI: 0.87, 1.18), cardiovascular mortality (OR 1.10, 95% CI: 0.90, 1.34), cancer mortality (OR 1.03, 95% CI: 0.82, 1.29) or breast cancer mortality (OR 1.03, 95% CI: 0.29, 3.67). HRT was associated with a significant reduction in deaths from causes other than cardiovascular disease and cancer (OR 0.67, 95% CI: 0.51, 0.88).
Younger women: 17 RCTs (n=4,141) were identified. The mean study duration was 3.66 years, and the mean baseline ages were 53.9 and 53.8 years in the treatment and control groups, respectively. In studies with a mean baseline age of less than 60 years, HRT was associated with a significant reduction in total mortality (OR 0.61, 95% CI: 0.39, 0.95), but there was no statistically significant difference between HRT and control in cardiovascular mortality (OR 0.68, 95% CI: 0.22, 2.15), cancer mortality (OR 0.69, 95% CI: 0.59, 1.08) or other mortality (OR 0.44, 95% CI: 0.17, 1.13).
Older women: 13 RCTs (n=22,567) were identified. The mean study duration was 4.66 years, and the mean baseline ages were 64.6 and 66.8 years in the treatment and control groups, respectively. In studies with a mean baseline age of more than 60 years, there was no statistically significant difference between HRT and control in total mortality (OR 1.03, 95% CI: 0.9, 1.18), cardiovascular mortality (OR 1.11, 95% CI: 0.91, 1.36) or cancer mortality (OR 1.07, 95% CI: 0.84, 1.37). HRT was associated with a significant reduction in mortality from other causes (OR 0.68, 95% CI: 0.56, 0.91).
HRT was associated with a significantly lower mortality in younger compared with older women (p=0.03). The results were similar when using different cut-off ages.
A regression analysis comparing three age groups showed a significant trend towards increasing mortality with baseline age. There was no significant difference in total mortality between unopposed oestrogen and combined oestrogen-progestin treatment for all ages combined or in younger or older women. The results for total mortality for all ages, younger and older women were little changed after excluding lower quality (C score) studies. No evidence of significant statistical heterogeneity was found for any of the analyses (p>0.7). The funnel plots showed no evidence of publication bias.
HRT reduced mortality in studies with a mean baseline age lower than 60 years but not in studies with a mean baseline age greater than 60 years.
The review addressed a clear question that was defined in terms of the participants, intervention, outcomes and study design. Several relevant sources were searched and attempts were made to minimise language bias. No specific attempts were made to minimise publication bias. Validity was assessed using defined criteria but the results were not reported in full. Methods were used to minimise reviewer errors and bias in the study selection, validity assessment and data extraction processes. Statistical heterogeneity was assessed and appropriate methods were used to pool the studies. This was generally a well-conducted review. The data appeared to support the authors' conclusions but population and treatment differences between the included studies, and low event rates, particularly in the 'younger' group, mean that the conclusions might not be reliable.
Implications of the review for practice and research
Practice: The authors stated that decisions about the use of HRT should be made on an individual basis, taking into account the woman's age, symptoms and risk factors for disease. Research: The authors stated that more large RCTs would be required to determine the optimal age to start HRT and the optimal treatment duration.
Santa Clara Valley Medical Centre; University of California, San Francisco.
Salpeter S R, Walsh J M, Greyber E, Ormiston T M, Salpeter E E. Mortality associated with hormone replacement therapy in younger and older women: a meta-analysis. Journal of General Internal Medicine 2004; 19(7): 791-804
Other publications of related interest
Salpeter SR, Walsh JM, Greyber E, Salpeter EE. Brief report: coronary heart disease events associated with hormone therapy in younger and older women. A meta-analysis. J Gen Intern Med 2006;21:363-6.
These additional published commentaries may also be of interest. Furberg CD, Psaty BM. Review: hormone replacement therapy may reduce the risk for death in younger but not older postmenopausal women. ACP J Club 2005;142:1. Furberg CD, Psaty BM. Review: hormone therapy may reduce the risk of death in younger but not older postmenopausal women. Evid Based Med 2005;10:10.
Subject indexing assigned by NLM
Age Factors; Aged; Cardiovascular Diseases /mortality; Cause of Death; Female; Hormone Replacement Therapy /mortality; Humans; Middle Aged; Neoplasms /mortality; Postmenopause; Randomized Controlled Trials as Topic
Database entry date
This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.