Send to

Choose Destination
See comment in PubMed Commons below
Maturitas. 2004 Mar 15;47(3):175-83.

Variability of breast cancer risk in observational studies of hormone replacement therapy: a meta-regression analysis.

Author information

  • 1Institute of Clinical Pharmacology, Humboldt-University and Institute of Pharmacoepidemiology and Technology Assessment, Berlin, Germany.



A re-analysis of data from 51 epidemiological studies reported a significant 14% increase in the risk of breast cancer associated with the use of hormone replacement therapy (HRT). Unlike randomized trials, these observational studies varied in design and methods. This study was conducted to explore the impact of study design factors on the rate ratio.


We performed a meta-regression analysis of 39 epidemiological studies of HRT and breast cancer. The rate ratio of breast cancer associated with ever use of HRT was evaluated in relation to study design, study period, country, primary study objective, method of exposure measurement, age control, adjustment factors related to reproduction and menopause, and the presence of breast cancer surveillance. We used stepwise multiple regression analysis, weighted by the inverse of the variance of the logarithm of the rate ratio, to estimate ratios of rate ratios for these factors.


Exposure measured by personal interview and/or medical record review was associated with a 14% lower rate ratio estimate as compared with telephone interview or self-administered questionnaire (P = 0.018). Among studies that did not adjust for age at menopause, the rate ratio was 12% lower if the primary objective was HRT effect than not (P = 0.016), while it was 43% higher among studies that adjusted for age at menopause (P = 0.042). An index that included as six desirable design features, breast cancer surveillance, matching of controls, more reliable exposure information, adjustment for age at menopause and reproductive risk factors, and as primary objective the effect of HRT suggests that studies with none of these properties would lead to a rate ratio estimate of 1.14 (95% CI: 1.00-1.29) while studies with all six properties would produce a rate ratio of 0.98 (95% CI: 0.83-1.15).


Design factors of epidemiological studies could be an alternative explanation for the reported 14% increase in the risk of breast cancer associated with the use of HRT.

[PubMed - indexed for MEDLINE]
PubMed Commons home

PubMed Commons

How to join PubMed Commons

    Supplemental Content

    Full text links

    Icon for Elsevier Science
    Loading ...
    Write to the Help Desk