The discrepancy between observational studies and randomized trials of menopausal hormone therapy: did expectations shape experience?

Ann Intern Med. 2003 Dec 2;139(11):923-9. doi: 10.7326/0003-4819-139-11-200312020-00011.


Differences between observational and randomized studies of the effects of menopausal hormone therapy (HT) on coronary heart disease (CHD) have been attributed to the fact that women who choose to use HT tend to be healthier than those who do not. Although this bias should affect all clinical outcomes with modifiable risk factors, estimates for stroke and pulmonary embolism were unaffected. The authors sought possible explanations for this isolated discrepancy in CHD findings. Unlike the randomized Women's Health Initiative (WHI) trial, the observational Nurses' Health Study (NHS) did not try to detect silent myocardial infarctions. Many women present with atypical ischemic symptoms. Hormone therapy users who believe that HT reduces CHD risks might not interpret ischemic symptoms as related to CHD, might not seek medical attention, and might present differently to their physicians, all of which could lead to more unrecognized myocardial infarctions among HT users in the NHS. In addition, persons completing death certificates and NHS physicians interpreting death certificates were not blinded to the use of HT. If persons assigning cause of death knew the patient had used HT and believed that HT prevented CHD, they might have been more likely to assign a condition other than CHD as the cause of death. If HT users were 20% less likely to have their infarctions recognized and their deaths attributed to CHD, a true increase in CHD due to HT use would appear to be a reduction in CHD. Combining these reporting biases with socioeconomic differences between users and nonusers could explain discrepancies. Beliefs held by patients, clinicians, and investigators might have affected the ascertainment of CHD outcomes in observational studies.

Publication types

  • Research Support, Non-U.S. Gov't
  • Research Support, U.S. Gov't, P.H.S.

MeSH terms

  • Bias
  • Coronary Disease / mortality
  • Coronary Disease / prevention & control*
  • Female
  • Hormone Replacement Therapy*
  • Humans
  • Models, Statistical
  • Observation*
  • Postmenopause*
  • Randomized Controlled Trials as Topic / standards*
  • Research Design / standards*
  • Risk Factors
  • Socioeconomic Factors