Table 4. Summary of Evidence

Benefits and Harms	Key Questions	Evidence Codes*	Quality of Evidence**
Potential Benefits Does HRT reduce risks for:	1. CHD and CVD incidence?	I, II-2	Fair-good: most studies are observational and have important biases; when confounders are considered, apparent benefits for current users are not supported; trial data from WHI indicates increased risk further undermining validity of observational studies.
	2. CHD and CVD mortality?	I, II-2	Fair-good: results based on observational studies with biases; both observational and trial data indicate no increase or decrease in risk.
	3. Stroke incidence?	I, II-2	Fair-good: results based on observational studies with biases; observational and trial data suggest increased risk.
	4. Stroke mortality?	I, II-2	Fair-good: observational studies indicated reduced risk for stroke mortality, although trial data did not support this finding.
	5. Colorectal cancer?	I, II-2	Poor-good: results are based on observational studies that were primarily designed for other outcomes; findings from the WHI are not significant when the analysis is adjusted.
	6. Low bone density?	I	Good: many good-quality RCTs are consistent and demonstrate benefit; limited by short duration of trials, bone density is an intermediate outcome.
	7. Fractures?	I, II-2	Fair-good: RCTs—few trials available, none is definitive because of limitations of methods although benefit is supported. Cohort studies—several good-quality cohort studies are consistent and demonstrate benefit; limited by healthy user bias.
	8. Decline in cognitive function?	I, II-2	Fair-poor: studies enlist different patient populations and measure many different outcomes; results for symptomatic women are different from asymptomatic women. Duration of studies is too short to be meaningful. Difficult to draw any conclusions because outcome measures are so diverse.
	9. Dementia?	II-2	Fair-poor: although the meta-analysis supports a protective effect, methodologic limitations and biases exist in individual studies (e.g., healthy user effect, use of proxy interviews, historical data obtained from subjects with dementia).
Potential Harms Does HRT increase risks for:	1. Venous thromboembolism?	I, II-2	Poor-good: RCTs—venous thromboembolism is a secondary outcome, groups were randomized for cardiac outcomes, method of outcome assessment was not reported. Case-control—quality ratings range from poor to good; analysis based on small numbers of cases, important confounders such as smoking not considered in some studies. The consistency of the findings for an increased risk support the relationship.
	2. Breast cancer incidence?	I, II-2	Poor-good: increased risk with current use of long duration was supported by observational data and WHI trial; despite biases of the observational studies, the consistency of this finding provides stronger evidence for an association.
	3. Breast cancer mortality?	II-2	Poor-good: observational and trial data indicate that mortality is not increased.
	4. Endometrial cancer?	II-2	Poor-good: results are based on observational studies only, although results are consistent and demonstrated dose-response relationships.
	5. Cholecystitis?	I, II-2	Poor-good: increased risk was reported from RCTs and observational studies, but was not a finding in every study; results demonstrated dose-response relationships.

*Study Design Categories
I: Randomized, controlled trials
II-1: Controlled trials without randomization
II-2: Cohort or case-control analytic studies
II-3: Multiple time series, dramatic uncontrolled experiments
III: Opinions of respected authorities, descriptive epidemiology

**Quality of evidence ratings based on criteria developed by the U.S. Preventive Services Task Force (Harris, 2001).

Return to Document