Richard Brand

CONTEXT: Postmenopausal vaginal bleeding is a common clinical problem. Endovaginal ultrasound (EVUS) is a noninvasive diagnostic test that may help determine which women should undergo endometrial biopsy. OBJECTIVE: To determine the accuracy of EVUS in detecting endometrial disease in postmenopausal women with vaginal bleeding according to hormone replacement use. DATA SOURCES: Literature search of English-language and non-English-language articles published from 1966 through November 1996 using MEDLINE and by a manual search of bibliographies of published articles. STUDY SELECTION: Studies were included if they prospectively collected EVUS measurements of endometrial thickness prior to obtaining endometrial tissue for histologic evaluation in postmenopausal women with vaginal bleeding. Of 85 studies that included data on EVUS and endometrial histology, 35 were included in the meta-analysis and included 5892 women. DATA EXTRACTION: Articles were reviewed and independently selected and abstracted by 2 reviewers. Disagreement was resolved by consensus. DATA SYNTHESIS: The overall summary mean weighted estimates of sensitivity and specificity were calculated for thresholds of endometrial thickness from 3 to 10 mm. Using a 5-mm threshold to define abnormal endometrial thickening, 96% (95% confidence interval [CI], 94%-98%) of women with cancer had an abnormal EVUS result, whereas 92% (95% CI, 90%-93%) of women with endometrial disease (cancer, polyp, or atypical hyperplasia) had an abnormal result. This did not vary by hormone replacement use. However, the number of women with normal histology who had an abnormal EVUS result did vary by hormone replacement use. In women who were not using hormone replacement therapy, 593 (8%) with normal histological findings had an abnormal EVUS result (specificity, 92%; 95% CI, 90%-94%), whereas 1544 (23%) using hormone replacement therapy had an abnormal EVUS result (specificity, 77%; 95% CI, 75%-79%). For a postmenopausal woman with vaginal bleeding with a 10% pretest probability of endometrial cancer, her probability of cancer is 1% following a normal EVUS result. CONCLUSION: Endovaginal ultrasound has a high sensitivity for detecting endometrial cancer and other endometrial disease and can reliably identify postmenopausal women with vaginal bleeding who are highly unlikely to have significant endometrial disease so that endometrial sampling may be unnecessary.

The hypothesis that triglyceride is a cause of coronary heart disease, although unconfirmed and never universally accepted, has nonetheless strongly influenced the practice of preventive medicine. We have examined the epidemiologic association between triglyceride and coronary heart disease to evaluate the validity of inferring that there is a causal relation between the two. Neither the evidence from published studies nor an analysis of data from the Western Collaborative Group Study provides strong support for the causal hypothesis. Information from other scientific disciplines is also meager, contrasting with the coherence of diverse evidence supporting the hypothesis that cholesterol is a cause of coronary heart disease. These arguments fall short of disproving the belief that lowering triglyceride will prevent coronary heart disease, especially since triglyceride and cholesterol are inextricably associated through mutual lipoprotein carriers. But we propose that the ethics of preventive medicine place the burden of proof on the proponents of intervention. We therefore recommend that widespread screening and treatment of healthy persons for hypertriglyceridemia be abandoned until more persuasive evidence becomes available.

The association between an a priori measure of social connections and five-year mortality from all causes, cardiovascular diseases (International Classification of Diseases, Eighth Revision (ICD-8) codes 390-458), and ischemic heart disease (ICD-8 codes 410-414) was studied in 13,301 men and women from eastern Finland who were first interviewed in 1972 or 1977. For men, there was a graded association between extent of social connections and mortality. In multivariate models with adjustment for age, smoking, serum cholesterol, mean weighted blood pressure, measures of prevalent illness, and other possible confounders, men who were in the two lowest quintiles of the social connections scale were at increased risk compared with those in the highest quintile (odds ratio (OR)all cause = 1.54, 95% confidence interval (CI) = 1.21-1.95; ORcardiovascular disease = 1.54, 95% CI = 1.11-2.13; ORischemic heart disease = 1.34, 95% CI = 0.94-1.90). No strong or consistent association was found for women. The association for men was modified by levels of blood pressure with the effect of low social connections greater at higher levels of blood pressure. In three separate analyses, there was no evidence for confounding or effect modification due to prevalent illness at baseline.