The positive predictive value of cervical smears in previously screened postmenopausal women: The heart and estrogen/progestin replacement study (HERS)

G. F. Sawaya, D. Grady, K. Kerlikowske, J. La Valleur, V. M. Barnabei, Katherine Miller-Bass, T. E. Snyder, J. H. Pickar, S. K. Agarwal, J. Mandelblatt

Research output: Contribution to journalArticle


Background: The benefits and risks of performing annual cervical smears on postmenopausal women are not well defined. The independent effect of hormone replacement therapy on development of cytologic abnormalities is unknown. Objective: To determine the positive predictive value of cervical smears in previously screened postmenopausal women and to determine the effect of oral estrogen plus progestin on incident cervical cytologic abnormalities. Design: Prospective cohort study (incidence) and randomized, double-blind, placebo-controlled trial (hormone therapy). Setting: 20 U.S. outpatient and community clinical centers. Participants: 2561 women with a uterus and normal cytologic characteristics at baseline. Interventions: Annual smears; oral conjugated equine estrogens, 0.625 mg/d, plus medroxyprogesterone acetate, 2.5 mg/d, or identical placebo. Measurements: Incident cytologic abnormalities (atypical squamous cells of undetermined significance, atypical glandular cells of undetermined significance, low-grade squamous epithelial lesion, and high-grade squamous epithelial lesion) and final histologic diagnoses. Results: The incidence of new cytologic abnormalities in the 2 years following a normal smear was 110 per 4895 person-years (23 per 1000 person-years [95% C, 18 to 27 per 1000 person-years]). Among the 103 women with known histologic diagnoses, one had mild to moderate dysplasia. The positive predictive value of any smear abnormality identified 1 year after a normal smear, therefore, was 0% (C, 0% to 5.0%) (0 of 78 women); the positive predictive value of abnormalities found within 2 years was 0.9% (C, 0.0% to 3.0%) (1 of 110 women). In hormonetreated compared with non-hormone-treated women, the incidence of cytologic abnormalities was nonsignificantly higher (relative hazard, 1.36 [C, 0.93 to 1.99]), largely because of a nonsignificant 58% greater incidence of atypical squamous cells of undetermined significance (relative hazard, 1.58 [C, 0.99 to 2.52]). Conclusions: Because of a poor positive predictive value, cervical smears should not be performed within 2 years of normal cytologic results in postmenopausal women. Therapy with oral estrogen plus progestin does not significantly affect the incidence of cytologic abnormalities.

Original languageEnglish (US)
Pages (from-to)942-950
Number of pages9
JournalAnnals of Internal Medicine
Issue number12
Publication statusPublished - Dec 19 2000
Externally publishedYes


ASJC Scopus subject areas

  • Medicine(all)

Cite this