Oral contraceptives and cancer of the cervix uteri. A meta-analysis.

AIM AND OBJECTIVE

Because the findings of epidemiologic studies of the relationship between oral contraceptive use and cervical cancer have not been consistent, we reanalyzed the relationship.

DESIGN

Meta-analysis of studies published to date.

SETTING AND SUBJECTS

Papers were located by searching the MEDLINE data base, supplemented by a hand search of all the references in the articles recovered.

MEASUREMENTS

Studies were graded as to quality. Two meta-analyses were performed: one including all the studies gathered and one including methodologically acceptable studies only. The method of Woolf was used to combine relative risks. Heterogeneity of the effect was assessed.

MAIN RESULTS

Fifty-one published studies were collected: 21 case-control, 18 cross-sectional, and 12 cohort designs. Twenty-one of these were considered as methodologically acceptable, but only 18 could be pooled. The main results observed were: relative risks was 1.52 (1.3-1.8) for dysplasia, 1.52 (1.3-1.8) for carcinoma in situ, and 1.21 (1.1-1.4) for invasive cancer. A significant linear dose-response effect was observed in dysplasia, carcinoma in situ, and invasive cervical cancer. Heterogeneity of the effect was present in some of the former estimates.

CONCLUSIONS

Oral contraceptive use may be a risk factor for all stages of the natural history of cervical cancer, which may imply an initiator effect. Limitations to this research are discussed.