Z Ye1, D B Thomas, R M Ray. 1. Fred Hutchinson Cancer Research Center, Seattle, WA 98177, USA.
Abstract
BACKGROUND: Although the possible influence of oral contraceptives on risk of cervical carcinoma in situ has been the subject of multiple prior investigations, the results have been inconsistent. METHODS: Data from a multinational, collaborative case-control study were analysed to investigate further this possible relationship. To assess potential screening bias, some statistical analyses were restricted to subgroups of cases with and without symptoms at the time of their diagnosis. RESULTS: Relative risk estimates in relation to various features of oral contraceptive use tended to be highest for asymptomatic disease, lowest for disease presenting with vaginal bleeding, and intermediate for disease presenting with other symptoms, suggesting the presence of a screening bias. In women with vaginal bleeding, who are least likely to have been detected by routine screening, no elevated risk of cervical carcinoma in situ was observed in relation to ever having used combined oral contraceptives, but there was an increased risk in users of over 60 months' duration. An increasing trend in risk with duration of use was most pronounced in these women who first used oral contraceptives in the past 5-10 years; and in women who used oral contraceptives for more than 60 months, risk declined with time since last use. CONCLUSION: These findings could reflect a reversible effect of long-term use of oral contraceptives at an intermediate stage in the carcinogenic process, or a non-causal relationship due to unidentified sources of bias or confounding.
BACKGROUND: Although the possible influence of oral contraceptives on risk of cervical carcinoma in situ has been the subject of multiple prior investigations, the results have been inconsistent. METHODS: Data from a multinational, collaborative case-control study were analysed to investigate further this possible relationship. To assess potential screening bias, some statistical analyses were restricted to subgroups of cases with and without symptoms at the time of their diagnosis. RESULTS: Relative risk estimates in relation to various features of oral contraceptive use tended to be highest for asymptomatic disease, lowest for disease presenting with vaginal bleeding, and intermediate for disease presenting with other symptoms, suggesting the presence of a screening bias. In women with vaginal bleeding, who are least likely to have been detected by routine screening, no elevated risk of cervical carcinoma in situ was observed in relation to ever having used combined oral contraceptives, but there was an increased risk in users of over 60 months' duration. An increasing trend in risk with duration of use was most pronounced in these women who first used oral contraceptives in the past 5-10 years; and in women who used oral contraceptives for more than 60 months, risk declined with time since last use. CONCLUSION: These findings could reflect a reversible effect of long-term use of oral contraceptives at an intermediate stage in the carcinogenic process, or a non-causal relationship due to unidentified sources of bias or confounding.