J ClinEpidcmblVol. 45, No. 10, pp. 1I II-1 118, 1992
08954356/92 SW0 + 0.00 Copyright 0 1992 Pergamon Press Ltd
Printed in Great Britain. All rights -cd
ORAL CONTRACEPTIVE USE AND CERVICAL INTRAEPITHELIAL NEOPLASIA ANN L. COKE&“*MARGARETF. MCCANN? BARBARAS. HULKA~ and LESLIEA. WALTON~ ‘Department of Epidemiology and Biostatistics, University of South Carolina, Columbia, SC 29208, 2Department of Epidemiology, University of North Carolina and “Department of Obstetrics and Gynecology, University of North Carolina, Chapel Hill, NC 25715, U.S.A. (Received in revised form 8 April 1992)
Ah&act-To explore the somewhat controversial relationship between oral contraceptives and pre-invasive cervical cancer, 103cases of biopsy-confirmed cervical intraepithelial neoplasia (CIN) II or CIN III were compared with 258 controls who had normal cervical cytology. Cases were slightly less likely than controls to have ever used oral contraceptives; the odds ratio, controlling for age, socioeconomic status, barrier method use, smoking history, age at first sexual intercourse, number of sex partners, current marital status, and number of Pap smears, was 0.7 (95% CI 0.3-1.6). Recency, latency, duration, and age at first oral contraceptive use were evaluted and in no instance was oral contraceptive use positively associated with CIN. This study adds to the body of knowledge that oral contraceptives are not associated with pre-invasive cervical cancer. Further, if oral contraceptive users continue to he regularly screened, their risk of developing the more invasive lesions should he very low. Cervix neoplasms Sex behavior
Contraceptive
agents
INTRODUmION Numerous case-control studies have investigated the relationship between oral contraceptive (OC) use and pre-invasive (l-l I] and invasive cervical cancer [2, 12-141, yet there is no consensus as to the etiologic role OCs play in cervical neoplasia. Early case-control studies were reassuring [l-3, 12, 131, but these studies did not control for the important potential confounder, sexual behavior. Additionally, these studies were conducted less than 10 years after combined OCs became widely available, and thus sufficient time may not have elapsed from first use of OCs to observation of the first O&elated carcinogenic cervical changes. The results of recent case-control studies of OC use *Author for correspondence.
Risk
Epidemiology
Smoking
and invasive cervical cancer, controlling for sexual behavior, are mixed. Five studies found an increased risk of cancer associated with OC use [8, 15-181 while six other studies reported no increased risk attributable to OC use [7,9, 10,19,20,21]. Three recent prospective studies indicate an increased risk of invasive cervical cancer with extended OC use [22-241. This case-control study was designed to assess the relationships between cervical intraepithelial neoplasia and oral contraceptives, barrier methods of contraception, cigarette smoking, and sexually transmitted diseases. METHODS Subjects Between September 1987 and November 1988, 103 University of North Carolina Hospitals
1111
1112
ANN L. CQKERet al.
(UNCH) Dysplasia Clinic patients with newlydiagnosed, biopsy-confirmed CIN II or CIN III were enrolled as cases. The study inclusion criteria are as follows: 1845 years old, black or white, non-pregnant North Carolina residents, and mentally competent to complete the structured interview. Women who had a history of treatment with cervical cryotherapy, laser therapy, cone biopsy, cauterization, or hysterectomy were excluded. The diagnosis of CIN II or CIN III was histologically confirmed by the UNC Department of Surgical Pathology: 40 cases were CIN II and 63 cases were CIN III. Cases were selected from a dysplasia clinic because a Pap smear was required to define case/control status. Although selecting cases and controls from the same cohort of women receiving Pap smear screening would have been methodologically desirable, this approach would have been extremely inefficient and thus not feasible. The controls were 258 UNCH Family Practice Center patients scheduled to receive a routine Pap smear and found to have normal cervical cytology. All women who had scheduled a routine physical examination which included a Pap smear and who met the inclusion criteria were asked to participate in this study. The controls were also recruited from September 1987, to November 1988. Since all cases were heterosexually active, only heterosexually active controls were included in the analysis. The 258 controls represent all consenting, eligible women with normal cervical cytology enrolled from this clinic during the study period. This control group was selected because these patients received care at the same (UNC) medical facility as the cases and, as such, Pap smears from both cases and controls could be interpreted by the same cytology laboratory. Requiring that Pap smears from both cases and controls be interpreted by the same cytology laboratory reduced the potential for disease misclassification which could have occurred if interpretations from more than one cytology laboratory were compared. However, requiring that all Pap smears are read by the same cytology laboratory in this case, imposed the additional criteria that both cases and controls receive care at a large referral hospital. Only 8.0% of cases (9 of 112) and 8.2% of controls (23 of 281) refused study participation.
Table 1 presents the demographic profile of participating subjects compared with refusers by case/control status. Among both cases and controls, women less than 25 years of age were more likely to refuse study participation than women older than 25. No difference in refusal status by race was observed among cases, but black potential controls were more likely to refuse than whites. No information was available for non-participants regarding other nondemographic risk factors for CIN because these women refused study participation and, thus, the interview. Data collection Interviews. All subjects participated in a 15min structured telephone or in-person interview which was conducted by one of three trained female interviewers. If a subject did not have easy access to a home telephone, she was interviewed in her respective clinic. Since a greater percentage of cases did not have telephones a greater percentage of cases (24.8%) were interviewed in their respective clinic compared with contols (3.7%). The format of the in-person interview was identical to that of the telephone interview. The following topics were covered in the questionnaire: demographic profile (age, race, education, occupation), sexual experience (number of male sex partners, number of male sex partners as a teen, age at first sexual intercourse), active and passive smoking exposure, Pap smear history, history of treatment for CIN, and self-reported sexually transmitted disease history. The Hollingshead Index was used as a proxy for socioeconomic status (SES). This index uses the subject’s occupation and education (inherent in the occupational grouping) to rank the SES from l-9 with 9 being the highest SES.
Table 1. Demographic characteristics of study participants and refusers CIN II/III
Controls
Participants n = 103 (“/I
Refusers n=9 w)
Participants n =258 w)
Refusers n =23 (“/)
Age
08954356/92 SW0 + 0.00 Copyright 0 1992 Pergamon Press Ltd
Printed in Great Britain. All rights -cd
ORAL CONTRACEPTIVE USE AND CERVICAL INTRAEPITHELIAL NEOPLASIA ANN L. COKE&“*MARGARETF. MCCANN? BARBARAS. HULKA~ and LESLIEA. WALTON~ ‘Department of Epidemiology and Biostatistics, University of South Carolina, Columbia, SC 29208, 2Department of Epidemiology, University of North Carolina and “Department of Obstetrics and Gynecology, University of North Carolina, Chapel Hill, NC 25715, U.S.A. (Received in revised form 8 April 1992)
Ah&act-To explore the somewhat controversial relationship between oral contraceptives and pre-invasive cervical cancer, 103cases of biopsy-confirmed cervical intraepithelial neoplasia (CIN) II or CIN III were compared with 258 controls who had normal cervical cytology. Cases were slightly less likely than controls to have ever used oral contraceptives; the odds ratio, controlling for age, socioeconomic status, barrier method use, smoking history, age at first sexual intercourse, number of sex partners, current marital status, and number of Pap smears, was 0.7 (95% CI 0.3-1.6). Recency, latency, duration, and age at first oral contraceptive use were evaluted and in no instance was oral contraceptive use positively associated with CIN. This study adds to the body of knowledge that oral contraceptives are not associated with pre-invasive cervical cancer. Further, if oral contraceptive users continue to he regularly screened, their risk of developing the more invasive lesions should he very low. Cervix neoplasms Sex behavior
Contraceptive
agents
INTRODUmION Numerous case-control studies have investigated the relationship between oral contraceptive (OC) use and pre-invasive (l-l I] and invasive cervical cancer [2, 12-141, yet there is no consensus as to the etiologic role OCs play in cervical neoplasia. Early case-control studies were reassuring [l-3, 12, 131, but these studies did not control for the important potential confounder, sexual behavior. Additionally, these studies were conducted less than 10 years after combined OCs became widely available, and thus sufficient time may not have elapsed from first use of OCs to observation of the first O&elated carcinogenic cervical changes. The results of recent case-control studies of OC use *Author for correspondence.
Risk
Epidemiology
Smoking
and invasive cervical cancer, controlling for sexual behavior, are mixed. Five studies found an increased risk of cancer associated with OC use [8, 15-181 while six other studies reported no increased risk attributable to OC use [7,9, 10,19,20,21]. Three recent prospective studies indicate an increased risk of invasive cervical cancer with extended OC use [22-241. This case-control study was designed to assess the relationships between cervical intraepithelial neoplasia and oral contraceptives, barrier methods of contraception, cigarette smoking, and sexually transmitted diseases. METHODS Subjects Between September 1987 and November 1988, 103 University of North Carolina Hospitals
1111
1112
ANN L. CQKERet al.
(UNCH) Dysplasia Clinic patients with newlydiagnosed, biopsy-confirmed CIN II or CIN III were enrolled as cases. The study inclusion criteria are as follows: 1845 years old, black or white, non-pregnant North Carolina residents, and mentally competent to complete the structured interview. Women who had a history of treatment with cervical cryotherapy, laser therapy, cone biopsy, cauterization, or hysterectomy were excluded. The diagnosis of CIN II or CIN III was histologically confirmed by the UNC Department of Surgical Pathology: 40 cases were CIN II and 63 cases were CIN III. Cases were selected from a dysplasia clinic because a Pap smear was required to define case/control status. Although selecting cases and controls from the same cohort of women receiving Pap smear screening would have been methodologically desirable, this approach would have been extremely inefficient and thus not feasible. The controls were 258 UNCH Family Practice Center patients scheduled to receive a routine Pap smear and found to have normal cervical cytology. All women who had scheduled a routine physical examination which included a Pap smear and who met the inclusion criteria were asked to participate in this study. The controls were also recruited from September 1987, to November 1988. Since all cases were heterosexually active, only heterosexually active controls were included in the analysis. The 258 controls represent all consenting, eligible women with normal cervical cytology enrolled from this clinic during the study period. This control group was selected because these patients received care at the same (UNC) medical facility as the cases and, as such, Pap smears from both cases and controls could be interpreted by the same cytology laboratory. Requiring that Pap smears from both cases and controls be interpreted by the same cytology laboratory reduced the potential for disease misclassification which could have occurred if interpretations from more than one cytology laboratory were compared. However, requiring that all Pap smears are read by the same cytology laboratory in this case, imposed the additional criteria that both cases and controls receive care at a large referral hospital. Only 8.0% of cases (9 of 112) and 8.2% of controls (23 of 281) refused study participation.
Table 1 presents the demographic profile of participating subjects compared with refusers by case/control status. Among both cases and controls, women less than 25 years of age were more likely to refuse study participation than women older than 25. No difference in refusal status by race was observed among cases, but black potential controls were more likely to refuse than whites. No information was available for non-participants regarding other nondemographic risk factors for CIN because these women refused study participation and, thus, the interview. Data collection Interviews. All subjects participated in a 15min structured telephone or in-person interview which was conducted by one of three trained female interviewers. If a subject did not have easy access to a home telephone, she was interviewed in her respective clinic. Since a greater percentage of cases did not have telephones a greater percentage of cases (24.8%) were interviewed in their respective clinic compared with contols (3.7%). The format of the in-person interview was identical to that of the telephone interview. The following topics were covered in the questionnaire: demographic profile (age, race, education, occupation), sexual experience (number of male sex partners, number of male sex partners as a teen, age at first sexual intercourse), active and passive smoking exposure, Pap smear history, history of treatment for CIN, and self-reported sexually transmitted disease history. The Hollingshead Index was used as a proxy for socioeconomic status (SES). This index uses the subject’s occupation and education (inherent in the occupational grouping) to rank the SES from l-9 with 9 being the highest SES.
Table 1. Demographic characteristics of study participants and refusers CIN II/III
Controls
Participants n = 103 (“/I
Refusers n=9 w)
Participants n =258 w)
Refusers n =23 (“/)
Age
