Morrison, Charles S. PhD*; Bright, Patricia PhD*†; Wong, Emelita L. DRPH*; Kwok, Cynthia MSPH*; Yacobson, Irina MD*; Gaydos, Charlotte A. PhD‡; Tucker, Heidi T. MPH*; Blumenthal, Paul D. MD§‖
HORMONAL CONTRACEPTIVES ARE among the most widely used family planning methods worldwide. Over 100 million women use oral contraceptives (OCs) and an estimated 20 to 30 million women use injectable depot-medroxyprogesterone acetate (DMPA). Moreover, DMPA use is increasing rapidly in many developing countries with a high prevalence of sexually transmitted infections (STIs).
Chlamydia and gonorrhea are among the most prevalent bacterial STIs. Annually, approximately 150 million new cases of chlamydia and gonorrhea occur worldwide.1 These bacterial infections are responsible for cervicitis and a number of important sequelae, including pelvic inflammatory disease (PID) and infertility. Inflammatory STIs, including chlamydia and gonorrhea, could also increase the risk of HIV acquisition.2–4 Thus, understanding the role that hormonal contraception might play in the transmission of these cervical infections is important.
A review of 29 cross-sectional studies examining OC use and chlamydial infection calculated a pooled unadjusted odds ratio of 1.9 (95% confidence interval [CI], 1.7–2.1).5 Three of 4 prospective studies have found an increased risk of chlamydial infection among OC users.6–9 The 2 prospective studies that provided adjusted risk estimates found almost identical results, reporting hazard ratios (HR) of 1.7 (95% CI, 1.1–2.8)6 and 1.8 (95% CI, 1.1–2.9),9 respectively. These studies also found similar results for the impact of OC use on the acquisition of gonorrhea (HR, 1.7; 95% CI, 1.1–2.86; HR, 1.4; 95% CI, 0.9–2.19). Sparse data exist on the effect of DMPA use on cervical infection risk. The sole prospective study reported adjusted hazard ratios of 1.6 (95% CI, 1.1–2.4) and 1.1 (95% CI, 0.8–1.6) for acquisition of chlamydial and gonococcal infections, respectively.9
Most research examining the impact of hormonal contraception on the acquisition of cervical infections has had major methodologic shortcomings. Studies have been primarily cross-sectional, have had insensitive measures of chlamydial infection, and most have failed to control for potential confounding factors. Therefore, although current studies suggest that OC use could increase the risk of chlamydial infection, high-quality prospective studies are needed to confirm this association and to examine the role of DMPA use on the acquisition of cervical infections.
One mechanism by which hormonal contraception might increase cervical infection risk is through the promotion of cervical ectopy. Cervical ectopy occurs when the columnar epithelium found in the endocervical canal extends out onto the ectocervix. It is typified by a single layer of glandular cells that lie close to the underlying vascular cervical stroma. Cervical ectopy is common in adolescents,8,10–12 pregnant women,10 and in women using oral contraceptives.6,8,11,12
Cervical ectopy could increase acquisition of some STI. Many studies have reported an association between cervical ectopy and either prevalent7,8,12,13 or incident6,8 chlamydial infections, whereas others have reported an association between ectopy and HPV infection,14 HIV infection,15 and CMV infection.12,16 Cervical ectopy, however, does not appear to be associated with gonococcal infection.6,13
The objectives of this study were to measure the effect of combined oral contraceptives and DMPA on the acquisition of chlamydial and gonococcal infection and to evaluate cervical ectopy as a mediator of any such association.
Materials and Methods
This research study was approved by the Protection of Human Subjects Committee of Family Health International, by the Institutional Review Board on Research Involving Human Subjects of the School of Public Health at the University of North Carolina, and by the National Medical Division of Planned Parenthood Federation of America. All women provided written informed consent before study participation.
Study Population and Procedures
We recruited women seeking gynecologic care at 2 Planned Parenthood of Maryland health centers from May 1996 through August 1999. One was located in the inner city with a largely minority clientele, whereas the other was suburban with a predominantly white, college-aged clientele.
To ensure that there were no alternate influences on the hormonal milieu of the cervix resulting from exogenous hormones or pregnancy, women ages 15 to 45 years, who had not used oral contraceptives or other hormonal contraception during the last complete menstrual cycle (including no DMPA injection in the previous 120 days) and who were not pregnant nor desired pregnancy during the following year were eligible. Women who had had an abortion in the prior week or a history of cervical cancer, hysterectomy, cone biopsy, or recent cryotherapy were ineligible.
At the time of enrollment, study participants were allowed to choose to initiate OCs, DMPA, or not to use a hormonal method. Thus, the study used a prospective cohort rather than a randomized, controlled trial design. This decision was made because most participants came to the clinics to initiate contraception and had a particular contraceptive method in mind. Women who freely choose a contraceptive method usually have higher contraceptive continuation and lower pregnancy rates. Thus, we believe that it was not ethically or practically feasible to use a randomized design.
Study visits were conducted at enrollment and at 3, 6, and 12 months follow up. A standardized interview was used to collect sociodemographic, sexual behavior, reproductive history, and contraceptive use data at each visit. Episodes of contraceptive use were recorded by month and were verified against clinic records.
At each visit, we conducted a standardized pelvic examination noting abnormalities in the external genitalia, and the vaginal and cervical epithelium. Vaginal pH, cervical friability, quantity and color of vaginal and cervical discharge were also noted. To sample for Chlamydia trachomatis, a swab was inserted into the endocervix and rotated for 15 to 30 seconds. It was then broken off and remained in the collection kit buffer during transport. For Neisseria gonorrhoeae culture, a cotton swab was placed in the cervical os and rotated 360° twice. Endocervical secretions and cells were inoculated directly onto Modified Thayer-Martin (MTM) selective agar medium and placed into a sealed extinction bag.
Methods for measuring cervical ectopy have been described in detail elsewhere.17 Briefly, after application of 4% acetic acid to the cervix, the clinician placed a vinyl dot of a known diameter (0.476 cm) on the face of the cervix. She then took 3 cervical photographs using a 35-mm Dental Eye II Camera with a ring flash (Yashica Inc., Somerset, NJ) before removing the vinyl dot. The photographic slides were digitized using Sigma Scan imaging analysis software (Sigma Scan; Jandel Scientific, San Rafael, CA). A trained independent rater used a cursor to trace the area of columnar epithelium on the digitized image. The rater then traced the diameter of the vinyl dot. The absolute area of ectopy was calibrated by comparing the outlined area of ectopy against the diameter of the vinyl dot.
Cervical swabs were processed and tested for C. trachomatis by ligase chain reaction (LCx; Abbott Laboratories, Abbott Park, IL) according to the manufacturer’s instructions. Samples were evaluated as negative, positive, or equivocal. Equivocal specimens were repeated according to the manufacturer’s instructions. To prevent contamination, specimen processing and amplification and detection stages were conducted in separate rooms; strict quality control measures were used.
Gonorrhea testing was performed by the Baltimore City Health Department. Specimens for gonorrhea culture were plated directly on MTM media (BBL, Sparks, MD). Colonies with appropriate morphology were tested by Gram stain, oxidase reaction, β lactamase and production. Presumptive cultures were confirmed by modified carbohydrate utilization test Gonocheck II (E-Y Laboratories, San Mateo, CA).
Analysis Population and Variable Definition
The analysis population consisted of enrolled women with at least 1 follow-up visit and with either negative baseline STI test results or positive baseline STI results that were then treated and confirmed negative (i.e., negative test of cure) before contributing person-time to the study. Because pregnancy dramatically changes the hormonal milieu and because most women who became pregnant were confined to those not using hormonal methods, we believed that including participants while pregnant could bias the study results. Thus, participants who became pregnant before their first follow-up visit were excluded. The outcome variable was the number of days from baseline to the earlier of: 1) date of diagnosis of first cervical infection; 2) the date of last study contact; or 3) the date of the last visit before a woman became pregnant. The outcome variable was calculated for chlamydial and/or gonococcal infections. To estimate the effects of time-varying contraceptive exposure, we analyzed the data by dividing a participant’s time into segments that correspond to the time between study visits. In dividing the participant’s time into segments, switch in contraceptive use as well as changes in other time-varying covariates such as condom use and number of sex partners were captured. For example, a study participant who had 3 follow-up visits had 3 segments with the first segment being the period from baseline to the first follow up. Contraceptive use during each segment was recorded; thus, a study participant could have been an OC user during the first segment, a control in the second segment, and then again an OC user in the last segment. Thus, these changes were represented by the time-varying contraceptive exposure variable used in both bivariable and multivariable analyses.
Contraceptive exposure was defined as the proportion of a segment a contraceptive was used. For women using DMPA, length of exposure was calculated as 120 days since date of last injection.
The absolute area of the cervix with ectopy was evaluated as a dichotomous variable representing the upper 75% versus the lower 25% (essentially some vs. no ectopy) of the distribution during the follow-up period. The cutoff point was 0.04 cm2 of the cervix with ectopy.
Tests of association between covariates and cervical infections were calculated for bivariable analyses using general chi-squared test statistics adjusted for clustering of visits within women (SUDAAN software).18 For multivariable analyses, standard Cox regression was used when covariates were time-invariant, whereas counting process Cox regression19 was used for time-varying covariates using the SAS system.20
Covariates were included in a final model based on the following criteria: age, race, and site and measures of contraceptive exposure were retained regardless of their statistical association with cervical infection. Although interactions between contraception and age, race, site, and sexual behavior were evaluated, interactions were not included in multivariable models as a result of a lack of precision in the estimates. Sexual behavior and reproductive health history variables were assessed for potential confounding by estimating whether their inclusion (vs. exclusion) resulted in a greater than 10% change in the coefficient of the contraceptive exposure variables.
Of the 1013 women screened, 1003 women were eligible and enrolled into the study; 850 participants returned for follow up. Thirty-one of these 850 women were excluded because they became pregnant by their first follow-up visit or did not have a follow-up visit with a valid chlamydia or gonorrhea test result (Table 1). Of the remaining 819 participants, 354 initiated OCs, 114 initiated DMPA, and 351 did not use hormonal contraception during the first month of the study. Most study participants (65%) did not change their contraceptive method throughout the study. Of the 1988 segments included in analyses, OCs and DMPA were used exclusively in 36.3% and 12.5% of segments, respectively, 37.7% represented control segments, and 13.5% were segments with mixed contraceptive exposure. The mean follow up for the 819 participants was 337 days.
The 819 participants were similar to the 184 not included in the analysis in terms of sociodemographic factors (age, marital status, ethnicity, enrollment site), reproductive (menarche, coitarche, prior contraceptive use) and STI history, and clinical signs of STI. However, women not included in the analysis had riskier recent sexual practices, including a higher proportion with multiple partners (20% vs. 16%;P = 0.02), and a lower proportion that always used condoms (20% vs. 33%;P < 0.01). They also had a somewhat higher baseline prevalence of cervical infections than those included in the analysis (7.6% vs. 5.3%;P = 0.21).
Participant Characteristics at Enrollment
Participants included in the analysis were primarily single and not living with a partner (77%), high school graduates (79%), and nulliparous (75%). Approximately two thirds were less than 25 years old, including 15% between 15–17 years old; 52% were white and 43% were African-American (Table 2). Approximately one half of participants had 6 or more lifetime sexual partners and 79% reported using male condoms during the previous 3 months. Almost two thirds had previously used OCs and 14% had previously used DMPA. At baseline, 39% of women had high vaginal pH (≥5.0), 27% had a friable cervix, and 5.3% had a chlamydial or gonococcal infection. The median area of cervical ectopy was 0.21 cm2 and the median proportion of the cervix with ectopy was 6.1%.
Although equal numbers of women participated from inner-city (n = 402) and suburban health centers (n = 417), inner-city participants were more likely to initiate DMPA use (18% vs. 13%) or not use hormonal contraception (50% vs. 24%;P <0.01). They were also more likely to be ≥25 years old (45% vs. 26%;P <0.01), black (70% vs. 17%;P <0.01), and parous (41% vs. 10%;P <0.01). Numbers of reported sex partners were similar, but fewer inner-city participants used condoms consistently (26% vs. 40%;P <0.01). At baseline, more inner-city (6.7%) than suburban participants (3.8%) had a chlamydial or gonococcal infection (P = 0.06).
Participants using OCs were significantly more likely than DMPA users or controls to be young, white, and nulliparous (Table 2). Although the proportion of women with multiple sex partners was similar across exposure groups, OC and DMPA users had sex more frequently than control group participants. At baseline, more controls than OC or DMPA users had signs of STI, including abnormal vaginal discharge, a friable cervix, and cervical discharge (Table 2). However, DMPA users had a somewhat higher baseline prevalence of cervical infections than OC users or controls.
Sexual Risk Factors Among Contraceptive Groups During Follow Up
During follow up, risky sexual behaviors and signs of STI were somewhat higher in the control group than among OC or DMPA users. Control group participants were more likely to have multiple sex partners, to have a new sex partner, and to report having a sex partner with a possible STI than other participants (Table 3). They were also more likely to have abnormal vaginal discharge and vaginal pH and bacterial vaginosis. Conversely, OC and DMPA users had sex more frequently and used condoms less consistently during follow up than control group participants. A higher percentage of OC users had cervical ectopy than DMPA users or control group participants. Participants switching methods during a segment had values that generally fell between those for the 3 contraceptive exposure groups for time-varying risk factors.
Association of Baseline and Time-Varying Covariates With Cervical Infection
During the study, 45 participants acquired a cervical infection (incidence density = 6.2 per 100 women-years [wy]), including 37 chlamydial infections and 14 gonococcal infections. Only 3 of these infections represented reinfections in women with a prevalent infection at baseline (2 controls, 1 DMPA user). Young age, lower education, nonwhite ethnicity, and inner-city site were strongly associated with incident cervical infection in bivariable analysis (Table 4). Among time-varying variables, multiple sex partners, inconsistent condom use, and vaginal douching were related to cervical infection acquisition, whereas cigarette smoking was not (Table 4). Signs of STI, including abnormal vaginal discharge, cervical friability, and vaginal pH ≥5.0, were associated with increased cervical infection risk. Women with a larger area of the cervix with ectopy had a moderately higher risk of infection.
Analysis of Hormonal Contraception and Cervical Infections
During follow up, there were 11 cervical infections in 285 woman-years among participants using OCs (3.9 per 100 wy), 12 cervical infections in 88 woman-years among DMPA users (13.7 per 100 wy), and 22 cervical infections in 369 woman-years among control participants (6.0 per 100 wy). In bivariable survival analysis, OC use during follow up was not associated with cervical infection risk (HR, 0.7; 95% CI, 0.3–1.5), whereas DMPA use was associated with cervical infection risk (HR, 2.8; 95% CI, 1.3–6.2) (Table 4).
In multivariable analysis, we did not find a significant association between OC use and cervical infection using a continuous measure of contraceptive exposure (HR, 1.5; 95% CI, 0.6–3.5). However, DMPA users had a 3-fold increase in the risk of cervical infection compared with those not using DMPA (HR, 3.6; 95% CI, 1.6–8.5) (Table 5). Young age, nonwhite ethnicity, inner-city site, and having multiple sex partners remained significantly associated with cervical infection acquisition after adjusting for other risk factors. Participants reporting consistent condom use had a decreased risk of infection, but the association was not statistically significant (HR, 0.5; 95% CI, 0.2–1.4).
We reanalyzed the data after limiting them to those segments (approximately 80%) in which no switching occurred between contraceptive groups. Again, although we found no significant association between COC use and acquisition of cervical infections (HR, 1.7; 95% CI, 0.7–4.1), DMPA users continued to be at increased cervical infection risk (HR, 3.3; 95% CI, 1.3–8.1).
We also applied our final multivariable model (outlined previously) to a chlamydia-only outcome variable (37 events). We found similar results to those outlined here for the combined cervical infection variable with DMPA (HR, 4.3; 95% CI, 1.7–11.1), but not OC use (HR, 1.9; 95% CI, 0.7–4.8), to be significantly associated with acquisition of a chlamydial infection.
Cervical Ectopy as a Mediating Factor
We hypothesized that cervical ectopy might mediate the association between hormonal contraception and cervical infection risk. Adding time-varying cervical ectopy to the model had little impact on the effect of DMPA (from HR = 3.6–3.8) or OC use (from HR = 1.7–1.8) on cervical infection risk. However, cervical ectopy could confer an independent risk of cervical infection acquisition (HR, 2.3; 95% CI, 0.9–6.0).
Our findings suggest that DMPA use is associated with a 3-fold increase in the risk of acquiring cervical chlamydia or gonorrhea. We found no significant increase in the risk of cervical infection among oral contraceptive users. Young age, nonwhite ethnicity, inner-city recruitment, and having multiple sex partners were also independent risk factors for acquiring a cervical infection.
We also found that cervical ectopy was not an important mediator of the hormonal contraception–cervical infection association. However, women with cervical ectopy could have an independent risk of acquiring a cervical infection.
Our finding of an increased risk of cervical infection among DMPA users agrees with 2 recently published studies, including the only prospective study that found an increased risk of chlamydial but not gonococcal infections among DMPA users.9,21 Our finding of no significant association between OC use and cervical infections contrasts with 3 of 4 previous prospective studies. This could be the result of differences in study populations (family planning clients vs. sex workers) or because we considered only women initiating hormonal contraception. Nevertheless, our adjusted hazard ratios of 1.5 for the effect of OC use on cervical infection acquisition is similar to the risk ratios of 1.7 to 1.8 reported by 2 of these studies.6,9 Therefore, a modest increase in risk of cervical infections among OC users cannot be ruled out.
Our finding that ectopy is not an important mediator of the hormonal contraception–cervical infection association was surprising to us. Nevertheless, this finding agrees with the Kenyan study that found that increased susceptibility to chlamydial infections associated with hormonal contraceptive use was independent of any increase in ectopy caused by hormonal contraception.9 The studies also agree that cervical ectopy could independently confer an increased risk of chlamydial infection.
Hypotheses for the physiological mechanism by which DMPA (progestin) increases susceptibility to cervical infection include:
1. DMPA leads to a hypoestrogenic state. Thinning of the vaginal epithelium associated with a hypoestrogenic state has been associated with increased transmission of simian immunodeficiency virus (SIV) in macaques.22 Although DMPA use does not appear to result in a substantial decrease in vaginal epithelial thickness in humans, a reduction in hydrogen–peroxide-producing lactobacilli could result in a corresponding increase in vaginal pH23–25 and increased susceptibility to vaginal and cervical infection.26,27 However, in our study, DMPA use was not associated with higher vaginal pH or bacterial vaginosis.
2. Animal studies suggest that estrogen and progesterone could enhance growth and persistence of genital chlamydial infections, although the relevance of these models to humans is unclear.9,21,28
3. Analogous to pregnancy, estrogen and progesterone could depress the immune system, including local immune factors at the cervix. However, the impact of immunologic factors on the acquisition of cervical infections remains unclear.28,29
Our study has a number of important strengths. First, it was specifically designed to evaluate the association between hormonal contraception and cervical infection; thus, we enrolled large numbers of women initiating OC and DMPA use (no prior OC and DMPA use for a minimum of 1 and 4 months, respectively) and not using hormonal contraception. Second, because we enrolled only women initiating hormonal contraception (and few women used hormonal contraception in the 6 months before enrollment), it is unlikely that prior hormonal contraceptive use had an important effect on study results. Third, hormonal contraceptive use was measured carefully, on a monthly basis, with self-reports validated against clinic records; the specific OC type was always recorded. Fourth, a very sensitive measure of chlamydial infection was used (ligase chain reaction), and women were free of cervical infections before contributing any person-time to the study. Likewise, we used a sensitive and highly reliable measure of cervical ectopy.17 Finally, we enrolled a diverse population of women seeking reproductive health services (various ages, ethnic groups, and so on). This is important because the dynamics of cervical infections could differ by population groups. Also, many STI studies are done in sex worker populations and might not be generalizable to the large majority of women who do not engage in commercial sex.
Several factors could have biased the study results. Foremost is that for ethical and practical reasons, we did not believe we could randomize women to contraceptive groups and thus differences in prognostic factors between groups could confound study results. For example, women in the OC group tended to be younger, white, not to have reported a partner with an STI, and had less cervical and vaginal abnormalities at baseline. However, a large confounding effect on study results is unlikely, particularly for the findings regarding DMPA use. First, data were carefully collected on many possible confounding factors. They were individually evaluated in multivariable modeling and retained if they appreciably influenced the measure of effect for hormonal contraception. This process resulted in important changes in our measure of effect of OC use on cervical infection between bivariable (HR, 0.7; 95% CI, 0.3–1.5) and multivariable (HR, 1.5; 95% CI, 0.6–3.5) models. Second, as previously discussed, the OC group was clearly the most different, whereas the DMPA and control groups were relatively similar on most risk factors. We would thus expect any residual confounding to primarily influence the effect of OC use on cervical infection acquisition. Third, few incident infections were diagnosed among women with prevalent infections at enrollment. Thus, most incident infections were not the result of reinfection by sex partners who were infected at study enrollment. Finally, during follow up, women in the hormonal groups reported less exposure to sex partners with a possible STI than control group participants. This suggests that any residual bias might underestimate the risks for acquiring a cervical infection among hormonal contraceptive users.
Study results could also have been influenced by incomplete follow up if participants with riskier sexual behaviors or greater exposure to cervical infections had differential rates of study completion among the various contraceptive groups. Among women included in the analysis, however, participants with incomplete study status (n = 159) were similar to study completers (n = 660) on most characteristics and completion status did not differ by contraceptive group. Thus, among women included in the analysis, study results should not be biased by study completion rates.
This study addresses the effect of hormonal contraception on acquisition of chlamydial and gonococcal infections. The effect of hormonal contraception on the acquisition of other STI, and particularly HIV infection, remains unclear. This study raises the possibility that an increased risk of cervical infections among DMPA users could mediate a possible DMPA–HIV association.
The findings of this article underscore the need to counsel all women who use hormonal contraception and are not in a mutually monogamous relationship to use condoms consistently and correctly. However, if our findings in respect to DMPA use and cervical infection risk are corroborated by further high-quality research, contraceptive counseling for clients initiating or continuing hormonal contraception in high-prevalence STI settings might need to be adjusted to reflect these findings.
2. Plummer FA, Simonsen JN, Cameron DW, et al. Cofactors in male–female sexual transmission of human immunodeficiency virus type 1. J Infect Dis 1991; 163:233–239.
3. Laga M, Manoka A, Kivuvu M, et al. Non-ulcerative sexually transmitted diseases as risk factors for HIV-1 transmission in women: Results from a cohort study. AIDS 1993; 7:95–102.
4. Martin HL Jr, Nyange PM, Richardson BA, et al. Hormonal contraception, sexually transmitted diseases, and risk of heterosexual transmission of human immunodeficiency virus type 1. J Infect Dis 1998; 178:1053–1059.
5. Cottingham J, Hunter D. Chlamydia trachomatis and oral contraceptive use: A quantitative review. Genitourin Med 1992; 68:209–216.
6. Louv WC, Austin H, Perlman J, et al. Oral contraceptive use and the risk of chlamydial and gonococcal infections. Am J Obstet Gynecol 1989; 160:396–402.
7. Avonts D, Sercu M, Heyerick P, et al. Incidence of uncomplicated genital infections in women using oral contraception or an intrauterine device: A prospective study. Sex Transm Dis 1990; 17:23–29.
8. Rahm VA, Odlind V, Pettersson R. Chlamydia trachomatis in sexually active teenage girls. Factors related to genital chlamydial infection: A prospective study. Genitourin Med 1991; 67:317–321.
9. Baeten JM, Nyange PM, Richardson BA, et al. Hormonal contraception and risk of sexually transmitted disease acquisition: Results from a prospective study. Am J Obstet Gynecol 2001;185:380–385.
10. Singer A. The cervical epithelium during puberty and adolescence. In: Jordan JA, Singer A, eds. The Cervix. London: WB Saunders Co, 1976:87–97.
11. Goldacre MJ, Loudon N, Watt B, et al. Epidemiology and clinical significance of cervical erosion in women attending a family planning clinic. BMJ 1978; 1:748–750.
12. Critchlow CW, Wolner-Hanssen P, Eschenbach DA, et al. Determinants of cervical ectopia and of cervicitis: Age, oral contraception, specific cervical infection, smoking, and douching. Am J Obstet Gynecol 1995; 173:534–543.
13. Chacko MR, Lovchik JC. Chlamydia trachomatis infection in sexually active adolescents: Prevalence and risk factors. Pediatrics 1984; 73:836–840.
14. Toon PG, Arrand JR, Wilson LP, et al. Human papillomavirus infection of the uterine cervix of women without cytological signs of neoplasia. BMJ (Clin Res Ed) 1986; 293:1261–1264. Martin HL Jr, Nyange PM, Richardson BA, et al. Hormonal contraception, sexually transmitted diseases, and risk of heterosexual transmission of human immunodeficiency virus type 1. J Infect Dis 1998; 178:1053–1059.
15. Plourde PJ, Pepin J, Agoki E, et al. Human immunodeficiency virus type 1 seroconversion in women with genital ulcers. J Infect Dis 1994; 170:313–317.
16. Collier AC, Handsfield HH, Ashley R, et al. Cervical but not urinary excretion of cytomegalovirus is related to sexual activity and contraceptive practices in sexually active women. J Infect Dis 1995; 171:33–38. Louv WC, Austin H, Perlman J, et al. Oral contraceptive use and the risk of chlamydial and gonococcal infections. Am J Obstet Gynecol 1989; 160:396–402.
17. Morrison CS, Bright P, Blumenthal PD, et al. Computerized planimetry versus clinical assessment for the measurement of cervical ectopia. Am J Obstet Gynecol 2001; 184:1170–1176.
18. Shah BV, Barnwell BG, Bieler GS. SUDAAN User’s Manual, release 7.5. Research Triangle Park, NC: Research Triangle Institute, 1977.
19. Andersen PK, Gill RD. Cox’s regression model for counting processes. Ann Stat 1982; 10:1100–1120.
20. SAS/STAT User’s Guide, version 8. Cary, NC: SAS Institute Inc, 1999.
21. Jacobson DL, Peralta L, Farmer M, et al. Relationship of hormonal contraception and cervical ectopy as measured by computerized planimetry to chlamydial infection in adolescents. Sex Transm Dis 2000; 27:313–319. Singer A. The cervical epithelium during puberty and adolescence. In: Jordan JA, Singer A, eds. The Cervix. London: WB Saunders Co, 1976:87–97.
22. Marx PA, Spira AI, Gettie A, et al. Progesterone implants enhance HIV vaginal transmission and early virus load. Nat Med 1996; 2:1084–1089.
23. Miller L, Patton DL, Meier A, et al. Depomedroxyprogesterone-induced hypoestrogenism and changes in vaginal flora and epithelium. Obstet Gynecol 2000; 96:431–439.
24. Bahamondes L, Trevisan M, Andrade L, et al. The effect upon the human vaginal histology of the long-term use of the injectable contraceptive Depo-Provera. Contraception 2000; 62:23–27.
25. Mauck CK, Callahan MM, Baker J, et al. The effect of one injection of Depo-Provera on the human vaginal epithelium and cervical ectopy. Contraception 1999; 60:15–24.
26. Donders G, De Wet HG, Hooft P, et al. Lactobacilli in Papanicolaou smears, genital infections, and pregnancy. Am J Perinatol 1993; 10:358–361.
27. Taha TE, Hoover DR, Dallabetta GA, et al. Bacterial vaginosis and disturbances of vaginal flora: Association with increased acquisition of HIV. AIDS 1998; 12:1699–1706.
28. Sonnex C. Influence of ovarian hormones on urogenital infection. Sex Transm Infect 1998; 74:11–19. Morrison CS, Bright P, Blumenthal PD, et al. Computerized planimetry versus clinical assessment for the measurement of cervical ectopia. Am J Obstet Gynecol 2001; 184:1170–1176.
29. Schuurs A, Geurts T, Goorissen E. Immunologic effects of estrogens, progestins, and estrogen-progestin combinations. In: Goldzieher J, Fotherby K, eds. Pharmacology of the Contraceptive Steroids. New York: Raven Press, 1994:379–399.