Worldwide more than 15 million women are human immunodeficiency virus (HIV)–infected, the vast majority of whom are of reproductive age. In the United States 6,000–7,000 HIV-infected women give birth every year, whereas globally there are now close to half a million children with HIV infections that were acquired by mother-to-child transmission.1 Many HIV-infected women are coinfected with human papilloma virus (HPV), the causative virus of cervical cancer.2,3 Various factors, including patient age,4 oral contraceptive use,5 and smoking,6 have been linked to HPV carriage. While there is evidence that hormonal environment5 and parity7 may be linked to cervical cancer, the effect of pregnancy on carriage of HPV has not been fully defined. Additionally, certain types of HPV (eg, types 6 and 11), by dint of their association with juvenile laryngeal papillomatosis, might have unique consequences in pregnancy. Therefore, any effect of pregnancy on the course of HPV disease would have major public health significance.
Although several investigators have compared rates of HPV between pregnant and nonpregnant women, these studies have tended to be cross-sectional and have generally focused on HIV-uninfected women. Most studies reported higher rates of HPV infection in pregnant than in nonpregnant women. However, because of their predominantly cross-sectional design, it is hard to say whether reported differences between carriage rates in pregnant and nonpregnant women reflect differences between women who choose to have children and those who do not or whether the differences they report reflect a biologic effect of pregnancy. Speculation regarding the causes of high rates of HPV infection reported among pregnant women in some studies has focused on the immunosuppressive state induced by pregnancy and changes in the hormonal milieu.8,9 With this background, it is difficult to know whether to expect that pregnancy will have little effect on risk of HPV in HIV-positive women, because they are already immune compromised, or whether there will be a potentiating effect.
It is concern regarding the latter possibility, that HIV-positive women are at especially high risk of oncogenic HPV infection,2,3,10–13 and particularly HPV infection with high copy number, a possible risk factor for persistent HPV infection leading to disease, that prompted our investigation. Therefore, we studied the prevalence, incident detection, and HPV copy number in two of the largest cohorts of HIV-infected pregnant women in the United States. To minimize concern that differences in the characteristics of women who become pregnant and those who do not might explain the results, women in our study essentially acted as their own controls, ie, we assessed these women before pregnancy (women in Women’s Interagency HIV Study [WIHS]), during pregnancy (WIHS and Women and Infants Transmission Study [WITS]), and shortly after pregnancy (WIHS and WITS) to estimate how pregnancy relates to the natural history of HPV infection.
MATERIALS AND METHODS
This study included women who had pregnancies in the WITS, and the WIHS, two national multicenter cohorts. The WITS is a prospective observational study of HIV-positive pregnant women and their infants. Between December 1989 and September 2004, 3,229 pregnant HIV-1–infected women and their infants were recruited by WITS investigators at study centers in Illinois (Chicago), Massachusetts (Boston and Worcester), New York City, and Puerto Rico (San Juan). In 1991 and 1993 Brooklyn and Texas (Houston), respectively, became WITS Clinical Sites. After providing informed written consent, women were enrolled at any time during pregnancy and up to 7 days after delivery. Infants were enrolled at birth or within 7 days of birth. Of these women, 450 had adequate volumes of cervical vaginal lavage (CVL) available, after unrelated studies were performed, to allow HPV to be assessed and were included in the current analysis.
The WIHS is a prospective study of the natural history of HIV infection in women that enrolled 2,056 HIV-seroprevalent and 554 at-risk seronegative women participants at six clinical consortia (Bronx, NY; Brooklyn, NY; Chicago; Los Angeles; San Francisco; Washington) between October 1994 and November 1995. Among these women, 178 HIV-infected women became pregnant at some time during the study, and were included in the current analysis. Details of recruitment and methods for WITS14 and WIHS15 have been published elsewhere and are summarized below. Each institution’s institutional review board in both WIHS and WITS approved the study. Together, 628 HIV-infected pregnant women enrolled in either WIHS or WITS contributed data to our substudy.
Study visits for women enrolled in WITS occurred during the following pregnancy and postpartum time periods: less than 18 weeks of gestation (visit 1), 19–31 weeks of gestation (visit 2), 32 weeks or more of gestation (visit 3), at delivery (visit 4), 2 months postpartum (visit 5), 6 months postpartum (visit 6), and every 6 months thereafter. At each visit, a detailed medical and behavioral questionnaire was administered, a physical examination was performed, and specimens, including venous blood, Pap test, and cervicovaginal lavage for HPV DNA testing were obtained. Obstetric data were obtained from medical record abstraction. In WIHS, participants had follow-up visits every 6 months that included an interviewer-administered questionnaire, a physical examination, and the collection of specimens, including Pap test and cervicovaginal lavage for HPV DNA testing. Steps were taken in the analysis to make the data from the two studies similar in terms of the amount of follow-up time, and to place these data on a common time metric (see information below on statistical methods).
In both studies, exfoliated cervical cells for HPV DNA testing were obtained by CVL before Pap test, as follows: using a syringe equipped with a 2-inch, 18-gauge catheter, 10 mL of sterile normal saline were sprayed against the cervical os and the exocervix. Using the same syringe, the fluid was aspirated from the posterior vaginal fornix and transferred to a 15-mL sterile polypropylene tube. If the volume recovered was less than 6 mL, a second lavage, using 5 mL of sterile normal saline was conducted and added to the 15 mL tube. The CVL was held on ice until processing within 6 hours. In the laboratory, the CVL was vortexed gently to evenly distribute cells and then aliquoted under sterile conditions. Once processed, specimens were kept at –70°C until tested.
Antiretroviral treatment was not prescribed by the study protocol in either cohort but was left to the discretion of the clinicians and patients. Most ZDV use before March 1994 reflected use for maternal health considerations (as opposed to use for prevention of maternal-child transmission). Women’s antiretroviral treatment use was classified as follows: highly active retroviral therapy (HAART) including three or more drugs, one of which was either a protease inhibitor or a nonnucleoside reverse transcriptase inhibitor; “combination therapy“ was defined as either two drugs, one of which could be a protease inhibitor or nonnucleoside reverse transcriptase inhibitor, or two or more drugs that did not include a protease inhibitor or nonnucleoside reverse transcriptase inhibitor, or the combination of AZT, 3TC, and abacavir; ”monotherapy" was defined as treatment with a single drug. Cigarette smoking was a self-reported variable.
The HPV DNA polymerase chain reaction (PCR) assay methods used have been described in detail elsewhere.2 In brief, HPV DNA was detected using L1 consensus primer MY09/MY11/HMB01.2,16,17 Control primer set PC04/GH20, amplifying a 268-bp cellular β-globin DNA fragment, was included in each assay to serve as an internal control for amplification. Following proteinase K digestion, 2–10 μL of each cell digest was used in reactions containing 10 mM Tris-HCL, 50 mM KCl, 4 mM MgCl2, 200 μM of each deoxyribonucleotide triphosphate, 2.5 units Taq DNA polymerase, and 0.5 μM of each primer. There were 40 amplification cycles (95°C for 20 seconds, 55°C for 30 seconds, and 72°C for 30 seconds, with a 5 minute extension period at 72°C on the last cycle). Amplified material was then detected using filters individually hybridized with biotinylated type-specific oligonucleotide probes for multiple HPV types including HPV 2, 6, 11, 13, 16, 18, 26, 31, 32, 33, 34, 35, 39, 40, 42, 45, 51, 52, 53, 54, 55, 56, 57, 58, 59, 61, 62, 63, 64, 66, 67, 68, 69, 70, 71 (AE8), 72, 73 (PAP238A), 74, 81 (AE7), 83(PAP291), 82(W13B and AE2), 84 (PAP155), 85 (AE5),88, 89 (AE6), and 92, as described.2,18,19 For these analyses, oncogenic and nononcogenic types were defined in accordance with recently published data that classified types using both phylogenetic and epidemiologic criteria. Oncogenic types were HPV-16, -18, -31, -33, -35, -39, -45, -51, -52, -56, -58, -59, -68, and 73, and Nononcogenic types included 6, 11, 13, 26, 32, 34, 40, 42, 53, 54, 55, 57, 61, 62, 64, 66, 67, 69, 70, 71, 72, 81, 82, 83, 84, 85, and 89.20
HPV copy number by individual HPV type was estimated using the signal intensity observed for each type-specific oligonucleotide probe on the HPV DNA dot blot hybridization filter. Specifically, signal intensity was graded using a 1–5 scale according to the radius of the hybridization circle (the “dot density”), through visual comparison of the radius size with a standardized set of dot densities shown on a reference card. Each dot-density result was graded without knowledge of other relevant study data, including the hybrid capture 2 relative light unit ratios. Dot blot signal intensity has been validated as a measure of HPV viral load by comparison with a quantitative TaqMan (Applied Biosystems, Foster City, CA) real-time, HPV16 type-specific PCR assay in two different laboratories (J. Palefsky and P. Gravitt, personal communication). In the WIHS itself, the correlation between PCR dot blot intensity and hybrid capture 2 relative light unit ratios was moderate (Pearson correlation 0.56) in baseline samples, and dot density results were significantly associated with the presence and grade of neoplasia, pointing to the clinical relevance of such data (Palefsky JM, Harris T, Fazzari M, Ahdieh-Grant L, da Costa M, Anastos K, et al. Quantitation of cervicovaginal HPV DNA level in HIV-positive women and its associations with time to clearance of HPV infection and incidence of cervical lesions. International Papillomavirus Meeting, Vancouver, Canada, April 30 to May 6, 2005). The reproducibility of grading the signal intensity on the 1–5 scale was also formally evaluated between two laboratories (J. Palefsky, University of California at San Francisco, and R. Burk, Albert Einstein College of Medicine), and the kappa agreement was shown to be high (κ 0.76) (Palefsky et al 2005).
Blood samples were collected in tubes with heparin or ethylenediaminetetraacetic acid, transported at room temperature to the local laboratory, and separated into serum, plasma, and peripheral blood mononuclear cell aliquots within 24 hours of collection. The CD4 counts were determined in real time by flow cytometric immunophenotyping using standardized monoclonal antibody panels (Becton Dickinson, San Jose, CA). All laboratories followed National Institute of Allergy and Infectious Diseases cytometry protocols and participated in the National Institute of Allergy and Infectious Diseases monthly flow cytometry quality assurance program.21 Plasma HIV-1 RNA was measured in virology quality assurance–certified laboratories according to AIDS Clinical Trial Group virology quality assurance recommendations.22,23
The prevalence of HPV DNA was expressed as the percentage of women with adequate HPV test results (ie, those in whom amplification of β-globin was detected—an indicator that the PCR performed satisfactorily). Incident detection of HPV was defined as a positive test result for any HPV type that was not present at baseline or at any other earlier visit in a given woman. We use the term incident detection, rather than incidence, because the relative contributions of newly acquired HPV infections and reactivation of previously acquired latent HPV infections cannot be estimated in a population with many years of sexual activity, regardless of the HIV status of that population.
Human papilloma virus prevalence, incident detection, and HPV copy number were compared between pregnancy and the periods 1) before (prepregnancy) and 2) after pregnancy (postpregnancy). The comparisons were made by estimating risk ratios (RRs; ie, the probability of HPV prevalence, incident detection, or high compared with low HPV copy number during pregnancy divided by the analogous probability during nonpregnancy). Specifically, these RRs were estimated using marginal binary response–generalized estimating equation regression models to control for the fact that the data reflected repeated observations of the same women over time, and all models were additionally adjusted for covariates thought to affect HPV natural history: age, CD4+ T cell count, HIV RNA level, number of lifetime sexual partners (more than 10), gestational age, number of parity, smoking, antiretroviral therapy, oral contraceptives, and a “period effect” variable (to address the introduction of HAART during the period of observation, ie, controlling for temporal confounding previously detected in WITS.24). Time-dependent covariates were used for all variables that changed over time, such as CD4 T cell count, and use of HAART.25 CD4 cells and HIV RNA levels were stratified into classes, shown to be associated with risk of HPV infection in previous studies.25 CD4 were classified as greater than 500 cells per mm, 200–500 cells per mm, and fewer than 200 cells per mm. Viral load was stratified into four classes, namely, less than 4,000 copies per mL, 4,000≤20,000 copies per mL, 20,001–100,000 copies per mL, and more than 100,000 copies per mL. The regression models assumed an exchangeable correlation structure to adjust standard errors for repeated measurements, but these models are robust as to the assumed correlation structure.26 The repeated measurements consisted of multiple concurrent HPV infections involving the same woman assessed at multiple study visits. Human papilloma virus types were grouped into two categories, oncogenic and nononcogenic types.20
Since the visit schedule of WITS and WIHS differed in both the number of HPV measurements taken on the women and the follow-up time (in WITS CVL samples were only obtained out to 6 months postpartum therefore the maximal follow-up was 15 months), time intervals were defined that would be identical for participants in both studies. We defined the intervals to be one fifth of a year, or 73 days. If more than one visit appeared within one interval, only the first visit was included in the analysis. This happened in fewer than 5% of the intervals. Furthermore, WITS had a maximum of 15-month follow-up time, whereas WIHS subjects typically had many years of follow-up data. To make these two studies more comparable, follow-up time for WIHS women was limited to the first six visits, approximately 2.5 years. These six visits were selected so that they began at the two prepregnancy visits that were closest in proximity to the first pregnancy visit. By defining a single set of time intervals for the two studies, prevalences in women in both studies are separated by a common time metric. For the analysis of HPV copy number the analysis conditioned on HPV being present. Posteriori power estimates for all analyses were calculated with power simulations that used bootstrap methods27 using SAS 8 software (SAS Institute Inc. Cary, NC).
Table 1 describes the characteristics of the combined cohort. The prevalence of high-risk and low-risk types of HPV were 32.8% and 33.3%, respectively. Slightly less than half of women (45.8%) were current smokers and 30.4% had more than ten lifetime sexual partners. A relatively small percentage of women had advanced HIV disease, with 8.9% having a CD4 count less than 200 mm3 and 8.0% having an HIV viral load more than 100,000 copies. Slightly more than 20% of women were on HAART at the baseline visit.
Table 2 shows the relationships between pregnancy and our several HPV endpoints, namely HPV prevalence, incident detection, and HPV copy number. The prevalence and copy number of oncogenic and nononcogenic HPV did not significantly differ between pregnancy and either the prepregnancy or postpregnancy periods. Incident HPV detection was significantly lower for both oncogenic and nononcogenic HPV during pregnancy compared with the postpregnancy (relative risk [RR] 0.534, 95% confidence interval [CI] 0.390–0.732, P<.001 and RR 0.577, 95% CI 0.428–0.779, P<.001, respectively), but not compared with the prepregnancy period.
As a precaution against potential temporal confounding, we modified our analyses to include the interaction between pregnancy phase with a blocking variable time era: pre-HAART era compared with the post-HAART era. This interaction was not significant in either the HPV prevalence or HPV incidence analyses suggesting that the inferences were homogeneous over time. In the HPV intensity analysis, this interaction was significant at the .05 level: P=.026. This suggests a difference may exist between the relative risk of HPV in prepregnancy compared with during pregnancy in the two time eras. The estimated risk ratios were RR 1.202, 95% CI 0.9111–1.5858, P=.219 in the pre-HAART era and RR 0.8151, 95% CI 0.6606–1.057, P=.057 in the post-HAART era. Hence, although there is weak evidence of temporal confounding across time, the qualitative inference remains homogeneous across time: in both time eras there is no difference in the probability of high copy number during pregnancy compared with during prepregnancy.
Because our analysis failed to detect increases in HPV prevalence during pregnancy compared with nonpregnancy, we posteriori determined the minimal detectable effect that our sample size had 80% power to detect. Using bootstrap methods we found that we had 80% power to detect an RR equal to or greater than 1.25 for either oncogenic or nononcogenic HPV types.
In a large cohort study of HPV infection in pregnant HIV-infected women, we found little evidence to suggest that pregnancy increases the risk of HPV infection or HPV copy number (among those already infected with HPV). In fact, although the prevalence and HPV copy number did not vary significantly with pregnancy, the incident detection of both nononcogenic and oncogenic types of HPV was statistically significantly lower during pregnancy.
Previous studies of the relationship between pregnancy (or parity) and HPV,7,28,29 have typically been small and cross-sectional, were overwhelmingly focused on HIV-seronegative women, and reported conflicting results. For example, Chen29 compared the prevalence and genotype distribution of cervical HPV infection between 308 pregnant and 308 nonpregnant women and did not find any significant difference between groups by HPV type. Conversely, Hernandez-Giron9 screened 274 pregnant and 1,060 age-matched nonpregnant women in Mexico for high-risk HPV infection. High-risk HPV DNA was detected in 37.2% of pregnant women and in 14.2% of nonpregnant women. In multivariate analysis, pregnancy was associated with an increased risk for HPV infection (odds ratio 3.5; 95% CI 2.7–4.9).
Nonetheless, multiparity is one of the few fairly consistently observed cofactors for progression to cervical cancer. Castellesague and colleagues7 in a review of key studies on high-grade squamous intraepithelial lesions and cervical cancer conducted among HPV-positive women concluded that high parity, smoking, and, less consistently, long-term oral contraceptive use were cofactors that might modulate the risk of progression from HPV infection to high-grade squamous intraepithelial lesion or cervical cancer. From a public health point of view, they felt that parity was the behavioral cofactor explaining the highest proportion of cervical cancer cases among HPV-infected women.
Given this background it was unclear whether pregnancy might be a period of extraordinary risk of oncogenic HPV infection among HIV-positive women. Hypotheses to explain the high rates of HPV infection reported among pregnant women in some studies focused on the immunosuppressive state induced by pregnancy and changes in the hormonal milieu. Even if correct, whether that would result in a strong synergistic rise in immunologic susceptibility to oncogenic HPV infection or instead no change among HIV-positive women (since they are already immune compromised) could not be predicted. Furthermore, we were skeptical of previously reported cross-sectional associations of HPV infection with pregnancy, in particular because women who comprised the pregnancy or high parity groups were generally not the same as those in their nonpregnant or low parity groups. Thus a variety of factors that differentiates those groups could have contributed to their findings. In this connection, we observed that when we included the 2,340 WIHS participants who were not pregnant during follow-up in our analyses (data not shown), we also found a significant increase in prevalence of high-risk types of HPV during pregnancy, but the finding reflected differences in the ages of the pregnant and nonpregnant groups, more than any other factor. Whereas when we limited our assessment to only women who had a pregnancy during follow-up in WIHS, our results were null, a reassuring finding.
An important limitation of the current study is that we did not have detailed recent sexual history on all participants and so were limited in our ability to comment on whether the lower incidence of HPV in pregnancy, which we found, was based on immune changes or behavior changes. It is unclear, however, why this would be different in our study from that found in earlier investigations, and we posteriori demonstrated that we had the power to detect a 25% increase in prevalence associated with pregnancy, a much smaller increase than has been reported in association with other factors influencing rates of HPV infection.30 We also note that in previous studies we have been able to detect an association of oncogenic HPV infection with cigarette smoking, an exposure that is thought to affect risk of HPV through the induction of an immunosuppressive state much like that hypothesized for pregnancy.
In any case, whatever limitations our study may have had, it is clear from these data that pregnancy is not a period of extraordinary vulnerability to oncogenic HPV infection among HIV-positive women. We think that these results will be reassuring to HIV-positive women of reproductive age and to their physicians.
2. Palefsky JM, Minkoff H, Kalish LA, Levine A, Sacks HS, Garcia P, et al. Cervicovaginal human papillomavirus infection in human immunodeficiency virus-1 (HIV)-positive and high-risk HIV-negative women. J Natl Cancer Inst 1999;91:226–36.
3. Ahdieh L, Klein RS, Burk R, Cu-Uvin S, Schuman P, Duerr A, et al. Prevalence, incidence, and type-specific persistence of human papillomavirus in human immunodeficiency virus (HIV)-positive and HIV-negative women. J Infect Dis 2001;184:682–90.
4. Ho GY, Bierman R, Beardsley L, Chang CJ, Burk RD. Natural history of cervicovaginal papillomavirus infection in young women. N Engl J Med 1998;338:423–8.
5. Shields TS, Falk RT, Herrero R, Schiffman M, Weiss NS, Bratti C, et al. A case–control study of endogenous hormones and cervical cancer. Br J Cancer 2004;90:146–52.
6. Minkoff H, Feldman JG, Strickler HD, Watts DH, Bacon MC, Levine A, et al. Relationship between smoking and human papillomavirus infections in HIV-infected and -uninfected women. J Infect Dis 2004;189:1821–8.
7. Castellsague X, Munoz N. Chapter 3: Cofactors in human papillomavirus carcinogenesis–role of parity, oral contraceptives, and tobacco smoking. J Natl Cancer Inst Monogr 2003;(31):20–8.
8. Michelin D, Gissmann L, Street D, Potkul RK, Fisher S, Kaufmann AM, et al. Regulation of human papillomavirus type 18 in vivo: effects of estrogen and progesterone in transgenic mice. Gynecol Oncol 1997;66:202–8.
9. Hernandez-Giron C, Smith JS, Lorincz A, Lazcano E, Hernandez-Avila M, Salmeron J. High-risk human papillomavirus detection and related risk factors among pregnant and nonpregnant women in Mexico. Sex Transm Dis 2005;32:613–8.
10. Minkoff H, Feldman J, DeHovitz J, Landesman S, Burk R. A longitudinal study of HPV carriage in human immunodeficiency virus–infected and human immunodeficiency virus–uninfected women. Am J Obstet Gynecol 1998;178:982–986.
11. Minkoff H, Feldman J, Burk R, Eisenberger Matityahu D, Clarke L. Prevalence and incidence of gynecologic disorders among women infected with human immunodeficiency virus. Am J Obstet Gynecol 1999;180:824–36.
12. Ellerbrock TV, Chiasson MA, Bush TJ, Sun XW, Sawo D, Brudney K, et al. Incidence of cervical squamous intraepithelial lesions in HIV-infected women. JAMA 2000;283:1031–7.
13. Gallagher B, Wang Z, Schymura MJ, Kahn A, Fordyce EJ. Cancer incidence in New York State acquired immunodeficiency syndrome patients. Am J Epidemiol 2001;154:544–56.
14. Sheon AR, Fox HE, Rich KC, Stratton P, Diaz C, Tuomala R, et al. The Women and Infants Transmission Study (WITS) of maternal-infant HIV transmission: study design, methods, and baseline data. J Womens Health 1996;5:69–78.
15. Barkan SE, Melnick SL, Preston-Martin S, Weber K, Kalish LA, Miotti P, et al. The Women’s Interagency HIV Study. WIHS Collaborative Study Group. Epidemiology. 1998;9:117–25.
16. Hildesheim A, Schiffman MH, Gravitt PE, Glass AG, Greer CE, Zhang T, et al. Persistence of type-specific human papillomavirus infection among cytologically normal women. J Infect Dis 1994;169:235–40.
17. Tachezy R, Van Ranst MA, Cruz Y, Burk RD. Analysis of short novel human papillomavirus sequences. Biochem Biophys Res Commun 1994;204:820–7.
18. Burk RD, Ho GY, Beardsley L, Lempa M, Peters M, Bierman R. Sexual behavior and partner characteristics are the predominant risk factors for genital human papillomavirus infection in young women. J Infect Dis 1996;174:679–89.
19. Castle PE, Wacholder S, Lorincz AT, Scott DR, Sherman ME, Glass AG, et al. A prospective study of high-grade cervical neoplasia risk among human papillomavirus-infected women. J Natl Cancer Inst 2002;94:1406–14.
20. Munoz N, Bosch FX, de Sanjose S, Herrero R, Castellsague X, Shah KV, et al. Epidemiologic classification of human papillomavirus types associated with cervical cancer. N Engl J Med 2003;348:18–527.
21. Calvelli T, Denny T, Paxton H, Gelman R, Kagan J. Guideline for flow cytometric immunophenotyping; a report from the National Institute of Allergy and Infectious Diseases, Division of DAIDS. Cytometry 1993;14:702–15.
22. Hollinger B., ACTG virology manual for HIV laboratories Version 2.1. Bethesda, (MD): National Institutes of Health; 1993.
23. Shearer WT, Quinn TC, LaRussa P, Lew JF, Mofenson L, Almy S, et al. Viral load and disease progression in infants infected with human immunodeficiency virus type 1. Women and Infants Transmission Study Group. N Engl J Med 1997;336:1337–42.
24. Neu N, Leighty R, Adeniyi-Jones S, Diaz C, Handelsman E, Kaufman G, et al. Immune parameters and morbidity in hard drug and human immunodeficiency virus-exposed but uninfected infants. Pediatrics 2004;113:1260–6.
25. Strickler HD, Burk RD, Fazzari M, Anastos K, Minkoff H, Massad LS, et al. Natural history and possible reactivation of human papillomavirus in human immunodeficiency virus-positive women. J Natl Cancer Inst 2005;97:577–86.
26. Zeger SL, Liang KY. Longitudinal data analysis for discrete and continuous outcomes. Biometrics 1986;42:121–30.
27. Efron B, Tibshirani RJ. An introduction to the bootstrap. New York (NY): Chapman & Hall; 1993. p. 37–381.
28. Castle PE, Walker JL, Schiffman M, Wheeler CM. Hormonal contraceptive use, pregnancy and parity, and the risk of cervical intraepithelial neoplasia 3 among oncogenic HPV DNA-positive women with equivocal or mildly abnormal cytology. Int J Cancer 2005; 117:1007–12.
29. Chan PK, Chang AR, Tam WH, Cheung JL, Cheng AF. Prevalence and genotype distribution of cervical human papillomavirus infection: comparison between pregnant women and non-pregnant controls. J Med Virol 2002 Aug;67:583–8.
30. Strickler HD, Palefsky JM, Shah KV, Anastos K, Klein RS, Minkoff H, et al. Human papillomavirus type 16 and immune status in human immunodeficiency virus-seropositive women. J Natl Cancer Inst 2003;95:1062–71.
WITS Principal investigators, study coordinators, program officers and funding include: Clemente Diaz, Edna Pacheco-Acosta (University of Puerto Rico, San Juan, PR; U01 AI 034858); Ruth Tuomala, Ellen Cooper, Donna Mesthene (Boston–Worcester Site, Boston, MA; 9U01 DA 015054); Phil LaRussa, Alice Higgins (Columbia Presbyterian Hospital, NY, NY; U01 DA 015053); Sheldon Landesman, Edward Handelsman, Ava Dennie (State University of New York, Brooklyn, NY; U01 HD 036117); Kenneth Rich, Delmyra Turpin (University of Illinois at Chicago, Chicago, IL; U01 AI 034841); William Shearer, Susan Pacheco, Norma Cooper (Baylor College of Medicine, Houston, TX; U01 HD 041983); Joana Rosario (National Institute of Allergy and Infectious Diseases, Bethesda, MD); Robert Nugent, (National Institute of Child Health and Human Development, Bethesda, MD); Vincent Smeriglio, Katherine Davenny (National Institute on Drug Abuse, Bethesda, MD); and Bruce Thompson (Clinical Trials & Surveys Corp., Baltimore, MD, N01 AI 085339). Scientific Leadership Core: Kenneth Rich (PI), Delmyra Turpin (Study Coordinator) (1 U01 AI 050274–01)
The WIHS Collaborative Study Group includes centers (Principal Investigators) at New York City–Bronx Consortium (Kathryn Anastos); Brooklyn, NY (Howard Minkoff); Washington DC Metropolitan Consortium (Mary Young); The Connie Wofsy Study Consortium of Northern California (Ruth Greenblatt); Los Angeles County/Southern California Consortium (Alexandra Levine); Chicago Consortium (Mardge Cohen); and Data Coordinating Center (Stephen Gange).