Since 2006 the Advisory Committee on Immunization Practices has recommended the human papillomavirus (HPV) vaccine for routine administration in girls aged 11–12 years with catch-up vaccination for women through age 26 years.1 There are currently three HPV vaccine formulations: the bivalent vaccine, quadrivalent (4vHPV) vaccine, and nonvalent vaccine, targeting an increasing number of HPV subtypes. In the United States, 4vHPV was the primary vaccine administered until the 2015 licensure of the nonvalent one. Despite suboptimal coverage during adolescence, HPV vaccination has been associated with significant population-level decreases in HPV prevalence.2–4
Pregnancy is listed as a precaution for HPV vaccination.1,5 Because approximately 10% of women 18–26 years of age experience pregnancy each year,6 and routine pregnancy testing before vaccination is not advised, inadvertent vaccination during pregnancy does occur. To date, data on HPV vaccine exposures during or around the time of pregnancy have not raised concerns.7–11 Using automated data from the Vaccine Safety Datalink, our group previously reported inadvertent 4vHPV vaccination during pregnancy or periconception occurred in 1.5% of pregnancies among women aged 13–27 years; these exposures were not associated with adverse maternal or neonatal outcomes including preeclampsia, gestational diabetes, preterm birth, small for gestational age, and birth defects.12
Prior studies on HPV vaccination and risks for fetal demise have been limited by small numbers,8,10 lack of a comparison group,9 or inability to conduct detailed case adjudication.7 Goals of the current study were to evaluate risks for spontaneous abortion after 4vHPV vaccination during pregnancy or peripregnancy within the Vaccine Safety Datalink.
MATERIALS AND METHODS
This retrospective, observational cohort study was conducted within the Vaccine Safety Datalink. The Vaccine Safety Datalink is a collaborative effort between the Centers for Disease Control and Prevention's Immunization Safety Office and several large integrated health care systems to monitor the safety of vaccines and conduct studies about rare and serious adverse events after immunizations administered within the United States.13
Data for this study came from the seven Vaccine Safety Datalink sites (Kaiser Permanente Northern California, Kaiser Permanente Southern California, Kaiser Permanente Northwest, Kaiser Permanente Washington, Kaiser Permanente Colorado, Marshfield Clinic, and HealthPartners) with available data files. Adolescent and young women aged 12–27 years, who had pregnancies ending between January 2008 and November 19, 2014, were identified based on claims, electronic health, and birth records using a validated algorithm.14 Continuous insurance coverage for the period starting 8 months before the last menstrual period (LMP) through the end of pregnancy was required. To be included in the final cohort for analysis, at least one dose of 4vHPV had to be administered at a Vaccine Safety Datalink site during a prespecified risk window: distal (16–22 weeks before the LMP), peripregnancy (the 42 days before the LMP), and during pregnancy (LMP through 19 weeks of gestation) (Fig. 1). Risk windows were chosen based on review of existing literature7,10,15 and potential for biological plausibility, allowing for a washout period between the peripregnancy and during pregnancy risk windows and the distal comparison, the control period. To accommodate data availability and to reduce the number of charts for review, eligible dates for HPV administration varied by site. For HealthPartners, all 4vHPV doses were administered from January 1, 2008, to December 31, 2012; for Kaiser Southern California all 4vHPV doses were administered from January 1, 2012, to December 31, 2013; at the remaining Vaccine Safety Datalink sites, 4vHPV doses were administered from January 1, 2008, to December 31, 2013. Vaccines were identified from standardized Vaccine Safety Datalink files and based on claims, electronic health records, and immunization registry data.
Pregnancy outcomes were initially identified through the previously described and validated pregnancy algorithm.14 This algorithm assigns gestational age for live births based on birth records. For spontaneous abortions and stillbirths, because there are no birth records, the algorithm assigns default gestational ages of 10 and 28 weeks, respectively. In a second step, spontaneous abortion and stillbirth outcomes were confirmed through medical chart review and the algorithm-assigned gestational ages were replaced by more accurate values based on the date of LMP and the estimated due date (EDD) obtained from manual chart review. Cases classified through the automated algorithm as “unknown abortion” also underwent chart review to identify additional spontaneous abortions. Consistent with prior studies, spontaneous abortions occurring before 6 weeks of gestation were excluded because these are often difficult to confirm and many of these early pregnancy losses are unrecognized.16 The following pregnancy outcomes were also excluded: therapeutic abortions, ectopic pregnancies, gestational trophoblastic disease, and multiple gestations. In addition, women with pharmacy dispensings or claims for medications increasing risk of spontaneous abortion, starting 6 months before pregnancy and through the end of pregnancy, were excluded (Table 1). The final cohort consisted of women with spontaneous abortions occurring between 6 and less than 20 weeks of gestation, stillbirths occurring between 20 and 43 weeks of gestation, and live births.
Chart reviews were conducted by trained chart abstractors at all Vaccine Safety Datalink sites with data entered into Research Electronic Data Capture.17 Data collected included LMP, EDD, urine or serum human chorionic gonadotropin levels, onset of spontaneous abortion symptoms, ultrasound results, and pathology results. As described previously, all spontaneous abortions, unknown abortions, and stillbirths were initially identified using the pregnancy algorithm and assigned a default gestational age at fetal demise; however, more precise classification of 4vHPV into the three exposure windows was determined once more accurate clinical data were available. Based on the adjudication results and verified EDD or LMP dates, eligible pregnancies were selected in each of the three eligible exposure windows. To start, using a wide exposure window, those with one or more doses of 4vHPV administered up to 29 weeks before the algorithm-assigned pregnancy start or LMP through the end of pregnancy underwent chart review. Based on LMP and EDD data entered during chart review, pregnancy start and end dates were adjusted. Spontaneous abortions with 4vHPV exposures in wide exposure windows (distal 24–14 weeks and peripregnancy 56 days to LMP) or for the period starting 2 weeks before and ending 2 weeks after the final exposure windows underwent adjudication. After adjudication, spontaneous abortions and stillbirths with 4vHPV exposures in the final exposure windows for analysis, distal (16–22 weeks before the LMP), peripregnancy (within 42 days before the LMP), and during pregnancy (LMP through 19 weeks of gestation) were retained (Fig. 1). Patients with more than one 4vHPV dose in an eligible exposure window were assigned an exposure window with “during pregnancy” exposures as highest priority and “distal vaccination” as lowest priority.
Spontaneous abortion outcomes were confirmed based on published criteria from the American College of Obstetricians and Gynecologists.18 Gestational age at fetal demise was assigned based on reported symptomatology, LMP and EDD, and ultrasound and pathology reports. Urine and serum human chorionic gonadotropin levels were also used to identify early spontaneous abortions and other pregnancy outcomes. All spontaneous abortions with ultrasound or pathology reports uploaded into Research Electronic Data Capture underwent further case adjudication by two obstetrician investigators (H.S.L. and S.S.S.) to confirm outcomes and assign gestational age at the time of fetal demise.
Covariates, including demographics, prepregnancy maternal comorbidities (eg, diabetes, hypertension, and systemic lupus erythematosus), maternal health care utilization (having one or more ambulatory visits and one or more inpatient visits in the year before pregnancy), 4vHPV dose number, and receipt of other vaccines before 20 weeks of gestation, and on the same day as peripregnancy or during pregnancy 4vHPV administrations, were assessed through electronic health record and administrative data. Obesity and smoking were assessed through multiple sources including International Classification of Diseases, 9th Revision codes, electronic health record data (for live births), and chart review (for spontaneous abortions and stillbirths).
We first conducted descriptive analyses for the covariates listed previously, comparing frequency distributions by exposure group. Spontaneous abortion risk after 4vHPV during pregnancy, peripregnancy, and peripregnancy or during pregnancy was compared with distal vaccination, the control period, using time-dependent covariate Cox regression models. This approach allowed us to account for both immortal time bias19 and to adjust for potential confounders. We also compared peripregnancy or during pregnancy 4vHPV exposures when administered concomitantly with other vaccines with distal vaccination using a time-dependent covariate Cox regression model. In addition, we compared spontaneous abortion risk after 4vHPV periconception with distal vaccination using standard Cox regression models. Two models for each association were computed; the first one included maternal age as a fixed effect and full adjusted models included as fixed effects maternal age, neighborhood poverty index based on census geocode, health care utilization in the previous year of pregnancy, smoking, obesity (body mass index [calculated as weight (kg)/[height (m)]2] 30 or greater), race, ethnicity, 4vHPV dose number (first, second, or third), and site.
We also explored whether there was a period around the time of conception when an increased risk was observed. For these exploratory analyses, we limited the cohort to women receiving the 4vHPV from 42 days before the LMP to 28 days after the LMP and used a generalized additive model to fit a smooth curve. For women with 4vHPV exposure during pregnancy, we explored timing from vaccination to spontaneous abortion by gestational week at vaccination. In additional to exploratory analyses, we evaluated among vaccinations administered at greater than 4 weeks of gestation whether there was an increased risk within 28 days after 4vHPV when compared with distal vaccination. In final exploratory analyses, we evaluated whether spontaneous abortion risks varied by dose number or with concomitant vaccination.
Our study was powered to detect hazard rate ratios of 1.37 or greater assuming 1,000 women with 4vHPV exposure during pregnancy and 1,000 during distal vaccination with 80% power at .05 significance level with a two-sided log-rank test. This study was approved by the institutional review boards of all participating sites with a waiver of informed consent.
Of 208,965 pregnancies in women aged 12–27 years, with continuous insurance enrollment from 8 months before pregnancy through pregnancy end, we identified 2,800 eligible pregnancies with 4vHPV exposure in the specified study exposure windows; 919 distal (33%), 986 (35%) peripregnancy, and 895 (32%) during pregnancy. Figure 2 shows the flowchart of inclusions and exclusions. For women vaccinated during pregnancy, the mean gestational age at vaccination was 3.6 weeks (SD 3.7 weeks). Mean age was 22.4 years in the distal and peripregnancy groups compared with 21.4 years among women vaccinated during pregnancy. Compared with distal vaccination, women receiving 4vHPV during pregnancy had later entry into prenatal care (14 weeks of gestation, SD 6.5 vs 12 weeks of gestation, SD 5.1 weeks of gestation, P<.001) and were more likely to receive other vaccines during pregnancy and before 20 weeks of gestation (47% vs 24%). The most common other vaccines administered during pregnancy were influenza, meningococcal, and the tetanus, diphtheria, and acellular pertussis vaccine. See Table 2 for other baseline comparisons by 4vHPV exposure group.
Of 774 spontaneous abortions initially identified using the pregnancy algorithm with 4vHPV exposure within 29 weeks of the LMP or during pregnancy, 528 were confirmed by chart review as a possible or probable spontaneous abortion (68% confirmation rate). In addition, 25 of 69 pregnancies with an “unknown abortion outcome” from the automated algorithm were confirmed by chart review as spontaneous abortions and 4 of 26 pregnancies identified with a stillbirth outcome based on the automated algorithm were confirmed by chart review as an spontaneous abortion occurring between 6 and 19 weeks of gestation. Of the 557 confirmed spontaneous abortions, 283 had a 4vHPV exposure in the final analysis windows and were retained in the cohort.
Among 919 women with distal 4vHPV exposure, 10.4% experienced a spontaneous abortion; unadjusted rates were 8.6% for vaccination during pregnancy and 11.2% for peripregnancy exposures (Table 3). In age-adjusted analyses, receipt of 4vHPV during or peripregnancy was not associated with increased risks for spontaneous abortion (age-adjusted hazard ratio [HR] 1.04, 95% CI 0.77–1.41 and 1.07, 95% CI 0.82–1.41, respectively). In fully adjusted models adjusting for age, poverty level, health utilization before pregnancy, obesity, history of smoking, and race, receipt of 4vHPV during or peripregnancy was also not associated with increased risks for spontaneous abortion (fully adjusted HR 1.10, 95% CI 0.81–1.51 and 1.07, 95% CI 0.81–1.41, respectively). Results were similar for a combined exposure window (during or peripregnancy) as compared with distal vaccination (age-adjusted HR 1.13, 95% CI 0.88–1.44, fully adjusted HR 1.16, 95% CI 0.90–1.48).
Concomitant vaccination or receipt of one or more additional vaccines on the same days as 4vHPV peripregnancy or during pregnancy did not increase the risk of spontaneous abortion (age-adjusted HR 1.05, 95% CI 0.76–1.46, fully adjusted HR 1.09, 95% CI 0.78–1.54) as compared with distal vaccination. Receipt of a second or third 4vHPV dose peripregnancy or during pregnancy, as compared with a first dose, was also not associated with increased spontaneous abortion risk (age-adjusted HR 1.11, 95% CI 0.83–1.48, fully adjusted HR 1.09, 95% CI 0.81–1.46; Table 4). Exploratory analyses did not show an increased risk of spontaneous abortion cluster based on time from vaccination or gestational age at vaccination (Figs. 3, 4). As shown in Figure 4, among women with 4vHPV exposure during pregnancy and spontaneous abortion, the time from vaccination to spontaneous abortion varied by the gestational week at vaccination. Analyses restricted to spontaneous abortions occurring within 28 days of 4vHPV exposure, as compared with distal vaccination, did not demonstrate an increased risk (fully adjusted HR 0.93, 95% CI 0.48–1.83).
In this observational cohort study, we found no association between 4vHPV exposure during or peripregnancy and risk of spontaneous abortion. This study supports current clinical practice, in which pregnancy testing is not routinely performed before vaccination.
Findings from our study are consistent with data from prelicensure clinical trials and postlicensure safety assessments. In the 4vHPV prelicensure clinical trials, approximately 15% of patients became pregnant.20 However, only 6% of these pregnancies were within 30 days of vaccination. In this subgroup (n=266), rates for spontaneous abortion were 18.2% in women vaccinated with 4vHPV and 21% in the placebo group. Using data from the 4vHPV manufacturer-sponsored pregnancy registry, including 216 exposures within 30 days of the LMP and 1,060 exposures during the first trimester, Goss et al15 reported spontaneous abortion rates to be 4.2% and 6.6%, respectively. More recently, Scheller7 et al used registry data from Denmark including 463 pregnancies with 4vHPV exposures from 7 to 22 weeks of gestation and reported a HR of 0.71 (95% CI 0.45–1.14). Our approach differed in that we included vaccinations occurring before 7 weeks of gestation, along with peripregnancy exposures, and our outcomes were verifiable spontaneous abortions that occurred at 6 weeks of gestation or later.
Our study, which included 895 women with 4vHPV exposures during pregnancy and 986 peripregnancy exposures, adds to the literature regarding risks for spontaneous abortion after inadvertent vaccination. To date, the optimal risk window for evaluating of spontaneous abortion risk after vaccination is not known. The 4vHPV administrations that we captured were inadvertent and generally occurred early in pregnancy. Given that background spontaneous abortion rates peak early, at 6–7 weeks of gestation, and this period overlaps with inadvertent 4vHPV exposures, assessments of risk during pregnancy but within a specified risk window (eg, 28 days) would be prone to bias. Our exploratory evaluation of the time between 4vHPV and spontaneous abortion did not demonstrate a window with a statistically significant increased risk. Rather, any apparent clustering of spontaneous abortion by timing of vaccination was likely the result of changes in background spontaneous abortion risk by gestational week.21
Some prior studies have suggested risk of spontaneous abortion after maternal vaccination may increase with repeated antigen exposure. In postmarketing surveillance of bivalent HPV in the United Kingdom, receipt of two doses within a 4- to 5-week period was associated with increased risk (HR 2.55, 95% CI 1.09–5.93).10 Similarly, in a case–control study by Donahue et al,22 risks for spontaneous abortion were highest in women who received influenza vaccine containing A/H1N1pdm2009 (pH1N1) antigen in both the current and prior influenza seasons. During the time of our study, 4vHPV was administered as a three-dose vaccine series. We required 8 months of continuous enrollment before the LMP. We may not have captured earlier vaccinations and thus could not be certain whether the dose identified in the current study was a first, second, or third dose. Nevertheless, as compared with a first dose, receipt of a second or third 4vHPV dose peripregnancy or during pregnancy was not associated with increased risk of spontaneous abortion.
Additional limitations to this research should be noted. First, women vaccinated during pregnancy were most likely unaware that they were pregnant. Thus, they may have differed from women who were aware that they were pregnant in other ways that could affect spontaneous abortion risk. We were unable to assess use of cigarettes or alcohol during pregnancy in this population. Second, as with any observational study, there was potential for misclassification of exposure status. Uncertainty in LMP dating could affect whether an exposure was assigned as peripregnancy, during pregnancy, or distal compared with not eligible for inclusion. For pregnancies ending in a live birth, gestational age at delivery came primarily from birth records. For stillbirths and spontaneous abortions, gestational age was based on chart review and expert adjudication. However, even with extensive review, in cases of spontaneous abortion, gestational age cannot always be confirmed and in these cases, estimates based on LMP were applied. Furthermore, the risk of spontaneous abortion after peripregnancy or during early pregnancy 4vHPV exposure may not be fully captured in our analysis, because spontaneous abortions before 6 weeks of gestation are often not clinically confirmed, and thus, for this, study they were excluded.16 A final limitation that as a result of our sample size, we were not able to rule out small increases in risk.
In conclusion, inadvertent 4vHPV exposure during or peripregnancy was not associated with statistically significant increased risk of spontaneous abortion. In February 2015, the nonvalent HPV vaccination, targeting five additional oncogenic HPV strains, was recommended by the Advisory Committee on Immunization Practices to be incorporated as part of existing HPV vaccine recommendations.5 Therefore, in future work, we will evaluate spontaneous abortion risk after inadvertent exposures with this new vaccine.
1. Markowitz LE, Dunne EF, Saraiya M, Lawson HW, Chesson H, Unger ER, et al. Quadrivalent human papillomavirus vaccine: recommendations of the Advisory Committee on Immunization Practices (ACIP). MMWR Recomm Rep 2007;56:1–24.
2. Drolet M, Bénard, Boily MC, Ali H, Baandrup L, Bauer H, et al. Population-level impact and herd effects following human papillomavirus vaccination programmes: a systematic review and meta-analysis. Lancet Infect Dis 2015;15:565–80.
3. Berenson AB, Hirth JM, Chang M. Change in human papillomavirus prevalence among U.S. women aged 18–59 years, 2009–2014. Obstet Gynecol 2017;130:693–701.
4. Oliver SE, Unger ER, Lewis R, McDaniel D, Gargano JW, Steinau M, et al. Prevalence of human papillomavirus among females after vaccine introduction—National Health and Nutrition Examination Survey, United States, 2003–2014. J Infect Dis 2017;216:594–603.
5. Petrosky E, Bocchini JA Jr, Hariri S, Chesson H, Curtis CR, Saraiya M, et al. Use of 9-valent human papillomavirus (HPV) vaccine: updated HPV vaccination recommendations of the advisory committee on immunization practices. MMWR Morb Mortal Wkly Rep 2015;64:300–4.
6. Ventura SJ, Curtin SC, Abma JC, Henshaw SK. Estimated pregnancy rates and rates of pregnancy outcomes for the United States, 1990–2008. Natl Vital Stat Rep 2012;60:1–21.
7. Scheller NM, Pasternak B, Molgaard-Nielsen D, Svanstrom H, Hviid A. Quadrivalent HPV vaccination and the risk of adverse pregnancy outcomes. N Engl J Med 2017;376:1223–33.
8. Forinash AB, Yancey AM, Pitlick JM, Myles TD. Safety of the HPV bivalent and quadrivalent vaccines during pregnancy. Ann Pharmacother 2011;45:258–62.
9. López-Fauqued M, Zima J, Angelo MG, Stegmann JU. Results on exposure during pregnancy from a pregnancy registry for AS04-HPV-16/18 vaccine. Vaccine 2017;35:5325–30.
10. Baril L, Rosillon D, Willame C, Angelo MG, Zima J, van den Bosch JH, et al. Risk of spontaneous abortion and other pregnancy outcomes in 15–25 year old women exposed to human papillomavirus-16/18 AS04-adjuvanted vaccine in the United Kingdom. Vaccine 2015;33:6884–91.
11. Panagiotou OA, Befano BL, Gonzalez P, Rodríguez AC, Herrero R, Schiller JT, et al. Effect of bivalent human papillomavirus vaccination on pregnancy outcomes: long term observational follow-up in the Costa Rica HPV Vaccine Trial. BMJ 2015 Sep 7.
12. Lipkind HS, Vazquez-Benitez G, Nordin JD, Romitti PA, Naleway AL, Klein NP, et al. Maternal and infant outcomes after human papillomavirus vaccination in the periconceptional period or during pregnancy. Obstet Gynecol 2017;130:599–608.
13. Baggs J, Gee J, Lewis E, Fowler G, Benson P, Lieu T, et al. The Vaccine Safety Datalink: a model for monitoring immunization safety. Pediatrics 2011;127(suppl 1):S45–53.
14. Naleway AL, Gold R, Kurosky S, Riedlinger K, Henninger ML, Nordin JD, et al. Identifying pregnancy episodes, outcomes, and mother-infant pairs in the Vaccine Safety Datalink. Vaccine 2013;31:2898–903.
15. Goss MA, Lievano F, Buchanan KM, Seminack MM, Cunningham ML, Dana A. Final report on exposure during pregnancy from a pregnancy registry for quadrivalent human papillomavirus vaccine. Vaccine 2015;33:3422–8.
16. Savitz DA, Hertz-Picciotto I, Poole C, Olshan AF. Epidemiologic measures of the course and outcome of pregnancy. Epidemiol Rev 2002;24:91–101.
17. Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Research electronic data capture (REDCap)—a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform 2009;42:377–81.
18. Early pregnancy loss. Practice Bulletin No. 150. American College of Obstetricians and Gynecologists. Obstet Gynecol 2015;125:1258–67.
19. Suissa S. Immortal time bias in pharmaco-epidemiology. Am J Epidemiol 2008;167:492–9.
20. Garland SM, Ault KA, Gall SA, Paavonen J, Sings HL, Ciprero KL, et al. Pregnancy and infant outcomes in the clinical trials of a human papillomavirus type 6/11/16/18 vaccine: a combined analysis of five randomized controlled trials. Obstet Gynecol 2009;114:1179–88.
21. Ammon Avalos L, Galindo C, Li DK. A systematic review to calculate background miscarriage rates using life table analysis. Birth Defects Res A Clin Mol Teratol 2012;94:417–23.
22. Donahue JG, Kieke BA, King JP, DeStefano F, Mascola MA, Irving SA, et al. Association of spontaneous abortion with receipt of inactivated influenza vaccine containing H1N1pdm09 in 2010–11 and 2011–12. Vaccine 2017;35:5314–22.