The Early Benefits of Human Papillomavirus Vaccination on Cervical Dysplasia and Anogenital Warts Leah M. Smith, MSca, Erin C. Strumpf, PhDa,b, Jay S. Kaufman, PhDa, Aisha Lofters, MD, PhDc, Michael Schwandt, MD, MPHd, Linda E. Lévesque, BScPhm, PhDe,f
Despite widespread promotion of quadrivalent human papillomavirus (qHPV) vaccination for young girls, there is limited information on the vaccine’s real-world effectiveness and none on the effectiveness of qHPV vaccination programs. We assessed the impact of the qHPV vaccine and Ontario’s grade 8 qHPV vaccination program on cervical dysplasia and anogenital warts (AGW). BACKGROUND:
abstract
By using administrative health databases of Ontario, Canada, we identified a population-based retrospective cohort of girls in grade 8 before (2005/2006–2006/2007) and after (2007/2008–2008/2009) program implementation. Vaccine exposure was ascertained in grades 8 to 9 and outcomes in grades 10 to 12. A quasi-experimental approach known as regression discontinuity was used to estimate absolute risk differences (RDs), relative risks (RRs), and 95% confidence intervals (CIs) attributable to vaccination and program eligibility (intention-to-treat analysis).
METHODS:
RESULTS: The cohort comprised 131 781 ineligible and 128 712 eligible girls (n = 260 493). We identified 2436 cases of dysplasia and 400 cases of AGW. Vaccination significantly reduced the incidence of dysplasia by 5.70 per 1000 girls (95% CI 29.91 to 21.50), corresponding to a relative reduction of 44% (RR 0.56; 95% CI 0.36 to 0.87). Program eligibility also had a significant protective effect on dysplasia: RD 22.32/1000 (95% CI 24.02 to 20.61); RR 0.79 (95% CI 0.66 to 0.94). Results suggested decreases in AGW attributable to vaccination (RD 20.83/1000, 95% CI 22.54 to 0.88; RR 0.57, 95% CI 0.20 to 1.58) and program eligibility (RD 20.34/1000, 95% CI 21.03 to 0.36; RR 0.81, 95% CI 0.52 to 1.25). CONCLUSIONS: This study provides strong evidence of the early benefits of qHPV vaccination among girls aged 14 to 17 years, offering additional justification for not delaying vaccination.
a Departments of Epidemiology, Biostatistics, and Occupational Health, and bEconomics, McGill University, Montreal, Quebec, Canada; cDepartment of Family and Community Medicine, St Michael’s Hospital, University of Toronto, Toronto, Ontario, Canada; dDepartment of Community Health and Epidemiology, University of Saskatchewan, Saskatoon, Saskatchewan, Canada; eDepartment of Public Health Sciences, Queen’s University, Kingston, Ontario, Canada; and fInstitute for Clinical Evaluative Sciences-Queen’s Health Services Research Facility, Kingston, Ontario, Canada
Ms Smith was involved in acquiring study data, played a major role in the conception, design, and interpretation of the study, carried out the statistical analyses, and drafted the manuscript; Drs Strumpf and Kaufman played major roles in the conception, design, analysis, and interpretation of the study, and critically reviewed the manuscript; Drs Lofters and Schwandt were involved in design of the study and critically reviewed the manuscript; Dr Lévesque played the principal role in acquiring the study data, played a major role in the conception, design, analysis, and interpretation of the study, and critically reviewed the manuscript; and all authors were involved in obtaining funding for this study, approved the final manuscript as submitted, and agree to be accountable for all aspects of this work. www.pediatrics.org/cgi/doi/10.1542/peds.2014-2961 DOI: 10.1542/peds.2014-2961 Accepted for publication Feb 24, 2015
WHAT’S KNOWN ON THIS TOPIC: Clinical trials of the quadrivalent human papillomavirus vaccine show it to be highly efficacious in preventing vaccine-type–specific cervical dysplasia and anogenital warts, but few studies have assessed its effects in the real world and none have done so at the program/population level. WHAT THIS STUDY ADDS: This study provides strong evidence of the early benefits of quadrivalent human papillomavirus vaccination on reductions in cervical dysplasia and possible reductions in anogenital warts among girls aged 14 to 17 years, offering additional justification for not delaying vaccination until girls are older.
Downloaded from pediatrics.aappublications.org at UCSF LIBRARY on April 27, 2015
PEDIATRICS Volume 135, number 5, May 2015
ARTICLE
The human papillomavirus (HPV) is a sexually transmitted infection that affects most individuals at some point during their lives.1 Although the vast majority of these infections selfresolve without clinical sequelae, others can lead to important health consequences, including anogenital warts (AGW) and cancers of the cervix, anus, and penis.2 In 2006, Canada was among the first countries to license Gardasil (Merck, Whitehouse Station, NJ), a quadrivalent human papillomavirus (qHPV) vaccine that protects against 4 types of HPV that cause 70% of cases of cervical cancer and at least 90% of AGWs.3–5 As randomized controlled trials of the vaccine showed it to be highly efficacious in preventing vaccine-type–specific precancerous cervical lesions and AGW,6,7 the vaccine received a great deal of attention from the scientific and medical communities, as well as from the general public.8 Currently, the qHPV vaccine is available in .100 countries, many of which also have national qHPV immunization programs.9 These programs are generally aimed at immunizing young girls before the onset of sexual activity when the possibility of previous HPV infection is low (eg, ages 9–13 years).9,10 Despite widespread promotion of the vaccine and the popularity of largescale qHPV vaccination programs, levels of vaccine uptake have been lower than expected in a number of jurisdictions.9,11,12 In part, low qHPV vaccine acceptance has been attributed to parental concerns about the effects of the vaccine, especially in the young age group targeted for vaccination.13,14 In addition, some parents are deciding to delay vaccination for their daughters until a later age, as they perceive their daughter’s likelihood of sexual activity as low.15 Given the tendency of parents to underestimate their child’s sexual experience,16 there are important concerns that delaying
e1132
vaccination might result in missed opportunities for prevention. In addition, low vaccine coverage may jeopardize the expected public health benefits and cost-effectiveness of qHPV vaccine programs. As there are currently few studies on the real-world effects of the qHPV vaccine17–21 and none on the population-level impact of qHPV vaccine programs on the burden of disease, we undertook a populationbased, retrospective cohort study to assess the impact of the qHPV vaccine and Ontario’s qHPV vaccination program on the incidence of cervical dysplasia and AGW among adolescent girls.
METHODS This study was approved by the Research Ethics Boards of Queen’s University, McGill University, and Sunnybrook Health Sciences Centre.
Study Setting Our study is based in Ontario, Canada’s most populous province, which began offering all 3 doses of the vaccine, free-of-charge, to all grade 8 girls in September of 2007.22 Doses are administered primarily through school-based immunization clinics, but girls also may receive the vaccine from a physician or at their health unit at no cost. Before September 2012, eligible girls had until the end of their grade 8 year to initiate the vaccine series and until the end of grade 9 to complete it. During our study period, girls who were not eligible for the school-based program (eg, in grade 8 before 2007) could obtain the vaccine series for ∼$400.
Data Sources Data for this study were obtained from Ontario’s population-based administrative health and immunization databases, which are housed at the Institute for Clinical Evaluative Sciences.23 We used (1) the Registered Persons Database
(RPDB) for sociodemographics, (2) the Ontario Health Insurance Plan (OHIP) database for fee-for-service claims by physicians, (3) the Discharge Abstract Database for hospitalizations, (4) the Same-Day Surgery database for same-day surgeries, (5) the National Ambulatory Care Reporting System (NACRS) for emergency department visits, and (6) the Immunization Records Information System (IRIS) for vaccinations. In these databases, residents are represented by a unique, encrypted identifier, enabling anonymized, individual-level record linkage across databases and time. These databases are generated by Ontario’s universal health insurance programs and have been used extensively in health research, including in the postmarketing evaluation of drug and vaccine effects.24–29
Study Population and Cohort Formation We identified a population-based cohort of all girls in grade 8 in Ontario during the 2005/ 2006–2008/2009 school years. This includes 2 years of girls who were eligible for publicly funded qHPV vaccination (ie, in grade 8 in 2007/ 2008 and 2008/2009) and, for the purposes of comparison, 2 years of girls who were ineligible (ie, in grade 8 in 2005/2006 and 2006/2007). As school grade is not available in these databases, grade was determined based on birth year because .96% of girls enter grade 8 in their 13th year of life.30 Specifically, we used the RPDB to identify all girls born in 1992, 1993, 1994, and 1995 and, through record linkage with IRIS, restricted these birth cohorts to girls whose immunization records were available at the time of this study and who were residing in Ontario on September 1 of 2005, 2006, 2007, and 2008, respectively. September 1 of grade 8 defined cohort entry, and cohort members were followed until the earliest of the following: date of
Downloaded from pediatrics.aappublications.org at UCSF LIBRARY on April 27, 2015
SMITH et al
death, occurrence of a study outcome, or March 31 of grade 12 (end of follow-up).
Measurement and Analysis Approach Observational studies of vaccine effects are notoriously vulnerable to confounding bias because individuals who opt for vaccination tend to have different health histories and behaviors than those who do not, and these characteristics are difficult to identify and quantify.31–34 To address this methodological challenge, we used the regression discontinuity design (RDD), a quasi-experimental approach for evaluating the causal effects of interventions in a way that accounts for this type of observed and unobserved confounding, thus facilitating reliable causal inference.35–37 The RDD has been used extensively in fields like health economics,37–39 and is becoming increasingly popular in epidemiology, given the advantages it offers over standard regression adjustment.40,41 In this study, the RDD enabled us to exploit the fact that girls were eligible for free qHPV vaccination (ie, assigned to treatment) based on whether they were in grade 8 before or after the September 2007 program implementation date (ie, born December 31, 1993 or earlier versus January 1, 1994 or later), which induced discontinuity in the probability of qHPV vaccination at the eligibility cutoff (Supplemental Appendix 1A). As such, birth date acted as the factor (“forcing variable”) that influenced whether a girl was exposed to vaccination. Essentially, the goal of the RDD was to estimate any corresponding discontinuity in the probability of the outcomes between eligibility groups at the same cutoff (Supplemental Appendix 1B). The value of the RDD over more traditional designs relies on the assumption that because the program implementation date and eligibility criteria were based on administrative decisions, the exact location of the
cutoff is random, making girls directly on either side of the cutoff similar with respect to measured and unmeasured confounders. In the RDD, this comparability is assumed to decrease with increasing distance from the cutoff, so this notion is taken into account in the analysis.
Baseline Characteristics We identified a number of characteristics relating to sociodemographics, vaccination history, health service use, and medical history. These were compared between eligibility groups and across levels of the forcing variable.
Forcing Variable The forcing variable (ie, birth date) was categorized into 3-month intervals, referred to as “birth year quarters.” Therefore, girls born October 1, 1993, to December 31, 1993, defined the “ineligible” side of the cutoff and girls born January 1, 1994, to March 31, 1994, defined the “eligible” side. Girls born earlier/later than these dates were represented with increasing distance from this cutoff on the ineligible/eligible sides (Supplemental Appendix 2).
Exposure To evaluate the impact of the qHPV vaccination program, exposure status was based solely on program eligibility, thus providing an intention-to-treat definition of vaccination. To assess the impact of the vaccine on our outcomes, qHPV vaccine receipt also was taken into account. The qHPV vaccine exposure was defined as receipt of all 3 doses between September 1 of grade 8 and August 31 of grade 9.
Outcomes Incident cases of detected cervical dysplasia and AGW were identified between September 1 of grade 10 and March 31 of grade 12. Cases were “incident” if they occurred after at least 730 days (cervical dysplasia) or 365 days (AGW) event-free. Because algorithms for ascertaining these outcomes have not been validated, we created 3 definitions of each (broad, possible, and probable) in consultation with substantive experts in the field to reflect different degrees of potential misclassification (Supplemental Appendices 3 and 4).
Statistical Analyses To assess program impact, local linear regression was used to estimate the association between program eligibility and each outcome of interest.35,36 Specifically, regression functions were estimated on each side of the eligibility cutoff, and the difference/discontinuity between intercepts represented the absolute impact of program eligibility on the outcome. To assess vaccine impact, a second stage was added to also estimate the association between program eligibility and qHPV vaccine exposure. The absolute effect of the vaccine was determined based on the ratio of the 2 discontinuities: eligibility-outcome and eligibilityvaccine. This 2-stage analysis is analogous to an instrumental variable approach.35 Similarly, 1- and 2-stage log-binomial regression models were used to estimate the relative impact of program eligibility and vaccination on each outcome. Finally, the “number needed to treat” (NNT) of each outcome was estimated based on the reciprocal of the absolute risk difference (RD) and 95% confidence limits.42 All analyses were based on the entire study cohort; however, girls born in 1993 and 1994 were weighted twice as heavily as girls born in 1992 and 1995 because girls closest to the cutoff are most comparable (ie, lower potential for confounding). Similarly, we controlled for birth quarter because we found that girls born early (or late) in the year were most comparable between birth years (Supplemental Appendix 5). A number of sensitivity analyses were performed to evaluate the robustness of our results, including
Downloaded from pediatrics.aappublications.org at UCSF LIBRARY on April 27, 2015
PEDIATRICS Volume 135, number 5, May 2015
e1133
one that modified the “exposed” definition to include girls who received only 2 doses of the vaccine, as 2 doses may be sufficient to provide immunity.43 The other assumptions of the RDD were tested quantitatively or qualitatively36 and were found to hold. SAS statistical software version 9.3 (SAS Institute, Inc, Cary, NC) was used for data management and Stata version 12 (Stata Corp, College Station, TX) for analyses.
RESULTS We identified a cohort of 260 493 girls, of whom almost half were eligible for free, school-based qHPV vaccination (Fig 1). Girls were 12.7 to 13.7 years old at cohort entry and were followed for an average of 4.6 years. Eligibility groups were similar with respect to a number of baseline characteristics, including frequency of outpatient physician visits and previous receipt of the
FIGURE 1
measles-mumps-rubella and diphtheria, tetanus, and pertussis vaccines. There were small differences in such characteristics as neighborhood income quintile, hepatitis B vaccination, and indicators of previous sexual behavior (Table 1). Of the 75 848 cohort members who received at least 1 dose of the qHPV vaccine in grades 8 to 9, 88% completed the 3-dose series. As expected, program eligibility had a major impact on this exposure. Specifically, 50.6% of eligible girls were classified as qHPV vaccine exposed compared with ,1% of ineligible girls. Figure 2 illustrates the discontinuity in qHPV vaccination between groups. We identified 2436 (0.94%) cohort members with incident cervical dysplasia (Table 2). Although there was no apparent discontinuity at the eligibility cutoff when assessed across birth year quarters (Fig 3A), collapsing into birth years revealed a clear drop in risk between groups
Cohort flow diagram. *At the time of this study, 2 of Ontario’s 36 IRIS databases (representing approximately 22% of Ontario’s population) had not yet been transferred to the Institute for Clinical Evaluative Sciences and were therefore unavailable for use in this study. IRIS records also were unavailable if the girl emigrated from Ontario before starting kindergarten or immigrated to Ontario after completing high school. †A girl’s IRIS record was defined as “up to date” if it had been modified 30 days before cohort entry or later. Otherwise, it was assumed the girl had moved out of our study area before cohort entry.
e1134
(Fig 3B), demonstrating the importance of controlling for birth quarter in the analyses. Indeed, the primary analysis (which adjusted for birth timing) revealed a statistically significant reduction in detected dysplasia attributable to program eligibility: RD 22.32 cases per 1000 girls, 95% confidence interval (CI) 24.02 to 20.61. An even stronger effect was observed for the vaccine: RD 25.70 per 1000 girls, 95% CI 29.91 to 21.50. These results indicate that 1 case of cervical dysplasia was prevented for every 431 (95% CI 248 to 1639) girls eligible for publicly funded vaccination and for every 175 (95% CI 101 to 667) girls who received the vaccine. The absolute RDs corresponded to a 21% (95% CI 6% to 34%) and 44% (95% CI 13% to 64%) relative risk (RR) reduction for eligibility and vaccination, respectively. RR reductions were similar for possible dysplasia and probable dysplasia, whereas absolute RDs decreased and NNTs increased with increasing specificity of the dysplasia definition (Table 3). We ascertained 400 cases of AGW (Table 2). Figure 4 A and B depict this risk by birth year quarter and by birth year, respectively. The results suggested reductions in detected AGW attributable to program eligibility and vaccination, both on the absolute scale (RD 20.34 per 1000, 95% CI 21.03 to 0.36; RD 20.83, 95% CI 22.54 to 0.88) and the relative scale (RR 0.81, 95% CI 0.52 to 1.25; RR 0.57, 95% CI 0.20 to 1.58); however, these results were not statistically significant. Findings indicated even greater RR reductions for more specific outcome definitions of AGW, whereas absolute risk reductions were similar across definitions (Table 3). Results for both cervical dysplasia and AGW were robust in sensitivity analyses, including analyses that adjusted for additional baseline characteristics (Supplemental Appendix 6), defined HPV vaccine exposure based on
Downloaded from pediatrics.aappublications.org at UCSF LIBRARY on April 27, 2015
SMITH et al
DISCUSSION
TABLE 1 Baseline Characteristics of Study Cohort a
Characteristic
Percentage
Sociodemographicsb Mean age at cohort entry (SD) Birth timing Q1: January–March Q2: April–June Q3: July–September Q3: October–December Residency Urban Rural Missingc Income quintile 1st (lowest) 2nd 3rd 4th 5th (highest) Missingc Vaccination historyd MMRe DTPe Hepatitis Be All 3 vaccines Frequency of health care usef,g Hospitalizations None (0) $1 Days (mean, SD) Same-day surgeries None (0) $1 Emergency department visits None (0) 1 $2 Outpatient visits 0–1 2–5 6–12 $13 Medical history Cancerf Mental health diagnosisf Indicators of sexual behaviorf,h Down syndromed Congenital malformationsd Intellectual disabilitiesd
Ineligible, n = 131 781)
Eligible, n = 128 712)
13.17 (0.28)
13.17 (0.28)
24.3 26.1 25.7 23.9
24.2 26.1 25.8 23.9
85.3 14.0 0.7
85.8 13.5 0.6
16.6 18.4 20.6 22.0 21.4 1.0
15.0 17.8 21.1 23.1 22.1 0.9
97.9 98.0 84.1 83.0
98.2 98.3 82.0 81.1
98.0 2.0 7.4, 15.6
98.2 1.8 8.0, 18.2
97.7 2.4
97.8 2.2
70.7 18.1 11.2
71.1 17.8 11.1
25.0 22.6 25.1 27.4
25.8 22.8 24.5 26.9
0.71 9.5 0.68 0.48 12.4 0.72
0.74 9.7 0.73 0.53 11.8 0.73
DTP, diphtheria, tetanus, and pertussis; MMR, measles-mumps-rubella; Q, quarter. a Figures are expressed as percentages except where indicated otherwise. b Obtained at cohort entry. c Missing because postal code in RPDB inaccurate or unavailable. d Obtained between birth and cohort entry. e At least 1 dose. f Obtained in the 2 years before cohort entry. g Categories of health care use were determined based on the frequency distribution of the data. h Composite endpoint of sexually transmitted infections, cervical dysplasia, Papanicolaou tests, and pregnancy, all of which are clinical indicators of previous sexual behavior.
receipt of at least 2 doses, and restricted the exposure ascertainment window to grade 8
and the outcome ascertainment window to grades 9 to 12 (Supplemental Appendix 7).
This is the first observational study of the benefits of the qHPV vaccine to use a methodological approach that permits causal inference, providing new, strong evidence of the positive health impact of qHPV vaccination on health outcomes. In particular, our analyses revealed a protective effect of the qHPV vaccine program on cervical dysplasia among girls aged 14 to 17 years and, although imprecise, we also observed apparent reductions in the incidence of AGW. Estimates of vaccine impact for both outcomes were even stronger than those of program impact. All results speak to the value of offering qHPV vaccination to young girls through a population-based program. Our use of the RDD enabled us to evaluate the intention-to-treat effects of qHPV vaccination for the first time outside of clinical trials. These analyses have particular public health value, as they provide measures of the population-level impact of the qHPV vaccine program on cervical dysplasia and AGW. Importantly, our findings revealed that despite relatively low qHPV vaccine coverage in Ontario (∼50% among eligible girls), meaningful benefits were nevertheless observed in this young age group at the population level. These results are promising for other jurisdictions across Canada12 and beyond9 that have also struggled with low HPV vaccine coverage. Nevertheless, estimates of program impact were considerably weaker than estimates of vaccine impact. For example, we observed almost 2.5 times greater absolute reductions in cervical dysplasia attributable to the vaccine compared with the program. Given that the benefits of the program impact are largely dependent on the proportion of girls vaccinated, these differences were not surprising, and they help demonstrate the important implications of vaccine coverage on reducing the burden of HPV-related
Downloaded from pediatrics.aappublications.org at UCSF LIBRARY on April 27, 2015
PEDIATRICS Volume 135, number 5, May 2015
e1135
FIGURE 2 Probability of qHPV vaccine exposure,* by birth quarter. Q1: born January–March; Q2: born April– June; Q3: born July–September; Q4: born October–December. *Receipt of 3 doses of the qHPV vaccine between September 1 of grade 8 and August 31 of grade 9.
diseases in the population. Indeed, a thorough appreciation of the interplay between the program- and vaccine-level effects in any region will be crucial to improving qHPV vaccine policy and practice in ways that maximize the health and economic benefits of this vaccine. To date, observational studies of cervical dysplasia and AGW have reported estimates of vaccine effectiveness based on analyses comparing vaccinated and unvaccinated girls. Our results of vaccine impact are generally consistent with previous research on cervical dysplasia.17–20 Although all observational studies have reported more conservative estimates than randomized controlled trials,7 this is not surprising given that trials
estimated treatment effects under ideal conditions. This does, however, highlight the importance of observational studies in determining the real-world effectiveness of this vaccine. The only nonecologic observational study of AGW reported an effect greater than reported here; however, the estimates generally fell within our CIs.21 The difference in results between studies may be partly attributable to residual confounding between vaccinated and unvaccinated groups in the earlier study. Regardless, given the low risk of diagnosed AGW in our study’s young age group and the resulting imprecision of our estimates, additional studies are needed to further elucidate these findings.
TABLE 2 Cumulative Incidence of Cervical Dysplasia and AGWsa Outcome
Frequency (percent) Ineligible
Eligible
Total, n = 260 493
1992, n = 66 653 1993, n = 65 128 1994, n = 64 818 1995, n = 63 894 Dysplasia Broad Possible Probable AGWs Broad Possible Probable a
698 (1.05) 444 (0.67) 232 (0.35)
720 (1.11) 478 (0.73) 229 (0.35)
555 (0.86) 358 (0.55) 161 (0.25)
463 (0.72) 277 (0.43) 121 (0.19)
2436 (0.94) 1557 (0.60) 743 (0.29)
119 (0.18) 68 (0.10) 62 (0.09)
120 (0.18) 59 (0.09) 54 (0.08)
91 (0.14) 32 (0.05) 30 (0.05)
70 (0.11) 31 (0.05) 27 (0.04)
400 (0.15) 190 (0.07) 173 (0.07)
Ascertained between September 1 of grade 10 and March 31 of grade 12.
e1136
A major strength of this study is the use of population-based administrative databases, which enabled us to conduct a large cohort study with limited potential for selection, recall, or response biases. An additional benefit is that our HPV vaccination data were highly sensitive (99.8%; 95% CI 99.3 to 99.9) and specific (97.7%; 95% CI 96.3 to 98.7) to girls’ vaccination status,44 meaning the impact of any exposure misclassification was likely negligible among eligible girls. Conversely, it is possible that we were missing information on HPV vaccines obtained by ineligible girls through private means. Although we expect this percentage to be low, it would nevertheless have caused us to underestimate the impact of the vaccine on our outcomes. The potential for herd immunity among ineligible girls would also have underestimated the benefits of the vaccine in this study. However, given the young age group and limited follow-up time of the study population, as well as the difference in calendar years of follow-up between eligibility groups, this effect was likely negligible. Another limitation is the potential for outcome misclassification, as measures of cervical dysplasia and AGW have not been validated. To assess this potential for error, we created 3 levels of each outcome to reflect increasing levels of outcome sensitivity. Indeed, the fact that the RR of broad AGW was biased toward the null compared with the more sensitive definitions suggests the broad definition was misclassified. This is not surprising, as this definition included a diagnostic code that also could be used to capture warts on other body sites. On the other hand, the consistency in relative estimates across definitions of dysplasia and between definitions of possible and probable AGW suggests that most of our outcome definitions did not result in misclassification. Another limitation of our data are
Downloaded from pediatrics.aappublications.org at UCSF LIBRARY on April 27, 2015
SMITH et al
FIGURE 3 A, Risk of broad dysplasia,* by birth quarter. Q1: born January–March; Q2: born April–June; Q3: born July–September; Q4: born October–December. B, Risk of broad dysplasia,* by birth year. *Ascertained September 1 of Grade 10 to March 31 of Grade 12.
they detected only outcomes of girls who followed up with their physician after an abnormal screening result or girls who received care for AGWs through physicians reimbursed through OHIP. As such, some incident cases may have been missed. Because this underascertainment would have
affected both groups equally, it would not have affected our relative estimates; however, our absolute estimates of effect are likely conservative. Another limitation is that we could not distinguish between dysplasia and cervical ectropion/erosion in the OHIP
TABLE 3 Impact of qHPV Vaccination on the Risk of Cervical Dysplasia and AGWs (n = 260 493) RDs, per 1000 Girls (95% CI)a Risk Ratio (95% CI)b Dysplasia Program impact Broad Possible Probable Vaccine impact Broad Possible Probable AGWs Program impact Broad Possible Probable Vaccine impact Broad Possible Probable
NNT (95% CI)c
22.32 (–4.02 to –0.61) 21.70 (–3.08 to –0.32) 20.79 (–1.74–0.16)
0.79 (0.66 to 0.94) 0.76 (0.61 to 0.95) 0.76 (0.56 to 1.05)
431 (248 to 1639) 588 (325 to 3125) 1266 (575 to ‘ to NNH 6250)
25.70 (–9.91 to –1.50) 24.19 (–7.59 to –0.80) 21.94 (–4.28 to 0.39)
0.56 (0.37 to 0.85) 0.50 (0.30 to 0.82) 0.51 (0.23 to 1.12)
175 (101 to 667) 239 (132 to 1250) 515 (234 to ‘ to NNH 2564)
20.34 (–1.03 to 0.36) 20.34 (–0.81 to 0.13) 20.29 (–0.74 to 0.16)
0.81 (0.52 to 1.25) 0.60 (0.31 to 1.15) 0.63 (0.32 to 1.24)
2941 (971 to ‘ to NNH 2778) 2941 (1235 to ‘ to NNH 7692) 3448 (1351 to ‘ to NNH 6250)
20.83 (–2.54 to 0.88) 20.84 (–1.99 to 0.31) 20.72 (–1.82 to 0.0.38)
0.57 (0.24 to 1.35) 0.31 (0.07 to 1.41) 0.34 (0.05 to 2.14)
1205 (394 to ‘ to NNH 1136) 1190 (503 to ‘ to NNH 3226) 1389 (549 to ‘ to NNH 2632)
NNT, number needed to treat; NNH, number needed to harm. To address the effect of birth timing we observed, we used the entire bandwidth of data (ie, all observations in the 1992 to 1995 birth cohorts), and included birth timing as a covariate in the model. In all analyses, the birth cohorts closest to the cutoff (1993 and 1994) were weighting twice as heavily as those farthest from the cutoff (1992 and 1995). a Estimated using local linear regression models. b Estimated using log-binomial regression models. c Reported as recommended by Altman (1998).42 When absolute RDs are not statistically significant, the limits of the CI for the NNT reflect both the possibility of benefit (NNT) and the possibility of harm (NNH). Moreover, the NNT is infinity (‘) when the absolute RD is 0. To convey the continuity of the CI, Altman42 suggests that both the NNT and the NNH be represented in 1 interval and that 0 be represented by infinity.
database. Because any resulting misclassification would have affected eligibility groups equally, it would have caused us to underestimate the magnitude of reductions. Also, although we could not report on the stage of dysplasia, given the young age of the study cohort and the long latency between HPV infection and cervical cancer,45 it is safe to assume the vast majority of cases of dysplasia represent atypical squamous cells or low-grade squamous intraepithelial lesions (ie, mild dysplasia). Future studies are needed to monitor for progression to high-grade lesions and cancer. Although the RDD addresses confounding by factors affecting an individual’s decision to receive the vaccine, it is conceptually vulnerable to confounding by factors that might have differentially affected eligibility groups, such as changes in cervical cancer screening practice or changes in sexual behavior caused by the implementation of the vaccination program. However, a recent study in this population provided strong evidence against changes in sexual behavior,29 and screening practices likely did not change in Ontario until January 1, 2013, when physicians were no longer reimbursed for screening girls younger than 21.
Downloaded from pediatrics.aappublications.org at UCSF LIBRARY on April 27, 2015
PEDIATRICS Volume 135, number 5, May 2015
e1137
FIGURE 4 A, Risk of broad AGW,* by birth quarter. Q1: born January–March; Q2: born April–June; Q3: born July–September; Q4: born October–December. B, Risk of broad AGW,* by birth year. *Ascertained September 1 of Grade 10 to March 31 of Grade 12.
Before that date, physicians typically made prescriptions for the oral contraceptive pill contingent on undergoing annual screening, meaning a large proportion of sexually active teenage girls received Papanicolaou tests. Because the vast majority of followup time was complete by January 2013, this policy change would not have had a meaningful impact on our results. We provide absolute measures of effect given their importance in determining the public health and economic impact of this intervention, but because absolute differences are a function of the underlying baseline risk, their generalizability to other populations is unclear. Therefore, our analyses should be repeated in a range of settings and over time, and
readers should use the relative measures when comparing our results to other studies. It is also important to recognize that our intention-to-treat estimates were influenced by the relatively low qHPV vaccine coverage in Ontario, meaning estimates of program impact are likely be of greater magnitude in jurisdictions with higher coverage. Conversely, the relative measures of vaccine impact take into account actual vaccine exposure and are therefore likely more generalizable to other jurisdictions offering free qHPV vaccination to girls who are largely sexually naive.
CONCLUSIONS This study provides new, strong evidence of the impact of qHPV
vaccination on reductions in cervical dysplasia among adolescent girls. Although imprecise, we also observed apparent reductions in AGWs. The fact that these benefits were observed in such a young age group strengthens current recommendations that vaccination should occur at an early age. As such, policy makers and physicians can use these findings to substantiate arguments that delaying vaccination may result in missed opportunities for prevention. In addition, cost-effectiveness studies should be updated to incorporate real-world estimates of program- and vaccine-level effectiveness and coverage to provide more accurate assessments of the value of qHPV vaccination.
Address correspondence to Leah M. Smith, MSc, Department of Public Health Sciences, Abramsky Hall, Queen’s University Kingston, ON K7L 3N6, Canada. E-mail: [email protected] PEDIATRICS (ISSN Numbers: Print, 0031-4005; Online, 1098-4275). Copyright © 2015 by the American Academy of Pediatrics FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose. FUNDING: Supported by a grant from the Canadian Institutes of Health Research (MOP130339). The funder had no role in the design or conduct of this study; the collection, analysis, or interpretation of the data; the preparation, review, or approval of the manuscript; nor in the decision to submit the manuscript for publication. This study was supported by the Institute for Clinical Evaluative Sciences, which is funded by an annual grant from the Ontario Ministry of Health and Long-Term Care. The opinions, results and conclusions reported in this paper are those of the authors and are independent from the funding sources. No
e1138
Downloaded from pediatrics.aappublications.org at UCSF LIBRARY on April 27, 2015
SMITH et al
endorsement by the Institute for Clinical Evaluative Sciences or the Ontario Ministry of Health and Long-Term Care is intended or should be inferred. Ms Smith was supported by a Canadian Institutes of Health Research Fredrick Banting and Charles Best Canada Graduate Scholarships Doctoral Award, Dr Strumpf by a Chercheure Boursier Junior 1 from the Fonds de la Recherche du Québec–Santé and the Ministère de la Santé et des Services Sociaux du Quebec, Dr Kaufman by the Canada Research Chairs program, Dr Lofters by a Canadian Cancer Society Research Institute Career Development Award in Cancer Prevention (grant 702114), and Dr Lévesque by the Canadian Foundation for Innovation –Leaders Opportunity Fund. POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
REFERENCES 1. Trottier H, Franco EL. The epidemiology of genital human papillomavirus infection. Vaccine. 2006;24(suppl 1):S1–S15 2. Parkin DM, Bray F. Chapter 2: The burden of HPV-related cancers. Vaccine. 2006; (24 suppl 3):S3/11–25 3. Muñoz N, Bosch FX, Castellsagué X, et al. Against which human papillomavirus types shall we vaccinate and screen? The international perspective. Int J Cancer. 2004;111(2):278–285 4. Greer CE, Wheeler CM, Ladner MB, et al. Human papillomavirus (HPV) type distribution and serological response to HPV type 6 virus-like particles in patients with genital warts. J Clin Microbiol. 1995; 33(8):2058–2063 5. Parry J. Vaccinating against cervical cancer. Bull World Health Organ. 2007; 85(2):89–90 6. Rambout L, Hopkins L, Hutton B, Fergusson D. Prophylactic vaccination against human papillomavirus infection and disease in women: a systematic review of randomized controlled trials. CMAJ. 2007;177(5):469–479 7. Lu B, Kumar A, Castellsagué X, Giuliano AR. Efficacy and safety of prophylactic vaccines against cervical HPV infection and diseases among women: a systematic review & meta-analysis. BMC Infect Dis. 2011;11:13 8. Haas M, Ashton T, Blum K, et al. Drugs, sex, money and power: an HPV vaccine case study. Health Policy. 2009;92(2-3): 288–295 9. Markowitz LE, Tsu V, Deeks SL, et al. Human papillomavirus vaccine introduction—the first five years. Vaccine. 2012;30(suppl 5):F139–F148 10. National Advisory Committee on Immunization (NACI). Statement on human papillomavirus vaccine. An Advisory Committee Statement (ACS). Can Commun Dis Rep. 2007;33(ACS-2):1–31
11. Rahman M, Laz TH, Berenson AB. Geographic variation in human papillomavirus vaccination uptake among young adult women in the United States during 2008-2010. Vaccine. 2013; 31(47):5495–5499 12. Colucci R, Hryniuk W, Savage C. HPV Vaccination Programs in Canada. Are we hitting the mark? In: The Cancer Advocacy Coalition of Canada, Report Card, 2008. Available at: www. canceradvocacy.ca/reportcard/2008/ HPV%20Vaccination%20Programs%20in %20Canada.pdf. Accessed March 18, 2015 13. Ogilvie G, Anderson M, Marra F, et al. A population-based evaluation of a publicly funded, school-based HPV vaccine program in British Columbia, Canada: parental factors associated with HPV vaccine receipt. PLoS Med. 2010;7(5): e1000270 14. Dorell C, Yankey D, Jeyarajah J, et al. Delay and refusal of human papillomavirus vaccine for girls, national immunization survey-teen, 2010. Clin Pediatr (Phila). 2014;53(3):261–269 15. Perkins RB, Clark JA, Apte G, et al. Missed opportunities for HPV vaccination in adolescent girls: a qualitative study. Pediatrics. 2014; 134(3). Available at: www.pediatrics.org/ cgi/content/full/134/3/e666 16. Liddon NC, Leichliter JS, Markowitz LE. Human papillomavirus vaccine and sexual behavior among adolescent and young women. Am J Prev Med. 2012; 42(1):44–52 17. Gertig DM, Brotherton JM, Budd AC, Drennan K, Chappell G, Saville AM. Impact of a population-based HPV vaccination program on cervical abnormalities: a data linkage study. BMC Med. 2013;11:227 18. Baldur-Felskov B, Dehlendorff C, Munk C, Kjaer SK. Early impact of human papillomavirus vaccination on cervical
neoplasia–nationwide follow-up of young Danish women. J Natl Cancer Inst. 2014; 106(3):djt460 19. Mahmud SM, Kliewer EV, Lambert P, Bozat-Emre S, Demers AA. Effectiveness of the quadrivalent human papillomavirus vaccine against cervical dysplasia in Manitoba, Canada. J Clin Oncol. 2014;32(5):438–443 20. Crowe E, Pandeya N, Brotherton JM, et al. Effectiveness of quadrivalent human papillomavirus vaccine for the prevention of cervical abnormalities: case-control study nested within a population based screening programme in Australia. BMJ. 2014;348: g1458 21. Leval A, Herweijer E, Ploner A, et al. Quadrivalent human papillomavirus vaccine effectiveness: a Swedish national cohort study. J Natl Cancer Inst. 2013; 105(7):469–474 22. Ministry of Health and Long-Term Care. Ontario’s HPV vaccination program. 2013. Available at: www.health.gov.on.ca/en/ ms/hpv/. Accessed March 13, 2014 23. Institute for Clinical Evaluative Sciences. Data discovery, better health. Available at: www.ices.on.ca/Data-and-Privacy/ ICES-data/Types-of-Data. Accessed March 1, 2014 24. Lipscombe LL, Gomes T, Lévesque LE, Hux JE, Juurlink DN, Alter DA. Thiazolidinediones and cardiovascular outcomes in older patients with diabetes. JAMA. 2007;298(22): 2634–2643 25. Lipscombe LL, Lévesque L, Gruneir A, et al. Antipsychotic drugs and hyperglycemia in older patients with diabetes. Arch Intern Med. 2009;169(14): 1282–1289 26. Kwong JC, Tanuseputro P, Zagorski B, Moineddin R, Chan KJ. Impact of varicella vaccination on health care outcomes in Ontario, Canada: effect of a publicly
Downloaded from pediatrics.aappublications.org at UCSF LIBRARY on April 27, 2015
PEDIATRICS Volume 135, number 5, May 2015
e1139
funded program? Vaccine. 2008;26(47): 6006–6012 27. Juurlink DN, Stukel TA, Kwong J, et al. Guillain-Barré syndrome after influenza vaccination in adults: a population-based study. Arch Intern Med. 2006;166(20): 2217–2221 28. Smith LM, Brassard P, Kwong JC, Deeks SL, Ellis AK, Lévesque LE. Factors associated with initiation and completion of the quadrivalent human papillomavirus vaccine series in an Ontario cohort of grade 8 girls. BMC Public Health. 2011;11:645 29. Smith LM, Kaufman JS, Strumpf EC, Levesque LE. Effect of human papillomavirus (HPV) vaccination on clinical indicators of sexual behaviour among adolescent girls: the Ontario Grade 8 HPV Vaccine Cohort Study. CMAJ. 2015;187(2):E74–E81 30. Lévesque LE, Smith LM, Perry AG, et al. The Ontario Grade 8 HPV Vaccine Cohort Study: A Feasibility and Validity Evaluation. In: Ontario Vaccine Sciences Symposium. Toronto, ON; November 24, 2011 31. Fine PE, Chen RT. Confounding in studies of adverse reactions to vaccines. Am J Epidemiol. 1992;136(2):121–135 32. Chen RT, Davis RL, Rhodes PH. Special methodological issues in pharmacoepidemiology studies of vaccine safety. In: Strom BL, ed. Pharmacoepidemiology. 4th ed. Sussex, UK: John Wiley & Sons, Ltd; 2005:455–485
e1140
33. Jackson LA, Jackson ML, Nelson JC, Neuzil KM, Weiss NS. Evidence of bias in estimates of influenza vaccine effectiveness in seniors. Int J Epidemiol. 2006;35(2):337–344 34. Nelson JC, Jackson ML, Weiss NS, Jackson LA. New strategies are needed to improve the accuracy of influenza vaccine effectiveness estimates among seniors. J Clin Epidemiol. 2009;62(7): 687–694 35. Hahn J, Todd P, Van der Klaauw W. Identification and estimation of treatment effects with a regressiondiscontinuity design. Econometrica. 2001; 69:201–209 36. Imbens GW, Lemieux T. Regression discontinuity designs: A guide to practice. J Econom. 2008;142:615–635 37. Cook TD. ‘‘Waiting for Life to Arrive’’: A history of the regression-discontinuity design in psychology, statistics and economics. J Econom. 2007;142:636–654
1–74. Available at: www.nber.org/papers/ w18309. Accessed March 18, 2015 40. O’Keeffe AG, Geneletti S, Baio G, Sharples LD, Nazareth I, Petersen I. Regression discontinuity designs: an approach to the evaluation of treatment efficacy in primary care using observational data. BMJ. 2014;349:g5293 41. Bor J, Moscoe E, Mutevedzi P, Newell ML, Bärnighausen T. Regression discontinuity designs in epidemiology: causal inference without randomized trials. Epidemiology. 2014;25(5): 729–737 42. Altman DG. Confidence intervals for the number needed to treat. BMJ. 1998; 317(7168):1309–1312 43. Dobson SR, McNeil S, Dionne M, et al. Immunogenicity of 2 doses of HPV vaccine in younger adolescents vs 3 doses in young women: a randomized clinical trial. JAMA. 2013;309(17): 1793–1802
38. Kadiyala S, Strumpf EC. How Effective is Population-Based Cancer Screening? Regression Discontinuity Estimates from the U.S. Guideline Screening Initiation Ages (July 1, 2011). Available at: http:// dx.doi.org/10.2139/ssrn.1893793. Accessed March 18, 2015
44. Smith LM, Lévesque LE, Nasr M, Perry AG. Validity of the Immunization Record Information System (IRIS) database for epidemiologic studies of the human papillomavirus (HPV) vaccine. Canadian Journal of Clinical Pharmacoepidemiology. 2010;17: e90–e127
39. Meyer BD, Wherry LR. Saving Teens: Using a Policy Discontinuity to Estimate the Effects of Medicaid Eligibility. Cambridge Massachusetts: The National Bureau of Economic Research; 2012:
45. Moscicki AB, Schiffman M, Kjaer S, Villa LL. Chapter 5: Updating the natural history of HPV and anogenital cancer. Vaccine. 2006;(24 suppl 3): S3/42–51
Downloaded from pediatrics.aappublications.org at UCSF LIBRARY on April 27, 2015
SMITH et al
The Early Benefits of Human Papillomavirus Vaccination on Cervical Dysplasia and Anogenital Warts Leah M. Smith, Erin C. Strumpf, Jay S. Kaufman, Aisha Lofters, Michael Schwandt and Linda E. Lévesque Pediatrics; originally published online April 27, 2015; DOI: 10.1542/peds.2014-2961 Updated Information & Services
including high resolution figures, can be found at: http://pediatrics.aappublications.org/content/early/2015/04/21 /peds.2014-2961
Supplementary Material
Supplementary material can be found at: http://pediatrics.aappublications.org/content/suppl/2015/04/2 1/peds.2014-2961.DCSupplemental.html
Permissions & Licensing
Information about reproducing this article in parts (figures, tables) or in its entirety can be found online at: http://pediatrics.aappublications.org/site/misc/Permissions.xh tml
Reprints
Information about ordering reprints can be found online: http://pediatrics.aappublications.org/site/misc/reprints.xhtml
PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly publication, it has been published continuously since 1948. PEDIATRICS is owned, published, and trademarked by the American Academy of Pediatrics, 141 Northwest Point Boulevard, Elk Grove Village, Illinois, 60007. Copyright © 2015 by the American Academy of Pediatrics. All rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275.
Downloaded from pediatrics.aappublications.org at UCSF LIBRARY on April 27, 2015
The Early Benefits of Human Papillomavirus Vaccination on Cervical Dysplasia and Anogenital Warts Leah M. Smith, Erin C. Strumpf, Jay S. Kaufman, Aisha Lofters, Michael Schwandt and Linda E. Lévesque Pediatrics; originally published online April 27, 2015; DOI: 10.1542/peds.2014-2961
The online version of this article, along with updated information and services, is located on the World Wide Web at: http://pediatrics.aappublications.org/content/early/2015/04/21/peds.2014-2961
PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly publication, it has been published continuously since 1948. PEDIATRICS is owned, published, and trademarked by the American Academy of Pediatrics, 141 Northwest Point Boulevard, Elk Grove Village, Illinois, 60007. Copyright © 2015 by the American Academy of Pediatrics. All rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275.
Downloaded from pediatrics.aappublications.org at UCSF LIBRARY on April 27, 2015
Despite widespread promotion of quadrivalent human papillomavirus (qHPV) vaccination for young girls, there is limited information on the vaccine’s real-world effectiveness and none on the effectiveness of qHPV vaccination programs. We assessed the impact of the qHPV vaccine and Ontario’s grade 8 qHPV vaccination program on cervical dysplasia and anogenital warts (AGW). BACKGROUND:
abstract
By using administrative health databases of Ontario, Canada, we identified a population-based retrospective cohort of girls in grade 8 before (2005/2006–2006/2007) and after (2007/2008–2008/2009) program implementation. Vaccine exposure was ascertained in grades 8 to 9 and outcomes in grades 10 to 12. A quasi-experimental approach known as regression discontinuity was used to estimate absolute risk differences (RDs), relative risks (RRs), and 95% confidence intervals (CIs) attributable to vaccination and program eligibility (intention-to-treat analysis).
METHODS:
RESULTS: The cohort comprised 131 781 ineligible and 128 712 eligible girls (n = 260 493). We identified 2436 cases of dysplasia and 400 cases of AGW. Vaccination significantly reduced the incidence of dysplasia by 5.70 per 1000 girls (95% CI 29.91 to 21.50), corresponding to a relative reduction of 44% (RR 0.56; 95% CI 0.36 to 0.87). Program eligibility also had a significant protective effect on dysplasia: RD 22.32/1000 (95% CI 24.02 to 20.61); RR 0.79 (95% CI 0.66 to 0.94). Results suggested decreases in AGW attributable to vaccination (RD 20.83/1000, 95% CI 22.54 to 0.88; RR 0.57, 95% CI 0.20 to 1.58) and program eligibility (RD 20.34/1000, 95% CI 21.03 to 0.36; RR 0.81, 95% CI 0.52 to 1.25). CONCLUSIONS: This study provides strong evidence of the early benefits of qHPV vaccination among girls aged 14 to 17 years, offering additional justification for not delaying vaccination.
a Departments of Epidemiology, Biostatistics, and Occupational Health, and bEconomics, McGill University, Montreal, Quebec, Canada; cDepartment of Family and Community Medicine, St Michael’s Hospital, University of Toronto, Toronto, Ontario, Canada; dDepartment of Community Health and Epidemiology, University of Saskatchewan, Saskatoon, Saskatchewan, Canada; eDepartment of Public Health Sciences, Queen’s University, Kingston, Ontario, Canada; and fInstitute for Clinical Evaluative Sciences-Queen’s Health Services Research Facility, Kingston, Ontario, Canada
Ms Smith was involved in acquiring study data, played a major role in the conception, design, and interpretation of the study, carried out the statistical analyses, and drafted the manuscript; Drs Strumpf and Kaufman played major roles in the conception, design, analysis, and interpretation of the study, and critically reviewed the manuscript; Drs Lofters and Schwandt were involved in design of the study and critically reviewed the manuscript; Dr Lévesque played the principal role in acquiring the study data, played a major role in the conception, design, analysis, and interpretation of the study, and critically reviewed the manuscript; and all authors were involved in obtaining funding for this study, approved the final manuscript as submitted, and agree to be accountable for all aspects of this work. www.pediatrics.org/cgi/doi/10.1542/peds.2014-2961 DOI: 10.1542/peds.2014-2961 Accepted for publication Feb 24, 2015
WHAT’S KNOWN ON THIS TOPIC: Clinical trials of the quadrivalent human papillomavirus vaccine show it to be highly efficacious in preventing vaccine-type–specific cervical dysplasia and anogenital warts, but few studies have assessed its effects in the real world and none have done so at the program/population level. WHAT THIS STUDY ADDS: This study provides strong evidence of the early benefits of quadrivalent human papillomavirus vaccination on reductions in cervical dysplasia and possible reductions in anogenital warts among girls aged 14 to 17 years, offering additional justification for not delaying vaccination until girls are older.
Downloaded from pediatrics.aappublications.org at UCSF LIBRARY on April 27, 2015
PEDIATRICS Volume 135, number 5, May 2015
ARTICLE
The human papillomavirus (HPV) is a sexually transmitted infection that affects most individuals at some point during their lives.1 Although the vast majority of these infections selfresolve without clinical sequelae, others can lead to important health consequences, including anogenital warts (AGW) and cancers of the cervix, anus, and penis.2 In 2006, Canada was among the first countries to license Gardasil (Merck, Whitehouse Station, NJ), a quadrivalent human papillomavirus (qHPV) vaccine that protects against 4 types of HPV that cause 70% of cases of cervical cancer and at least 90% of AGWs.3–5 As randomized controlled trials of the vaccine showed it to be highly efficacious in preventing vaccine-type–specific precancerous cervical lesions and AGW,6,7 the vaccine received a great deal of attention from the scientific and medical communities, as well as from the general public.8 Currently, the qHPV vaccine is available in .100 countries, many of which also have national qHPV immunization programs.9 These programs are generally aimed at immunizing young girls before the onset of sexual activity when the possibility of previous HPV infection is low (eg, ages 9–13 years).9,10 Despite widespread promotion of the vaccine and the popularity of largescale qHPV vaccination programs, levels of vaccine uptake have been lower than expected in a number of jurisdictions.9,11,12 In part, low qHPV vaccine acceptance has been attributed to parental concerns about the effects of the vaccine, especially in the young age group targeted for vaccination.13,14 In addition, some parents are deciding to delay vaccination for their daughters until a later age, as they perceive their daughter’s likelihood of sexual activity as low.15 Given the tendency of parents to underestimate their child’s sexual experience,16 there are important concerns that delaying
e1132
vaccination might result in missed opportunities for prevention. In addition, low vaccine coverage may jeopardize the expected public health benefits and cost-effectiveness of qHPV vaccine programs. As there are currently few studies on the real-world effects of the qHPV vaccine17–21 and none on the population-level impact of qHPV vaccine programs on the burden of disease, we undertook a populationbased, retrospective cohort study to assess the impact of the qHPV vaccine and Ontario’s qHPV vaccination program on the incidence of cervical dysplasia and AGW among adolescent girls.
METHODS This study was approved by the Research Ethics Boards of Queen’s University, McGill University, and Sunnybrook Health Sciences Centre.
Study Setting Our study is based in Ontario, Canada’s most populous province, which began offering all 3 doses of the vaccine, free-of-charge, to all grade 8 girls in September of 2007.22 Doses are administered primarily through school-based immunization clinics, but girls also may receive the vaccine from a physician or at their health unit at no cost. Before September 2012, eligible girls had until the end of their grade 8 year to initiate the vaccine series and until the end of grade 9 to complete it. During our study period, girls who were not eligible for the school-based program (eg, in grade 8 before 2007) could obtain the vaccine series for ∼$400.
Data Sources Data for this study were obtained from Ontario’s population-based administrative health and immunization databases, which are housed at the Institute for Clinical Evaluative Sciences.23 We used (1) the Registered Persons Database
(RPDB) for sociodemographics, (2) the Ontario Health Insurance Plan (OHIP) database for fee-for-service claims by physicians, (3) the Discharge Abstract Database for hospitalizations, (4) the Same-Day Surgery database for same-day surgeries, (5) the National Ambulatory Care Reporting System (NACRS) for emergency department visits, and (6) the Immunization Records Information System (IRIS) for vaccinations. In these databases, residents are represented by a unique, encrypted identifier, enabling anonymized, individual-level record linkage across databases and time. These databases are generated by Ontario’s universal health insurance programs and have been used extensively in health research, including in the postmarketing evaluation of drug and vaccine effects.24–29
Study Population and Cohort Formation We identified a population-based cohort of all girls in grade 8 in Ontario during the 2005/ 2006–2008/2009 school years. This includes 2 years of girls who were eligible for publicly funded qHPV vaccination (ie, in grade 8 in 2007/ 2008 and 2008/2009) and, for the purposes of comparison, 2 years of girls who were ineligible (ie, in grade 8 in 2005/2006 and 2006/2007). As school grade is not available in these databases, grade was determined based on birth year because .96% of girls enter grade 8 in their 13th year of life.30 Specifically, we used the RPDB to identify all girls born in 1992, 1993, 1994, and 1995 and, through record linkage with IRIS, restricted these birth cohorts to girls whose immunization records were available at the time of this study and who were residing in Ontario on September 1 of 2005, 2006, 2007, and 2008, respectively. September 1 of grade 8 defined cohort entry, and cohort members were followed until the earliest of the following: date of
Downloaded from pediatrics.aappublications.org at UCSF LIBRARY on April 27, 2015
SMITH et al
death, occurrence of a study outcome, or March 31 of grade 12 (end of follow-up).
Measurement and Analysis Approach Observational studies of vaccine effects are notoriously vulnerable to confounding bias because individuals who opt for vaccination tend to have different health histories and behaviors than those who do not, and these characteristics are difficult to identify and quantify.31–34 To address this methodological challenge, we used the regression discontinuity design (RDD), a quasi-experimental approach for evaluating the causal effects of interventions in a way that accounts for this type of observed and unobserved confounding, thus facilitating reliable causal inference.35–37 The RDD has been used extensively in fields like health economics,37–39 and is becoming increasingly popular in epidemiology, given the advantages it offers over standard regression adjustment.40,41 In this study, the RDD enabled us to exploit the fact that girls were eligible for free qHPV vaccination (ie, assigned to treatment) based on whether they were in grade 8 before or after the September 2007 program implementation date (ie, born December 31, 1993 or earlier versus January 1, 1994 or later), which induced discontinuity in the probability of qHPV vaccination at the eligibility cutoff (Supplemental Appendix 1A). As such, birth date acted as the factor (“forcing variable”) that influenced whether a girl was exposed to vaccination. Essentially, the goal of the RDD was to estimate any corresponding discontinuity in the probability of the outcomes between eligibility groups at the same cutoff (Supplemental Appendix 1B). The value of the RDD over more traditional designs relies on the assumption that because the program implementation date and eligibility criteria were based on administrative decisions, the exact location of the
cutoff is random, making girls directly on either side of the cutoff similar with respect to measured and unmeasured confounders. In the RDD, this comparability is assumed to decrease with increasing distance from the cutoff, so this notion is taken into account in the analysis.
Baseline Characteristics We identified a number of characteristics relating to sociodemographics, vaccination history, health service use, and medical history. These were compared between eligibility groups and across levels of the forcing variable.
Forcing Variable The forcing variable (ie, birth date) was categorized into 3-month intervals, referred to as “birth year quarters.” Therefore, girls born October 1, 1993, to December 31, 1993, defined the “ineligible” side of the cutoff and girls born January 1, 1994, to March 31, 1994, defined the “eligible” side. Girls born earlier/later than these dates were represented with increasing distance from this cutoff on the ineligible/eligible sides (Supplemental Appendix 2).
Exposure To evaluate the impact of the qHPV vaccination program, exposure status was based solely on program eligibility, thus providing an intention-to-treat definition of vaccination. To assess the impact of the vaccine on our outcomes, qHPV vaccine receipt also was taken into account. The qHPV vaccine exposure was defined as receipt of all 3 doses between September 1 of grade 8 and August 31 of grade 9.
Outcomes Incident cases of detected cervical dysplasia and AGW were identified between September 1 of grade 10 and March 31 of grade 12. Cases were “incident” if they occurred after at least 730 days (cervical dysplasia) or 365 days (AGW) event-free. Because algorithms for ascertaining these outcomes have not been validated, we created 3 definitions of each (broad, possible, and probable) in consultation with substantive experts in the field to reflect different degrees of potential misclassification (Supplemental Appendices 3 and 4).
Statistical Analyses To assess program impact, local linear regression was used to estimate the association between program eligibility and each outcome of interest.35,36 Specifically, regression functions were estimated on each side of the eligibility cutoff, and the difference/discontinuity between intercepts represented the absolute impact of program eligibility on the outcome. To assess vaccine impact, a second stage was added to also estimate the association between program eligibility and qHPV vaccine exposure. The absolute effect of the vaccine was determined based on the ratio of the 2 discontinuities: eligibility-outcome and eligibilityvaccine. This 2-stage analysis is analogous to an instrumental variable approach.35 Similarly, 1- and 2-stage log-binomial regression models were used to estimate the relative impact of program eligibility and vaccination on each outcome. Finally, the “number needed to treat” (NNT) of each outcome was estimated based on the reciprocal of the absolute risk difference (RD) and 95% confidence limits.42 All analyses were based on the entire study cohort; however, girls born in 1993 and 1994 were weighted twice as heavily as girls born in 1992 and 1995 because girls closest to the cutoff are most comparable (ie, lower potential for confounding). Similarly, we controlled for birth quarter because we found that girls born early (or late) in the year were most comparable between birth years (Supplemental Appendix 5). A number of sensitivity analyses were performed to evaluate the robustness of our results, including
Downloaded from pediatrics.aappublications.org at UCSF LIBRARY on April 27, 2015
PEDIATRICS Volume 135, number 5, May 2015
e1133
one that modified the “exposed” definition to include girls who received only 2 doses of the vaccine, as 2 doses may be sufficient to provide immunity.43 The other assumptions of the RDD were tested quantitatively or qualitatively36 and were found to hold. SAS statistical software version 9.3 (SAS Institute, Inc, Cary, NC) was used for data management and Stata version 12 (Stata Corp, College Station, TX) for analyses.
RESULTS We identified a cohort of 260 493 girls, of whom almost half were eligible for free, school-based qHPV vaccination (Fig 1). Girls were 12.7 to 13.7 years old at cohort entry and were followed for an average of 4.6 years. Eligibility groups were similar with respect to a number of baseline characteristics, including frequency of outpatient physician visits and previous receipt of the
FIGURE 1
measles-mumps-rubella and diphtheria, tetanus, and pertussis vaccines. There were small differences in such characteristics as neighborhood income quintile, hepatitis B vaccination, and indicators of previous sexual behavior (Table 1). Of the 75 848 cohort members who received at least 1 dose of the qHPV vaccine in grades 8 to 9, 88% completed the 3-dose series. As expected, program eligibility had a major impact on this exposure. Specifically, 50.6% of eligible girls were classified as qHPV vaccine exposed compared with ,1% of ineligible girls. Figure 2 illustrates the discontinuity in qHPV vaccination between groups. We identified 2436 (0.94%) cohort members with incident cervical dysplasia (Table 2). Although there was no apparent discontinuity at the eligibility cutoff when assessed across birth year quarters (Fig 3A), collapsing into birth years revealed a clear drop in risk between groups
Cohort flow diagram. *At the time of this study, 2 of Ontario’s 36 IRIS databases (representing approximately 22% of Ontario’s population) had not yet been transferred to the Institute for Clinical Evaluative Sciences and were therefore unavailable for use in this study. IRIS records also were unavailable if the girl emigrated from Ontario before starting kindergarten or immigrated to Ontario after completing high school. †A girl’s IRIS record was defined as “up to date” if it had been modified 30 days before cohort entry or later. Otherwise, it was assumed the girl had moved out of our study area before cohort entry.
e1134
(Fig 3B), demonstrating the importance of controlling for birth quarter in the analyses. Indeed, the primary analysis (which adjusted for birth timing) revealed a statistically significant reduction in detected dysplasia attributable to program eligibility: RD 22.32 cases per 1000 girls, 95% confidence interval (CI) 24.02 to 20.61. An even stronger effect was observed for the vaccine: RD 25.70 per 1000 girls, 95% CI 29.91 to 21.50. These results indicate that 1 case of cervical dysplasia was prevented for every 431 (95% CI 248 to 1639) girls eligible for publicly funded vaccination and for every 175 (95% CI 101 to 667) girls who received the vaccine. The absolute RDs corresponded to a 21% (95% CI 6% to 34%) and 44% (95% CI 13% to 64%) relative risk (RR) reduction for eligibility and vaccination, respectively. RR reductions were similar for possible dysplasia and probable dysplasia, whereas absolute RDs decreased and NNTs increased with increasing specificity of the dysplasia definition (Table 3). We ascertained 400 cases of AGW (Table 2). Figure 4 A and B depict this risk by birth year quarter and by birth year, respectively. The results suggested reductions in detected AGW attributable to program eligibility and vaccination, both on the absolute scale (RD 20.34 per 1000, 95% CI 21.03 to 0.36; RD 20.83, 95% CI 22.54 to 0.88) and the relative scale (RR 0.81, 95% CI 0.52 to 1.25; RR 0.57, 95% CI 0.20 to 1.58); however, these results were not statistically significant. Findings indicated even greater RR reductions for more specific outcome definitions of AGW, whereas absolute risk reductions were similar across definitions (Table 3). Results for both cervical dysplasia and AGW were robust in sensitivity analyses, including analyses that adjusted for additional baseline characteristics (Supplemental Appendix 6), defined HPV vaccine exposure based on
Downloaded from pediatrics.aappublications.org at UCSF LIBRARY on April 27, 2015
SMITH et al
DISCUSSION
TABLE 1 Baseline Characteristics of Study Cohort a
Characteristic
Percentage
Sociodemographicsb Mean age at cohort entry (SD) Birth timing Q1: January–March Q2: April–June Q3: July–September Q3: October–December Residency Urban Rural Missingc Income quintile 1st (lowest) 2nd 3rd 4th 5th (highest) Missingc Vaccination historyd MMRe DTPe Hepatitis Be All 3 vaccines Frequency of health care usef,g Hospitalizations None (0) $1 Days (mean, SD) Same-day surgeries None (0) $1 Emergency department visits None (0) 1 $2 Outpatient visits 0–1 2–5 6–12 $13 Medical history Cancerf Mental health diagnosisf Indicators of sexual behaviorf,h Down syndromed Congenital malformationsd Intellectual disabilitiesd
Ineligible, n = 131 781)
Eligible, n = 128 712)
13.17 (0.28)
13.17 (0.28)
24.3 26.1 25.7 23.9
24.2 26.1 25.8 23.9
85.3 14.0 0.7
85.8 13.5 0.6
16.6 18.4 20.6 22.0 21.4 1.0
15.0 17.8 21.1 23.1 22.1 0.9
97.9 98.0 84.1 83.0
98.2 98.3 82.0 81.1
98.0 2.0 7.4, 15.6
98.2 1.8 8.0, 18.2
97.7 2.4
97.8 2.2
70.7 18.1 11.2
71.1 17.8 11.1
25.0 22.6 25.1 27.4
25.8 22.8 24.5 26.9
0.71 9.5 0.68 0.48 12.4 0.72
0.74 9.7 0.73 0.53 11.8 0.73
DTP, diphtheria, tetanus, and pertussis; MMR, measles-mumps-rubella; Q, quarter. a Figures are expressed as percentages except where indicated otherwise. b Obtained at cohort entry. c Missing because postal code in RPDB inaccurate or unavailable. d Obtained between birth and cohort entry. e At least 1 dose. f Obtained in the 2 years before cohort entry. g Categories of health care use were determined based on the frequency distribution of the data. h Composite endpoint of sexually transmitted infections, cervical dysplasia, Papanicolaou tests, and pregnancy, all of which are clinical indicators of previous sexual behavior.
receipt of at least 2 doses, and restricted the exposure ascertainment window to grade 8
and the outcome ascertainment window to grades 9 to 12 (Supplemental Appendix 7).
This is the first observational study of the benefits of the qHPV vaccine to use a methodological approach that permits causal inference, providing new, strong evidence of the positive health impact of qHPV vaccination on health outcomes. In particular, our analyses revealed a protective effect of the qHPV vaccine program on cervical dysplasia among girls aged 14 to 17 years and, although imprecise, we also observed apparent reductions in the incidence of AGW. Estimates of vaccine impact for both outcomes were even stronger than those of program impact. All results speak to the value of offering qHPV vaccination to young girls through a population-based program. Our use of the RDD enabled us to evaluate the intention-to-treat effects of qHPV vaccination for the first time outside of clinical trials. These analyses have particular public health value, as they provide measures of the population-level impact of the qHPV vaccine program on cervical dysplasia and AGW. Importantly, our findings revealed that despite relatively low qHPV vaccine coverage in Ontario (∼50% among eligible girls), meaningful benefits were nevertheless observed in this young age group at the population level. These results are promising for other jurisdictions across Canada12 and beyond9 that have also struggled with low HPV vaccine coverage. Nevertheless, estimates of program impact were considerably weaker than estimates of vaccine impact. For example, we observed almost 2.5 times greater absolute reductions in cervical dysplasia attributable to the vaccine compared with the program. Given that the benefits of the program impact are largely dependent on the proportion of girls vaccinated, these differences were not surprising, and they help demonstrate the important implications of vaccine coverage on reducing the burden of HPV-related
Downloaded from pediatrics.aappublications.org at UCSF LIBRARY on April 27, 2015
PEDIATRICS Volume 135, number 5, May 2015
e1135
FIGURE 2 Probability of qHPV vaccine exposure,* by birth quarter. Q1: born January–March; Q2: born April– June; Q3: born July–September; Q4: born October–December. *Receipt of 3 doses of the qHPV vaccine between September 1 of grade 8 and August 31 of grade 9.
diseases in the population. Indeed, a thorough appreciation of the interplay between the program- and vaccine-level effects in any region will be crucial to improving qHPV vaccine policy and practice in ways that maximize the health and economic benefits of this vaccine. To date, observational studies of cervical dysplasia and AGW have reported estimates of vaccine effectiveness based on analyses comparing vaccinated and unvaccinated girls. Our results of vaccine impact are generally consistent with previous research on cervical dysplasia.17–20 Although all observational studies have reported more conservative estimates than randomized controlled trials,7 this is not surprising given that trials
estimated treatment effects under ideal conditions. This does, however, highlight the importance of observational studies in determining the real-world effectiveness of this vaccine. The only nonecologic observational study of AGW reported an effect greater than reported here; however, the estimates generally fell within our CIs.21 The difference in results between studies may be partly attributable to residual confounding between vaccinated and unvaccinated groups in the earlier study. Regardless, given the low risk of diagnosed AGW in our study’s young age group and the resulting imprecision of our estimates, additional studies are needed to further elucidate these findings.
TABLE 2 Cumulative Incidence of Cervical Dysplasia and AGWsa Outcome
Frequency (percent) Ineligible
Eligible
Total, n = 260 493
1992, n = 66 653 1993, n = 65 128 1994, n = 64 818 1995, n = 63 894 Dysplasia Broad Possible Probable AGWs Broad Possible Probable a
698 (1.05) 444 (0.67) 232 (0.35)
720 (1.11) 478 (0.73) 229 (0.35)
555 (0.86) 358 (0.55) 161 (0.25)
463 (0.72) 277 (0.43) 121 (0.19)
2436 (0.94) 1557 (0.60) 743 (0.29)
119 (0.18) 68 (0.10) 62 (0.09)
120 (0.18) 59 (0.09) 54 (0.08)
91 (0.14) 32 (0.05) 30 (0.05)
70 (0.11) 31 (0.05) 27 (0.04)
400 (0.15) 190 (0.07) 173 (0.07)
Ascertained between September 1 of grade 10 and March 31 of grade 12.
e1136
A major strength of this study is the use of population-based administrative databases, which enabled us to conduct a large cohort study with limited potential for selection, recall, or response biases. An additional benefit is that our HPV vaccination data were highly sensitive (99.8%; 95% CI 99.3 to 99.9) and specific (97.7%; 95% CI 96.3 to 98.7) to girls’ vaccination status,44 meaning the impact of any exposure misclassification was likely negligible among eligible girls. Conversely, it is possible that we were missing information on HPV vaccines obtained by ineligible girls through private means. Although we expect this percentage to be low, it would nevertheless have caused us to underestimate the impact of the vaccine on our outcomes. The potential for herd immunity among ineligible girls would also have underestimated the benefits of the vaccine in this study. However, given the young age group and limited follow-up time of the study population, as well as the difference in calendar years of follow-up between eligibility groups, this effect was likely negligible. Another limitation is the potential for outcome misclassification, as measures of cervical dysplasia and AGW have not been validated. To assess this potential for error, we created 3 levels of each outcome to reflect increasing levels of outcome sensitivity. Indeed, the fact that the RR of broad AGW was biased toward the null compared with the more sensitive definitions suggests the broad definition was misclassified. This is not surprising, as this definition included a diagnostic code that also could be used to capture warts on other body sites. On the other hand, the consistency in relative estimates across definitions of dysplasia and between definitions of possible and probable AGW suggests that most of our outcome definitions did not result in misclassification. Another limitation of our data are
Downloaded from pediatrics.aappublications.org at UCSF LIBRARY on April 27, 2015
SMITH et al
FIGURE 3 A, Risk of broad dysplasia,* by birth quarter. Q1: born January–March; Q2: born April–June; Q3: born July–September; Q4: born October–December. B, Risk of broad dysplasia,* by birth year. *Ascertained September 1 of Grade 10 to March 31 of Grade 12.
they detected only outcomes of girls who followed up with their physician after an abnormal screening result or girls who received care for AGWs through physicians reimbursed through OHIP. As such, some incident cases may have been missed. Because this underascertainment would have
affected both groups equally, it would not have affected our relative estimates; however, our absolute estimates of effect are likely conservative. Another limitation is that we could not distinguish between dysplasia and cervical ectropion/erosion in the OHIP
TABLE 3 Impact of qHPV Vaccination on the Risk of Cervical Dysplasia and AGWs (n = 260 493) RDs, per 1000 Girls (95% CI)a Risk Ratio (95% CI)b Dysplasia Program impact Broad Possible Probable Vaccine impact Broad Possible Probable AGWs Program impact Broad Possible Probable Vaccine impact Broad Possible Probable
NNT (95% CI)c
22.32 (–4.02 to –0.61) 21.70 (–3.08 to –0.32) 20.79 (–1.74–0.16)
0.79 (0.66 to 0.94) 0.76 (0.61 to 0.95) 0.76 (0.56 to 1.05)
431 (248 to 1639) 588 (325 to 3125) 1266 (575 to ‘ to NNH 6250)
25.70 (–9.91 to –1.50) 24.19 (–7.59 to –0.80) 21.94 (–4.28 to 0.39)
0.56 (0.37 to 0.85) 0.50 (0.30 to 0.82) 0.51 (0.23 to 1.12)
175 (101 to 667) 239 (132 to 1250) 515 (234 to ‘ to NNH 2564)
20.34 (–1.03 to 0.36) 20.34 (–0.81 to 0.13) 20.29 (–0.74 to 0.16)
0.81 (0.52 to 1.25) 0.60 (0.31 to 1.15) 0.63 (0.32 to 1.24)
2941 (971 to ‘ to NNH 2778) 2941 (1235 to ‘ to NNH 7692) 3448 (1351 to ‘ to NNH 6250)
20.83 (–2.54 to 0.88) 20.84 (–1.99 to 0.31) 20.72 (–1.82 to 0.0.38)
0.57 (0.24 to 1.35) 0.31 (0.07 to 1.41) 0.34 (0.05 to 2.14)
1205 (394 to ‘ to NNH 1136) 1190 (503 to ‘ to NNH 3226) 1389 (549 to ‘ to NNH 2632)
NNT, number needed to treat; NNH, number needed to harm. To address the effect of birth timing we observed, we used the entire bandwidth of data (ie, all observations in the 1992 to 1995 birth cohorts), and included birth timing as a covariate in the model. In all analyses, the birth cohorts closest to the cutoff (1993 and 1994) were weighting twice as heavily as those farthest from the cutoff (1992 and 1995). a Estimated using local linear regression models. b Estimated using log-binomial regression models. c Reported as recommended by Altman (1998).42 When absolute RDs are not statistically significant, the limits of the CI for the NNT reflect both the possibility of benefit (NNT) and the possibility of harm (NNH). Moreover, the NNT is infinity (‘) when the absolute RD is 0. To convey the continuity of the CI, Altman42 suggests that both the NNT and the NNH be represented in 1 interval and that 0 be represented by infinity.
database. Because any resulting misclassification would have affected eligibility groups equally, it would have caused us to underestimate the magnitude of reductions. Also, although we could not report on the stage of dysplasia, given the young age of the study cohort and the long latency between HPV infection and cervical cancer,45 it is safe to assume the vast majority of cases of dysplasia represent atypical squamous cells or low-grade squamous intraepithelial lesions (ie, mild dysplasia). Future studies are needed to monitor for progression to high-grade lesions and cancer. Although the RDD addresses confounding by factors affecting an individual’s decision to receive the vaccine, it is conceptually vulnerable to confounding by factors that might have differentially affected eligibility groups, such as changes in cervical cancer screening practice or changes in sexual behavior caused by the implementation of the vaccination program. However, a recent study in this population provided strong evidence against changes in sexual behavior,29 and screening practices likely did not change in Ontario until January 1, 2013, when physicians were no longer reimbursed for screening girls younger than 21.
Downloaded from pediatrics.aappublications.org at UCSF LIBRARY on April 27, 2015
PEDIATRICS Volume 135, number 5, May 2015
e1137
FIGURE 4 A, Risk of broad AGW,* by birth quarter. Q1: born January–March; Q2: born April–June; Q3: born July–September; Q4: born October–December. B, Risk of broad AGW,* by birth year. *Ascertained September 1 of Grade 10 to March 31 of Grade 12.
Before that date, physicians typically made prescriptions for the oral contraceptive pill contingent on undergoing annual screening, meaning a large proportion of sexually active teenage girls received Papanicolaou tests. Because the vast majority of followup time was complete by January 2013, this policy change would not have had a meaningful impact on our results. We provide absolute measures of effect given their importance in determining the public health and economic impact of this intervention, but because absolute differences are a function of the underlying baseline risk, their generalizability to other populations is unclear. Therefore, our analyses should be repeated in a range of settings and over time, and
readers should use the relative measures when comparing our results to other studies. It is also important to recognize that our intention-to-treat estimates were influenced by the relatively low qHPV vaccine coverage in Ontario, meaning estimates of program impact are likely be of greater magnitude in jurisdictions with higher coverage. Conversely, the relative measures of vaccine impact take into account actual vaccine exposure and are therefore likely more generalizable to other jurisdictions offering free qHPV vaccination to girls who are largely sexually naive.
CONCLUSIONS This study provides new, strong evidence of the impact of qHPV
vaccination on reductions in cervical dysplasia among adolescent girls. Although imprecise, we also observed apparent reductions in AGWs. The fact that these benefits were observed in such a young age group strengthens current recommendations that vaccination should occur at an early age. As such, policy makers and physicians can use these findings to substantiate arguments that delaying vaccination may result in missed opportunities for prevention. In addition, cost-effectiveness studies should be updated to incorporate real-world estimates of program- and vaccine-level effectiveness and coverage to provide more accurate assessments of the value of qHPV vaccination.
Address correspondence to Leah M. Smith, MSc, Department of Public Health Sciences, Abramsky Hall, Queen’s University Kingston, ON K7L 3N6, Canada. E-mail: [email protected] PEDIATRICS (ISSN Numbers: Print, 0031-4005; Online, 1098-4275). Copyright © 2015 by the American Academy of Pediatrics FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose. FUNDING: Supported by a grant from the Canadian Institutes of Health Research (MOP130339). The funder had no role in the design or conduct of this study; the collection, analysis, or interpretation of the data; the preparation, review, or approval of the manuscript; nor in the decision to submit the manuscript for publication. This study was supported by the Institute for Clinical Evaluative Sciences, which is funded by an annual grant from the Ontario Ministry of Health and Long-Term Care. The opinions, results and conclusions reported in this paper are those of the authors and are independent from the funding sources. No
e1138
Downloaded from pediatrics.aappublications.org at UCSF LIBRARY on April 27, 2015
SMITH et al
endorsement by the Institute for Clinical Evaluative Sciences or the Ontario Ministry of Health and Long-Term Care is intended or should be inferred. Ms Smith was supported by a Canadian Institutes of Health Research Fredrick Banting and Charles Best Canada Graduate Scholarships Doctoral Award, Dr Strumpf by a Chercheure Boursier Junior 1 from the Fonds de la Recherche du Québec–Santé and the Ministère de la Santé et des Services Sociaux du Quebec, Dr Kaufman by the Canada Research Chairs program, Dr Lofters by a Canadian Cancer Society Research Institute Career Development Award in Cancer Prevention (grant 702114), and Dr Lévesque by the Canadian Foundation for Innovation –Leaders Opportunity Fund. POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
REFERENCES 1. Trottier H, Franco EL. The epidemiology of genital human papillomavirus infection. Vaccine. 2006;24(suppl 1):S1–S15 2. Parkin DM, Bray F. Chapter 2: The burden of HPV-related cancers. Vaccine. 2006; (24 suppl 3):S3/11–25 3. Muñoz N, Bosch FX, Castellsagué X, et al. Against which human papillomavirus types shall we vaccinate and screen? The international perspective. Int J Cancer. 2004;111(2):278–285 4. Greer CE, Wheeler CM, Ladner MB, et al. Human papillomavirus (HPV) type distribution and serological response to HPV type 6 virus-like particles in patients with genital warts. J Clin Microbiol. 1995; 33(8):2058–2063 5. Parry J. Vaccinating against cervical cancer. Bull World Health Organ. 2007; 85(2):89–90 6. Rambout L, Hopkins L, Hutton B, Fergusson D. Prophylactic vaccination against human papillomavirus infection and disease in women: a systematic review of randomized controlled trials. CMAJ. 2007;177(5):469–479 7. Lu B, Kumar A, Castellsagué X, Giuliano AR. Efficacy and safety of prophylactic vaccines against cervical HPV infection and diseases among women: a systematic review & meta-analysis. BMC Infect Dis. 2011;11:13 8. Haas M, Ashton T, Blum K, et al. Drugs, sex, money and power: an HPV vaccine case study. Health Policy. 2009;92(2-3): 288–295 9. Markowitz LE, Tsu V, Deeks SL, et al. Human papillomavirus vaccine introduction—the first five years. Vaccine. 2012;30(suppl 5):F139–F148 10. National Advisory Committee on Immunization (NACI). Statement on human papillomavirus vaccine. An Advisory Committee Statement (ACS). Can Commun Dis Rep. 2007;33(ACS-2):1–31
11. Rahman M, Laz TH, Berenson AB. Geographic variation in human papillomavirus vaccination uptake among young adult women in the United States during 2008-2010. Vaccine. 2013; 31(47):5495–5499 12. Colucci R, Hryniuk W, Savage C. HPV Vaccination Programs in Canada. Are we hitting the mark? In: The Cancer Advocacy Coalition of Canada, Report Card, 2008. Available at: www. canceradvocacy.ca/reportcard/2008/ HPV%20Vaccination%20Programs%20in %20Canada.pdf. Accessed March 18, 2015 13. Ogilvie G, Anderson M, Marra F, et al. A population-based evaluation of a publicly funded, school-based HPV vaccine program in British Columbia, Canada: parental factors associated with HPV vaccine receipt. PLoS Med. 2010;7(5): e1000270 14. Dorell C, Yankey D, Jeyarajah J, et al. Delay and refusal of human papillomavirus vaccine for girls, national immunization survey-teen, 2010. Clin Pediatr (Phila). 2014;53(3):261–269 15. Perkins RB, Clark JA, Apte G, et al. Missed opportunities for HPV vaccination in adolescent girls: a qualitative study. Pediatrics. 2014; 134(3). Available at: www.pediatrics.org/ cgi/content/full/134/3/e666 16. Liddon NC, Leichliter JS, Markowitz LE. Human papillomavirus vaccine and sexual behavior among adolescent and young women. Am J Prev Med. 2012; 42(1):44–52 17. Gertig DM, Brotherton JM, Budd AC, Drennan K, Chappell G, Saville AM. Impact of a population-based HPV vaccination program on cervical abnormalities: a data linkage study. BMC Med. 2013;11:227 18. Baldur-Felskov B, Dehlendorff C, Munk C, Kjaer SK. Early impact of human papillomavirus vaccination on cervical
neoplasia–nationwide follow-up of young Danish women. J Natl Cancer Inst. 2014; 106(3):djt460 19. Mahmud SM, Kliewer EV, Lambert P, Bozat-Emre S, Demers AA. Effectiveness of the quadrivalent human papillomavirus vaccine against cervical dysplasia in Manitoba, Canada. J Clin Oncol. 2014;32(5):438–443 20. Crowe E, Pandeya N, Brotherton JM, et al. Effectiveness of quadrivalent human papillomavirus vaccine for the prevention of cervical abnormalities: case-control study nested within a population based screening programme in Australia. BMJ. 2014;348: g1458 21. Leval A, Herweijer E, Ploner A, et al. Quadrivalent human papillomavirus vaccine effectiveness: a Swedish national cohort study. J Natl Cancer Inst. 2013; 105(7):469–474 22. Ministry of Health and Long-Term Care. Ontario’s HPV vaccination program. 2013. Available at: www.health.gov.on.ca/en/ ms/hpv/. Accessed March 13, 2014 23. Institute for Clinical Evaluative Sciences. Data discovery, better health. Available at: www.ices.on.ca/Data-and-Privacy/ ICES-data/Types-of-Data. Accessed March 1, 2014 24. Lipscombe LL, Gomes T, Lévesque LE, Hux JE, Juurlink DN, Alter DA. Thiazolidinediones and cardiovascular outcomes in older patients with diabetes. JAMA. 2007;298(22): 2634–2643 25. Lipscombe LL, Lévesque L, Gruneir A, et al. Antipsychotic drugs and hyperglycemia in older patients with diabetes. Arch Intern Med. 2009;169(14): 1282–1289 26. Kwong JC, Tanuseputro P, Zagorski B, Moineddin R, Chan KJ. Impact of varicella vaccination on health care outcomes in Ontario, Canada: effect of a publicly
Downloaded from pediatrics.aappublications.org at UCSF LIBRARY on April 27, 2015
PEDIATRICS Volume 135, number 5, May 2015
e1139
funded program? Vaccine. 2008;26(47): 6006–6012 27. Juurlink DN, Stukel TA, Kwong J, et al. Guillain-Barré syndrome after influenza vaccination in adults: a population-based study. Arch Intern Med. 2006;166(20): 2217–2221 28. Smith LM, Brassard P, Kwong JC, Deeks SL, Ellis AK, Lévesque LE. Factors associated with initiation and completion of the quadrivalent human papillomavirus vaccine series in an Ontario cohort of grade 8 girls. BMC Public Health. 2011;11:645 29. Smith LM, Kaufman JS, Strumpf EC, Levesque LE. Effect of human papillomavirus (HPV) vaccination on clinical indicators of sexual behaviour among adolescent girls: the Ontario Grade 8 HPV Vaccine Cohort Study. CMAJ. 2015;187(2):E74–E81 30. Lévesque LE, Smith LM, Perry AG, et al. The Ontario Grade 8 HPV Vaccine Cohort Study: A Feasibility and Validity Evaluation. In: Ontario Vaccine Sciences Symposium. Toronto, ON; November 24, 2011 31. Fine PE, Chen RT. Confounding in studies of adverse reactions to vaccines. Am J Epidemiol. 1992;136(2):121–135 32. Chen RT, Davis RL, Rhodes PH. Special methodological issues in pharmacoepidemiology studies of vaccine safety. In: Strom BL, ed. Pharmacoepidemiology. 4th ed. Sussex, UK: John Wiley & Sons, Ltd; 2005:455–485
e1140
33. Jackson LA, Jackson ML, Nelson JC, Neuzil KM, Weiss NS. Evidence of bias in estimates of influenza vaccine effectiveness in seniors. Int J Epidemiol. 2006;35(2):337–344 34. Nelson JC, Jackson ML, Weiss NS, Jackson LA. New strategies are needed to improve the accuracy of influenza vaccine effectiveness estimates among seniors. J Clin Epidemiol. 2009;62(7): 687–694 35. Hahn J, Todd P, Van der Klaauw W. Identification and estimation of treatment effects with a regressiondiscontinuity design. Econometrica. 2001; 69:201–209 36. Imbens GW, Lemieux T. Regression discontinuity designs: A guide to practice. J Econom. 2008;142:615–635 37. Cook TD. ‘‘Waiting for Life to Arrive’’: A history of the regression-discontinuity design in psychology, statistics and economics. J Econom. 2007;142:636–654
1–74. Available at: www.nber.org/papers/ w18309. Accessed March 18, 2015 40. O’Keeffe AG, Geneletti S, Baio G, Sharples LD, Nazareth I, Petersen I. Regression discontinuity designs: an approach to the evaluation of treatment efficacy in primary care using observational data. BMJ. 2014;349:g5293 41. Bor J, Moscoe E, Mutevedzi P, Newell ML, Bärnighausen T. Regression discontinuity designs in epidemiology: causal inference without randomized trials. Epidemiology. 2014;25(5): 729–737 42. Altman DG. Confidence intervals for the number needed to treat. BMJ. 1998; 317(7168):1309–1312 43. Dobson SR, McNeil S, Dionne M, et al. Immunogenicity of 2 doses of HPV vaccine in younger adolescents vs 3 doses in young women: a randomized clinical trial. JAMA. 2013;309(17): 1793–1802
38. Kadiyala S, Strumpf EC. How Effective is Population-Based Cancer Screening? Regression Discontinuity Estimates from the U.S. Guideline Screening Initiation Ages (July 1, 2011). Available at: http:// dx.doi.org/10.2139/ssrn.1893793. Accessed March 18, 2015
44. Smith LM, Lévesque LE, Nasr M, Perry AG. Validity of the Immunization Record Information System (IRIS) database for epidemiologic studies of the human papillomavirus (HPV) vaccine. Canadian Journal of Clinical Pharmacoepidemiology. 2010;17: e90–e127
39. Meyer BD, Wherry LR. Saving Teens: Using a Policy Discontinuity to Estimate the Effects of Medicaid Eligibility. Cambridge Massachusetts: The National Bureau of Economic Research; 2012:
45. Moscicki AB, Schiffman M, Kjaer S, Villa LL. Chapter 5: Updating the natural history of HPV and anogenital cancer. Vaccine. 2006;(24 suppl 3): S3/42–51
Downloaded from pediatrics.aappublications.org at UCSF LIBRARY on April 27, 2015
SMITH et al
The Early Benefits of Human Papillomavirus Vaccination on Cervical Dysplasia and Anogenital Warts Leah M. Smith, Erin C. Strumpf, Jay S. Kaufman, Aisha Lofters, Michael Schwandt and Linda E. Lévesque Pediatrics; originally published online April 27, 2015; DOI: 10.1542/peds.2014-2961 Updated Information & Services
including high resolution figures, can be found at: http://pediatrics.aappublications.org/content/early/2015/04/21 /peds.2014-2961
Supplementary Material
Supplementary material can be found at: http://pediatrics.aappublications.org/content/suppl/2015/04/2 1/peds.2014-2961.DCSupplemental.html
Permissions & Licensing
Information about reproducing this article in parts (figures, tables) or in its entirety can be found online at: http://pediatrics.aappublications.org/site/misc/Permissions.xh tml
Reprints
Information about ordering reprints can be found online: http://pediatrics.aappublications.org/site/misc/reprints.xhtml
PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly publication, it has been published continuously since 1948. PEDIATRICS is owned, published, and trademarked by the American Academy of Pediatrics, 141 Northwest Point Boulevard, Elk Grove Village, Illinois, 60007. Copyright © 2015 by the American Academy of Pediatrics. All rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275.
Downloaded from pediatrics.aappublications.org at UCSF LIBRARY on April 27, 2015
The Early Benefits of Human Papillomavirus Vaccination on Cervical Dysplasia and Anogenital Warts Leah M. Smith, Erin C. Strumpf, Jay S. Kaufman, Aisha Lofters, Michael Schwandt and Linda E. Lévesque Pediatrics; originally published online April 27, 2015; DOI: 10.1542/peds.2014-2961
The online version of this article, along with updated information and services, is located on the World Wide Web at: http://pediatrics.aappublications.org/content/early/2015/04/21/peds.2014-2961
PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly publication, it has been published continuously since 1948. PEDIATRICS is owned, published, and trademarked by the American Academy of Pediatrics, 141 Northwest Point Boulevard, Elk Grove Village, Illinois, 60007. Copyright © 2015 by the American Academy of Pediatrics. All rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275.
Downloaded from pediatrics.aappublications.org at UCSF LIBRARY on April 27, 2015
