Background
Human papillomavirus vaccination may result in lowered intention to be screened for cervical cancer, potentially leading to gaps in screening coverage and avoidable cervical cancer diagnoses.
Objective
The purpose of this study was to examine the association between human papillomavirus vaccination and subsequent cervical cancer screening initiation and adherence to recommended screening intervals to detect gaps in screening coverage and inform future prevention efforts.
Study Design
A retrospective cohort study was conducted in 2 distinct cohorts of female members of Kaiser Permanente Southern California, which is a large integrated healthcare delivery system. Papanicolaou screening initiation was evaluated in women who reached 21 years from 2010–2013. Adherence to recommended screening intervals was evaluated in women who were 25–30 years old in 2010. All women were observed to the end of 2013 for the evaluation of their screening behaviors. History of human papillomavirus vaccination and Papanicolaou screening were obtained from electronic medical records. Adherence to recommended screening intervals was measured as ≥85% vs <85% of the observed “screening up-to-date” person-time. Multivariable Cox and logistic regression models were used to examine associations between vaccination history and screening initiation and interval adherence. Demographic characteristics, gynecologic health history, healthcare use, and characteristics of women’s primary care providers were included as potential confounders in the analyses.
Results
There were 27,352 and 41,328 women included in the screening initiation and screening interval adherence analyses, respectively. In comparison with unvaccinated women, adjusted hazard ratios (95% confidence intervals [CIs]) for screening initiation among women who had been vaccinated against human papillomavirus were 1.19 (95% CI, 1.11–1.28), 1.44 (95% CI, 1.34–1.53), and 1.57 (95% CI, 1.50–1.65) for 1, 2, and ≥3 doses, respectively. Adjusted odds ratios for screening interval adherence were 0.93 (95% CI, 0.83–1.04), 1.73 (95% CI, 1.52–1.97), and 2.29 (95% CI, 2.05–2.56), for 1, 2, and ≥3 doses, respectively.
Conclusion
Women who had been vaccinated against human papillomavirus in this community-based, integrated healthcare setting were more likely to be screened for cervical cancer than were unvaccinated women. Our findings underscore the need for targeted interventions among unvaccinated women who may be disproportionally affected by cervical cancer, despite the presence of population-based screening programs.
Immunization with the human papillomavirus (HPV) vaccine does not eliminate the need for routine cervical cancer screening because currently available HPV vaccines do not offer complete protection against all oncogenic HPV types. Also, vaccination does not always occur at the recommended ages (ie, 11–12 years old), which increases the chance of exposure to oncogenic HPV before vaccination. Furthermore, many of those who receive the vaccine are not vaccinated according to the recommended dosing schedule, which raises concerns of compromised protection. As a consequence of these factors, it remains important for women, regardless of their vaccination status, to continue to undergo cervical cancer screening.
A study reported that almost all women (96%) who participated in the 2008 US Health Information National Trend Survey were aware of the need to continue cervical cancer screening after HPV vaccination. However, a previous report suggested that vaccination may lower women’s perception of their risk for the development of HPV-related diseases, affecting their motivation to be screened for cervical cancer after vaccination. This is concerning because a recent simulation study suggested that missed cervical intraepithelial neoplasia II-III cases could progress to cancer if vaccinated women are less likely to get screened. Although the findings of several surveys suggest that vaccination is generally not associated with a reduced willingness for screening, actual screening behaviors after receipt of the HPV vaccine have not been examined comprehensively. Two studies in the United States have reported a positive association between vaccination and subsequent screening, but these studies were not designed to elucidate the nature of this relationship, which requires accounting for important differences in clinical and utilization histories between women with and without vaccination. To address these gaps in the literature, we examined the relationship between HPV vaccination and subsequent screening initiation and adherence to recommended screening intervals among female members of Kaiser Permanente Southern California (KPSC). We took advantage of KPSC’s comprehensive electronic medical records to assess women’s vaccination and screening behaviors and their detailed relevant clinical characteristics.
Materials and Methods
Study setting and population
KPSC is the largest integrated healthcare delivery system in Southern California, serving >4 million members who are broadly representative of the racial/ethnic and socioeconomic diversity of the population in this geographic area. KPSC provides comprehensive healthcare services to its members; by nature of the prepaid managed care system, members have comparable access to healthcare. Preventive measures that include HPV vaccination and cervical cancer screening are offered without additional out-of-pocket cost. KPSC’s cervical cancer screening guidelines generally follow the national guidelines that recommend the initiation of screening at age 21 years. In 2004, KPSC adopted HPV cotesting for women at ≥30 years old and HPV reflex testing for women who are 21–29 years old. Subsequent screening is recommended every 3 years. This study was approved by the KSPC Institutional Review Board. The Strengthening the Reporting of Observational Studies in Epidemiology statements were followed for the preparation of this article.
A retrospective cohort study design was used to examine the association between HPV vaccination and subsequent cervical cancer screening initiation and adherence to recommended screening intervals. Two separate, nonoverlapping cohorts of women were identified to examine these behaviors between calendar years 2010 and 2013. The eligibility requirements for entry into the 2 cohorts were guided by age-related cervical cancer screening recommendations and the timing of the approval and subsequent availability of the HPV vaccine.
For the evaluation of screening initiation, all female KPSC members who reached age 21 years between 2010 and 2013 were eligible for inclusion. Follow-up evaluation started for these women at age 21 years (study baseline), which is the age for screening initiation as recommended by current clinical guidelines. To ensure complete capture of vaccination history, women were excluded if they did not have continuous KPSC membership between June 2006 when the HPV vaccine was first approved by the US Food and Drug Administration and the study baseline. Women were also excluded if, before baseline, they had a hysterectomy in which the cervix was removed, they had cervical cancer, or they had initiated cytology-based screening before the guideline-recommended age of 21 years. Women were observed from study baseline to the initiation of Papanicolaou screening, the end of the observation period (December 31, 2013), or a censoring event (ie, HPV vaccination after baseline, hysterectomy, termination of KPSC membership, or death), whichever came first.
For the evaluation of adherence to recommended screening intervals, all female KPSC members who reached any age between 25 and 30 years in 2010 were eligible for inclusion. This age range was chosen to ensure that all subjects were screening eligible (ie, ≥21 years old) and vaccine eligible (ie, ≤26 years old) at the time when the vaccine was introduced. Follow-up evaluation started for these women in year 2010 (study baseline). The same exclusion criteria described earlier were also applied to this cohort, except that women with Papanicolaou screening before baseline were not excluded. Women were followed from study baseline to December 31, 2013, or censoring events (ie, HPV vaccination after baseline, an abnormal Papanicolaou test result [which would place the woman on a surveillance schedule instead of a screening schedule], hysterectomy, termination of KPSC membership, or death), whichever came first.
Data collection
The exposure of interest was HPV vaccination and included the number of doses received. The outcomes of interest were the initiation of Papanicolaou screening and adherence to screening interval. In addition, information on the following potential confounders was collected: race/ethnicity; census block-level education and income level; Medicaid enrollment; primary care provider’s specialty and sex; healthcare use in the year before baseline, which included the number of office visits, any emergency room visit or hospitalization, and influenza vaccination; gynecologic history that included history of oral contraceptive use, pregnancy, and sexually transmitted infections (STIs), specifically chlamydia, gonorrhea, syphilis and herpes; and primary medical center. All data were obtained from KPSC’s electronic medical records that include clinical databases on immunization, laboratory tests, diagnoses, procedures, pharmacy, utilization (health care use), and providers.
Statistical analysis
The distributions of demographic and clinical characteristics of the study cohorts were calculated by HPV vaccination status; differences were assessed with the t -test or chi-square test. The incidence rate of screening initiation was calculated by HPV vaccination status. The associations between HPV vaccination status (yes vs no and by dose) and screening initiation were examined with the use of bivariate and multivariable Cox models that were adjusted for all potential confounders described earlier (categories/forms of each confounder that were adjusted for are shown in Table 1 ). Stratified analyses by race/ethnicity were also conducted.
| Variable | Women who reached age 21 years from 2010–2013 | ||
|---|---|---|---|
| Had human papillomavirus vaccine (n=17,485) | No human papillomavirus vaccine (n=9867) | P value | |
| Race/ethnicity, n (%) | <.01 | ||
| White | 4,989 (28.5) | 3081 (31.2) | |
| Black | 1,827 (10.4) | 1315 (13.3) | |
| Hispanic | 7,245 (41.4) | 2836 (28.7) | |
| Asian/Pacific Islander | 2,129 (12.2) | 1070 (10.8) | |
| Other/unknown | 1,295 (7.4) | 1565 (15.9) | |
| Census block income and education level, n (%) | |||
| Median annual household income in the census block | <.01 | ||
| ≤$45,000 | 3,825 (21.9) | 2046 (20.8) | |
| $45,001–$80,000 | 8,588 (49.1) | 4774 (48.3) | |
| >$80,000 | 5,072 (29.0) | 3047 (31.0) | |
| Percent of adults in the census block with high school degree or higher | <.01 | ||
| 0–50% | 1,109 (6.3) | 536 (5.4) | |
| 51–75% | 4,705 (26.9) | 2319 (23.6) | |
| 76–100% | 11,671 (66.7) | 7012 (71) | |
| Medicaid enrollment, n (%) | 264 (1.5) | 113 (1.1) | .01 |
| Mean length of membership, y±standard deviation | 15.0±5.3 | 14.1±5.6 | <.01 |
| Primary care provider characteristics, n (%) | |||
| Specialty | <.01 | ||
| Pediatrics | 282 (1.6) | 467 (4.7) | |
| Family medicine | 12,967 (74.2) | 6993 (70.9) | |
| Internal medicine | 3,811 (21.8) | 1986 (20.1) | |
| Other/unknown | 425 (2.4) | 421 (4.3) | |
| Female | 11,762 (67.3) | 5898 (59.8) | <.01 |
| Healthcare use within 12 mo before baseline | |||
| Any hospitalization, n (%) | 412 (2.4) | 179 (1.8) | .03 |
| Any emergency room visit, n (%) | 2,323 (13.3) | 1054 (10.7) | <.01 |
| Any outpatient visit, n (%) | 13,719 (78.5) | 6140 (62.2) | <.01 |
| No. of outpatient visits, mean±standard deviation | 0.1±0.7 | 0.1±1.0 | .20 |
| Flu vaccination within 12 mo before baseline, n (%) | 4,017 (23.0) | 986 (10.0) | <.01 |
| Gynecologic history before baseline, n (%) | |||
| Oral contraceptive use | 5,496 (31.4) | 1573 (15.9) | <.01 |
| Pregnancy | 419 (2.4) | 166 (1.7) | <.01 |
| Sexually transmitted infections | 657 (3.8) | 189 (1.9) | <.01 |
| Human papillomavirus vaccination history before baseline, n (%) | |||
| 1 Dose | 3,220 (18.4) | — | |
| 2 Doses | 3,478 (19.9) | — | |
| ≥3 Doses | 10,787 (61.7) | — | |
Although current cervical cancer screening guidelines recommend screening initiation at age 21 years, earlier guidelines also recommended screening for those sexually active for 3 years before age 21 years. Thus, in a sensitivity analysis, screening initiation at age 18 years was examined in a separate cohort of women who reached 18 years between 2010 and 2013. The same exclusion criteria were applied to this cohort as in the primary analysis. Furthermore, because the requirement of continuous membership from 2006 led to exclusion of a large proportion of initially eligible women; an additional sensitivity analysis that included all women, regardless of their length of membership, was also conducted.
A person-time–based approach for measuring adherence to preventive services, called the Prevention Index (PI) was used to evaluate adherence to cervical cancer screening intervals. Details of the PI method have been described elsewhere. Briefly, the PI represents the proportion (range, 0–100%) of “screening up-to-date” person-time in the study period. For example, Papanicolaou screening is recommended every 3 years. Thus, for a study subject, the 36 months after a Papanicolaou test is considered “up-to-date” person-time, and the time between month 37 and receipt of the next Papanicolaou test is considered “not up-to-date” person-time. For this study, the PI for Papanicolaou screening was calculated for each woman in the interval adherence cohort between 2010 and 2013, which took into account the date of their Papanicolaou screening before study baseline. The distribution of the PI was then calculated by HPV vaccination status. Because the distribution of the PI was skewed highly toward 100%, a dichotomized outcome of 85% was used as the cutoff in the analysis, which was the mean among unvaccinated women. The association between HPV vaccination (yes/no and by dose) and screening adherence was evaluated with the use of bivariate and multivariable logistic regression and adjusted for potential confounders ( Table 2 ) . Stratified analyses by race/ethnicity were also conducted. In a sensitivity analysis, the analyses were repeated and were restricted to women who did not have an abnormal Papanicolaou screening result before study baseline to exclude those women who might have received a Papanicolaou test for surveillance rather than for screening. Furthermore, an additional sensitivity analysis that included all women, regardless of their length of membership, was also conducted. All analyses were conducted with the use of SAS software (version 9.3; SAS Institute Inc, Cary, NC).
| Variable | Women aged 25–30 in 2010 | ||
|---|---|---|---|
| Had human papillomavirus vaccine (n=6951) | No human papillomavirus vaccine (n=34377) | P value | |
| Race/ethnicity, n (%) | <.01 | ||
| White | 2295 (33.0) | 9,928 (28.9) | |
| Black | 528 (7.6) | 2,889 (8.4) | |
| Hispanic | 2870 (41.3) | 16,138 (46.9) | |
| Asian/Pacific Islander | 669 (9.6) | 2,967 (8.6) | |
| Other/unknown | 589 (8.5) | 2,455 (7.1) | |
| Census block income and education level, n (%) | |||
| Median annual household income in the census block | <.01 | ||
| ≤$45,000 | 1545 (22.2) | 8,652 (25.2) | |
| $45,001–$80,000 | 3508 (50.5) | 17,576 (51.1) | |
| >$80,000 | 1898 (27.3) | 8,149 (23.7) | |
| Percent of adults in the census block with high school degree or higher | <.01 | ||
| 0–50% | 581 (8.4) | 3,686 (10.7) | |
| 51–75% | 1646 (23.7) | 8,987 (26.1) | |
| 76–100% | 4724 (68) | 21,704 (63.1) | |
| Medicaid enrollment | 248 (3.6) | 1,248 (3.6) | .80 |
| Length of membership, y±standard deviation | 10.3±8.0 | 8.4±6.6 | <.01 |
| Primary care provider characteristics, n (%) | |||
| Specialty | <.01 | ||
| Family medicine | 5335 (76.8) | 24,934 (72.5) | |
| Internal medicine | 1448 (20.8) | 8,403 (24.4) | |
| Other/unknown | 168 (2.4) | 1,040 (3.0) | |
| Female | 4311 (62.1) | 19,576 (57.3) | <.01 |
| Healthcare use within 12 mo before baseline | |||
| Any hospitalization, n (%) | 525 (7.6) | 4,345 (12.6) | <.01 |
| Any emergency room visit, n (%) | 1306 (18.8) | 5,922 (17.2) | <.01 |
| Any outpatient visit, n (%) | 6606 (95.0) | 30,629 (89.1) | <.01 |
| No. of outpatient visits, mean±standard deviation | 0.3±1.2 | 0.5±1.7 | <.01 |
| Flu vaccination within 12 mo before baseline, n (%) | 1933 (27.8) | 8,677 (25.2) | <.01 |
| Gynecologic history before baseline, n (%) | |||
| Oral contraceptive use | 4782 (68.8) | 18,518 (53.9) | <.01 |
| Pregnancy | 1659 (23.9) | 15,557 (45.3) | <.01 |
| Sexually transmitted infections | 932 (13.4) | 2,903 (8.4) | <.01 |
| Abnormal Papanicolaou results | 404 (5.8) | 1,396 (4.1) | <.01 |
| Human papillomavirus vaccination history before baseline, n (%) | |||
| 1 Dose | 1709 (24.6) | – | |
| 2 Doses | 1810 (26.0) | – | |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
