A Mail-Based HPV Self-Collection Program to Increase Cervical Cancer Screening in Appalachia: Results of a Group Randomized Trial

Abstract

Background:

Despite the promise of mail-based human papillomavirus (HPV) self-collection programs for increasing cervical cancer screening, few have been evaluated in the United States. We report the results of a mail-based HPV self-collection program for underscreened women living in Appalachia.

Methods:

We conducted a group randomized trial during 2021-2022 in the Appalachian regions of Kentucky, Ohio, Virginia, and West Virgnia. Participants were women ages 30-64 who were underscreened for cervical cancer and from a participating health system. Participants in the intervention group (n=464) were mailed an HPV self-collection kit followed by telephone-based patient navigation (if needed), and participants in the usual care group (n=338) were mailed a reminder letter to get a clinic-based cervical cancer screening test. Generalized linear mixed models compared cervical cancer screening between study groups.

Results:

Overall, 14.9% of participants in the intervention group and 5.0% of participants in the usual care group were screened for cervical cancer. The mail-based HPV self-collection intervention increased cervical cancer screening compared to the usual care group (OR=3.30, 95% CI: 1.90–5.72, p=0.005). One or more high-risk HPV types were detected in 10.5% of the returned HPV self-collection kits. Among participants in the intervention group who patient navigators attempted to contact, 44.2% were successfully reached.

Conclusions:

HPV self-collection increased cervical cancer screening, and future efforts are needed to determine how to optimize such programs, including the delivery of patient navigation services.

Impact:

Mail-based HPV self-collection programs are a viable strategy for increasing cervical cancer screening among underscreened women living in Appalachia.

Introduction

Persistent infection with high-risk human papillomavirus (HPV) types, mainly types 16 and 18, causes nearly all cases of cervical cancer (1). Cervical cancer is largely preventable through regular screening and appropriate follow-up care for precancerous lesions, and most cases of cervical cancer occur among underscreened women (2, 3). For women ages 30-65, the United States Preventive Services Task Force (USPSTF) currently recommends cytology (i.e., Pap testing) alone every three years, high-risk HPV testing alone every five years, or a combination of cytology and high-risk HPV testing every five years (4). The high-risk HPV testing included in the current recommendations involves samples collected by healthcare providers in a clinical setting, though it is also possible for women to collect their own samples (i.e., HPV self-collection). Past research has shown that the sensitivity and specificity of self-collected samples is comparable to provider-collected samples for detecting cervical disease (5, 6). HPV self-collection is currently not a recommended cervical cancer screening strategy in the US, though in May 2024, the US Food and Drug Administration approved HPV self-collection in a healthcare setting (i.e., women would self-collect a sample in a healthcare setting) as a cervical cancer screening option (7). The National Cancer Institute also recently started the Last Mile Initiative to provide additional evidence about the accuracy and clinical effectiveness of HPV self-collection (8).

In addition to HPV self-collection in a healthcare setting, another approach for using HPV self-collection to increase cervical cancer screening is through mail-based programs. In such programs, women are sent an HPV self-collection kit in the mail, asked to collect a cervicovaginal sample at home by themselves, and then mail the sample back to be tested for the detection of HPV. Mail-based HPV self-collection programs conducted outside of the US have been successful, with pooled estimates indicating that about 20% of women who were sent an HPV self-collection kit got screened for cervical cancer (9, 10). Most of these women were screened by completing and returning their HPV self-collection kit, with fewer becoming screened by receiving a clinic-based test after a self-collection kit was mailed to them (9, 10). Given this success, multiple countries have integrated mail-based HPV self-collection into national cervical cancer screening programs (11).

In the US, formative research has showed that most women would be willing to complete an HPV self-collection kit at home and return it via mail (12–15). Studies then implemented mail-based programs in the US and have produced promising yet varied results for underscreened women (16–22). One potential reason for this variability is differences in how the programs were implemented. Some programs took an “opt-in” approach for the distribution of HPV self-collection kits and often included research-focused interactions (e.g., completion of a consent form, a baseline survey) prior to kit distribution. Cervical cancer screening in these programs ranged from 64% to 80% (16–20). Other programs took a “mail-to-all” approach and distributed HPV self-collection kits without research-focused interactions. Cervical cancer screening in these programs was less than 40% (21, 22). The latter approach may better align with how mail-based HPV self-collection programs would occur in a clinical setting in the US (i.e., outside of a research study), so it is important that additional programs with this approach are evaluated.

The current study reports the results of a group randomized trial of the HOME (Health Outcomes through Motivation and Education) Initiative, which was a mail-based HPV self-collection program for underscreened women in four Appalachian states. This program used a mail-to-all approach for distributing HPV self-collection kits to women. Results will therefore not only add to the evidence base for this type of mail-based HPV self-collection program in the US but will also help guide the implementation of future programs.

Materials and Methods

Overview

The HOME Initiative was one of three initiatives included in a larger research effort, called Take CARE (Clinical Avenues to Reach Health Equity), that was designed to address the cervical cancer burden in the Appalachian region of four states (Kentucky, Ohio, Virginia, and West Virgnia). Appalachia is a largely rural geographic region in the US that is underserved and has existing disparities related to cervical cancer (23–25). For example, a report showed the cervical cancer incidence rates in the Appalachian regions of Kentucky (12.4 cases per 100,000 population), Ohio (10.9 cases per 100,000 population), and West Virginia (11.1 cases per 100,000 population) were all higher than the national rate (7.7 cases per 100,000 population) for the same data years (24). The goal of the HOME Initiative was to implement and evaluate a mail-based HPV self-collection program to increase cervical cancer screening in Appalachia via a group randomized trial. The program described in this report was implemented during 2021-2022.

Take CARE partnered with ten health systems with health center locations in the Appalachian regions of Kentucky, Ohio, Virginia, and West Virgnia. Nine of these health systems participated in this portion of the HOME Initiative, with one health system not taking part primarily due to challenges stemming from the coronavirus disease 2019 (COVID-19) pandemic. For Take CARE, health systems were stratified by state and then randomized in a 1:1 ratio to one of two study groups: the intervention group or the usual care group. All health center locations within a given health system were randomized to the same study group. The Institutional Review Board (IRB) at The Ohio State University approved this study, including a waiver of informed consent, and served as the IRB of record. The randomized trial is registered at ClinicalTrials.gov (NCT04411849) and reporting was based on the CONSORT guidelines.

Participants

Health systems used medical records to identify individuals meeting the following eligibility criteria: a) female; b) ages 30-64; c) not within cervical cancer screening guidelines for this age range (i.e., no Pap test in the last three years or clinic-based high-risk HPV test in the last five years); d) not pregnant; e) intact cervix; f) no history of invasive cervical cancer; g) seen in the health system within the last two years; and h) had a working telephone. Age 64 was the upper age limit so women did not “age out” of the recommended ages for cervical cancer screening (4) during the study.

We randomly sampled potentially eligible individuals from de-identified lists created by the health systems. Each randomly sampled person was sent a study eligibility mailing. If an individual did not meet the above eligibility criteria, they were asked to return a form indicating ineligibility. Individuals were considered to be confirmed eligible and participants in the study, unless an ineligibility form was received. A total of 464 women from health systems in the intervention group and 338 women from health systems in the usual care group participated in this study and were sent the study materials described below. Participants were not aware of which study group their health system had been randomized to.

Study Materials

Intervention Group

Participants from health systems in the intervention group were mailed an HPV self-collection kit. The self-collection device included in the kit was a brush developed by Preventive Oncology International (POI) (26), and the device packaging included instructions for use and return. Additional materials included in the kit were an introductory brochure about HPV self-collection (that addressed common questions/concerns women in Appalachia previously reported about self-collection (20, 27)), an information booklet about cervical cancer that was based on materials from the Centers for Disease Control and Prevention (28), and a postage-paid box for returning the device after use. These materials were reviewed for clarity and appropriateness during focus groups of community members in each state prior to the group randomized trial. The HPV self-collection kit and laboratory testing (described below) were free of charge to participants, and no incentive was provided to participants for returning their kit.

Participants who had not returned their HPV self-collection kit within two weeks of distribution were contacted by a patient navigator (PN) to remind them about the kit and assist with kit return, as patient navigation can improve cancer screening behaviors (29). PNs were existing health system staff who had been trained by the study team. PNs attempted to contact participants by telephone, making up to three contact attempts. As needed, PNs provided information about cervical cancer screening and HPV self-collection, answered participants’ questions about their HPV self-collection kit, and provided social support. To help ensure consistency, PNs followed scripts developed by the study team during phone calls.

All returned HPV self-collection kits were sent to Atila Biosystems (Mountain View, CA) to be tested for the detection of high-risk HPV types. The laboratory used the Atila AmpFire HPV test that can detect 15 high-risk HPV types (16, 18, 31, 33, 35, 39, 45, 51, 52, 53, 56, 58, 59, 66 and 68) (30). For each participant who returned an HPV self-collection kit, health systems mailed them a letter that included the HPV detection results and an interpretation of the results. Results in the letter were categorized as either: (a) one or more high-risk HPV types detected; (b) high-risk HPV types not detected; or (c) inadequate sample. For participants who had one or more high-risk HPV types detected, the letter indicated the importance of receiving follow-up care and that a PN would contact them to help schedule a follow-up appointment. For participants who did not have high-risk HPV types detected, the letter indicated that no follow-up care is needed and encouraged them to continue to get screened for cervical cancer in the future. Any participant who collected an inadequate sample was offered a replacement kit.

Usual Care Group

Participants from health systems in the usual care group were mailed a reminder letter to get a clinic-based cervical cancer screening test and an information booklet about cervical cancer (i.e., the same booklet as the intervention group). We considered a reminder letter to be “usual care” since many health systems regularly use reminder letters for cervical cancer screening (31).

Measures

Our primary outcome was whether participants got screened for cervical cancer within approximately six months after study materials were sent. Using criteria similar to those from another recent HPV self-collection trial in the US (18), each participant was categorized as “screened” or “not screened.” We categorized a participant as “screened” if they met at least one of the following criteria: a) returned an HPV self-collection kit and did not have a high-risk HPV type detected (intervention group only); b) returned an HPV self-collection kit, had one or more high-risk HPV types detected, and then attended a follow-up appointment (intervention group only); or c) received a clinic-based screening test (intervention group and usual care group). For participants who returned an HPV self-collection kit and had one or more high-risk HPV types detected, we required them to attend a follow-up appointment to be categorized as “screened” given the importance of these women receiving additional care. All participants not meeting one or more of the above criteria were categorized as “not screened.” Data on clinic-based screening and follow-up appointments were based on medical records, with this information abstracted by staff members at the participating health systems using a de-identified data collection form that was created by the study team.

During the patient navigation process, PNs documented information about their contact with participants and any actions taken by the PNs during these interactions. These data were recorded by the PNs via an online form created by the study team. Lastly, health systems completed a survey prior to the implementation of the mail-based HPV self-collection program that examined health system-level characteristics. To help ensure confidentiality, the study team did not have access to individual-level data from medical records on the demographic characteristics of participants.

Data Analysis

Using an intent-to-treat approach, we compared cervical cancer screening between the two study groups. We used a generalized linear mixed model (GLMM) with a logit link to make this comparison, as this model was able to account for the correlation between participants from the same health system (32). The GLMM produced an odds ratio (OR) and 95% confidence interval (CIs). The study was designed to have 90% power assuming a screening prevalence of 10% in the usual care group, a screening prevalence of 25% in the intervention group, an intraclass correlation of 0.017 for outcomes from the same health system, and a two-tailed alpha of 0.05. Lastly, we descriptively examined data from the patient navigation process and characteristics of the health systems. All analyses were conducted using Stata version 15.0 (Statacorp, College Station, TX). Given the study’s focus on cervical cancer screening and that all participants were identified as female in health systems’ medical records, sex as a biological variable was not integrated into the analytic plan.

Data Availability

Deidentified data may be made available upon request.

Results

Characteristics of Health Systems

Across the nine participating health systems, the average number of physicians per system was seven and the average number of other healthcare providers (e.g., physician assistants, nurse practitioners, nurses) per system was 33 (Table 1). The average number of female patients ages 30-64 (i.e., the study’s age range) per system was 3,453, and the health systems reported on average that 28% of their patients had private health insurance. The health systems indicated an average wait time of five days for women to get an appointment to receive a clinic-based cervical cancer screening test. All health systems indicated that they offered HPV testing (on samples collected by healthcare providers) as part of their current cervical cancer screening process.

Table 1.

Characteristics of the participating health systems

	Intervention Systems (n=5)	Usual Care Systems (n=4)	Total (n=9)
	Mean (min., max.)	Mean (min., max.)	Mean (min., max.)
Number of physicians	8 (1, 19)	6 (2, 13)	7 (1, 19)
Number of other healthcare providers (e.g., physician assistants, nurses)	32 (13, 66)	34 (15, 61)	33 (13, 66)
Number of female patients ages 30-64 (i.e., the study’s age range)	3016 (796, 7718)	3999 (1525, 7422)	3453 (796, 7718)
Percent of patients who have private health insurance	27 (18, 40)	28 (25, 34)	28 (18, 40)
Average wait time in days for women to get an appointment for cervical cancer screening	6 (4, 7)	4 (0, 7)	5 (0, 7)
Average wait time in days for women to get an appointment for follow-up care after an abnormal cervical cancer screening test	5 (4, 7)	3 (1, 5)	4 (1, 7)

Note. Table reports the mean (minimum value [min.], maximum value [max.]) for each health system-level characteristic. One additional health system from the larger Take CARE (Clinical Avenues to Reach Health Equity) research effort did not take part in this study primarily due to challenges from the coronavirus disease 2019 (COVID-19) pandemic. Data for this health system are not included in the table.

Cervical Cancer Screening

Among participants in the usual care group, 5.0% (17/338) were categorized as screened for cervical cancer by receiving a clinic-based screening test (Figure 1). Among participants in the intervention group, 14.9% (69/464) were categorized as screened for cervical cancer. The 69 participants screened in the intervention groups included: a) 50 participants who returned their HPV self-collection kit and did not have a high-risk HPV type detected in their sample; b) two participants who returned their HPV self-collection kit, had one or more high-risk HPV types detected in their sample, and then attended a follow-up appointment; and c) 17 participants who did not return their HPV self-collection kit but received a clinic-based screening test. The mail-based HPV self-collection intervention increased cervical cancer screening in comparison to the usual care group (OR=3.30, 95% CI: 1.90–5.72, p=0.005).

Figure 1.

Cervical cancer screening among the intervention group and the usual care group.

Figure 1 shows the percent of participants in the intervention group (n=464) and the usual care group (n=338) who were categorized as screened for cervical cancer for the HOME (Health Outcomes through Motivation and Education) Initiative.

HPV Self-Collection Kit Return and HPV Detection

Overall, 12.3% (57/464) of participants in the intervention group returned their HPV self-collection kit. This included the 52 participants described above who returned their kits and were subsequently categorized as screened because a high-risk HPV type was not detected or because a high-risk HPV type was detected and a follow-up appointment was attended, as well as five additional participants. Of these five additional participants, four returned their HPV self-collection kit, had one or more high-risk HPV types detected in their sample, but did not attend a follow-up appointment; and one who returned an HPV self-collection kit but had an inadequate sample (and did not subsequently return a replacement HPV self-collection kit). These additional five participants were not categorized as screened since they did not meet one or more of the criteria for our primary outcome.

Among the 57 participants who returned their HPV self-collection kit, the median time from kit distribution to sample return was 21 days (mean=40.6 days). Among these 57 participants, one or more high-risk HPV types was detected in six samples (10.5%). This included type 16 (two samples), type 53 (one sample), type 59 (one sample), types 18 and 56 (one sample), and types 58 and 59 (one sample).

Patient Navigation

Patient navigation was attempted for 412 participants in the intervention group. A total of 182 (44.2%) of these participants were successfully contacted at least once by a PN, with a few contacted more than once, resulting in a total of 190 unique interactions across the 182 contacted participants. The most common actions taken by PNs during interactions were providing education about cervical cancer screening/HPV self-collection (28.9%), advising participants about how to complete or return the HPV self-collection kit (13.7%), and providing help with appointments at a health center (4.2%). A total of 13 participants returned their HPV self-collection kit after a PN had successfully contacted them or left a message for them.

Discussion

The HOME Initiative used a mail-to-all approach to test an HPV self-collection program among underscreened women in Appalachia. To our knowledge, our program was the first of its type to be implemented in Appalachia, a largely rural geographic area with existing cervical cancer disparities (23–25). The HOME Initiative increased cervical cancer screening in comparison to a usual care group, with about 15% of women in the intervention group getting screened and about 5% of women in the usual care group getting screened. Similar to past studies (9, 10), most women in the intervention group became screened by returning their HPV self-collection kit, with a smaller number receiving a clinic-based screening test. About 10% of women who returned an HPV self-collection kit had one or more high-risk HPV types detected in their sample, which aligns well with a recent meta-analysis that reported a pooled prevalence estimate of 11.1% for high-risk HPV in self-collection studies (10). Also similar to past studies (10), nearly all women who returned an HPV self-collection kit were able to successfully collect an adequate specimen. Taken together, our results provide further support that HPV self-collection is feasible for women to perform and that mail-based self-collection programs can increase cervical cancer screening.

In comparing the magnitude of our increase in cervical cancer screening with results from other mail-based HPV self-collection programs, it is necessary to consider study design (i.e., mail-to-all vs. opt-in approaches). Our increase in screening is fairly comparable to those from studies from both within and outside of the US that also took a mail-to-all approach for distributing HPV self-collection kits (9, 10, 22). For example, two studies that were conducted with an integrated healthcare delivery system in the western US reported that their mail-to-all programs increased cervical cancer screening, in comparison to a control group, by 17% and 10%, respectively (21, 22). In contrast, the increase found in our study was not nearly as large as those reported for US studies that used an opt-in approach (16–20). However, the denominators for these opt-in studies were the number of women who chose to receive a self-collection kit and often completed other research-focused interactions (e.g., completion of a consent form) prior to kit distribution, and not the larger number of women who were likely eligible but either did not respond to recruitment efforts, chose not to enroll in the study, or did not complete required research-focused interactions. To account for differences in study design, studies have directly compared mail-to-all and opt-in approaches and have typically found that mail-to-all approaches are more successful in increasing cervical cancer screening (10, 33), though the success of opt-in approaches may vary based on the mode of communication that women use to opt-in (e.g., mail, online) (34, 35). Future efforts are needed to further understand how to optimize the design of mail-based HPV self-collection programs while also considering other factors that can impact such programs (e.g., cost, logistical and personnel requirements, sustainability).

It is important to also consider the COVID-19 pandemic when interpreting the results of the current study. Our trial began during a timeframe when several clinic-based cancer screening rates, including those for cervical cancer screening, remained below pre-pandemic rates (36). This suggests that the pandemic may have hindered clinic-based screening in our study to some degree. It is less clear how the pandemic potentially affected women’s decisions about whether to complete an HPV self-collection kit in our study. At-home stool tests for colorectal cancer screening increased during our study’s timeframe (36), but people in the US may be more familiar and comfortable with the concept of at-home stool tests compared to HPV self-collection. Lastly, the COVID-19 pandemic may have affected attendance at follow-up appointments among women who had a positive self-collection test result, as only two of six HPV-positive participants in our study attended a follow-up appointment based on their medical records. We acknowledge this sample size is modest, but we observed follow-up adherence among HPV-positive women that was much lower than what was reported in a recent meta-analysis (pooled estimate of follow-up adherence=79.0%) (10). However, there is variability in follow-up adherence across studies and our estimate of adherence is actually quite similar to a past study among another underserved population living in a mostly rural area in the US (i.e., African American women living in the Mississippi Delta region) (37). As the delivery of healthcare services continues to rebound from the pandemic, it will be important for future mail-based HPV self-collection programs in rural areas to closely monitor follow-up adherence and determine if our finding was driven in large part by the COVID-19 pandemic or if follow-up adherence is an ongoing issue in such areas. If it appears to be an ongoing issue, it will then become critical to identify strategies for improving follow-up adherence among HPV-positive women in rural areas.

Fewer than half of participants in the intervention group were successfully contacted by a patient navigator, which we think may be attributable to a few reasons. First, patient navigation occurred by telephone, and the ability to successfully contact participants is a common challenge with phone-based navigation (38). Second, the PNs began attempting to contact participants about two weeks after kit distribution. Some participants may have already discarded their kit or forgotten about receiving the kit by this time and therefore not responded to contact attempts by the PN. Third, the materials sent with the kit included information that may have answered a lot of the questions and concerns that participants had about HPV self-collection, thereby reducing women’s perceived need to speak with a PN. The materials were based on those used successfully in a past HPV self-collection program (20) and further revised for this trial based on input from community members. Given that patient navigation has been shown to improve health outcomes across the cancer continuum (29), future efforts should examine how to improve the delivery of patient navigation services in mail-based HPV self-collection programs, including determining if additional modes of contact (e.g., patient portal platforms, text message) reach more participants and determining when contact attempts should optimally begin in relation to kit distribution.

There are some potential limitations to the current study. The study team did not have access to individual-level data from medical records on the demographic characteristics of participants, and it is possible that some medical records information accessed by health systems was not fully up-to-date (e.g., participants’ mailing addresses and phone numbers). It is also possible that some participants received a clinic-based cervical cancer screening test or attended a follow-up appointment at a different healthcare system and this information was not captured in the outcome data. All participants had been seen in a participating health system within the last two years, although the study team did not have data on the number or recency of visits within this timeframe. Future efforts should attempt to also include women who may not have a medical home or have not recently engaged with a healthcare system. This includes determining how patient navigation can be used to reach such women to help increase cervical cancer screening and appropriate follow-up care, if needed. Study materials were only available in English, though over 90% of residents of this geographic area report speaking English “very well” or better (39).

There is increasing interest in HPV self-collection as a cervical cancer screening approach in the US. Results of the current study show that a mail-based HPV self-collection program is a viable strategy for increasing cervical cancer screening among underscreened women living in the Appalachian region of the US. Future efforts are needed to determine how to better implement such programs, including the implications of using a mail-to-all or opt-in approach for distributing HPV self-collection kits and how patient navigation services can be improved. This information can then help optimize the design and implementation of future mail-based HPV self-collection programs in the US.