Abstract

Background

Human papillomavirus (HPV) self-sampling is recognized as a feasible option for enhancing screening for cervical cancer, particularly among hard-to-reach women. The magnitude of the effectiveness of screening participation under different invitation strategies was reported. This review seeks to compare the effectiveness of invitation strategies in increasing screening participation of HPV self-sampling across diverse study settings.

Methods

A systematic literature search was conducted in Embase, MEDLINE, and PubMed in April 2023. Articles were included if (1) their target participants were aged between 25 and 70 years; (2) participants in the intervention arm were randomized to receive HPV self-sampling devices through various invitation strategies; (3) participants in the control arm who either received invitations for cervical cancer screening other than HPV self-sampling or opportunistic screening as usual care; (4) studies that provided sufficient data on screening participation in HPV self-sampling as outcome measured. The study design of the included articles was limited to randomized controlled trials.

Results

A total of 15 articles were included in this review. Invitation strategies of disseminating HPV self-sampling devices included opt-out and opt-in. Meta-analysis revealed screening participation in the self-sampling group was significantly greater than control arm (OR 3.43, 95% CI 1.59鈥�7.38), irrespective of the invitation strategy employed. Among invitation strategies, opt-out appeared to be more effective on increasing screening participation, compared to control and opt-in strategy (opt-out vs. control OR 3.91, 95% CI 1.82鈥�8.42; opt-in vs. control OR 1.34, 95% CI 0.28鈥�6.39).

Conclusions

Opt-out strategy is more successful at improving screening participation compared to opt-in and routine invitation to cervical screening. It is therefore a promising way to improve participation in cervical cancer screening. The findings of this review provide important inputs to optimize strategies for inviting women to participate in vaginal HPV self-sampling across the study setting, thus improving participation in cervical cancer screening.

Background

Cervical cancer is a highly preventable malignancy among women with vaccination and regular screening. The nationwide cervical screening programs in developed countries has led to a significant decline in both of its incidence and mortality. [1, 2] However, it is still ranked the fourth most common female cancer in the world due to low screening coverage. [2, 3] Suboptimal screening participation remains a global challenge to the success of screening programme and the majority of cervical cancer occur in women who are either never-screened or underscreened [3]. The causal relationship between persistent infection with high-risk human papillomavirus (hrHPV) and the development of cervical cancer was well established [4], Human papillomavirus (HPV) testing has therefore become an emerging option for cervical cancer prevention in the past decade as the timely detection of hrHPV infection enables early management of possible cervical intraepithelial lesions and prevent progression of cervical cancer [1, 5]. HPV testing is performed through clinician-collected or self-collected; the latter is known as vaginal HPV self-sampling which involves the use of a self-sampling device for collecting cervicovaginal samples by women themselves.

Recently, vaginal HPV self-sampling is promoted to tackle the barriers to conventional cytology and clinician-collected sampling, such as psychological distress, invasion of privacy, cost of attending clinic [3]. Studies have shown that the sensitivity of vaginal HPV self-sampling is similar to clinician-collected sampling in detecting cervical intraepithelial neoplasia grade two or worse (CIN2+) when polymerase chain reaction (PCR) based tests are used [5,6,7]. The demonstrated accuracy of vaginal HPV self-sampling underpins its potential by adopting HPV testing as primary screening test [3]. Compared to clinician-collected HPV testing, women offered vaginal HPV self-sampling have shown greater participation, particularly among hard-to-reach populations [8,9,10]. A systematic review in 2024 showed there was a high acceptability, feasibility and sustainability by adopting vaginal HPV self-sampling as an alternative cervical cancer screening tool [11]. Besides, local studies have suggested that the low awareness and literacy of HPV self-sampling was the possible hurdle to increase the screening participation [12]. Invitation strategies of vaginal HPV self-sampling are therefore essential for successful cervical screening programme. Existing invitation strategies were generally categorized into 鈥渙pt-out鈥� and 鈥渙pt-in鈥�. The former was also known as 鈥渟end-to-all鈥�, involving direct dissemination of self-sampling devices to all target women for cervicovaginal sampling, whilst the latter required women to indicate their request to obtain sampling devices. Current evidence revealed that various invitation strategies for disseminating HPV self-sampling devices resulted in substantial variation in screening participation. Among studies involving multiple arms for various invitation strategies, it revealed that opt-out strategy potentially resulted in higher screening participation of cervical screening, compared to opt-in [13, 14]. This study aimed to assess the impact of invitation strategies on screening participation and to identify the most effective strategy on increasing screening participation across the study settings. The findings provide insights into informing the tailored policy of cervical cancer prevention and combating cervical cancer through improving its screening participation.

Methods

Search strategy and selection criteria

A systematic literature search was conducted through electronic databases, including Embase, MEDLINE and PubMed, in April 2023. Specific search terms and MeSH headings were employed in individual databases to ensure a robust search of the literature and capture the conceptual ideas of the research topic. The search terms included HPV, cervical cancer, screening, DNA testing, self-sampling, self-collection specimen, uptake, participation and coverage. The full search strategy was shown in appendix A. Studies were considered when the following eligible criteria were fulfilled: women aged between 25 and 70 years; participants in the intervention arm who were randomized to receive self-sampling devices through various invitation strategies for vaginal HPV testing; participants in the control arm who either received invitations for cervical cancer screening other than vaginal HPV self-sampling or opportunistic screening as usual care; studies that provided sufficient data on screening participation in vaginal HPV self-sampling as outcome measured. The study design of the included studies was restricted to randomized controlled trials as it is the most robust for establishing causal relationships. The inclusion criteria were not limited by the study setting or language. To ensure the inclusion of the most recent evidence on the effectiveness of HPV self-sampling for improving screening participation, only articles published from 1 January 2013 to 31 December 2023 were considered. Two authors independently reviewed titles and abstracts to identify studies for full-text screening. The full text of all studies independently reviewed for their eligibility. Throughout the screening process, discrepancies in the screening decisions between the two authors were resolved by discussion and consensus. Pre-specified analysis was adopted in this review by outlining the objectives, outcomes and anticipated statistical procedures in this review prior to the implementation. Random-effects meta-analysis was used to estimate the effect of screening participation across different invitation strategies. All statistical analyses were performed using Review Manager (RevMan). Cochrane Risk of Bias 2 (RoB 2) tool for randomized trials was adopted for quality assessment of the included studies [15].

Research question

In this review, screening participation of HPV self-sampling refers to the return of collected cervicovaginal specimens within the designated period of an individual study, while invitation strategy is defined as the process of disseminating materials to facilitate women鈥檚 participation in HPV self-sampling. Our research question is as follows: 鈥淲hich of the invitation strategies are considered the most effective on increasing participation in cervical cancer screening with respect to the study setting?鈥� This systematic review was presented in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) statement 2020 [16].

Results

We identified a total of 477 records, including 466 records from electronic databases and 11 records from manual searching. There were 154 relevant records extracted after titles and abstracts screening. Seventy-eight records were excluded for irrelevant variables according to the research question, including not provide sufficient data on screening participation of HPV self-sampling (n鈥�=鈥�27), study designs other than randomized controlled trials (n鈥�=鈥�48), urine HPV self-sampling (n鈥�=鈥�3). At the completion of full-text screening, sixty-one records were further excluded as 48 of those did not fulfil eligibility criteria of this review and the remaining 13 articles did not report specific data on screening participation of HPV self-sampling. A total of 15 articles were eventually included in this review [13, 14, 17,18,19,20,21,22,23,24,25,26,27,28,29]. The flowchart of the retrieval of records in the PRISMA diagram was presented (Fig.听1).

Study characteristics

The 15 studies included 94,908 participants, with individual sample sizes ranging from 400 to 19,851. All the included studies were published between 2013 and 2022. Twelve of these studies were conducted in high-income countries, including six each in European and Western countries. The latter included Argentina, Australia, Canada, United Kingdom and United States. The remaining three studies were conducted in low- and middle-income countries (LMICs) which were Ethiopia, Nigeria and Uganda. Studies adopting opt-out strategy were primarily conducted in Western countries (n鈥�=鈥�5, 33.3%), following by Europe (n鈥�=鈥�3, 20.0%) and LMICs (n鈥�=鈥�3, 20.0%). Conversely, there was only one study in Western countries adopted an opt-in approach only [20]. Studies with multiple arms involving both opt-out and opt-in approaches were exclusively conducted in Europe (n鈥�=鈥�3, 20.0%). Of the included studies, participants from 10 studies (66.7%) targeted nonattenders, under screened and never-screened, while participants in the remaining 5 studies (33.3%) were recruited from general population. One study conducted in LMICs targeted women with low socioeconomic status [21]. The age of the study participants ranged from 25 to 70 years. The mean age of participants from 6 of the included studies was 47.16 while the remaining studies reported proportion of participants in each age group.

Fig. 1

Study flow chart. The process of article selection was illustrated in accordance with the PRISMA flow diagram 2020. A literature search was conducted in selected electronic databases and manual searching, with 477 records initially identified. Fifteen randomized trials were included in the review

All participants in the self-sampling arm were provided with HPV self-sampling devices through various invitation strategies, including opt-out (send-to-all) and opt-in. Over 70% of the included studies (n鈥�=鈥�11, 73.3%) employed opt-out strategy by directly offering participants with self-collection devices through direct mail (n鈥�=鈥�6, 54.5%) or door-to-door method (n鈥�=鈥�4, 36.4%) by research assistants, community health workers, healthcare professionals, or outreach workers. However, one study unclearly reported how the participants in self-sampling arm were provided with devices under opt-out strategy [21]. Only one study (6.67%) employed opt-in strategy, where participants are required to fill in postal order form to obtain the devices [20]. Additionally, three studies (20%) involved multiple arms, by including opt-in and opt-out strategies into intervention arms separately. Of these opt-out arms, devices were exclusively offered to participants through direct mail while participants in control arm were provided with various channels for placing their orders, including online platforms (email and website), mail, phone and short message service. Among the included studies, the opt-out strategy was identified as the most common method for disseminating self-sampling devices. The characteristics of included studies were presented in Table听1.

Table 1 Characteristics of included studies

Various types of self-sampling devices were utilized, including swabs [17, 20,21,22,23,24,25, 27, 28] (n鈥�=鈥�9, 60%), brushes [13, 17,18,19, 26] (n鈥�=鈥�5, 33.3%), and lavage-based devices [18] (n鈥�=鈥�1, 6.67%). Among these, one study employed more than one type of sampling device [18]. Nevertheless, one study did not explicitly specify the type of sampling device used [29]. Of the included studies, women from self-sampling arms primarily performed self-sampling at home, except for one study where women performed in a private area of a health post [19]. Participants in the control arm were primarily provided with invitation for cytology screening as usual care. One study conducted in LMICs provided participants with the option for hospital-based sampling by trained nurses [21] whereas women from two other studies conducted in LMICs were offered with appointments for visual inspection with acetic acid (VIA) [19, 22]. The length of measuring screening participation of HPV self-sampling ranged from 3 to 12 months. Five of the included studies (33.3%) assessed participation of cervical cancer screening within 6 months after invitation letter was sent [13, 17, 25, 26, 28] whilst six studies accepted screening participation within the study period [19, 22,23,24, 27, 29]. Only two studies measured screening participation at multiple time points [20, 25].

The overall methodological quality of the 15 included studies was moderate to good according to the RoB 2 tool [15]. The risk of bias in the randomization process was generally low, while four studies (26.7%) were categorized as some concerns. One study mentioned to adopt Zelen鈥檚 design for randomization without detailed discussion [20]. Contamination between treatment groups likely occurs as participants were fully aware of treatment allocation. Another study did not include details of randomization and selection bias would be a concern [24]. Simple randomization was employed to two intervention arms in 1:1 ratio in one study. However, the allocation concealment and blinding were unclearly discussed [23]. All included studies were deemed low risk in remaining four domains, including deviations from the intended interventions, missing outcome data, measurement of the outcome and selection of the reported result. The findings of the quality assessment of the included studies and reported outcomes are presented in Table听2.

Table 2 Quality assessment of all included studies

Impact of invitation strategy on screening participation

Fourteen of the included studies (93.3%) compared the effectiveness of opt-out strategy on disseminating devices between self-sampling arms and control arms. Of these studies adopting opt-out strategy, irrespective of the approach of disseminating devices, it was shown that participants in self-sampling arm generally demonstrated greater screening participation than control arm. The pooled odds ratio of screening participation between self-sampling arm and control arm under opt-out strategy was 3.91, 95% CI 1.82鈥�8.42 (Fig.听3), indicating participants in self-sampling arms receiving devices under opt-out strategy were approximately 4-fold greater in the likelihood of performing cervical cancer screening compared to control arms. Under opt-out strategy, the dissemination approach can be further categorized into direct mailing and door-to-door. The pooled odds ratio of screening participation between self-sampling arm and control arm was 2.28, 95% CI 1.03鈥�5.04 (direct mailing vs. control) and 9.64, 95% CI 2.51-37.00 (door-to-door vs. control) respectively (Figs.听4 and 5). Door-to-door dissemination through opt-out strategy was therefore demonstrated to be more effective in increasing screening participation compared to direct-mailing.

Four of the included studies compared the effectiveness of opt-in strategy on disseminating devices between self-sampling arm and control arm. Regardless of the availability of channels for ordering devices, screening participation among women in self-sampling arm were generally greater than control arm. The pooled odds ratio of screening participation between self-sampling arm and control arm under opt-in strategy was 1.34, 95% CI 0.28鈥�6.39 (Fig.听6). As the confidence interval crossed 1, there was no statistically significant difference in screening participation between opt-in arms and control arms.

Fig. 2

Forest plot for the odds ratios of screening participation between self-sampling arm and control arm. The overall pooled effect of 15 studies was OR3.43, 95% CI 1.59鈥�7.38, indicating that women from self-sampling arm receiving devices through various invitation strategies were 3.4-fold greater in the likelihood of performing cervical cancer screening than control arm

Fig. 3

Forest plot for the odds ratios of screening participation between opt-out arm and control arm. The overall pooled effect of 14 studies was OR3.91, 95% CI 1.82鈥�8.42, indicating that women from self-sampling arm receiving devices through opt-out strategy were approximately 4-fold greater in the likelihood of performing cervical cancer screening than control arm

Fig. 4

Forest plot of the odds ratios of screening participation between direct-mailing (opt-out) arm and control arm. The overall pooled effect of the 9 studies was OR2.28 (95% CI 1.03鈥�5.04), indicating that women from direct-mailing arm receiving devices through opt-out strategy were approximately 2.3-fold greater in the likelihood of performing cervical cancer screening than control arm

Fig. 5

Forest plot of the odds ratios of screening participation between door-to-door (opt-out) arm and control arm. The overall pooled effect of the 4 studies was OR9.64 (95% CI 2.51-37.00), indicating that women from door-to-door arm receiving devices through opt-out strategy were 9.6-fold greater in the likelihood of performing cervical cancer screening than control arm

Fig. 6

Forest plot of the odds ratios of screening participation between opt-in arm and control arm. The overall pooled effect of the 4 studies was OR1.34 (95% CI 0.28鈥�6.39), indicating thatthere was no statistically significant difference in screening participation between opt-in arm and control arm

Impact of HPV self-sampling on screening participation

Women in self-sampling arm generally demonstrated greater screening participation than women in the control arm, except for the findings of one study conducted by Zehbe, Jackson [27] which reported that there was no significant difference in screening participation (p鈥�=鈥�0.628) between the self-sampling arm and the control arm, with values of 20.0% and 14.3%, respectively. Among the included studies, the pooled odds ratio of screening participation in self-sampling arms compared to control arms was 3.43, 95% CI 1.59鈥�7.38 (Fig.听2), indicating that the likelihood of performing cervical cancer screening of women in the self-sampling arms was 3.4-fold greater than the control arms.

Impact of demographics on screening participation

Ten of the included studies (66.7%) reported socioeconomic data, including income level, employment status and health insurance status of participants as baseline demographics. Regarding the predictors of screening participation of cervical screening services, Modibbo, Iregbu [18] indicated that socioeconomic status and educational level were not significantly associated with screening participation, with reported p鈥塿alues of 0.861 and 0.894, respectively. Of the included studies, 12 studies recruited participants aged 30 years or above, while participants from the remaining 3 studies were aged 25 years or above. The pattern of screening participation of HPV self-sampling across age groups varied. Nearly half of included studies (n鈥�=鈥�7, 46.7%) reported middle-aged women in self-sampling arms aged between 40 and 60 years demonstrated the highest participation [13, 18, 20, 21, 23, 24, 26]. Considering the study setting, over 70% of the included studies (n鈥�=鈥�11, 73.3%) were conducted in high-income settings, 1 in middle-income setting, 3 in LMICs and low-income setting according to the classification of the World Bank [30]. The findings of studies conducted in high-income settings revealed that women in self-sampling arms generally demonstrated greater screening participation than control arms, except for four studies [18, 20, 26, 28]. Studies conducted in LMICs and low-income settings showed screening participation in self-sampling arms was consistently greater than control arms. Vaginal HPV self-sampling was therefore found as a feasible option to increase participation of cervical screening services, especially in resource-constraint settings.

Impact of educational materials on screening participation

More than half of the included studies (n鈥�=鈥�9, 60%) provided women in self-sampling arms with educational materials, including leaflet, pictorial instructions and video, whereas four studies (26.7%) did not explicitly report the use of educational materials for self-sampling arms. Of these educational materials, pictorial instructions were the most commonly used. One study conducted by Winer, Lin [28] did not provide description of educational materials used. In addition to educational materials, two studies offered onsite education by community health workers to women in self-sampling arms [17, 22]. Both studies reported that women in self-sampling arms showed significantly greater screening participation compared to other studies involving educational materials. Besides, one study in LMICs involved direct supervision by trained healthcare practitioners to women during self-sampling procedure, with a statistically significant difference in screening participation (p鈥�<鈥�0.001) between self-sampling arm and control arm [19].

Discussion

Cervical cancer is a largely preventable cancer in females by vaccination and regular screening, however, it remains a significant public health concern especially in LMICs. Although the incidence and mortality of cervical cancer has been substantially declined due to the introduction of national screening programmes. screening participation is still suboptimal. In response to this challenge, vaginal HPV self-sampling has been suggested as a feasible alternate of cervical cancer screening to improve the screening participation. In accordance with the global strategy to eliminate cervical cancer launched by the World Health Organisation (WHO) [31], HPV testing has been proven as a superiority than conventional cytology or VIA, in term of its sensitivity and reproducibility. HPV testing has comparable accuracy in detecting cervical intraepithelial neoplasia grade two or worse (CIN2+). Despite the proven accuracy and feasibility of HPV self-sampling in increasing screening participation, worldwide use of HPV self-sampling remains very low. Until 2021, there was solely 48 countries adopting HPV-based screening as primary screening approach and majority of these countries are situated in Europe and America [3]. In addition, recent systematic reviews clearly demonstrated that HPV self-sampling is an acceptable method of cervical cancer screening in reaching never screened and under-screened women as it overcomes those barriers in conventional cytology, namely psychological distress, discomfort, inconvenience and privacy [7, 32].

The current findings suggested that screening participation in HPV self-sampling arms was generally greater than control arms when women were provided with self-sampling devices, regardless of the invitation strategy employed. Considering the effectiveness of invitation strategies on screening participation, opt-out strategy appears to bemore effective on reaching women and increasing their screening participation compared to control and opt-in. This finding aligned with previous findings from several recent systematic reviews by Yeh, Kennedy [33] and Costa, Verberckmoes [34]. Among opt-out strategy, direct mailing of self-sampling devices was found as the most common approach in high-income settings. It is likely attributed to the organized postal system that enables effective and reliable delivery to women living in geographically diverse area. On the contrary, all included studies conducting in LMICs, women in self-sampling arm were offered devices through door-to-door approach. It can be explained by their collected specimens rely on the assistance of healthcare workers for transportation in the rural settings. The highest screening participation was observed when women received self-sampling devices from outreach workers providing onsite instruction with standard script and diagram through door-to-door approach [22]. It can be possibly explained by the face-to-face situation enables women to directly clarify their concerns regarding self-sampling procedures. It also allows outreach workers to provide women with reassurance and support during the interaction which makes women less likely in the failure of completing self-sampling procedures. Although substantial heterogeneity across included studies was observed, vaginal HPV self-sampling had generally positive impact on screening participation irrespective of the invitation strategy. These findings further support existing evidence on the feasibility of vaginal HPV self-sampling as the alternate option of cervical cancer screening to increase its screening participation. Besides, nine of the included studies in high-income settings disseminated self-sampling devices to home addresses of participants through direct mail [13, 14, 18, 23,24,25,26,27,28], their findings consistently demonstrated that screening participation in self-sampling arm significantly greater than the control arm. It implies that direct mailing is acceptable and effective approach to reach women in increasing screening participation. Although screening participation involves a complex process of health maintenance behaviour, surprisingly, none of the included studies measured screening participation in the longitudinal manner. Only one study assessed screening participation in HPV self-sampling at 12 months [20]. Currently, limited longitudinal studies pinpointed the compliance of HPV self-sampling as a regular cervical screening approach. It provides insights on examining the feasibility of HPV self-sampling for sustaining regular screening behaviour in the future studies. This review compares the effectiveness of invitation strategies of disseminating self-sampling devices on improving screening participation across the study setting. It ascertains vaginal HPV self-sampling is a feasible way to improve participation in cervical screening, especially when self-sampling devices were disseminated by door-to-door approach through opt-out strategy.

Several limitations should be addressed in this review. Majority of studies included various types of educational materials, namely written instructions with diagrams on self-sampling procedures, while two of the included studies involved direct supervision of women in the intervention arms during self-sampling procedures. Nevertheless, variation of its impact on screening participation has not been clearly discussed. Further studies appraising the external factors associated with screening participation are therefore necessary. The considerable statistical heterogeneity across the studies implied that the results of individual studies were poorly overlapped which leaded to substantial variation of screening participation reported by the included studies. In addition, little is known about the experiences of end users during self-sampling procedures. This study provides insights for future qualitative studies to explore the experience of vaginal HPV self-sampling across female population. Our strengths in this review were reflected by the inclusion of rigorously conducted studies. There was also no restriction on language or study setting to ensure our extensive search on this topic.

Conclusions

In conclusion, the findings of this review underpin that opt-out appears to be more effective on increasing screening participation, compared to control and opt-in strategy. To enhance the generalizability of the findings, pilot intervention studies should be conducted to address the viability and appropriateness of the invitation strategy of vaginal HPV self-sampling to the target population and study setting. Various invitation strategies for the dissemination of HPV self-sampling devices were adopted and presented with a substantial impact on screening participation. Women receiving devices through opt-out strategy appears to have greater screening participation than women from opt-in arms and usual care. Follow-up studies assessing the feasibility of vaginal HPV self-sampling for maintaining compliance with cervical screening are warranted as current literature has primarily examined screening behaviour on the cross-sectional basis. Finally, future qualitative studies exploring the experience and attitudes towards HPV self-sampling among the end-users are valuable. The findings of this review provide important inputs to optimize invitation strategies of vaginal HPV self-sampling across the study setting and thus improving participation in cervical cancer screening.