Cervical cancer, the fourth most prevalent malignancy and fourth leading cause of cancer-related mortality in women worldwide [1], poses a significant threat to global women's health. The World Health Organization (WHO) estimates 604,000 new cases and 349,000 deaths globally in 2022, with cervical cancer ranking as the most frequent female malignancy in 25 countries [2]. In China, 2022 cancer registry data from the National Cancer Center reported 150,700 new cervical cancer cases among 4,824,700 total cancer diagnoses [3].

Persistent infection with high-risk human papillomavirus (HR-HPV) genotypes (particularly HPV 16/18) is the primary etiological factor for cervical carcinogenesis. These oncogenic variants induce dysregulation of cervical epithelial cell cycle control mechanisms, ultimately leading to malignant transformation [4–6]. Notably, HPV 16/18 account for 70% of cervical cancer cases globally, while regional data from Henan Province, China, reveals a 75.7% detection rate of HPV 16/58/52 among cervical cancer patients [7]. The disproportionately higher mortality rates in developing countries, attributable to limited screening resources, underscore the critical role of HPV-based screening in cervical cancer prevention [8].

HPV genotyping has emerged as a cornerstone of cervical cancer screening, demonstrating superior sensitivity and negative predictive value compared to conventional cytological methods such as Pap smears [9]. This molecular approach enables earlier detection of high-risk HPV infections, thereby facilitating timely interventions to reduce cervical cancer incidence and mortality. Furthermore, HPV genotyping provides critical information for risk stratification through specific high-risk genotype identification, supporting personalized clinical management and surveillance protocols [10].

In January 2023, China's National Health Commission launched the Accelerating Cervical Cancer Elimination Action Plan (2023–2030), mandating nationwide improvements in cervical cancer prevention systems and comprehensive treatment capabilities. Concurrently, the Chinese Guidelines for Cervical Cancer Screening (I) (2023) established HR-HPV nucleic acid testing as the primary screening modality [11]. Despite these advancements, current HR-HPV screening implementation in China remains suboptimal, with significant regional disparities in coverage rates. These challenges primarily stem from limited healthcare accessibility and variable patient acceptance of screening procedures.

Self-sampling for rapid testing has gained global traction in medical diagnostics, particularly during the COVID-19 pandemic [12]. Conventional cervical screening methods relying on clinician-collected Pap smears or HPV tests face limitations due to healthcare resource disparities and women's concerns regarding privacy and procedural discomfort, resulting in persistently low screening coverage in underserved regions [13]. The clinical utility of HPV self-sampling was first demonstrated in 1999 by Hillemanns et al., who reported 93% sensitivity for detecting cervical intraepithelial neoplasia grade 2 or worse (CIN2+) lesions [14].

Global implementation studies confirm that HPV self-sampling significantly enhances screening accessibility through decentralized specimen collection. Systematic reviews demonstrate that home-based self-sampling programs achieve screening participation rates several-fold higher than traditional clinic-based approaches [15, 16]. This strategy has proven particularly effective in hard-to-reach populations, such as Canada's Indigenous communities, where its convenience and privacy advantages overcome cultural barriers to screening participation [17]. Economic analyses further highlight the cost-effectiveness of self-sampling, prompting large-scale validation studies across low- and middle-income countries (LMICs) in the Americas, Asia, and Africa [18–22]. Recognizing its transformative potential, the World Health Organization (WHO) now endorses self-sampling as a cornerstone strategy for cervical cancer elimination in LMICs [23].

Despite these advancements, global adoption remains limited. A 2023 landscape analysis revealed that only 35% of 139 countries with formal screening guidelines promote HPV primary screening, with merely 17 nations (35% of HPV-adopting countries) officially recommending self-sampling [24]. In China, persistent challenges include suboptimal HR-HPV screening rates and pronounced regional disparities in implementation. The integration of self-sampling for HR-HPV nucleic acid detection could address these gaps by leveraging its operational simplicity and privacy protection features, thereby expanding screening coverage and advancing national cervical cancer prevention objectives. We are more interested in understanding whether women in economically developed regions of China, such as Shanghai, where medical conditions are advanced, have the same level of acceptance for self-sampling testing for high-risk HPV nucleic acid as women in low- and middle-income regions. What factors have a significant impact on Shanghai women's choice of self-sampling?

This study was structured to achieve three critical objectives: First, to clinically vali-date a novel cervical self-sampling device through comparative evaluation of high-risk HPV (HR-HPV) genotyping performance against clinician-collected specimens, including sensitivity, specificity, and concordance rate analyses. Second, to systematically assess patient acceptance patterns of self-sampling technology while identifying demographic, socioeconomic, and cultural determinants influencing screening participation rates, thereby generating evidence for protocol optimization. Third, to evaluate the integration potential of self-sampling as a complementary cervical cancer screening modality within China's healthcare system, with specific emphasis on its ability to expand population coverage and reduce geographic disparities in preventive care access.

This prospective paired-sample study employed a cross-sectional comparative design to assess the diagnostic validity of cervical self-sampling for HPV genotyping. Participants consecutively performed self-collected vaginal sampling followed by immediate physician-obtained cervical specimens, a sequential protocol eliminating potential confounding from prior clinical procedures on vaginal cytological integrity. Using physician-collected specimens as the diagnostic reference standard, self-sampling performance was quantified through sensitivity, specificity, and concordance analyses. A structured questionnaire systematically evaluated participant acceptability profiles of self-sampling technology, perceived procedural comfort levels, and methodological preferences between self-administered and clinician-based approaches.

This cross-sectional study was carried out at the Cervical Specialty Clinic of the Department of Gynecology, Shanghai East Hospital, between September 1, 2024 and February 28, 2025. Participants meeting the following criteria were recruited through convenience sampling:

The sample size was determined using the anticipated concordance (κ = 0.95) and 95% confidence interval width (± 0.05) from prior research, along with an expected HPV infection rate of 30%. This calculation yielded a required sample of 250 cases. Accounting for a 10% attrition rate, the final recruitment target was set at 275 participants.

Certified research nurses at Shanghai East Hospital's Cervical Specialty Clinic explained the study protocol and distributed informed consent forms to eligible participants. After obtaining consent, research assistants provided verbal guidance on self-sampling device operation, placed illustrated instructions in the dedicated hospital sampling area, and allowed participants to independently perform self-sampling using provided materials. Attending gynecologists then collected cervical specimens through clinician-administered sampling. All participants completed standardized questionnaires evaluating both sampling methods after specimen collection. (Fig. 1)

Both self-collected and clinician-collected specimens were preserved in cytological preservation solution (Cida Biotechnology, Guangzhou; Batch No. GZ20230812) at 2–8°C for 24–48 hours. Specimens were subsequently transported at ambient temperature (15–25°C) to Zhengzhou Aiwidi Medical Laboratory (an independent third-party laboratory) with a transit time not exceeding 72 hours.

HPV genotyping was performed using a PCR-reverse dot hybridization assay (Human Papillomavirus Genotyping Kit, Type 23; Yaneng Biotechnology, Shenzhen; Batch No. SZ20230915; Registration No. 20173400219). The assay detects 23 HPV subtypes including 18 high-risk types (16, 18, 31, 33, 35, 39, 45, 51, 52, 53, 56, 58, 59, 66, 68, 73, 82, 83) and 5 low-risk types (6, 11, 42, 43, 44, 81). All specimens were processed within 7 days of laboratory receipt. To minimize inter-batch variability, specimens arriving in the same shipment underwent concurrent analysis.

Following completion of both sampling methods, subjects completed a structured questionnaire containing five domains: (1) demographic characteristics (age, educational attainment, HPV vaccination history, cervical screening history); (2) procedural experience evaluation using 10-point scales (comfort level: 1 = least to 10 = most satisfied; pain intensity: 0 = no pain to 10 = worst pain) for both methods; (3) sampling preference (self-/ clinician-sampling/no preference) with rationale selection (privacy, comfort, professional expertise); (4) self-sampling implementation assessment (operational concerns, assistance requirements); and (5) cost-benefit analysis comparing method preferences under different pricing scenarios (approximately CNY200 for self-sampling versus CNY300 for clinician-sampling).

Statistical analyses adhered to the according-to-protocol principle, including subjects with complete paired HPV test results. Three analytical frameworks were implemented: (1) Concordance Evaluation using Kappa statistics (95% CI; interpretation thresholds: >0.8 = very good, 0.61–0.80 = good, 0.41–0.60 = moderate, 0.21–0.40 = fair, < 0.21 = slight) to quantify inter-method HPV detection agreement; (2) Diagnostic Performance Evaluation calculating self-sampling’s sensitivity, specificity, accuracy, PPV, and NPV (95% CI) against clinician-collected reference standard, with McNemar’s test comparing paired proportions; (3) Acceptability Profiling presenting categorical variables as frequencies (%) and continuous variables as mean ± SD/median(IQR), assessing demographic predictors via χ² tests (association strength by Cramér’s V: 0.1–0.3 = weak, 0.3–0.5 = moderate, > 0.5 = strong) and identifying self-sampling acceptability determinants through multivariable logistic regression. All analyses were conducted in SPSS 25.0 (IBM Corp.) with α = 0.05 defining statistical significance.

As outlined in Fig. 1, all of 936 women attending the gynecology outpatient clinic at Shanghai East Hospital underwent initial screening. Of these, 658 were excluded due to unmet inclusion criteria or refusal to participate, yielding 278 enrolled participants. During the study period, two participants were excluded for incomplete self-sampling procedures, resulting in 276 cases with full datasets for analysis. Additionally, 268 valid questionnaires were collected, while eight participants completed HPV testing but failed to submit their questionnaires.

Table 1 summarizes the cohort’s demographic and clinical profiles. The mean age was 43.6 ± 9.8 years, with the largest proportion (47.8%) aged 31–45 years. HPV vaccination coverage was 35.07%, and 60.45% reported prior HPV testing. Notably, 21.64% had never undergone cervical cancer screening.

As shown in Table 2, the statistical comparison of HPV test results between self-sampling and clinician sampling revealed a positive concordance rate of 91.4% (95% CI: 83.2–96.5%) and a negative concordance rate of 100% (95% CI: 98.2–100%). The overall concordance (P₀) was 97.5%, with a Kappa coefficient of 0.937 (95% CI: 0.89–0.98), indicating excellent agreement between the two methods.

Using clinician sampling results as the gold standard, the diagnostic performance of self-sampling is summarized in Table 3. Self-sampling demonstrated a sensitivity of 91.4% (95% CI: 83.2–96.5%), specificity of 100% (95% CI: 98.2–100%), accuracy of 97.5% (95% CI: 94.8–99.0%), positive predictive value of 100% (95% CI: 95.1–100%), and negative predictive value of 96.5% (95% CI: 93.0–98.5%).

Among the seven false-negative HPV test results, further analysis revealed that these cases primarily involved HPV type 58 (three cases), HPV type 16 (two cases), and HPV type 52 (two cases). This suggests potential differences in detection sensitivity across HPV subtypes with self-sampling, or possible contributing factors such as non-adherence to sampling protocols, inadequate sampling depth/volume, or lesion location variability.

The results of the questionnaire (N = 268) showed that 41.42% of the respondents preferred cervical self-sampling, 32.09% preferred clinician sampling, and 26.49% indicated that both methods were acceptable. After considering the cost factor (approximately CNY200 for self-sampling vs. approximately CNY 300 for clinician sampling plus time cost), the preference for self-sampling increased significantly to 56.72%, clinician sampling preference decreased to 19.78%, and the indifference rate was 23.51%. This significant change (up 15.3 percentage points) suggests that financial factors are an important consideration influencing the choice of patient sampling method.

The main reasons for respondents' preference for self-sampling were: a more comfortable and less embarrassing environment for at-home sampling (59.21%), the ability to control the sampling force by themselves to reduce pain (57.24%), protection of privacy (56.58%), and flexibility and convenience of time (41.45%). These proportions were similar, suggesting that multiple factors combined to influence patients' self-sampling preferences. The main reasons for choosing clinician sampling were that clinician sampling was more professional and reliable (73.58%) and the accuracy of hospital test reports was high (56.60%), while only 9.43 per cent of patients chose clinician sampling because they were afraid to do it by themselves.

In terms of comfort rating, 55.97% of the respondents considered self-sampling as more comfortable, only 23.13% considered clinician sampling as more comfortable and 20.90% considered both as equally comfortable. The comfort rating of the self-sampling product showed that 76.87% of the respondents rated it as ‘very satisfactory’ (10 points), while only 49.25% of the physician samplers rated it as ‘very satisfactory’, indicating that the self-sampling has a significant advantage in terms of product comfort (χ²=38.44, P < 0.001).

In terms of pain ratings, 67.91% of respondents indicated that self-sampling was less painful, while only 11.94% felt that physician-sampling was less painful, and 20.15% felt that the difference was not significant.

Self-sampling pain ratings showed that 71.64% of users rated no pain (0), 21.26% rated mild pain (1–3), and only 7.10% reported moderate or greater pain sensations (4–10). In comparison, clinician sampling was rated as pain-free by only 22.39% of users, 48.13% reported mild pain, and 29.48% reported more than moderate pain sensation. The difference in pain-free rates (71.64% vs. 22.39%) further confirms the clear advantage of self-sampling in mitigating the patient discomfort experience. (Fig. 2)

Chi-square analysis revealed that sampling mode preference was significantly associated with age, HPV vaccination status, prior screening history, time since last HPV test, subjective comfort during sampling, and self-sampling product comfort/pain scores (P < 0.05, Table 4), indicating that preferences are shaped by a combination of individual characteristics, healthcare behaviors, and sampling experiences.

Further multifactorial logistic regression analysis identified three independent predictors of self-sampling preference: comfort score (OR = 9.78, 95%CI: 5.12–18.67), age ≥ 46 years (OR = 2.81, 95%CI: 1.43–5.52), and absence of screening history (OR = 3.21, 95%CI: 1.58–6.53). Notably, HPV vaccination history (OR = 0.67, 95%CI: 0.37–1.22) showed no independent association with self-sampling preference.

The survey revealed that 91.04% of participants successfully performed cervical self-sampling without challenges and independently adhered to the provided instructions. Among the 8.96% reporting difficulties, operational uncertainties predominated: confusion about swab tip detachment or sample preservation (58.33%), uncertainty regarding sampling adequacy or removal timing (41.67%), ambiguity about insertion depth markers (16.67%), and questions about rotational technique during sampling (16.67%). These results highlight critical gaps in user comprehension, specifically in technical execution and procedural clarity, directly guiding targeted improvements to instructional materials and competency-based training protocols for these operational challenges.

Three key methodological strengths distinguish this investigation. First, the prospective matched-pair design enabled direct intra-individual comparison between self-collected and clinician-obtained specimens, effectively eliminating confounding from inter-subject variability through sequential dual sampling. Second, we systematically mapped the associations between critical demographic determinants (age, screening history) and sampling modality preferences, generating actionable evidence for stratified implementation strategies. Third, the study employed a novel self-sampling device ergonomically optimized to accommodate the anatomical characteristics of Chinese women's cervical canals, a technical refinement that may enhance sampling adequacy compared to conventional designs.

Guided by established evidence showing higher self-sampling acceptance among urban residents and individuals with advanced education [30], we strategically selected a Shanghai-based cohort to simultaneously evaluate diagnostic performance (via histopathological confirmation) and user acceptability metrics (through structured questionnaires). This dual assessment framework not only validated the device's clinical reliability but also quantified its social receptivity in a population with concentrated healthcare resources, thereby providing a comprehensive evidence base for scaling self-sampling programs in metropolitan settings.

This study has several limitations that warrant careful consideration. The statistical robustness may be constrained by the moderate sample size (N = 276), particularly in subgroup analyses. Conducted at a single medical center with predominantly urban participants, the findings may lack generalizability to rural and remote populations, thereby inadequately reflecting China's diverse demographic landscape. Furthermore, the self-sampling procedure was administered under researcher supervision rather than in unsupervised home environments, potentially overestimating accuracy in real-world implementation. The cross-sectional design also precludes assessment of long-term outcomes, including the sustained impact of self-sampling on CIN2 + detection rates and cervical cancer incidence reduction.

Additionally, the cost-related questions in the survey relied on hypothetical payment scenarios rather than actual financial decisions, which may inflate willingness-to-pay estimates. Future research should incorporate revealed-preference experiments with tiered pricing models for self-sampling kits to quantify price elasticity. Multilevel analyses are needed to examine geographic disparities in cost sensitivity (e.g., high-income vs. low-income regions) and evaluate insurance reimbursement frameworks (e.g., integration with public healthcare or commercial insurance coverage). Such investigations could identify economic thresholds to enhance affordability and expand HPV screening coverage, particularly through policy innovations that subsidize self-sampling costs or integrate home-based testing into national prevention programs.

This study provides empirical support for integrating cervical self-sampling as a complementary strategy to clinician-collected sampling, particularly in populations with suboptimal screening coverage. The inherent advantages of self-sampling—enhanced accessibility, privacy protection, and reduced procedural discomfort—position it as a viable intervention to improve HPV screening adherence.

A key finding reveals the critical role of cost differentials in modality selection: when self-sampling was priced at 200 CNY compared to 300 CNY (plus time costs) for clinic-based sampling, preference rates increased by 15.3 percentage points. This finding suggests that even in economically developed regions, price factors can influence respondents' choices. Of course, respondents in economically developed regions place greater emphasis on the opportunity cost of lost time than those in middle- and low-income regions. Therefore, through reasonable pricing strategies and medical insurance policy support, coupled with improvements in the efficiency of medical visits, self-sampling is expected to achieve broader coverage.

Three implementation pathways warrant prioritization: First, hospital-based distribution of self-sampling kits with postal return systems could decentralize screening access. Second, community health centers could operationalize standardized training protocols to ensure procedural competency. Third, mobile health platforms should be leveraged to provide real-time instructional support, particularly in resource-limited settings. Tailored communication strategies are equally vital—emphasizing privacy and convenience for younger demographics, while highlighting simplified protocols and minimal discomfort for older populations. Such multidimensional approaches could catalyze the integration of self-sampling into national cervical cancer prevention programs, ultimately advancing health equity across socioeconomic strata.

Building upon current evidence, five research priorities emerge requiring systematic investigation. First, large-scale multicenter validation studies should expand sample sizes across diverse geographic regions and demographic subgroups, particularly addressing variations by age, education level, and urban/rural residency to strengthen generalizability. Second, real-world accuracy assessments must evaluate self-sampling performance in unsupervised home environments, with dedicated focus on operational errors among populations with differing educational backgrounds. Third, the development of digital assistance tools warrants urgent attention, including mobile app-guided video tutorials for standardized sampling procedures and AI-driven quality control systems to overcome challenges in sample preservation and adequacy evaluation. Fourth, longitudinal clinical impact studies through sustained follow-up are needed to establish self-sampling's effectiveness in precancerous lesion detection and early cervical cancer diagnosis. Finally, comprehensive cost-benefit analyses comparing self-sampling with clinic-based methods should quantify direct medical costs, time expenditures, and systemic healthcare burdens.