This study adopts a qualitative approach following Consolidated Criteria for Reporting Qualitative Research (COREQ), the standardized and widely recognized checklist for qualitative research (Checklist 1), and conducted semistructured interviews with managers and implementers of cervical cancer screening programs from 14 institutions. To systematically explore the factors influencing universal cervical cancer screening, the interview guide was developed based on the MMHS. Thematic analysis was performed following a 6-step analytic framework [34], through which textual data were integrated, categorized, and synthesized to ultimately identify the factors affecting the achievement of universal cervical cancer screening coverage. The researchers’ reflections on the predefined statements are presented in detail in the Discussion section.

This study was conducted in Hubei province, China, from January to August 2024. A combination of purposive sampling and multistage stratified sampling was used to obtain an information-rich and representative sample. Hubei province was selected as the study area because it is the first region in China to implement province-wide universal cervical cancer screening.

The sampling procedure was conducted in 2 phases. In the first phase, purposive sampling was applied at the provincial and municipal levels to capture macro-level perspectives on policy-making and program management. Five key administrators were invited from the core management institutions involved in the program. These included 2 officials from the Maternal and Child Health Department of the Provincial/Municipal Health Commission, 2 project managers from the Provincial/Municipal Women’s Federation, and 1 project director from the Provincial Maternal and Child Health Hospital.

In the second phase, a multistage stratified sampling method was used to select county-level implementation sites to investigate influencing factors at the grassroots level. The multistage stratified sampling primarily considers factors such as screening rounds, cervical cancer detection rates, and follow-up rates. The detailed sampling flowchart is presented in Figure 2. Based on this framework, 9 county or district-level maternal and child health hospitals were selected as study sites. These hospitals are the sole organizing and implementing institutions for cervical cancer screening within their respective jurisdictions; therefore, their project managers possess comprehensive knowledge of local screening management and operations. One project manager directly responsible for the program was invited from each hospital to participate in an interview. Thus, this study included 14 key informants as interview participants, whose role distribution ensured multilevel perspective coverage from policy formulation to grassroots implementation.

All participants were informed of the researcher’s institutional affiliation and the purpose of the qualitative study, which was to understand the implementation of the cervical cancer screening program. Following the provision of verbal informed consent by all participants, the data were collected between January and August 2024. All semistructured interviews were conducted by the same researcher (the first author, holding a PhD in management) to ensure consistency in the interview process and interaction style. The interviewer was female, possessed 7 years of qualitative research experience, and had previously conducted specialized interviews related to cervical cancer screening, ensuring a deep understanding of the research context. Of the 14 participants, 2 had prior professional contact with the interviewer from a 2021 project; to mitigate potential bias toward official stances, the interviewer emphasized academic neutrality and proactively probed for challenges, while the remaining 12 were new contacts. Interviews were carried out either face-to-face or by telephone, in private offices or meeting rooms to protect participant confidentiality. Each interview lasted approximately 60 to 80 minutes and was audio-recorded.

Within 24 hours post interview, a research team member transcribed the audio recordings verbatim. A second researcher cross-verified transcripts against original recordings to ensure completeness and accuracy. We replaced all personal identifiers with unique codes to anonymize field notes, recordings, and documents. Transcripts were imported into the NVivo 12 (QSR International) software for organization and coding. During the dynamic determination process, we adopted the constant comparative method of “interviewing, coding, and comparing simultaneously.” Two professionals independently completed the transcription and coding of interviews; if there were discrepancies in coding, the experienced interview host made the final decision, which effectively avoided subjective biases caused by different researchers. Based on the inductive thematic analysis method under the MMHS framework, we have analyzed and identified the key influencing factors and have also obtained a comprehensive and systematic set of influencing factors.

Thematic saturation was achieved in this study, with no new themes emerging after the interview with the second participant. In the early stage of the study, the applicability of the MMHS framework was verified through a systematic literature review and expert consultations, and a semi-structured interview guide was developed based on this framework. The guide includes sequential questions targeting influencing factors at different levels, supplemented by keyword prompts to minimize potential information gaps, thereby ensuring the systematicity and completeness of the interview content (Multimedia Appendix 1). Coupled with the subsequent adoption of inductive thematic analysis, which enabled the rapid categorization of data into corresponding thematic categories, the study achieved thematic saturation efficiently.

Furthermore, the findings of the thematic analysis were presented and validated for feedback at a cervical cancer screening seminar with 2 provincial-level key informants, who were also participants in this study. Their perspectives were highly consistent with the study’s findings, and no revision suggestions were proposed.

Ethics approval was obtained from the Tongji Medical College, Huazhong University of Science and Technology (approval number [2023] IEC(A235)). Prior to the start of each interview—whether conducted face-to-face or via telephone—verbal informed consent was formally obtained from all participants, who were fully informed of the study purpose, procedures, potential risks, and right to withdraw at any time without any negative consequences before their consent was confirmed. All participant data were anonymized and treated in strict confidence. Each participant was offered a financial compensation of 500 RMB (US $70.60) as a recognition of the time they spent (60‐80 minutes per interview) and the professional responses they provided. A portion of participants voluntarily chose to participate on an unpaid basis, aiming to promote the smooth implementation of the full-coverage project to serve women’s health. The compensation was solely intended to offset participants’ time costs and acknowledge their professional inputs, without serving as an incentive to influence the authenticity of their responses.

No participants refused or withdrew during the process. Between January and August 2024, we conducted interviews with 14 administrators from the Health Commission (HC), Women’s Federation (WF), and maternal and child health (MCH) hospitals. Participants comprised 9 program managers from MCH hospitals, 2 program directors from provincial HC, and 2 representatives from provincial WF. Among the participants, 92.86% (n=13) reported having more than 10 years of work experience, with 42.86% (n=6) having over 20 years. Most participants held bachelor’s degrees, with 3 participants having completed postgraduate education. Detailed participant characteristics are provided in Table 1.

Through interviews, we have identified factors influencing the achievement of universal coverage in cervical cancer screening, which span 3 dimensions: structural, service-process, and outcome-related. Among these, government stewardship is considered the primary determinant, followed by the application of AI-assisted diagnostic technology, sufficient funding, and public health literacy, all of which jointly contribute to the realization of comprehensive screening coverage. During implementation, government-led initiatives provide strong support for the promotion of AI-assisted diagnostic technology, multisectoral collaboration, and funding allocation, while also facilitating public access to screening services. Concurrently, AI-assisted diagnostic technology helps alleviate shortages of trained screening technicians, establishes information management platforms, reduces screening costs, and thereby enhances institutional service capacity. To further advance the goal of universal coverage, it is essential to ensure sustained financial investment, improve public health literacy, and strengthen follow-up and treatment for positive cases.

Government stewardship serves as the critical enabler for implementing large-scale cervical cancer screening programs. In developed nations such as the United States, New Zealand, and the United Kingdom, cervical cancer prevention systems are legally mandated, whereas China adopts national policy-driven approaches [36]. Hubei’s universal screening program, integrated into the provincial “Public Service Initiatives,” ties screening targets to local government performance metrics, ensuring comprehensive policy support and fiscal allocation [37]. The program established a multisector framework led by the government, coordinated by the Women’s Federation, with participation from health care, financial, and labor unions. It is noteworthy that even when funding management and public health literacy failed to exert a significant positive influence, the government effectively achieved the phased objective of “testing all who should be tested” through the impetus of clearly defined assessment targets. In contrast, most LMICs lack both legal frameworks and substantive policy support for cervical cancer prevention [38]. Screening in LMICs remains opportunistic, lacking systematic implementation strategies. Thus, robust governmental commitment and policy infrastructure are prerequisites for population-wide screening.

AI-assisted screening technology represents a pivotal component in achieving large-scale cervical cancer screening. It substantially enhances screening efficiency and coverage while reducing costs, enabling Hubei province to complete population-wide screening within compressed timelines. In 2016, a single Chinese pathologist could manually process approximately 100 cytology screenings daily—a rate that would leave millions of women lifetime-unserved by screening programs [39]. However, AI technology allows a single pathologist to conduct roughly 10,000 screenings per day, significantly cutting down mass screening time. The technology also improves accessibility, allowing rural women to undergo sample collection at local clinics without travel burdens. Economically, AI-assisted screening demonstrates superior cost-effectiveness, reducing per-capita costs from 100-200 RMB (medical service fees) to 49 RMB, enabling broader eligibility for free screening. These advantages make universal screening coverage a feasible public health goal.

Financially, the government has allocated approximately 600 million RMB to advance the universal cervical cancer screening initiative. However, it is noteworthy that screening institutions continue to face challenges in funding security, manifested through 3 dimensions: insufficient total funding, lack of matching funds, and ambiguous allocation mechanisms. Funding shortages are particularly acute in LMICs, where systemic underinvestment persists [40,41]. Despite AI’s cost-saving potential, funding gaps endure due to insufficient government reimbursement and overreliance on singular revenue streams [42,43]. Sustainable implementation requires diversified financing mechanisms and evidence-based fiscal policies [44]. Current tiered matching-fund systems fail to alleviate local funding deficits, as reported by stakeholders. This mismatch disproportionately affects economically disadvantaged regions, where heavy reliance on central transfers, limited fiscal autonomy, and low economic capacity exacerbate funding inadequacies [45]. Solutions include needs-based allocation models incorporating regional disparities in fiscal capacity, population size, and geographic coverage. Long-term economic development remains critical for fiscal resilience [46]. China’s lump-sum public health funding—which includes screening budgets—lacks transparent suballocation guidelines [47]. Ambiguities persist regarding cost coverage and interagency fund distribution. Unclear allocation risks resource concentration, compromising efficiency and equity [48]. Standardized operational guidelines for fund distribution are thus imperative for program sustainability [49].

Population health literacy has emerged as a critical determinant of overall health status, reflecting growing public health awareness. It directly influences both health service utilization and treatment-seeking behaviors [50]. This study identified multidimensional barriers to screening participation: low awareness, embarrassment, geographic accessibility issues, and competing household priorities—necessitating incentives, such as gifts or transportation. Knowledge gaps about cervical cancer persist universally across income settings [51]. Despite understanding triennial screening recommendations, women often discontinue after negative results, reflecting insufficient risk perception. Targeted education interventions demonstrably improve screening knowledge and participation rates [52]. Sustainable screening expansion thus requires dual strategies: technological or service optimization and literacy-focused health promotion. Multimodal education can foster proactive health behaviors and sustained screening engagement [53,54].

In the initial phase of our research, we viewed AI technology as the primary driver for achieving universal cervical cancer screening coverage. We believed that addressing the supply bottleneck caused by a shortage of pathologists through technological means would enable rapid attainment of coverage targets. However, through systematic interview analysis, we discovered that while AI technology can effectively enhance service delivery capacity, its effectiveness is highly dependent on systemic support. Government leadership emerged as the most critical factor in achieving screening coverage. By integrating screening into government performance evaluation systems, introducing new technologies (AI-assisted diagnosis), and providing sustained financial support, the government effectively coordinated resources. It mobilized multiple departments—including WFs and MCH hospitals—to collaborate on outreach and organizational efforts, significantly boosting residents’ willingness to participate and actual screening rates. This reflection prompted a fundamental shift in our cognitive framework: moving from an initial “technology-centric” singular assumption to recognizing a multilayered, systematic advancement logic characterized by “government leadership, technological support, multistakeholder collaboration, and competency enhancement.” This shift lends our research conclusions greater explanatory power and policy relevance.

Improving cervical cancer screening coverage is a critical public health priority in China and other LMICs. The emergence of AI technology offers a promising solution to enhance the accessibility of these services. Within the context of AI applications in China’s basic public health services, this study systematically investigates the challenges faced in its practical implementation for increasing screening coverage. Furthermore, by applying the MMHS, we provide a comprehensive analysis of the determinants of screening, which offers a robust framework for developing the interview guide. However, a limitation of this study is that it only considers the influencing factors from the perspectives of project managers and providers. In actual program implementation, the perspective of the demand side is equally crucial, and its absence may result in factors affecting program execution being identified incompletely.

The achievement of comprehensive cervical cancer screening coverage is mainly influenced by government stewardship, AI-assisted screening, screening funding, and health literacy. Among these, government stewardship serves as the core driving force, specifically reflected in the inclusion of screening in government performance evaluations, which drives policy implementation, technological adoption, funding security, multisectoral collaboration, and activities aimed at improving residents’ health literacy and screening willingness. The application of AI-assisted diagnostic technology effectively mitigates constraints, such as shortages of trained screening technicians, limited funding, and insufficient service delivery capacity. To ensure the smooth implementation of the project and achieve preventive goals, coordinated efforts are required in 3 key areas: at the policy level, clear screening plans and supportive measures should be formulated; at the financial level, sufficient funding must be guaranteed, with coordinated multilevel and multidepartmental resources and clearly defined allocation mechanisms; at the service delivery level, diversified health education should be carried out to enhance residents’ health literacy, cultivate regular screening habits, and promote the application of AI technology.