The Zelen design, with randomisation occurring prior to obtaining participant consent, is employed to reduce potential bias.17 Participants in the usual care arm will not be informed about the self-sampling option, thus minimising performance bias and the ‘disappointment effect’ that might influence their willingness to undergo clinician-sampling. This approach enables the assessment of real-world screening behaviour.

Based on data from the National Population Health Survey 2022,3 the cervical cancer screening uptake rate in Singapore is currently 43%. A meta-analysis by Dr Gianfranco Di Gennaro et al provided a pooled relative ratio of 1.89 (95% CI 1.76 to 2.02) for self-sampling HPV DNA screening uptake compared with clinician-sampling.11 Using this data, our study estimates an 83.0% uptake rate in the intervention arm (self-sampling or clinician-sampling). Additionally, a local study by Lim et al reported HR-HPV detection rates of 20.1% for self-sampling and 21.0% for clinician-sampling.16 This translates to an estimated HR-HPV detection rate of 16.7% in the intervention arm and 9.0% in the usual care arm (clinician-sampling only). Assuming an alpha level of 0.05, 80% power, a 1:1 allocation ratio and accounting for a 10% loss to follow-up, the required sample size is calculated to be 325 participants per arm, for a total of 650 participants.

Participants will be identified during their primary care visits and referred to a research coordinator for further study procedures. A prerandomisation questionnaire will collect sociodemographic data (age, ethnicity, education level, marital status, employment status) and barriers to cervical cancer screening. The questionnaire will be available in English, Chinese and Malay. Data will be collected via hard copies or through FormSG, a secure digital platform.

Inclusion criteria:

Exclusion criteria:

After completing the questionnaire, participants will be randomised in a 1:1 ratio to either the intervention or usual care arm using sealed opaque envelopes (see figure 1). The randomisation sequence will be computer-generated, using block randomisation with varying block sizes of 2 and 4. The allocation sequence will be concealed from the research coordinator prior to assignment. This study will not use blinding, as it is a pragmatic RCT, designed to reflect the real-world conditions.

In accordance with the Zelen design, written informed consent will be obtained from participants after randomisation and prior to any cervical cancer screening procedures. This process ensures that participants are fully informed about the study’s purpose, procedures, potential risks and benefits. The detailed consent forms for nurse-sampling and self-sampling participants are provided as online supplemental file 1 and online supplemental file 2.

Participants in the Usual Care Arm will be offered clinician-sampled HPV DNA testing. Participants in the Intervention Arm will first be offered clinician-sampled HPV DNA testing, and if they decline, they will be offered self-sampled HPV DNA testing. Educational materials (videos and leaflets) will be provided to aid decision-making. Participants can complete their screening either on the day of recruitment or schedule it within 3 months.

After participants complete cervical cancer screening (whether through self-sampling or clinician-sampling), they will be invited to complete a post-screening questionnaire. This questionnaire is designed to assess their experiences with the screening process and their preferences for future screenings. It will be available in English, Chinese and Malay, and can be completed either as a hard copy or through FormSG, a secure digital platform.

Management of self-sampled HPV DNA test results will follow the process outlined in figure 2, ensuring prompt follow-up and appropriate care for participants:

Management of clinician-sampled HPV DNA test results will follow the process outlined in figure 3.

All colposcopy referrals will be directed to Kandang Kerbau Women’s and Children’s Hospital, the largest public hospital specialising in women’s healthcare in Singapore.

Patients and the public were not involved in the design, conduct, or reporting of this research.

All participant data will be de-identified using unique study IDs and stored securely on password-protected, encrypted servers. Only authorised personnel will have access to the data. Data will be subject to routine quality audits and will be stored for a minimum of 5 years following the conclusion of the study, in compliance with institutional policies and regulations.

Descriptive analyses will be conducted for both the intervention and usual care groups. Continuous variables will be summarised using means (with SD) or medians (with IQRs), depending on the data distribution. Categorical variables will be reported as frequencies and percentages. An intention-to-treat approach will be followed for all analyses.

The primary goal of this study is to determine if offering self-sampled HPV DNA testing alongside clinician-sampling increases the detection of HR-HPV DNA. A χ² test will be used to compare detection rates between the two arms, with statistical significance set at p<0.05.

The χ² test will also be used to evaluate differences between the intervention and usual care groups in terms of the proportion of participants who undergo cervical cancer screening, are referred for colposcopy, and are diagnosed and treated for CIN 2, CIN 3 or cervical cancer. Statistical significance will be considered at p<0.05.

A cost-effectiveness analysis (CEA) will be conducted to compare two cervical cancer screening strategies: (1) offering self-sampling alongside clinician-sampling and (2) offering clinician-sampling alone, from a societal perspective. The analysis will be based on data collected during the first 2 years of the RCT and supplemented with data from existing literature. This analysis will account for the costs of performing and processing both types of HPV DNA tests, as well as follow-up care for abnormal results. The incremental cost-effectiveness ratio will be calculated to determine the cost per additional cervical cancer screening conducted over a 5-year time horizon.

To assess the economic value of each screening strategy, a decision tree model will be used to simulate transitions between different health states (eg, normal, CIN1, CIN2, CIN3, cervical cancer) and their associated costs.18 Sensitivity analyses will be conducted to evaluate the robustness of the results by varying key parameters such as screening uptake, costs, adherence rates and treatment rates. These analyses will help identify which factors have the greatest influence on the cost-effectiveness of the screening strategies.

Descriptive statistics will be used to summarise the feasibility outcomes of self-sampling, incorporating both subjective and objective measurements.

For subjective feasibility, the distribution of Likert scale responses will be analysed to assess participants’ perceptions of self-sampling. Mean or median scores will be reported for factors such as ease of performing the test, discomfort, anxiety, embarrassment, unpleasantness and trust in test results. Additionally, the study will determine the proportion of participants who are willing to undergo cervical cancer screening again in 5 years and their preferred screening method (self-sampling or clinician-sampling).

For objective feasibility, the study will evaluate the proportion of successfully submitted self-collected samples and the proportion of valid self-collected samples.

Multivariate logistic regression will identify factors associated with the choice of nurse-sampling, self-sampling or non-screening, based on demographic, health-related and psychosocial variables collected via questionnaires.

Multivariate logistic regression will be conducted to assess factors associated with cervical cancer and CIN. Independent variables will include demographic characteristics, family history of cervical cancer, smoking status, use of contraceptives, immunosuppressive conditions, HPV vaccination status, sexual and reproductive history and screening frequency, as collected through participant questionnaires.

Adverse events (AEs) will be actively monitored throughout the study, particularly for any complications arising from self-sampled or clinician-sampled HPV DNA testing. Participants will be encouraged to report any adverse effects at any time during the study.

The expected AEs may include mild discomfort during the testing process or spotting following the sample collection. These are considered low-risk and typically resolve without intervention. However, severe vaginal bleeding or persistent spotting is considered an unexpected event. Participants will be advised to seek medical attention if such symptoms occur and to inform the study team immediately.

The study team will assess all AEs for severity and relation to the intervention. Follow-up care will be provided as necessary. A review of safety outcomes will be conducted by the principal investigator and co-investigators every 3 months to ensure ongoing participant safety.

Ethical approval has been granted by the NHG Domain Specific Review Board (NHG DSRB 2024/00194). The trial is registered with ClinicalTrials.gov (NCT06528184). Study results will be disseminated through peer-reviewed publications, presentations and reports to policymakers. The findings will be shared with the Ministry of Health to provide guidance on the potential inclusion of self-sampling in the national programme.

Consent obtained directly from patient(s).