Implementing a medical opinion leader-led management model to promote timely treatment initiation among screen-positive patients with upper gastrointestinal cancer in rural China: a cluster randomised controlled trial protocol

Abstract

Introduction

Upper gastrointestinal cancer (UGC), including oesophageal and gastric cancers, imposes a substantial health burden, particularly in China. Timely treatment after a positive screening result is critical, as treatment delays may contribute to disease progression and poorer outcomes. This study aims to evaluate the effectiveness of a management model led by medical opinion leaders (MOLs) in shortening the time to treatment initiation among screen-positive patients with UGC.

Methods and analysis

This multicentre, cluster randomised controlled trial will be conducted at 16 secondary hospitals in Shandong province, China. The participating hospitals will be randomly assigned to the intervention or control group in a 1:1 ratio. Screen-positive patients aged 18–80 years whose UGC diagnosis is confirmed by endoscopy and pathological examination will be recruited. At least 372 participants will be enrolled. The intervention group will receive an MOL-led management intervention comprising WeChat-based health education and face-to-face consultations. The primary outcome is the 30-day timely treatment rate, defined as the proportion of participants who initiate treatment within 30 days after pathological diagnosis. Secondary outcomes are the cumulative treatment initiation rates within 60, 90 and 180 days. Baseline information will be collected at enrolment, and treatment status will be assessed at 30, 60, 90 and 180 days.

Ethics and dissemination

The study protocol was reviewed and approved by the Science and Technology Ethics Committee of the School of Public Health, Shandong University (No. STEC-SPH-SDU-20250703). Written informed consent will be obtained from all participants. The study findings will be disseminated through publications in peer-reviewed journals, presentations at national and international academic conferences and communication with participating medical institutions and study participants.

Trial registration number

ChiCTR2500109389.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• This study uses a rigorous multicentre, prospective, randomised controlled design to evaluate the effectiveness of a medical opinion leader-led management model across 16 secondary hospitals.

• The 180-day follow-up period will enable treatment initiation to be assessed at 30, 60, 90 and 180 days, together with an evaluation of the intervention implementation process.

• Sample size is limited by the detection rate of screen-positive patients, requiring rolling rather than simultaneous enrolment.

• Although standardised selection, training and implementation procedures will be used, variation in medical opinion leaders’ engagement and practice styles may introduce heterogeneity in intervention delivery.

Introduction

Upper gastrointestinal cancer (UGC), primarily comprising oesophageal and gastric cancers, constitutes a major global health burden. In 2022, there were approximately 1.48 million new cases and 1.10 million deaths.^{1 2} Alarmingly, nearly half of these cases occurred in China, where the rates are significantly higher than in developed countries like Japan and South Korea.^{1 3 4} By 2050, it is projected that the new UGC cases and deaths will reach 2.76 million and 2.11 million, respectively, with 1.02 million new cases and 0.88 million deaths anticipated in China.^{5 6} UGC also contributes substantially to premature mortality, disability-adjusted life years and productivity losses.^{7 8}

Early detection and timely treatment are important strategies to reduce the burden of UGC,^9–11 with the 5-year survival rate increasing from 20% to 90%.^{4 5} Prolonged time to treatment initiation (TTI) among patients may permit disease or stage progression and is associated with poorer health outcomes^{12 13} and higher treatment costs.¹⁴ Delayed TTI is widely reported across countries, healthcare settings and cancer types. For instance, studies in the USA have shown that 44.3% of patients experienced delayed treatment initiation, with 35.3% of patients with colorectal cancer experiencing similar delays.^{15 16} A study conducted in Japan revealed that the proportion of patients experiencing waiting times exceeding 30 days for treatment was 53.7% for gastric cancer, 42.8% for colorectal cancer, 50.5% for lung cancer and 75.7% for breast cancer.¹⁷ In Malaysia, 54.2% of patients with cancer experienced a treatment delay exceeding 30 days.¹⁸ In Brazil, 18.8% of patients with breast cancer experienced a treatment delay exceeding 60 days.¹⁹ Research in China found that 28.9% of patients with gastric cancer experienced a treatment delay exceeding 30 days.²⁰ Definitions of delayed TTI varied across those studies, with thresholds ranging from 7 to 90 days; however, 30 days is the most commonly used cut-off value.^21–24 In China, national UGC guidance also states that a cut-off of 30 days is the optimal benchmark.²⁵

Previous studies have identified several factors associated with prolonged TTI, including age, disease stage, type of treatment, socioeconomic status, disease awareness and attitudes and recommendations from healthcare providers. Shan et al²⁶ found that patients with cancer with lower education levels, lower incomes and higher anxiety levels were more likely to experience delays in seeking treatment. Patients who are obese, older or have comorbidities typically face longer delays.¹⁴ Individuals with lower socioeconomic status often postpone seeking medical care, and significant differences in pathological and clinical variables can contribute to treatment delays.^{27 28} Furthermore, receiving a diagnosis and treatment at different institutions may increase the risk of treatment delay.¹⁹ Despite these factors, high-quality evidence regarding effective interventions for reducing delayed treatment initiation remains scarce. Because UGC may progress rapidly, specialised guidance and support are essential for facilitating the transition from a positive screening result to treatment initiation. In this context, obtaining guidance from trusted healthcare professionals becomes particularly crucial.

Key opinion leaders (KOLs), first introduced by scholars Lazarsfeld and Berelson in 1948 in the context of voter behaviour, are defined as individuals with specific expertise who can influence others’ decision-making and behaviours.²⁹ To date, KOLs have been shown to positively impact individual behaviours in fields such as social media and marketing.^30–32 In healthcare, medicalopinion leaders (MOLs) are healthcare professionals who use their clinical expertise and professional influence to shape patients’ attitudes and behaviours.³³ Studies have confirmed the role of KOLs in enhancing service quality and improving medication-related behaviours among patients with chronic diseases. They can enhance doctors’ awareness of new drugs and treatment methods.³⁴ Berner et al³⁵ found that the involvement of a local KOL was associated with a reduction in patients’ length of hospital stay by an average of 1.5 days. Regarding the identification of KOLs, Flynn et al³⁶ developed a new scale based on the self-assessment scale created by Childers³⁷ and evaluated its reliability and validity. Despite these findings, high-quality evidence evaluating MOL-led interventions for improving timely TTI among screen-positive patients with UGC remains lacking.

Therefore, we will conduct a multicentre, cluster randomised controlled trial to evaluate whether an MOL-led management model can shorten TTI and increase the proportion of screen-positive patients with UGC who initiate treatment within 30 days. Secondary objectives are to assess cumulative treatment initiation rates within 60, 90 and 180 days and to evaluate the implementation of the intervention.

Methods

Theoretical framework

The Stimulus-Organism-Response model, developed by Mehrabian and Russell, is a flexible framework for examining a variety of external and internal stimuli, such as perception, motivation, thinking and emotion, along with their resulting responses, which include intentions and behaviours.^{38 39} This theory has been increasingly applied in the health sector, informing efforts to optimise health-information-seeking and improve healthcare-seeking behaviour. In this study, the stimulus (S) is defined as an MOL-led management model, representing an external environmental factor. The organism (O) signifies the patient’s internalisation of the stimulus, while the response (R) denotes their treatment behaviour.

Study design and setting

This study is a multicentre, cluster randomised controlled trial that will be conducted at 16 secondary hospitals in Shandong province, China. This protocol is reported in accordance with the Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) 2025 checklist.⁴⁰ The participating hospitals will be randomised in a 1:1 ratio to the intervention and control groups, with the intervention based on a MOL-led management model intended to facilitate timely treatment initiation among screen-positive patients. The planned study period is from November 2025 to June 2026, encompassing participant recruitment and completion of the 6-month follow-up. The study will begin with an eligibility screening phase, followed by dynamic enrolment of eligible screen-positive patients over an 8-week period. Baseline data will be collected at enrolment (T0). Treatment status will subsequently be assessed at 30 days (T1), 60 days (T2), 90 days (T3) and 180 days (T4) after pathological diagnosis. The intervention will continue until the participant initiates treatment, as verified by medical records, or completes the 180-day observation period, whichever occurs first. A flowchart outlining the study procedures is presented in figure 1.

Figure 1. Flow diagram of the trial. MOL, medical opinion leader; ITT, intention-to-treat.

Identification of medical opinion leaders

At each intervention hospital, 10 healthcare professionals with experience in the screening, diagnosis or treatment of UGC will be invited to participate as MOL candidates. Candidates will first complete the Flynn et al self-assessment scale³⁶ (table 1). Negatively worded items will be reverse-scored so that a higher total score consistently indicates a greater level of opinion leadership. The three candidates with the highest self-assessment scores at each hospital (ie, the top 30%) will proceed to peer assessment.

Table 1. Self-Assessment Scale for MOL in upper gastrointestinal cancer.

Order	Scale item	Score*
1	My views on screening and diagnosis of upper gastrointestinal cancer do not seem to be valued by others.	7	6	5	4	3	2	1
2	When discussing information related to screening and diagnosis of upper gastrointestinal cancer with colleagues/patients, others consult me for information less frequently.	7	6	5	4	3	2	1
3	Others rarely ask for my advice on issues related to screening and diagnosis of upper gastrointestinal cancer.	7	6	5	4	3	2	1
4	Others take actions based on the opinions/information I provide regarding screening and diagnosis of upper gastrointestinal cancer.	7	6	5	4	3	2	1
5	In the field of screening and diagnosis of upper gastrointestinal cancer, I often persuade others to make decisions.	7	6	5	4	3	2	1
6	I frequently influence others’ views on screening and diagnosis of upper gastrointestinal cancer.	7	6	5	4	3	2	1

^*‘Strongly agree’ gives 7 points, ‘strongly disagree’ gives 1 point.

MOL, medical opinion leader.

For the peer assessment, 10 healthcare professionals from the same institution who are not MOL candidates will independently assess the shortlisted candidates using the study-specific peer-assessment instrument (see online supplementary material 1). Peer assessors will not have access to the candidates’ self-assessment scores. The instrument evaluates five domains: professional competence, influence, information dissemination, organisational ability, innovation. The mean peer-assessment score across all instrument items and peer assessors will be calculated for each shortlisted candidate.

The peer-assessment instrument will be developed through a literature review and multidisciplinary expert consultation. Its content validity will be evaluated by an expert panel, and pilot testing will be conducted before formal use. Internal consistency will be assessed using Cronbach’s alpha, and inter-rater reliability will be evaluated using the intraclass correlation coefficient.

Before participating in patient management, all selected MOLs will receive standardised training using a common intervention manual. The training will cover the clinical referral pathway, recommended timelines for treatment initiation, patient communication, delivery of WeChat-based health education, face-to-face consultation, documentation and confidentiality. MOLs will be required to meet the prespecified competency criteria before delivering the intervention; those who do not meet the criteria will receive additional training and reassessment. Intervention fidelity will subsequently be monitored using standardised implementation checklists and periodic supervision.

Participants and eligibility criteria

Patients will be eligible for inclusion if they: (1) are aged 18–80 years, regardless of sex; (2) have a positive result for an upper gastrointestinal lesion on endoscopic or pathological examination, defined as high-grade intraepithelial neoplasia or a more advanced lesion (see table 2); and (3) voluntarily agree to participate in the study and provide written informed consent. Patients will be excluded if they: (1) are undergoing post-treatment follow-up examinations; (2) are unable to provide complete diagnostic and treatment information; (3) have serious comorbidities; (4) are terminally ill; or (5) have participated in another clinical trial or are otherwise considered unsuitable for participation by the investigator.

Table 2. Pathology codes and result interpretation criteria.

Code	Diagnosis	Result	Stage	Code	Diagnosis	Result	Stage
1	Normal squamous epithelium	Negative	—	186	Barrett oesophagus	Negative	—
4	Basal cell hyperplasia	Negative	—	19	Indefinite for intraepithelial neoplasia in glandular epithelium	Negative	—
5	Mild oesophagitis	Negative	—	20	Low-grade intraepithelial neoplasia of glandular epithelium	Negative	—
6	Moderate oesophagitis	Negative	—	21	High-grade intraepithelial neoplasia of glandular epithelium	Positive	Early
7	Severe oesophagitis	Negative	—	22	Intramucosal adenocarcinoma (not otherwise specified)	Positive	Early
8	Low-grade squamous intraepithelial neoplasia	Negative	—	221	Intramucosal adenocarcinoma	Positive	Early
10	High-grade squamous intraepithelial neoplasia	Positive	Early	222	Submucosal adenocarcinoma	Positive	Early
11	Squamous intraepithelial neoplasia, NOS	Negative	—	23	Adenocarcinoma	Positive	Advanced
13	High-grade squamous intraepithelial neoplasia, suspicious for invasion	Positive	Early	24	Aden squamous carcinoma	Positive	Advanced
14	Intramucosal squamous cell carcinoma, not otherwise specified	Positive	Early	25	Neuroendocrine carcinoma (including small cell carcinoma and large cell neuroendocrine carcinoma)	Positive	Advanced
141	Intramucosal squamous cell carcinoma	Positive	Early	26	Neuroendocrine tumour (G1, G2, G3)	Positive	Advanced
142	Submucosal squamous cell carcinoma	Positive	Early	27	Carcinoma, NOS	Positive	Advanced
15	Squamous cell carcinoma	Positive	Advanced	28	Lymphoma	Positive	Advanced
16	Normal glandular epithelium	Negative	—	29	Mesenchymal tumour	Positive	Advanced
17	Non-atrophic gastritis	Negative	—	30	Squamous papilloma of the oesophagus	Negative	—
18	Atrophic gastritis/intestinal metaplasia, not otherwise specified	Negative	—	31	Fundic gland polyp	Negative	—
181	Mild-moderate atrophic gastritis	Negative	—	32	Hyperplastic polyp	Negative	—
182	Severe atrophic gastritis	Negative	—	33	Others	Negative	—
183	Mild-moderate intestinal metaplasia	Negative	—	99	Insufficient for diagnosis	Negative	—
184	Severe intestinal metaplasia	Negative	—	100	No biopsy required and no pathological diagnosis required	Negative	—

Endoscopic diagnosis is for reference only; the final diagnosis is determined by histopathological examination of biopsy specimens. Only high-grade lesions and early-stage cancers were included in this study.

NOS, not otherwise specified.

Intervention

Participants in the intervention group will receive the MOL-led management model, while participants in the control group will receive the usual management provided by their institutions. Given that the primary outcome of this study is the timely treatment rate within 30 days, and taking into account supporting literature, clinical management experience and practical feasibility, a 6-month (180-day) intervention period is deemed sufficient to assess changes in the treatment initiation of screen-positive patients with UGC, as well as the sustainability of the intervention effects. Details of the intervention are presented in figure 2.

Figure 2. Intervention measures, follow-up points and data collection overview.

Intervention group

The intervention measures delivered by MOLs to screen-positive participants include: (1) Following pathological confirmation, each participant will receive one face-to-face consultation to explain the positive result and provide treatment recommendations. (2) Participants who are able and willing to use WeChat independently will be invited to join a WeChat group for regular health education and follow-up management. For participants who cannot use WeChat independently because of limited digital literacy or physical limitations, a family member (such as a spouse or child) or another caregiver may join the WeChat group with the participant’s consent and assist the participant in receiving treatment-related information and completing follow-up assessments. All educational materials will be standardised and developed centrally by the project team. (3) Follow-up will be conducted weekly or every 2 weeks by telephone or WeChat, according to participants’ treatment status and support needs.

Control group

Participants in the control group will receive the usual management provided by their institutions.

Outcomes

Primary outcome

The timely treatment rate (TTR) is the primary outcome, defined as the proportion of enrolled screen-positive participants who initiate treatment within 30 days after pathological diagnosis. It will be calculated using the following formula:

$${\displaystyle {\displaystyle \begin{array}{l}\text{Timely Treatment Rate}=\\ \left({\displaystyle \frac{\text{Number of participants who initiate treatment within 30 days}}{\text{Total number of enrolled screen-positive participants}}}\right)\times 100\mathrm{\%}\end{array}}}$$

Secondary outcomes

The secondary outcomes are the cumulative treatment initiation rates within 60, 90 and 180 days after pathological diagnosis. These are defined as the proportions of enrolled screen-positive participants who initiate treatment within the corresponding time periods. The rate at each time point will be calculated separately using the following formula:

$${\displaystyle {\displaystyle \begin{array}{l}\text{Treatment Initiation Rate at time }t\\ =\left({\displaystyle \frac{\text{Number of participants who initiate treatment within }t\text{ days after pathological diagnosis}}{\text{Total number of enrolled screen-positive participants}}}\right)\\ \times 100\mathrm{\%}\end{array}}}$$

where t=60, 90 or 180 days.

Harms

No serious intervention-related harms are anticipated. Any unintended adverse events will be documented during follow-up and reported to the ethics committee in accordance with institutional requirements.

Sample size

The initial sample size was calculated using the formula for comparing two independent proportions. The sample size calculation formula is as follows:

$${\displaystyle {\displaystyle N=\frac{{\left[Z_{1-\alpha /2}\sqrt{2p(1-p)}+Z_{\beta }\sqrt{p_1(1-p_1)+p_2(1-p_2)}\right]}^2}{(p_1-p_2{)}^2}}}$$

Based on published literature,⁴¹ the timely treatment rate for the control group was set at p₁=0.70, while the intervention was expected to produce an absolute increase of 0.15, resulting in p₂=0.85. The combined proportion was calculated as $p=\frac{(p_1+p_2)}{2}$. With a two-sided significance level of $\alpha =0.05$ and a power of 1−β=0.80, it was calculated using the above formula that each group should include 121 participants, totaling 242 participants.

To adjust the sample size, the design effect (DE) is applied, calculated using the formula $DE=1+(m-1)\times ICC$, where m is the expected average number of participants per medical institution (estimated at $m=16$). Based on literature from multicentre randomised controlled trials, the intracluster correlation coefficient (ICC) is conservatively set at $0.02$, yielding a DE of $1+(16-1)\times 0.02=1.3$. After adjustment for clustering, the required sample size per group was: $N_{\text{corrected}}=121\times 1.3\approx 158$. Allowing for an anticipated attrition rate of 15%, the required sample size per group was 157.30/(1–0.15)=185.06, which was rounded up to 186 participants. Each institution is expected to enrol approximately 23–24 screen-positive participants, subject to flexible adjustments based on actual circumstances.

Assignment and randomisation

Medical institutions will serve as the units of randomisation. Before randomisation, the 16 participating institutions will be stratified according to geographical region (eastern, central or western Shandong) and local economic development level. Within each resulting stratum, institutions will be randomly assigned to the intervention or control group in a 1:1 ratio using a computer-generated randomisation sequence. Randomisation will be implemented through a centralised system by project researchers who are independent of the on-site principal investigators, thereby ensuring allocation concealment. Individual participants will not be randomised; all eligible participants will receive the intervention or control protocol assigned to their institution.

Blinding

Participants will not be explicitly informed of their group allocation. Outcome data will be extracted from medical records by trained researchers using standardised procedures to minimise assessment bias.

Data collection and statistical methods

Data will be collected, organised and quality-controlled by the research team using standardised procedures. Data that fail quality checks will be verified against source records. Analyses will follow a prespecified statistical analysis plan. The primary analysis will follow the intention-to-treat (ITT) principle and will use the full analysis set (FAS), including all eligible and consenting participants analysed according to their hospital allocation regardless of intervention adherence. The per-protocol set (PPS) will include participants who adhere to the assigned protocol and have no major protocol deviations. PPS analyses will be conducted as supportive sensitivity analyses, while all primary and secondary outcomes will be analysed primarily using the FAS.

Baseline characteristics will be summarised by randomised group. Continuous variables will be presented as mean (SD), and categorical variables as number and percentage. All adjusted regression models will include prespecified baseline prognostic covariates and randomisation stratification factors, as detailed in the statistical analysis plan.

The binary cumulative treatment initiation outcomes at 30, 60, 90 and 180 days will be analysed separately using mixed-effects logistic regression models. Group allocation will be included as the main fixed effect, and medical institution as a random intercept. Treatment effects will be reported as adjusted ORs with 95% CIs. For fixed-time binary outcomes, participants lost to follow-up before a given time point without documented treatment initiation will have missing outcome data at that time point. Methods for handling missing data and the corresponding sensitivity analyses are prespecified in the statistical analysis plan.

Time from pathological diagnosis to treatment initiation will also be analysed as a time-to-event outcome, with treatment initiation defined as the event. Participants who do not initiate treatment will be right-censored at 180 days or on the last date they are confirmed to remain untreated, whichever occurs first. Kaplan-Meier curves will present the cumulative probability of treatment initiation as 1−S(t). Cox proportional hazards regression with cluster-robust SEs will estimate HRs with 95% CIs while accounting for clustering within medical institutions. The proportional hazards assumption will be assessed using Schoenfeld residuals. No adjustment for multiplicity will be made for secondary outcomes, which will be interpreted as supportive.

Analyses will be performed using R V.4.3.3. All tests will be two-sided, and p values<0.05 will be considered statistically significant.

Discussion

To our knowledge, this study will be the first cluster randomised controlled trial to evaluate the effectiveness of an MOL-led management model in promoting timely treatment initiation among screen-positive patients with UGC.

The phenomenon of delayed TTI behaviour among patients with cancer is widely recognised internationally, with existing research confirming that delayed TTI significantly reduces patients with cancer’s survival. As delayed TTI increases, mortality rates worsen. Previous studies have reported associations between prolonged TTI and survival across several cancer types, including lung, breast, colorectal, cervical and ovarian cancers.^{12 17 21 42–46} An analysis of breast cancer data from the SEER–Medicare and National Cancer Database found that a longer interval from diagnosis to surgery was associated with an increased risk of mortality, with each additional time interval corresponding to a HR of 1.09.²¹ Another observational study reported a negative association between time to treatment initiation and survival across several cancer types, with delays exceeding 30 days associated with poorer survival and more pronounced reductions observed after 90 days.⁴³ Additionally, Gore JL et al⁴⁷ reported that treatment delays exceeding 12 weeks among patients undergoing radical cystectomy were associated with an increased risk of mortality (HR 1.38; 95% CI 1.18 to 1.62) and a decline in the 5 year survival rate from 67% to 58%. At the same time, studies have explored potential factors influencing TTI, including age, socioeconomic status, education level, annual household income and disease stage.^{19 22 23 48 49} Nevertheless, evidence regarding effective interventions to promote timely treatment initiation remains limited.

To address this evidence gap, we will evaluate the MOL-led management model designed to promote timely treatment initiation among screen-positive patients with UGC. This multicentre cluster randomised controlled trial will include 16 secondary hospitals in Shandong province, China. The participating hospitals will be randomly assigned to the intervention or control group in a 1:1 ratio. The protocol has been developed in accordance with the SPIRIT 2025 guidance.⁵⁰ The primary outcome is the 30-day timely treatment rate, defined as the proportion of participants who initiate treatment within 30 days after pathological diagnosis. Secondary outcomes are the cumulative treatment initiation rates within 60, 90 and 180 days.

This study will provide evidence on whether an MOL-led management model can promote timely treatment initiation among screen-positive patients with UGC. If effective, the model may offer a practical approach to strengthening patient education, treatment guidance and follow-up support in secondary hospitals. The findings may also inform the development of similar patient-centred management strategies for other cancers with established screening and treatment pathways.

This study has several potential limitations. First, because of the nature of the intervention, participants and MOLs cannot be blinded to group allocation, which may introduce performance bias. Nevertheless, treatment initiation will be determined from medical records using standardised procedures, which may reduce the risk of assessment bias. Second, the effectiveness of the intervention may depend on the appropriate identification, training and engagement of MOLs and may vary across institutions. To reduce this variability, MOLs will be selected using standardised self-assessment and peer-assessment procedures, receive standardised training and be monitored using implementation checklists. Third, the study will be conducted in 16 secondary hospitals in Shandong province, which may limit the generalisability of the findings to other healthcare settings. Finally, the 180-day follow-up period may be insufficient to assess the long-term sustainability of the intervention. Further studies with longer follow-up periods and broader settings will therefore be warranted.

In conclusion, this study will evaluate whether an MOL-led management model can promote timely treatment initiation among screen-positive patients with UGC by strengthening patient education, treatment guidance and follow-up support. The findings may provide evidence for developing patient-centred management strategies in China and other settings with similar healthcare contexts.

Ethics and dissemination

Ethical approval for this study was obtained from the Science and Technology Ethics Committee of the School of Public Health, Shandong University (No. STEC-SPH-SDU-20250703). Written informed consent will be obtained from all participants before enrolment (see online supplementary material 2). The study findings will be disseminated through publications in peer-reviewed journals and presentations at national and international academic conferences. Summaries of the findings will also be communicated to participating medical institutions and study participants in accessible language. No information that could identify individual participants will be disclosed.