Randomization

The study population will be recruited from the Taluks. 3 of the 6 clusters will be randomly exposed to the intervention (mHealth) and the remaining 3 to control (Conventional method) for Oral Cancer Screening by FHPs. The CHWs will be pre-assigned to the randomly allotted intervention (mHealth) or control (conventional method) clusters. The CHWs assigned to the mHealth clusters will be using the mobile phones with an oral cancer application that has decision algorithms for risk assessment, oral lesion diagnoses and follow up care. Remote specialists at the cancer centre at the district headquarters (CCRC) will aid in the diagnosis through the mobile-based approach. The participants in the control clusters will undergo oral cancer screening by CHWs using the conventional method (Fig. 1).

Inclusion and Exclusion criteria

Participants will include 1) men and women of 18 years or older who are at risk for developing oral cancer in the Ernakulam District of Kerala, 2) self-identified with current or previous use of tobacco and tobacco-related products, alcohol use, betel quid/areca nut use, 3) able and willing to give informed consent to willing to participate in the study. Individuals excluded would be those 1) suffering from an acute illness, 2) undergoing treatment for tuberculosis, 3) diagnosed or undergoing treatment for oral cancer, and 4) pregnant women.

Study Procedures

The CHWs will conduct Oral Cancer Awareness and Screening Programs at the selected Primary Health Centers (PHC) and Community Health Centers (CHC) in their assigned clusters using the mHealth or conventional methods. The participants, with a high prevalence of risk habits, will be chosen based on a questionnaire. The triaging will be based on the presence of any of the following three factors: a history of risk habits (smoking tobacco, chewing betel leaf or gutka [combination of areca nut with or without tobacco], or regular use of alcohol), older than 40 years, and clinical signs (nonhealing ulcers, red or white patches of the mouth that last more than three weeks, restriction of mouth opening, and swelling of the neck) [4], [6]. A single round of screening will be done, and the screen-detected positive patients would be followed up or appropriately treated. The study design is a hub-and-node based model (Fig. 2). The hub is located at the tertiary cancer center (CCRC, Kochi, India), and the nodes are the PHCs/CHCs under a taluk hospital. The project will be presented for approval by the appropriate institutional review boards and ethical committees. Informed consent will be obtained from all participants. The workflow will involve the use of an Android mobile phone and a screening software operational with a clinical decision algorithm for the risk assessment and diagnosis (Fig. 2). The gold standard diagnostic test for oral lesions is a core tissue biopsy. The subjects who are not diagnosed with any oral lesions will be monitored by the CHWs once every three months for one year by clinical examination and habit cessation counselling. Patients with oral lesions recommended for biopsy by remote expert’s review will be referred to oral cancer specialists in the PHCs/CHCs [4], [6]. Patients with non-neoplastic lesions were discharged after anti-tobacco counselling. In contrast, those with low-risk lesions will be counselled for habit cessation and follow-up. Patients with high-risk lesions and invasive cancer will be referred to the tertiary cancer centre for surgery and adjuvant treatment according to the established cancer treatment guidelines. Compliance with biopsy and the treatment at nodal and cancer centre will be checked. Fig. 3 shows the timelines for the project.

Technology platform

Mobile application which complies for best practices for securing data will be used. All communications between the mobile apps, server and other solution components are encrypted using encryption technologies. Authentication is typically managed via integration with a backend user directory. All data will be uploaded to a secure server, which is HIPPA compliant, A webpage will facilitate remote consultation, appointment schedule and referrals.

Workflow

In arm 1 (Interventional arm, mHealth), the software on mobile phone will generate patient registration number, date of registration, and demographic details. Each participant will be interrogated using a risk evaluation questionnaire incorporated in the application. The details will consist of demographics, risk factors, and symptoms. The data obtained through the mobile phones, along with photographs of the suspicious lesions, will be uploaded to the system through a secure server. The system will then create a queue that will be reviewed at the coordinating center by the oral cancer specialist. The specialist will review the image and will judged it as interpretable or not interpretable. The interpretable images will BE clinically stratified as nonneoplastic, potentially malignant, or malignant. For oral potentially malignant and malignant lesions, the patients will undergo biopsy. Text messages (also known as short messaging services, or SMS) with follow-up instructions will be sent to the respective CHWs. On receiving the message, the CHWs will refer the patients with suspected lesions to the nodal centers (PHCs/CHCs) for biopsy. In the control arm (Conventional method), all the above procedures including the registration, record keeping, the questionnaire, the clinical details will be done through a conventional paper-based method.

Histologic evaluation: Participants with oral lesions will be seen by the specialist at the health center. The gold standard diagnostic test for oral lesions is a core tissue biopsy. The histopathologic reporting of the OPMDs will be carried out or reviewed by specialist pathologists at the central pathology laboratory of the cancer center. The reporting will be carried out as per the World Health Organization classification; oral epithelial precursor lesions will be classified into categories (mild, moderate, and severe dysplasia; invasive carcinoma) [9]. For treatment decisions, we will adopt the binary system developed by Kujan and colleagues [10]. Accordingly, nondysplastic lesions will be combined with mild dysplasia as low-risk lesions, while moderate and severe dysplasia will be the high-risk lesions.

Sample Size

The data from the Trivandrum oral screening study by Sankaranarayan et al [3] (for the conventional approach) and the reports our previous publications on mHealth approach were used for calculation of the sample size. In the Trivandrum study, the intervention arm (the control arm in our study) could identify 2389 oral precursor lesions or cancers from an eligible high-risk population of 45,601 (5%). The pilot study from our group had a study population (n = 3440) included a targeted cohort (n = 2000) and an opportunistic cohort (n = 1440) screened by CHWs and dental professionals, respectively. A total of 236 (130(6%) in the targeted group and 106(7%) in the opportunistic group, lesions were detected out of the total 3440 screened (6%). In our earlier study that summarized the screening in different settings by mHealth approach [5], the prevalence of OPMDs ranged from 0.8 to 62% at different cohorts, with a mean of 12.6%. The sample size was calculated to detect at least a 2% difference between the arms. Assuming that 5% of the subjects in the reference population have the factor of interest, and after applying continuity correction, the study would require a minimum sample size of 2309 for each group, to achieve a power of 80% for detecting a difference in proportions of 0.02 between the two groups (test - reference group) at a two-sided p-value of 0.05. We inflated the sample size assuming a design effect of 2 for intra-cluster correlation, providing us with a sample 4618 in each group. Considering a dropout rate of 5%, the sample size would increase to 4848 for each group. (i.e. a total sample size of 9696 assuming equal group sizes).

We will enrol a minimum of 9696 subjects from all 6 clusters over 18 months. In each cluster, at least 1616 subjects at the selected sites (PHCs and CHCs) will undergo oral cancer screening by CHWs. Each cluster will be assigned to a group of CHWs (m Health/SOC) to screen at least 80 participants per visit. The number of referable lesions for biopsy will be approximately double of the expected number of positive biopsies (according to data from Sankaranarayan et al. [1]). Accordingly, in the mHealth arm, it will be 678(14%), and in the control arm, it will be 484(10%). The expected number of patients who will be referred for biopsy will be 1162 over 18 months).

Cost estimation

The cost estimation will include, the costs of the screening program (cost of recruiting health-care workers and screening the subjects), costs of diagnosing and treating the oral cancers detected, and of research activities. The cost from a societal perspective by including the cost of the time spent by the patient for undergoing diagnosis and treatment and the cost due to the productivity loss will also be included. An activity-based approach will be used to calculate the costs associated with the various components of the program. Activity-based costs will be derived by assigning the costs of the resources used to specific activities involved in implementing the screening program. These program activities included recruitment or invitation of screening participants, screening, data collection, research, and management and administration. Data on all the resources used will be obtained from the financial database that will be maintained by the screening program. Details of the staff employed by the program and of the equipment and supplies purchased will also be obtained from the database. The costs of equipment, staffing, consumables and travel allocated to program activities based on the expenditure incurred for each activity will be calculated. The costs of the mobile devices, the technology platform, the development and maintenance of the software and the cost of storing the data will be included. The cost of training involved in both the arms also will be included. Indirect costs and the overhead costs for program activities will be obtained. Also, the costs of biopsies and treatment were derived from the information contained in the program financial database and hospital records at the CCRC. The cost of a patient's time will be estimated from the time spent undergoing diagnostic tests and receiving treatment. Estimated daily wages of the subjects will be used. The average cost per individual in both intervention and control arms will be calculated, and the incremental cost of the intervention will be derived by comparing the two figures [8].

Effectiveness assessment

The effectiveness of the screening will be assessed using the number of cancers/high-risk dysplasia detected in each arm. At the end of the study, there will be data on 1. The number of high-risk dysplasia in each arm and the 2. The number of cancers in each arm. From this point, a decision-analytic model will be built till death or cure, based on the malignant transformation rates (available in the literature) of the dysplasia. Cost of treatment of the dysplasia (will be obtained from the study) cost of treatment of cancers (cost for early cancer will be obtained from the study, and that of the advanced cancers will be obtained from the hospital records). Mortality rates of oral cancers (available in the literature) Indirect costs of treatment period can be obtained from the study. Cost of the life lost is also available (the life expectancy rates of the district is available).

Incremental cost effectiveness ratio (ICER)

The incremental (i.e. the difference between intervention and control arms) cost per oral cancer/high-risk dysplasia detected and the incremental cost per life saved will be reported. We will conduct sensitivity and scenario analysis to evaluate the robustness of the findings from this cluster randomized trial. Probabilistic Sensitivity Analysis will be done.