The study, conducted between January 2017 and December 2017, was approved by the ethics committees of International Agency for Research on Cancer (IARC), France and GBH American Hospital, India.
Study design and setting
The community-based cross-sectional study was conducted by GBH Memorial Cancer Hospital (GBHMCH) and American International Institute of Medical Sciences (AIIMS) in the Udaipur district of Rajasthan, India. The study area is one of the backward rural districts of Rajasthan due to its low literacy level, semi-arid landscape, low level of agricultural production as well as its high tribal concentration. The study was conducted in Gogunda rural block having a total population of approximately 200,000 spread across 232 villages; the population in the villages ranging from 88 to 2,787 [6]. Gogunda has four primary health centers (PHC). Private health care is almost non-existent. Total 20 villages having an adult population of approximately 12,000, belonging to Gogunda block and served by a common PHC were selected to implement the study.
Selection and training of CHWs
Female CHWs (N=10) were recruited following the same criteria of selecting CHWs for other health programs in India. The incumbents were between 25 and 45 years of age, belonged to the local community and were educated at least up to secondary level. CHWs were trained over two weeks at GBHMCH. The training was structured to ensure both knowledge development and skill acquirement and was followed by an exit evaluation. After successful completion of training, each CHW performed the counseling and the screening procedures under direct supervision of the investigators at the GBHMCH out-patients for one week. A reorientation training was held after three months. Each CHW was provided with a kit that contained the following:
-
1.
A flipchart to educate the study participants
-
2.
A portable electronic scale and a measuring tape
-
3.
A digital upper arm blood pressure (BP) monitor (OmronTM Healthcare, Japan)
-
4.
A blood glucose meter (AccucheckTM, Roche, Germany)
-
5.
A torch and a packet of wooden spatula for oral visual examination
-
6.
A prosthetic silicone breast model with a palpable nodule inside to educate the women about breast cancer early detection
-
7.
DigeneTM Cervical Samplers, each containing a cervical brush and specimen transport medium
The weighing machines, BP monitors and glucometers were checked by the investigators during each monthly meeting of the CHWs. Each BP monitor was validated using a mercury sphygmomanometer. Each weighing machine was cross checked using standard weights. The blood glucose of a hospitalized patient was tested with the glucometer just before the same patient provided blood sample for estimation of blood glucose at the GBHMCH laboratory.
Selection of study participants
Men and women between 30 and 60 years of age and residing in the selected villages were eligible. Individuals suffering from debilitating illnesses and the pregnant women were excluded. The pregnant women were advised to attend antenatal check-ups and appropriate investigations as per the reproductive health care program guidelines. The study aimed to recruit 5,000 women and 2,000 men. The sample size was empirically decided to recruit enough number of men and women to evaluate the feasibility and efficacy of the pilot interventions. We recruited higher number of women to assess the performance of the self-collected samples for HPV test at home, a novel approach in Indian setting.
The CHWs worked in groups of two and systematically visited every household of the village. In every household the eligible men and women were invited to participate in the study and a written informed consent was obtained from each of them. The CHWs revisited the household on a later date if an eligible member was not available during the initial visit. Screening of men was stopped after recruiting 2000 participants, while screening of women continued till 5000 participants were recruited.
Procedures at home visit
Every study participant was educated by the CHWs about healthy lifestyle, the harmful effects of tobacco and alcohol and the symptoms of common cancers. The self-reported current tobacco and/or alcohol users (defined as those consumed tobacco and/or alcohol at least once in past 30 days) were counselled to quit. Weight was measured with light garments on and height was recorded without any shoes. BP was recorded with the subject sitting comfortably after at least 15 minutes of rest. If the initial recoding exceeded 140/90 mm Hg, a second measurement was obtained after 15 minutes. The lower of the two recordings was considered as final. Random blood glucose was estimated with blood sample drawn from a finger prick using sterile lancet and the displayed reading on the glucometer was recorded. A systematic oral visual examination (OVE) followed by palpation of an abnormal area was performed on every participant self-reporting to be ever tobacco and/or alcohol user. All women were made aware of the early symptoms of breast cancer and the benefits of early diagnosis. They were educated about the feel of a breast lump using the prosthetic breast model. The women provided self-collected upper vaginal sample for Human Papillomavirus (HPV) detection test. The CHW explained the steps of collection using a visual chart and the women collected the sample in the privacy of another room or toilet. At the end the CHW issued a health card documenting the test results. Participants with high BP. high glucose, abnormal OVE or breast symptoms were advised to attend the nearest PHC for further evaluation. An already known hypertensive or diabetic was advised to attend the PHC, if his/her BP or random glucose was high or if he/she was not taking medicines advised by a physician.
HPV detection test
The cervical samples were tested for 14 most common oncogenic HPV types using the CareHPVTM (Qiagen, Gaithersburg, USA) test. The test facility was set up at the GBHMCH laboratory. The CHWs deposited the samples to the laboratory once a week. Till that time, the samples were stored at room temperature. A technician was trained to run the CareHPV test following the manufacturer’s instructions. Samples with relative light unit (RLU)/cut-off ratio >1 were considered positive. The CHWs collected the reports from the laboratory during their weekly visits to GBHMCH and distributed them to the women.
Further evaluation at PHC
We followed the referral criteria mentioned in the Indian National Guidelines for high BP (>140/90 mm) and high random blood glucose (>140 mg/dl). At the PHC, all the referred participants had BP checked by a medical officer and had fasting and/or post-prandial blood glucose estimated. Hypertension was confirmed if BP was >140/90 mm Hg up on repeat checkup. An individual with a fasting glucose of >120 mg/dl and/or a post-prandial glucose of >140 mg/dl was diagnosed as a diabetic.
Individuals with abnormal OVE or women complaining of breast symptoms were examined at the PHC by a clinician trained at GBHMCH. Further evaluation (biopsy for suspected oral premalignant lesions or mammography/ultrasound for breast abnormalities) was performed at the GBHMCH based on the clinician’s recommendations.
The HPV-positive women were examined at the PHC by a trained gynecologist who performed visual inspection of cervix after application of 5% acetic acid (VIA). Women with positive VIA test were treated by thermal-ablation after taking a punch biopsy, if the lesion was suitable for ablative treatment (type I transformation zone, lesion purely ectocervical and occupying less than 75% of the ectocervix and no suspicion of cancer). The VIA-positive women who were not eligible for ablative treatment were referred to GBHMCH for colposcopy and further management. The HPV-positive but VIA-negative women were advised repeat checkup after 1 year.
The records of each participant were maintained in an online password protected database. The project coordinators (N=2) tracked the screen positive participants and reminded them over phone, if the follow up was not completed within a month of referral.
Statistical considerations
Statistical analysis was performed using the Stata 14.2 (StataCorp LP, Texas, USA) software. The distribution of participant socio-demographic, reproductive characteristics, tobacco and alcohol consumption, and clinical findings, overall and stratified by sex, were presented as numbers and proportions. The process measures and the intermediate outcomes were also presented as numbers and proportions. The process measures included numbers screened for hypertension, diabetes, oral, breast and cervical lesions; screen positivity; attendance for referral after an abnormality on any of the screening tests done; and treatment received. The intermediate outcomes included confirmed hypertension, confirmed diabetes, detection of oral, breast and cervical neoplasias during the further investigations. The detection rates of hypertension and diabetes were estimated as the proportion of the men and women with confirmed disease out of the total screened after exclusion of the known hypertensives or the diabetics. The positive predictive value (PPV) of the screening tests was estimated as the proportion of individuals with confirmed disease out of those undergoing diagnostic assessments. Because of the possible interaction between hypertension and diabetes, we defined an outcome with 4 categories including: participants with neither hypertension nor diabetes; those with hypertension but no diabetes; those with diabetes but no hypertension; and those with both. The last 3 categories then defined our studied endpoints. The effect of participant characteristics on the 3 endpoints was assessed using odds ratios (ORs) and their 95% confidence intervals (CIs) obtained from multinomial logistic regression models. Regression analysis models were adjusted for clustering on study village to allow for the possible correlation of within village responses, that is, the responses are independent between but not necessarily within study villages. Adjustment in the regression models was done by including all the cofactors assessed in the multivariate analysis. P for trend values were obtained to assess the dose-response effects of age, education and BMI grade in a multivariate model in which these factors were included as ordinal variables. For each endpoint, the group of participants with neither hypertension nor diabetes was used as the base outcome in the regression analyses.