The study will test a hypothesis that a community-based healthcare delivery intervention for primary and secondary prevention of stroke is feasible and will reduce stroke mortality in a rural community in Gadchiroli.
- To study the effect of the trial intervention on-
- stroke mortality
- all-cause and cardiovascular mortality
- percentage of hypertensive patients taking blood pressure medications in the community
- blood pressure control
- blood glucose control
- awareness about stroke and its risk factors
- Identify facilitators and barriers for the delivery of the intervention
- Document intervention delivery and costs of the intervention to provide insights for scale up if the intervention is found to be successful
The study is a community-based, two-arm, parallel-group, cluster-randomised controlled trial. As the intervention package is community-based, a cluster randomised design was preferred over individual randomisation.
Study site and population
The study will be conducted in Gadchiroli, one of the most underdeveloped districts of India . Gadchiroli is located in the state of Maharashtra in central India (Figure 1). According to the Indian National Census 2011, the total population of the district was 10,72,942. Ninety percent of the population of this district lives in rural areas, literacy rate is 66% and farming and manual labor are the predominant occupations in the district.
The government healthcare services remain an important source of healthcare in the district and provide care through one district hospital (DH), 12 rural or sub-district hospitals (SDH), 45 primary health centres (PHCs) and 365 sub-centres. Each SDH typically covers a population of 100,000 while one PHC covers a population of 20,000–30,000. Care for hypertension, diabetes and preventive care for stroke is available free of cost at the PHCs, SDHs and the DH through the NPCDCS. Screening of individuals with hypertension, diabetes and cancer at the village level through a community level worker called ASHA has recently started through this programme. In addition to the government-run health services, care is also provided by formal and informal private healthcare providers and some non-governmental organisations. After acute stroke, patients are either admitted to the district hospital, private hospitals or seek care from herbal medicine providers. Brain imaging is not easily available as there is only one computerized tomography (CT) scanner available in the entire district. In a population-based study conducted by us in villages of Gadchiroli, only 12% of stroke patients had brain imaging. Facilities to give intravenous tissue-plasminogen activator (t-PA) are also not available in the district.
SEARCH (Society for Education, Action and Research in Community Health) is a non-governmental organization working in Gadchiroli district since 1986. It has a service area of 86 villages distributed in 3 administrative blocks (Gadchiroli, Armori and Chamorshi, Figure 1) of the district. In these villages SEARCH has an active demographic surveillance system where all births and deaths are regularly recorded. Information on cause of death is obtained on all deaths through verbal autopsies. A population census is conducted every 10 years and the last census was conducted in 2015. A population register is maintained which is updated annually. SEARCH also runs a rural hospital where primary and secondary level outpatient and inpatient care is provided at a nominal cost and includes care for hypertension, diabetes and stroke. SEARCH, in collaboration with the state government of Maharashtra also runs a district wide campaign to reduce consumption of tobacco and alcohol by way to awareness generation, enforcement of legal restrictions on sale and community mobilisation.
Randomisation and blinding:
From the sampling frame of 86 villages in the service area of SEARCH, 32 villages will be randomly assigned to the intervention arm and 32 to the EUC arm (Figure 2) so that equal number of villages from each of the three blocks are assigned per arm. Allocation of villages will be done by the statistician at SEARCH using the random number generator in the statistical software Stata (College Station, TX, USA). Randomisation will be done before the participants are recruited.
Due to the nature of the intervention it is not possible to blind patients or the intervention implementers to the allocation group. However, the primary outcome assessors will be blinded to the allocation group.
Inclusion and exclusion criteria
Villages will be included in the study if a) they belonged to the service area of SEARCH and b) have population >400 so that the average village size of the selected villages is close to the average village population in India which is about 1250 . The highest population of a village in the service area is 2,372. Villages within five kilometers of Gadchiroli town will be excluded as the purpose of the study is to assess the impact of the intervention in a larger rural community which has difficulty in accessing healthcare.
Individuals from the selected villages will be recruited if they are resident of the village for >6 months and are ≥50 years of age at the time of screening. We selected this age group for the intervention as >90% of stroke deaths and prevalent strokes in rural Gadchiroli occurred in this age group[5,6]. In order to receive the treatment intervention the individual should have one of the following- a) hypertension, defined as- systolic blood pressure ≥140 mm Hg and/or diastolic blood pressure of ≥90 mm Hg or should be on antihypertensive medications at the time of screening by the CHWs which is further confirmed in the outreach clinic by a physician, b) diabetes, defined as- glucosuria during screening by the CHWs and random capillary blood glucose of ≥200 mg/dL in the outreach clinic or outside records showing fasting plasma glucose ≥ 126 mg/dL or a single random venous blood glucose value of ≥ 200 mg/dL or previous diagnosis of diabetes and taking treatment for it, c) has had stroke, defined using the World Health Organization’s (WHO) definition of stroke as a focal (or at times global) neurological impairment of sudden onset, and lasting > 24 hours (or leading to death), and of presumed vascular origin . Patients diagnosed with hypertension, diabetes or stroke will be invited to receive the trial intervention if they agree to provide a written or audio consent (for individuals who can not read or write). Those who are terminally ill will be excluded and those have symptoms of acute cardiovascular compromise (suspected acute cardiac chest pain, heart failure or cardiac arrhythmia with pulmonary oedema, hypotension or hypoxia) or kidney diseases ( oliguria, clinically suspected acidosis, pulomonary oedema) that can not be managed in primary care setting at the time of evaluation by the outreach physician will be referred for higher care and will be excluded until their clinical condition is unstable.
Individuals will be recruited in the trial from the villages selected in the intervention arm. A list of individuals ≥50 years of age at the time of commencement of the trial will be drawn from the census conducted by SEARCH in 2015. These individuals will be contacted by the trained CHWs after making a home visit and offered screening followed by treatment for hypertension, diabetes and secondary prevention of stroke.
The intervention package was designed considering the risk factors for stroke , local relevance of these risk factors, current barriers to treatment (Table 1), feasibility of delivering the intervention and preferences of the community regarding providers and their roles. The intervention will be community-based and all the services will be delivered at the level of the community[31–33]. Hypertension, diabetes, alcohol and tobacco use will be targeted. Obesity and lack of physical activity will not be targeted as they are not locally important risk factors given that the population is lean and is engaged in physical labour. Modification of diet and addressing psychosocial factors were not considered feasible given limited food choices and lack of availability of an effective intervention which could target psychosocial factors in this area. In a formative study conducted before starting the trial, people felt that it would be convenient for them if a physician could visit the village and then a village-level worker could follow up patients and provide counseling.
The intervention will include following four components
- House to house screening for hypertension, diabetes and stroke by community-level workers
Women community members aged 25–40 years with 7–12 years of school education will be selected, one each for villages with population ≤ 1500 and two for villages with population > 1500, after a rigorous selection process and will be trained as CHWs. The trained CHWs will visit every household with individuals ≥ 50 years of age and after obtaining informed consent will-
- measure their blood pressure using an electronic blood pressure monitor (Omron HEM–7121) after the patient is seated for 5 minutes. If the initial systolic blood pressure (SBP) is ≥ 140mm Hg and/or diastolic blood pressure (DBP) is ≥ 90mm Hg, another reading will be taken at 5 minutes after the first reading. The average of the two readings will be considered as the blood pressure of the patient.
- check urine glucose on all individuals≥ 50 years of age using urine dipsticks
- measure weight, height, waist and hip circumference
- refer patients with hypertension (defined above), glucosuria (urine glucose more than trace on urine dipstick) to the outreach clinic for evaluation and initiation of treatment.
Stroke patients will be screened using a validated questionnaire as described by us earlier . It had sensitivity of 85.7% and specificity of 99% in diagnosing stroke in the community where the trial is being conducted .Given the logistics of the trial, the initial house to house screening for stroke will be conducted during baseline assessment of stroke prevalence in the intervention arm (Figures 2 and 3) by the community-level stroke surveyors who will be a part of the evaluation team under the trial and will be separate from the intervention CHWs (Supplementary figure 1). This will be done to avoid the inconvenience caused to the target population due to double screening by the community-level surveyors and the CHWs. Briefly, the community-level surveyors will screen all individuals ≥50 years of age in the intervention villages for symptoms of stroke using questionnaire which will inquire if any of the family members ever had a)weakness on one side of the body, b) numbness on one side of the body, c) drooping of face on one side or d) slurring of speech. If any of these are present then the respondent will be asked d) if these were acute in onset and e) whether they lasted more than 24 hours. Individuals with one or more of the first four symptoms which were acute in onset and lasted >24 hours will be suspected of having stroke and will be referred to a stroke survey physician for confirmation of diagnosis.
- Evaluation, guideline-based treatment and follow up of patients with hypertension, diabetes and stroke by a mobile outreach clinic and referral where necessary
A mobile outreach clinic comprising of an outreach physician (OP), a pharmacist and a driver will visit each intervention village once in two to three months. The OP will evaluate patients referred by the CHWs, confirm diagnosis of hypertension by rechecking blood pressure, diagnosis of diabetes by checking random capillary blood glucose (RCBG) using a glucometer (Accu-Check Active) and clinically confirm the diagnosis of stroke using WHO’s definition of stroke as described before . Hypertension will be treated using hydrochlorothiazide, amlodipine and atenolol and diabetes with metformin and glipizide (Tables 2 and 3). At the beginning of the trial, a stroke survey physician will confirm the diagnosis stroke and a list of stroke patients will be provided to the OP to evaluate these patients and start treatment. Stroke patients will receive secondary prophylaxis using low dose atorvastatin (10mg) and enteric coated aspirin (75mg) as per the treatment algorithm of the study (Tables 2 and 3). Stroke patients who had loss of consciousness or seizures at the onset of symptoms will be presumed to have a hemorrhagic stroke. These patients and those with intra-cerebral hemorrhage on brain imaging will not be prescribed aspirin. The goal of the anti-hypertensive treatment will be to keep blood pressure <140/90 mm Hg among those with hypertension and <130/80 mm Hg among those with diabetes and chronic renal failure but to keep it > 100/60mm Hg. Among diabetes patients, the target will be to keep RCBG between 100–200mg/dL. Patients with stroke, if they have hypertension or diabetes, will be treated according to the guidelines for treatment of these conditions. All the trial medicines, urine and blood glucose testing will be provided free of cost to the patients during the study period. The OP will refer patients to the rural hospital of SEARCH if hypertension or diabetes can not be controlled with trial medications, if patients are unable to tolerate any of the trial medications or cardiac diseases (angina, atrial fibrillation or heart failure) or renal failure are clinically suspected. If screened individuals or patients who are being followed up develop acute stroke during the intervention period, they will be started on treatment for secondary prevention of stroke once they are back to their homes after receiving acute care for stroke. Also, individuals who did not have hypertension or diabetes at the time of screening but are incidentally diagnosed with hypertension or diabetes at a later date by the OP or another physician during the study period and wish to take treatment from the outreach clinic will be enrolled in the intervention and will receive treatment as per the study treatment algorithm. Patients will be allowed to continue medications for other ailments from their regular physicians. If patients wish to take medications for hypertension, diabetes or stroke from other physician but wish to get their blood pressure or blood glucose checked by the OP, they will be allowed to do so if they provided consent. Patients who wish to discontinue treatment provided in the trial will be free to do so at any point during the trial.
- Follow up by the CHW to ensure medication compliance, risk factor control and health education
The CHW will follow each patient with hypertension, diabetes and stroke once a month to assess if the patient is taking medications as prescribed, tolerating them and whether the blood pressure is adequately controlled (<140/90 mm Hg). The CHW will also counsel patients not to have added salt during meals and quit tobacco and alcohol. CHWs will have stock of trial medications available with them. If blood pressure control is inadequate or the patient is not tolerating a medication then the CHWs will inform the OP over phone and make modifications to the treatment as instructed. This will be done to ensure that side effects are addressed in timely manner and hypertension control is ensured before the next village visit of the OP. If a patient fails to attend the mobile clinic then the OP will instruct the CHW to provide medications prescribed during previous visit to ensure uninterrupted supply of medications. CHWs will also provide individual health education to patients regarding hypertension, diabetes and stroke using videos and animations in local language Marathi specifically designed for this purpose. These videos and animations will be shown using hand held computers. Patients with hypertension, diabetes or stroke who do not wish to take their medications from the mobile outreach clinic but consent to follow up by the CHWs will also be visited every month and provided health education and counseling. If screened individuals or patients being followed develop stroke they will be screened by the CHWs and referred to the OP if they meet screening criteria for stroke described earlier. Work of the CHWs will be supervised by two field supervisors by making visit to every CHW once every 15 days.
- A community awareness programme to provide health education about stroke and its risk factors
The awareness programme will be designed based on the insights gained in the formative study on the knowledge, attitudes and practices for stroke. The information on hypertension, diabetes and stroke will be provided through short movies and animations developed in the local language Marathi for this purpose. The awareness programmes will be conducted by field supervisors once every year in each intervention village. The supervisors will conduct these sessions in the evening after announcing the programme and it will be open for all villagers to attend.
Enhanced usual care: In this arm,allhouseholds will be provided information regarding harmful effects of and information to quit tobacco at the beginning of the intervention using information pamphlets. This was done as tobacco use is highly prevalent in the district .
The villagers in the intervention as well as the EUC villages will have access to the government’s NPCDCS programme. Villagers from both the arms will also have access to local practitioners in villages as well as private practitioners in Gadchiroli or towns which are block headquarters.
As this is a pragmatic trial and we are using standard medications, we anticipate adverse events commonly reported with these medications. We will use case definitions, standardised operating procedures and a reporting protocol to record all adverse events and treatment of these will be covered under the trial.
Primary and secondary outcomes and evaluation of outcomes
The primary outcome of the trial will be reduction in stroke mortality. We will compare stroke mortality in the two arms in the last 2.5 years of the intervention period. We selected stroke mortality as the primary outcome for two reasons—first, the trial was planned in an attempt to address a local public health problem i.e. high stroke mortality in rural Gadchiroli  and second, non-randomised studies of community-based interventions from Taiwan and Japan have shown the feasibility of reducing stroke mortality over a reasonably short period of time[35,36]. All deaths will be recorded by the demographic surveillance system. Stroke deaths will be determined using verbal autopsies as described before[5,28]. Verbal autopsies will be conducted on all deaths in the intervention and EUC arms during the intervention period using a verbal autopsy tool which has been validated in the Million Death Study. Verbal autopsies have relatively high sensitivity (≥75%) and specificity (>90%) in diagnosing stroke in validation studies conducted in India and other countries and remain an important tool to assess stroke mortality in resource poor settings[38,39]. Verbal autopsies will be coded by two trained physicians independently. If the two coders do not agree a third physician will adjudicate the cause of death. Verbal autopsy coders and adjudicators will be blinded to the identity and location of the deceased. For diagnosing death due to stroke verbal autopsy coders will use definition of stroke provided by the WHO. Deaths with diagnostic codes of I64 through I69 as per the tenth revision of the International Classification of Diseases, (ICD–10) will be counted as stroke deaths. For estimating stroke mortality rate per 100,000 population in individuals ≥50 years of age, stroke deaths in each arm between July 1st 2017 and December 31st 2019 will be considered.
The secondary outcomes include
- reduction in all-cause and cardiovascular (ICD 10 codes I00-I99) mortality in the intervention area, in the last 2.5 years of the intervention compared to that in the EUC arm over the same period. These will be estimated using the method of verbal autopsy as described above.
- percentage of hypertensive patients taking antihypertensive medicines
- blood pressure reduction in the intervention arm
- percentage of diabetic patients who have random capillary blood glucose of 200mg/dL or less in the intervention arm, and
- awareness about stroke and its risk factors- the respondents will be asked two questions. Which body organ is affected in stroke? and which diseases or factors increase the risk of stroke? The appropriate answer to the first question would be brain and for the second question would be any of the following correct responses namely hypertension, diabetes, obesity, cardiac ailments, lack of physical activity, tobacco use, alcohol use, consumption of excessive salt, lack of fruits and vegetables in the diet and mental stress.
The outcomes b,d and e will be assessed in the intervention and the EUC area at baseline and at the end of 3.5 years of intervention (Figure 3) through a survey conducted on a randomly selected subsample of individuals ≥ 50 years of age from both the arms (Figure 2). The sample will be drawn from the list of individuals in this age group available in the census conducted by SEARCH in 2015. The outcome c will be assessed at the baseline and at the end of 3.5 years among those on treatment in the intervention arm (Figure 3).
We will estimate prevalence of stroke in two arms at baseline and at the end of the intervention using a three-stage survey as described before . Briefly, a house-to-house screening for symptoms of stroke will be conducted using a well-validated questionnaire. Diagnosis of stroke will be made by a trained stroke survey physician using WHO’s clinical definition of stroke. Clinical diagnosis alone will be used when no supporting documents are available irrespective of whether the patient had any residual neurodeficit at the time of evaluation by the physician. Doubtful cases will be evaluated by an external neurologist. Evaluation of all study outcomes will be conducted by an evaluation team which will be separate from the study implementation team (Supplementary figure 1).
Strategies to reduce contamination
The likelihood of contamination will be low as villages are the units of randomization and are physically separated and the intervention will be limited to the residents of the intervention villages. The list of eligible village residents will be provided to the CHWs as well as the OP based on the population register maintained by SEARCH. CHWs and the OP will be actively instructed to provide medications only to those who belong to the intervention villages.
We calculated sample size based on the primary outcome of the trial which is reduction in stroke mortality. We hypothesized that the community-level intervention will produce at least 40% reduction in stroke mortality over the last 2.5 years of intervention. This assumption was based on a study in Taiwan where a community-based hypertension control programme resulted in 40% decrease in stroke mortality rate in the entire population over 3 years. As we plan to measure stroke mortality in only the high risk population (≥ 50 years of age) with relatively less access to care and the intervention uses strategies for both primary and secondary prevention, we assumed that the intervention will result in at least 40% reduction in stroke mortality in the given period. We used the formula by Hayes and Bennett to estimate the sample size. We assumed between-cluster correlation coefficient of variation (k) for stroke mortality to be 0.2 as such data were not available from population-based studies. The baseline crude stroke mortality rate was taken as 688/100,000 population based on our previous study in these villages . To detect 40 % reduction in baseline stroke mortality in a population ≥ 50 years of age with 80% power and at 5% level of significance and k of 0.2, we needed 27 villages per intervention group. After adjusting the sample size for 75% coverage, the required sample size was 32 villages per group.
In order to estimate the prevalence of risk factors for stroke, blood pressure and glucose control and awareness about stroke, a survey will be conducted on a random subsample of individuals ≥ 50 years of age at baseline and after the end of the intervention in both the arms. We calculated sample size for this survey assuming that 40% of individuals aged ≥ 50 years will have hypertension, and among these, at the end of the intervention, 65% and 35% of the hypertensive individuals in the intervention and EUC arm will have their blood pressure controlled (<140/90 mm Hg) respectively. Using 80% power, 5% level of significance, design effect of 1.6 and non-response of 20% the required sample size was 330 individuals per arm.
Data collection and management
Information on the individuals ≥ 50 years of age will be obtained from the census conducted by SEARCH. Information of stroke deaths will be collected using verbal autopsies and stroke prevalence will be estimated using a house to house survey as described above. Quantitative data will be collected in the field by the CHWs, field supervisors, OP and surveyors of the sample surveys using standardized pre-tested questionnaires. This will be done to assess indicators specified in Figure 3. We will monitor the progress of the study and response of the community to each component of the intervention using process indicators which will be captured through standardised management information system (MIS) reports. This will provide key information regarding operational aspects of the study which can be used for mid-course corrections and for scaling up of the intervention if the intervention is effective. Qualitative data will be collected to explore reasons for success or failure of various components of the intervention package. Four to six focus group discussions (FGDs) will be conducted with patients and their relatives and individual interviews will be conducted with the CHWs, field supervisors and OP to understand the implementation process and facilitators and barriers to the implementation of the intervention. After checking and entry of paper forms, they will be kept in locked cupboards. Data will be stored in databases on a central server in the research department of SEARCH and will be backed up at regular intervals. Qualitative data will be audio recorded and the audio files will be kept on a computer which is password protected. Access to intervention information will be limited to the study team. Databases will only be accessible to the data management team of the study. We will make the research data from this study available for use by other researchers after publication of trial results in academic journals and this information will be publicised through research manuscripts.
The analysis and presentation of results will be according to the Consolidated Standards of Reporting Trials (CONSORT) Statement for cluster randomised controlled trials. We will first compare baseline demographic features in the intervention and EUC villages such as total population, population ≥ 50 years of age, sex distribution in this population, socioeconomic and education status and the average distance from the district town in order to assess if there are any baseline imbalances after randomisation. We will also compare stroke mortality, point prevalence of stroke and the prevalence of selected risk factors for stroke such as hypertension, diabetes, use of tobacco and alcohol, body mass index, waist hip ratio and awareness about stroke in both the arms.
The impact of the intervention in the intervention villages will be analysed by comparing the primary and secondary outcomes in these two arms under the intention-to -treat principle with use of two-sided tests and significance level of 5%. Cluster level as well as individual level data will be used to compare primary and secondary outcomes. Results will be presented as effect sizes (difference in means between arms, odds ratios). Unadjusted and adjusted results will be presented. We will use regression methods applying generalized estimation equation or logistic regression with random effects models to account for the clustered design of the intervention. Age, sex, occupation, socioeconomic status, education, baseline stroke mortality will be used as covariates in the multivariate analyses based on their bivariate association with the outcomes of interest. We will also use Cox proportional hazards regression to compare time to first occurrence of death due to stroke, cardiovascular diseases as well as any-cause death between the two groups. Pre-planned subgroup analyses will include assessing the effect of the intervention by age, sex, socioeconomic status, education, individual risk factors (hypertension, diabetes, tobacco use, alcohol use, waist hip ratio, BMI), antihypertensive medication use, number of visits to the outreach clinic and number of visits by the CHW. These will be interpreted with caution. Pre-planned exploratory analysis will include a Bayesian analysis of the primary trial outcome. Given that we will be working closely with the community, we expect that the amount of missing data will not be large and hence we will not have to account for it in analyses. Data will be analysed using statistical software Stata Version 14(College Station, TX, USA).
Qualitative data will be transcribed and analysed using a thematic approach as described earlier .
All intervention related costs will be audited through the project accounting system. We will estimate the cost per stroke and all-cause death averted and the cost of delivering these interventions per 100,000 population per year.
Process evaluation will be conducted to understand-a) how the intervention was delivered, b) what were the barriers and facilitators for the implementation of the intervention and c) mechanisms through which the intervention may or may not have worked. Quantitative data from process indicators, review of various process documents and qualitative data obtain through FGDs and individual interviews with patients, their relatives, implementation team members will be used for this purpose.
Quality control measures
Quality control measures will be taken at all levels of intervention delivery, data collection and entry. These include- a) rigorous training and evaluation of the CHWs, field supervisors and the OP, b) use of pre-tested standardized questionnaires to collect data by the surveyors, the CHWs and the OP, c) refresher training of implementation team members at periodic intervals, d) cross- checking as well as re-survey of 5% of the screened population during screening as well as follow up visits by the CHWs to check for the accuracy of data collection, e) evaluation of adherence to treatment protocol by the OP, f) range check on data values and g) cross-checking of entered data. Blood pressure monitors, glucometers and weighing scales will be calibrated every one to three months and equipment showing errors as per pre-defined criteria will be replaced.
Interim analyses or stopping rules
As we will be using standard medications in the intervention, we have not planned for any interim analyses. However, we will conduct an urgent safety review of the study if in the intervention arm serious adverse events are reported by >5% or minor side effect are reported by > 25% study participants or the crude death rate in the intervention arm exceed one standard deviation above the baseline crude death rate (average of crude death rates over three years before the intervention) in the intervention villages.
The implementation of the trial will be overseen by the trial execution committee and a trial steering committee. Data and safety will be monitored by an independent external data safety and monitoring board (DSMB). It will review trial and adverse events data every six months.
Dissemination of findings
We will communicate the results of the trial to study participants, district health officials, state and national health ministries, the national chronic diseases programmes and the Indian Council of Medical Research. The study findings will be presented in the national, regional and international conferences. We plan to publish the results of the trial in a PubMed-listed peer-reviewed journal in an open access format.