|Home | About | Journals | Submit | Contact Us | Français|
About 25% of the Indian population experience common mental disorders (CMD) but only 15–25% of them receive any mental health care. Stigma, lack of adequate mental health professionals and mental health services account for this treatment gap, which is worse in rural areas. Our project evaluated task shifting and mobile–technology based electronic decision support systems to enhance the ability of primary care health workers to provide evidence–based mental health care for stress, depression, and suicidal risk in 30 remote villages in the state of Andhra Pradesh, India.
The Systematic Medical Appraisal Referral and Treatment (SMART) Mental Health project between May 2014 and April 2016 trained lay village health workers (Accredited Social Health Activists – ASHAs) and primary care doctors to screen, diagnose and manage individuals with common mental disorders using an electronic decision support system. An anti–stigma campaign using multi–media approaches was conducted across the villages at the outset of the project. A pre–post evaluation using mixed methods assessed the change in mental health service utilization by screen positive individuals. This paper reports on the quantitative aspects of that evaluation.
Training was imparted to 21 ASHAs and 2 primary care doctors. 5007 of 5167 eligible individuals were screened, and 238 were identified as being positive for common mental disorders and referred to the primary care doctors for further management. Out of them, 2 (0.8%) had previously utilized mental health services. During the intervention period, 30 (12.6%) visited the primary care doctor for further diagnosis and treatment, as advised. There was a significant reduction in the depression and anxiety scores between start and end of the intervention among those who had screened positive at the beginning. Stigma and mental health awareness in the broader community improved during the project.
The intervention led to individuals being screened for common mental disorders by village health workers and increase in mental health service use by those referred to the primary care doctor. The model was deemed feasible and acceptable. The effectiveness of the intervention needs to be demonstrated using more robust randomized controlled trials, while addressing the issues identified that will facilitate scale up.
Mental disorders is a major global public health problem and accounts for 8.5% of the total years of life lost due to premature death and years lived with disability globally . Between 13–50% of Indians suffer from common mental disorders (CMD) such as depression, stress and suicidal risk , but majority receive little or no care, with estimates from low– and middle–income countries (LMICs), such as India, suggesting that only 15–25% of affected individuals receive any treatment for their mental illness , resulting in a large ‘treatment gap’. The reasons for this gap are numerous, but include poor awareness about mental health, personal and community stigma related to mental illness and help seeking, lack of appropriate mental health services and trained mental health professionals [4,5]. Treatment gaps are more in rural populations , especially in Scheduled Tribe (ST) communities in India, which have particularly poor infrastructure and resources for health care delivery in general, and almost no capacity for providing mental health care.
The ST communities are identified as culturally or ethnographically unique by the Indian Constitution. They are populations with poorer health indicators and fewer health care facilities compared to non–ST rural populations, even when within the same state , and often live in demarcated geographical areas known as ST areas. In Andhra Pradesh, the state where the current study is based, infant mortality rate in ST areas and non ST areas was 94.1 and 54.0, respectively; and under–5 mortality rate was 112 and 63.2, respectively . Primary health care systems in ST areas are similar to those available in other rural areas with a tiered model involving sub–centers, primary health centers (PHCs) and district hospitals. Non–physician health care workers, called Accredited Social Health Activists (ASHAs), are a key resource for providing health care in rural settings. These female health care workers are community members with an average of 8–10 years of formal education, who mainly focus on maternal and child health by visiting households regularly and systematically. Each ASHA is responsible for about 1000 villagers. Primary care doctors are in–charge of the activities of a PHC which covers 20000–30000 population. They provide primary health care and refer any condition that they cannot manage, including mental disorders, to the next level of care at district hospitals.
One way to reduce treatment gap is by addressing the lack of trained mental health professionals in rural areas by using task shifting and training primary care health workers in the villages to manage CMD. A systematic review found that task shifting involving non–physician health workers is beneficial for a number of chronic conditions including mental disorders . The process can be facilitated by using screening and management protocols that could be used by them easily. Task shifting for ASHAs and primary care doctors involve training and provision of basic skills to identify and manage CMD, which otherwise would have been the responsibility of trained mental health professionals. Task shifting for ASHAs involves training on concepts about mental health and screening for CMD; and providing skills in basic mental health care. Task shifting for primary care doctors involve training in interviewing, diagnosing, and managing CMD using standardized guidelines and algorithms.
Integrating standardized protocols into algorithm based electronic decision support systems (EDSS) could facilitate task shifting, by making the protocols easier to administer. A number of systematic reviews have outlined the effectiveness of EDSS to deliver appropriate health care [9–13]. Computer–based decision support systems are effective in bringing about positive change , and provision of individualised recommendations using an EDSS have been found to be useful . Mobile technology based EDSS using commercial mobile networks leverages the increasing penetration of mobile phones across India, including rural India, increasing 3G connectivity that allows faster data sharing, and availability of cheaper smart phones/tablets. Prior research has highlighted the use of mHealth in communicable diseases and maternal and child health, but its research on use in mental illness is limited .
Another way to reduce treatment gap and increase demand for mental health services is by increasing knowledge about mental health and reducing stigma related to mental illness and help–seeking [4,5,16]. Research has shown that interventions especially those involving social contact with people with mental illness are effective in reducing stigma [17,18].
This paper reports on a “proof of concept” project – Systematic Medical Appraisal Referral and Treatment (SMART) Mental Health project – conducted in rural Andhra Pradesh, India. The project used task shifting supported by a mobile technology based mental health services delivery model for screening, diagnosing, and managing CMD. A campaign to increase mental health awareness and reduce stigma related to mental health and help–seeking was also implemented as part of the intervention. The aim was to ascertain the acceptability, feasibility and preliminary impact of the intervention, specifically on mental health services use, using mixed evaluation methods. This paper focuses on the key quantitative results.
The methods used have been outlined earlier , and are summarized below.
The project had two key objectives:
The primary outcome of the evaluation was to assess the change in proportion of mental health services use by individuals who were suffering from CMD. Other quantitative outcomes included changes in depression and anxiety scores, number of individuals screened, and number of screen positive individuals referred to the primary care doctor. The feasibility and acceptability of the intervention including process evaluation were assessed using qualitative interviews conducted at post–intervention, and will be reported separately.
The project was implemented in 30 villages associated with two PHCs, located in an ST area of the West Godavari district of Andhra Pradesh. All villages served by the PHCs were listed and a random selection of 15 villages from each PHC was made. The eligibility criteria for selecting villages were that all villages should have ASHAs and the population should be proportionate to the number of ASHAs. One village with a population >3000, was replaced by another one with a smaller population, keeping in mind that the average population was~400. The main source of livelihood was farming. In Andhra Pradesh, 5.5% of the total population belong to ST communities. The West Godavari district of Andhra Pradesh, where this project was conducted, has 3.4% ST population living in ST areas (http://aptribes.gov.in/statistics.htm), which are in the interior or more remote areas of the district. On extrapolating recent Census data from 2011 (http://censusindia.gov.in/pca/cdb_pca_census/Houselisting–housing–AP.html) and data collected through the government led Anganwadi program, more than two–third of the population belonged to the ST communities and other backward communities in the 30 villages. All eligible adults ≥18 years of age who gave consent to participate, were able to understand the questions and instructions, and were not limited by any severe physical disorder from accessing mental health services were invited to participate.
24 months, from May 2014 to April 2016.
Ethics approval was received from the Independent Review Committee of the Centre for Chronic Disease Control, New Delhi, India. Informed written consent was obtained from all participants and data were collected according to the Declaration of Helsinki on ethical research. Approval for conducting the project was obtained from the Health Department, Government of Andhra Pradesh, and the Integrated Tribal Development Agency was informed about the project. Approval was also obtained from all the local village administrations. Data are reported as per STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines for reporting observational studies .
All data were captured electronically, encrypted and stored on secure servers at the George Institute office in Hyderabad. All tablets and servers were password protected. Data on tablets could be accessed by user defined login. Additionally all applications were locked by the administrator so that only data and applications relevant to the project were accessible to ASHAs and doctors, and once data were submitted by ASHAs or doctors it could be changed only by the administrator in case of any errors. All data were cleaned by the data management team. Most of the coding was predefined at the time of developing the electronic data capture tools by the software developers in consultation with the researchers. Further modifications, as per need, were made by researchers. Only de–identified data were shared with researchers for analyses.
Villages in the ST areas are smaller in size compared to other rural villages in the West Godavari district; for 30 villages we estimated the population would be around 10000. We anticipated that approximately 7500 individuals will be aged ≥18 years based on the demographic profile. Based on our extensive previous work, we expected a response rate of 75%, or about 5600 participants. It was conservatively estimated that about 15% of consenting participants at baseline will have a CMD as determined by the screening tools, representing approximately 850 individuals. Studies have estimated that in developing countries only 15–25% of those with severe mental disorders receive treatment, and these numbers are even less for CMD . We conservatively assumed that 10% of individuals who screened positive will have sought medical care for mental disorders in the previous 12 months at baseline.
With these assumptions, a project involving 360 screen positive individuals would have 80% power at 2–tailed α=0.05, to detect a relative increase of mental health care utilization by as little as 30% (ie, from 10% to 13%) at follow–up. This further assumes up to 4% of discordant results between baseline and follow–up; that is, up to 0.5% who switch from utilizing services at baseline to no longer accessing services at follow–up. Other studies on provision of mental health services in primary care in India have found an intracluster correlation (ICC) of 0.03 . After adjusting for the ICC and 30 clusters/villages a sample size of 545 individuals (on average 18 per cluster) was estimated to provide 80% power.
The proportionate change in mental health services use following intervention was the primary outcome. Depression and anxiety scores among those who had scored ≥10 (cut off score for screening positive) on either the PHQ9 or GAD7 at the beginning of the intervention were compared with their scores at post–intervention phases using paired t–test. The cluster impact (village as a cluster) was explored by adding village as a random effect to a mixed model of change from baseline anxiety and depression score.
Table 1 outlines the population base and participants screened at different phases of the project. Training was provided to 21 ASHAs and 2 PHC doctors in managing CMD.
The socio–demographic and basic health parameters of the 5167 individuals screened at baseline is presented in Table 2. There were 775 (15.0%) individuals with a score of ≥5 for either depression/anxiety, corresponding to mild to severe depression/anxiety. At baseline, 200 (3.9%) individuals had a score of ≥10 on the depression/anxiety scale corresponding to moderate/severe types of depression/anxiety, and 224 had responded positively to the question on suicide. All individuals who screened positive were advised to seek medical care from the nearest doctor or health facility (as the intervention had not been deployed at this time). Family members of those who screened positive for suicide were specifically told about the risk and the need for seeking urgent medical care, after obtaining permission from the individual. Eighty–five (42.5%) individuals out of the 200 who had a score of ≥10 on the depression/anxiety scale, had responded positively to the question on suicide.
The commonest stressful events in the past year, associated with moderate or severe depression/anxiety, were suffering financial problems, suffering from major illness/injury either to self or loved ones, and death of loved ones (Table 3).
During the intervention, the ASHAs re–screened all eligible adults and identified 238 individuals (4.75%) as ‘screen positive’ (72.7% female). There were only 38 individuals who were commonly identified as screen positive at both the baseline survey by interviewers and screening done by ASHAs at the beginning of the intervention.
Thirty of the individuals identified as screen positive by ASHAs visited a doctor, with 19 visiting camps and 11 visiting the PHCs. Of these, 24 (80%) were female. Eighteen of these 30 individuals were diagnosed with a confirmed mental health problem by the mhGAP–IG tool (Figure 1). Due to non–availability of anti–depressant medication, the individual with moderate depression was referred to the district hospital. Psychological therapy consisting of discussions on stressors, involving social networks, participating in pleasurable activities and work were provided to all individuals with emotional stress, 2 individuals with suicidal risk, and one with bipolar disorder. They were all asked to follow-up later to assess progress. Others were referred to the district hospital for specialist care. The ASHAs completed a follow–up visit for almost 80% of screen positive individuals.
Only 2 (0.8%) out of the 238 screen positive individuals had received mental health treatment in the past. The increase in mental health service use in this population was from 0.8% at the beginning of intervention to 12.6% at the end of intervention. Due to the small numbers, we had not adjusted for the clustering effect of villages for this analysis.
Post–intervention data were collected from 232 out of the 238 individuals who had screened positive during the intervention (Table 4). Three people had died due to natural causes (none due to self–harm) and 3 could not be contacted even after repeated attempts.
Table 5 shows that both depression and anxiety scores reduced significantly at the end of the intervention, among those who had a score ≥10 on the depression/anxiety scales at the beginning of the intervention, as identified by ASHAs. After adjusting for clustering, there was a significant reduction in score of depression by 3.6 (SE 0.6, P<0.0001) and a reduction in anxiety score by 1.3 (SE 0.4, P=0.004), at the end of intervention.
The results from the anti–stigma campaign show that during the project period, the communities’ knowledge, attitude and behavior related to mental health  had consistently shown an improved lower score, especially for the attitude and behavior related questions (Table 6 and Table S1 in Online Supplementary Document(Online Supplementary Document)). The responses on the BACE–TS  are indicative of low baseline stigma which reduced even further during the project period (Table 7 and Table S2 in Online Supplementary Document(Online Supplementary Document)).
We found that about 5% individuals suffered from common mental health disorders in a rural, remote community in India. This population is considered to be particularly vulnerable due to remote location making traveling for help–seeking difficult, poorer health facilities and limited mental health facilities. The rates of CMD are somewhat lower than reported in a recent national mental health survey from India found that in community settings, CMD prevalence is around 10%, and it includes substance use disorders . The SMART Mental Health Project did not include substance use disorders under its definition for CMD. Moreover it included only moderate to severe depression/anxiety, hence the rates were even lower. If we Include mild depression/anxiety, the prevalence for CMD increases to 15%. This is similar to the lower value observed in earlier studies, where the prevalence of CMD was estimated to be between 13–50% , with the variability explained due to different study designs and instruments.
We implemented a mobile technology enabled mental health services model coupled with training of primary care health workers with the aim of increasing identification and referral of community members with CMD. We found the intervention of training, task shifting, and referral model to be feasible, acceptable to community and health care providers. It also led to identification of people with CMD in this rural community. Detailed qualitative process evaluation will be reported subsequently and will provide more details about feasibility and acceptability, but the fact that these could be implemented in this population and the intervention could be delivered using a predetermined strategy, provides preliminary indication about the feasibility and acceptability of the intervention. It appeared to lead to an increase in the proportion of individuals seeking mental health services. Few studies from India and other LMICs have focused on the mental health issues of ST communities leading to limited knowledge about the prevalence of mental disorders or availability of mental health services in such communities. Remoteness of ST areas and difficulties in conducting research are some of the reasons for this. Studies show that ST populations seek care overwhelmingly through public sector, which highlights the importance of strengthening the public health sector in such areas . To the best of our knowledge, this is the first project from India which reports on using innovative mobile–technology based mental health service delivery mechanisms in an ST area.
The project has a number of limitations. First, this is a pre–post design with no controls, hence the results need to be interpreted cautiously. Second, only 38 out of the screen positive individuals identified by interviewers at baseline were also identified by ASHAs. This suggests a poor inter–rater reliability. However, the reasons for such could be that some of the initial screen positive have had spontaneous remission. A systematic review showed that for major depressive disorder 23% have spontaneous remission within 3 months, and mild–to moderate depression has 20–30% higher remission than severe forms . The time difference for the two assessments in this project was 2–3 months, and a number of screen positive individuals were suffering mild to moderate depression with suicidal risk. Another explanation is the ‘retest effect’ where results from psychiatric research show that retesting using the same instrument can lead to attenuated results due to a number of reasons . Additionally, preliminary information from qualitative interviews of different stakeholders conducted at post–intervention stage reveal that some community members had not disclosed their symptoms to ASHAs due to apprehensions about the type of treatment they would be asked to undertake; fear that disclosing mental health symptoms to ASHAs could be inadvertently leaked within their community as ASHAs are part of the community; and lack of confidence in the PHC doctor’s ability to manage mental disorders (data not shown). Third, the 3–month intervention period was short and allowed only 30 screen positive individuals to access services. Our experience gained through ongoing research in other villages show that uptake of services increases gradually with time and 3 months was insufficient. The short time also prevented organizing more than 3 health camps, which were the main avenues to seek care by the community as they preferred to receive care closer to home which helped to reduce their traveling time and cost. Fourth, the short period of intervention also implies that while service contact information is available, information about adequacy and effectiveness of treatment is not available. While there was significant changes in depression/anxiety score, the sample size was small, making it difficult to generalize the results. This could also represent natural remission and regression to mean. Finally, though there was a need for anti–depressants in only one case, the lack of psychotropic medications at primary health center meant that the individual had to be referred to the district hospital.
More women were identified as screen positive compared to men and reflects the higher prevalence of depression in women found by others [32–34]. However, another reason could be that due to cultural norms, men are less likely to express symptoms of depression or anxiety leading to under–reporting . Marital status, education, and age were significantly associated with CMD, as had been observed earlier . Financial problems, illness/injury to self or to someone in the family, and death of a loved one were the commonest stressors associated with moderate/severe depression, similar to earlier research . Lifetime alcohol use had a high prevalence and some reasons suggested for higher consumption among tribal populations are increased poverty, illiteracy, increased stress, and peer pressure .
The SMART Mental Health project used three key strategies to provide mental health care. The first involved task shifting, where ASHAs and PHC doctors were trained in screening and managing CMD. The second involved developing and implementing a mobile technology enabled EDSS for use by ASHAs and doctors, to screen, diagnose, and manage CMD. The third involved implementation of an anti–stigma campaign.
Task shifting combined with health system restructuring is effective for management of non–communicable disorders, though its cost–effectiveness is still inconclusive . Task shifting is particularly relevant for mental health care in resource poor settings with few mental health professionals , and was reported to be effective and acceptable in an earlier study done in tribal areas . But in that study majority of cases who sought treatment suffered from severe mental disorders, which are easier to identify in the community, hence more likely to be treated when compared to CMD . Our project showed that ASHAs and PHC doctors were able to perform the task of screening and managing CMD using the EDSS.
Using the EDSS the ASHAs were able to screen and refer individuals with CMD to the PHC doctors, who in turn used an EDSS with the mhGAP–IG tool to diagnose and manage such individuals. Overall there was an increase in accessing mental health care from PHC doctors. The results also appeared to be beneficial and showed a significant decrease in the depression/anxiety scores at post–intervention, for those individuals who had scored moderate–severe depression at the beginning of the intervention. The results showed that a mobile–technology based mental health services delivery strategy can be implemented in this community using existing government resources.
This project helps to provide mental health services in rural and remote areas and provides some preliminary evidence suggesting that task shifting coupled with a mobile technology enabled EDSS can be used is such areas after suitable adaptations. mHealth enabled health care models have been criticized  for a lack of appropriate scaled up projects to follow the initial pilot projects. The aim of this project was to use the learnings from a proof–of–concept project to develop more robust studies which could be scaled up across larger areas. The EDSS provided a platform that not only helped ASHAs to screen for CMD, but also allowed them to follow them up to ensure treatment adherence. It also enabled the doctors to use an evidence–based guideline to manage the cases. One review identified EDSS as an important strategy to improve health care delivery, especially where algorithms were used to provide treatment plans and included patient and provider prompts . However, another review found that data from LMICs is limited, but a number of ongoing projects with a potential to provide valuable information about mHealth solutions in health care are available , and this project has the same potential.
Responses from the Knowledge, Attitude and Behavior questionnaire (Table 6) show that about 20% of the population knew someone with mental disorders and that may be a reason for the low level of baseline stigma in the population. About 7% of people responded that faith healers and religious leaders were the first point of contact for people with mental illness and this has been reported earlier from India . During the intervention period people’s attitudes and behaviors appeared to improve but not their scores on knowledge, and this is similar to other studies . Stigma related to help–seeking was low and is similar to earlier research from LMIC .
In conclusion, the SMART Mental Health project showed that the delivery of mobile based mental health services was possible in the community and preliminary evidence suggests an increase in mental health service use. Future research needs to use more robust randomized controlled trial methods to identify the effectiveness and cost–effectiveness of the program. In order to take this to scale, there needs to be greater involvement of the government at all levels and systems should be in place at PHCs to enable mental health services delivery including training of staff, provision of psychotropic medications, basic counselling services and a streamlined referral systems to the next level where specialist mental health services can be provided to those in need.
PKM is an Intermediate Career Fellow of the WT/DBT India Alliance. Shailaja Chilappagari of George Institute for Global Health India provided additional support for data analysis. Dr Jocalyn Clark (Assistant Professor, University of Toronto, and consultant to Grand Challenges Canada) provided valuable comments to earlier version of this draft. Dr Laurent Billot and Mr Anadya Prakash Tripathi of George Institute for Global Health provided additional comments to the paper and helped in refining the analyses.
Ethical standards: Ethics approval was obtained from the Independent Ethics Committee of the Centre for Chronic Disease Control, New Delhi. The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008.
Data sharing: Data is available from authors on request.
Funding: The SMART Mental Health Project is supported by the WT/DBT Fellowship (IA/I/13/1/500879) and a Grand Challenges Canada Grant (No: 0524–01–10) on which PKM is the Principal investigator.
Authorship contribution: Conceptualization: PKM, SK, SD, AP, VJ; data curation: PKM, SD, SK; formal analysis: PKM, VKV; funding acquisition: PKM; investigation: PKM, SK, SD; methodology: PKM, SK, SD, AP; project administration: PKM, SD, SK; resources: PKM, SD, SK; software: PKM, SK, SD; supervision: PKM, AP; validation: PKM, SK, SD; writing (original draft): PKM; writing (review and editing): PKM, SK, SD, AP, VJ.
Competing interests: The authors have completed the Unified Competing Interest form at www.icmje.org/coi_disclosure.pdf (available upon request from the corresponding author), and declare no conflict of interest.