Study design

The study was carried out as a component of the International Study of Discrimination and Stigma Outcomes (INDIGO) global research network [14]. Within INDIGO, the “PRogramme for Interventions addressing Mental healthcare knowledge, Attitudes and behavior in primaRY care” (PRIMARY) initiative aimed to develop, implement and evaluate interventions to improve knowledge, attitudes and behavior of PCPs towards service users in primary care settings.

The current study is one of the initial steps in the INDIGO-PRIMARY initiative. This is a qualitative situational analysis carried out in diverse participating sites that aimed to understand experiences in primary care centers. The study sought to cover the perspectives of PCPs, service users, and their families, as well as other stakeholders in primary care settings in Asia, Africa and Europe. Partner investigators and institutions in the broader INDIGO research network were given the opportunity to volunteer to participate in this multi-country qualitative situational analysis. The participating sites included four LMIC in South Asia, the Middle East, and North Africa (India, Nepal, Lebanon and Tunisia) as well as three HIC in Europe (Czech Republic, Hungary, and Italy). There were no exclusion criteria for sites. The only requirement was that partners would be able to implement the qualitative research without funding, and that a common qualitative protocol would be used. The study was carried out without project-specific funding in all sites except India, where a local grant was obtained. The unfunded nature of the study meant that feasibility considerations determined the recruitment strategies and sample sizes in each site.

Methods comprised the collection and analysis of qualitative data arising from original interviews and focus groups with PCPs, service users, and their families as well as other stakeholders. The qualitative data included in the analyses comprised largely of information collected purposely for the INDIGO-PRIMARY study. In Nepal, the study team had recently collected relevant qualitative information on the study subject [15–19], so these data were used and analyzed afresh for the purposes of this study. In the Czech Republic, new INDIGO-PRIMARY data collection was supplemented with comparable qualitative data from other local studies. All site leads carried out a documentary analysis of health policy and health system documents relevant to mental health in primary care in their site (these findings will be published in detail separately).

Qualitative data were collected by local researchers in the local languages in each site. Participation in the study was initiated after providing written informed consent. Care was taken to carry out interviews in settings to maintain privacy and not hinder participants’ free expression. After audio recording, interviews were transcribed by the researchers who had conducted them and analyzed in the local language. The exceptions were in Tunisia and Lebanon, where interviews were translated into English and then analyzed by international interdisciplinary research teams. Full details for the Consolidated Criteria for Reporting Qualitative Studies (COREQ) are provided for each site in S1 Table.

Participants

Participants for individual interviews and focus groups consisted of:

1. Primary Care Providers (PCPs): frontline providers working in participating primary care centers who had received general training and were not mental health specialists. These included professional providers (doctors, nurses, midwives) whose primary responsibilities were patient care, non-professional providers (such as community health workers and lay providers), and administrative and support staff. In addition, supervisors for primary care facilities, such as primary care managers or lead primary care clinicians, were included.
2. People with mental illness (service users) and family members: persons receiving treatment for a mental illness at the primary care centers and their family members. Service users were recruited from the primary care centers based on nomination from PCPs about patients who received care there for mental illness. In addition, because of other ongoing research and clinical activities, study site teams were familiar with service users who obtained mental healthcare in primary care settings, and these individuals were also recruited. For the purposes of this study, we have referred to service users with common mental disorders as those experiencing depression, anxiety, and posttraumatic stress disorder. Service users with severe mental disorders refers to those with schizophrenia or other psychotic conditions, as well as bipolar affective disorder. We use this convention for classification because it was a common categorization used by respondents. We also acknowledge that depression, anxiety, and posttraumatic stress disorder can be accompanied by severe symptomatology and severe disability. The common vs. severe distinction is not an indication of the impact the disorder on an individual’s life and wellbeing.

Efforts were made to ensure that participants in the sample were reasonably representative of PCPs and service users with experiences in mental healthcare integrated into primary care in the different settings. Interviews were also collected with a few mental health professionals collaborating with the primary care facilities and policy makers or health authority representatives. However, not all countries participating in INDIGO-PRIMARY were able to complete this. Therefore, the current analysis will only focus on findings involving comparable respondents across all sites (i.e., PCPs, service users, and their family members).

In the Czech Republic, researchers had at least a master’s degree or higher and they were affiliated with the National Institute of Mental Health in the Czech Republic. Some of the researchers had clinical training. For PCPs, recruitment took place and interviews were conducted in their primary care centers. For PCPs, purposive sampling was used to obtain individuals with different professional roles and clinical backgrounds. For service users, the researchers approached them and interviewed them in clinics. Service users were approached when researchers were available to attend clinics, and thus it was a sample of convenience. No one approached for the study refused. All interviews were audio recorded.

In Hungary, inclusion criteria for PCPs were professional training or a degree in health care and active contact with patients. Inclusion criteria for service users were diagnosis of a mental illness and active involvement in psychiatric treatment. In Hungary, the interviewer orally informed the study participants about the study and provided them with an information sheet and obtained a written form of consent. As the resources allowed us to use a convenience sample, the participants were contacted through the Hungarian team’s lead researcher (a psychiatrist) in the district of the Community Psychiatry Centre. The healthcare professionals of the primary care centers were interviewed by a clinical psychologist trainee. Respondents’ answers were transcribed during the interview on paper. The interview duration was 45–90 minutes. The service users and the family members were interviewed by the site’s senior psychiatrist. The interview duration was 60–90 minutes and with notes transcribed on paper. No video or audio recording was utilized to mitigate participants’ concerns of confidentiality breaches. All PCPs, service users, and family members agreed to be interviewed, and none of the participants refused to take part in the study.

In India, the study sites selected were in rural geographic areas. The study team approached the District Mental Health Programme consultant (a psychiatrist) for coordination of the study at the site and the project lead and the consultant together supervised the activities of the researcher at the study site. With the help of the District Mental Health Programme team, three rural primary care centers were selected out of sixty primary care centers based on the demographic details, referral rates and cooperation for the study. The team also provides supervision of mental health services at primary care centers and also trains PCPs on mental health. At the primary care centers, the study sample was selected based on purposive sampling: the available PCPs who were willing to participate in the study were included. Service users were identified and approached for the interviews with the help of primary care staff. Service users who were symptomatic or cognitively impaired were excluded. Two service users and one PCP refused to participate in the study. However, there was no dropout of participants. The research staff who had research experience underwent special training in qualitative research by the project lead, who has previous experience in qualitative research. An interview guide prepared by the INDIGO team was translated to the local language and used for the interviews. In-depth interviews with participants were carried out at the selected primary care centers and each interview lasted for 60 to 90 minutes. The interviews were audio recorded and the verbatims were translated and transcribed. The transcripts were reviewed by an independent researcher. Subsequently, the research team carried out the qualitative data analysis, and the project lead and co-lead who were experts in qualitative research supervised and reviewed the data analysis process.

In Italy, PCPs were those potentially involved in the care of patients with a mental illness. In Italy, all primary care providers are the first-line professionals for people with illnesses, as they are the gatekeepers for specialized mental health services. For service users, the only criterion was that he/she had to receive care for a mental health problem, regardless of diagnosis. For caregivers, the only criterion was that he/she had to be a first-degree relative of a person with a mental illness, regardless of diagnosis. Initially, potential participants were contacted by the lead researcher via telephone or e-mail and asked if they were interested in participating. In the case of consent, participants were subsequently approached by the local research team for the interviews. Participants were first given full details on the aims and the overall design of the study; informed consent forms were signed by participants (with the exception of telephone interviews). Interviews lasted on average 30–60 minutes. Interviews with PCPs, service users, and family members were conducted face-to-face. Only the interview with the primary care manager was made on the telephone.

In Lebanon, primary care centers providing mental health services were included and PCPs trained on mhGAP were recruited for interviews. The primary care department of the Ministry of Population and Health contacted the directors of the primary care centers. The director decided whether to accept for his/her staff to participate in the study. Service users receiving care at the primary care centers were invited to participate. Only service users in psychiatric emergencies were excluded from the option to participate. Participants were contacted by primary care clinic focal point employees who were responsible for explaining the study and obtaining verbal agreement from participants. A member of the local research team orally presented the study to participants and provided them with a participant information sheet to peruse. Except when the interview was performed over the phone, all participants then completed and signed an informed consent form. When the phone conversation was set, the interviewer repeated the explanation and the consent form to the participant. The interviewer ensured that the participant had enough time to ask questions in advance. Phone interviews were conducted for service users who preferred it, and consent was obtained verbally in these circumstances. The interview duration was 30 to 60 minutes. Although all staff agreed to be interviewed, some service users did not feel comfortable doing the interview because the first contact was done by primary care clinic focal persons. However, the exact number of service users who refused to participate was not recorded.

In Nepal, existing qualitative datasets with PCPs, service users and their family members, and policymakers [15–19] were used to extract information relevant to the INDIGO-PRIMARY topics of interest. Inclusion criteria for PCPs were those who received mhGAP training through the PRogramme for Improving Mental healthcarE (PRIME) project in Nepal [20] and who consented to participate in all research activities including training evaluations and qualitative interviews in the beginning of the training. Inclusion criteria for service users were: being identified as having one of the four priority disorders of the project (depression, alcohol use disorder, epilepsy, or psychosis), being above the age of 18 years, and having received treatment from the PCPs who received the mhGAP training. All the interviews were conducted by trained research assistants in the Chitwan district of Nepal, where the projects were being implemented. The participants were selected by the study teams in coordination with the PCPs from local primary care centers. In-depth interviews were conducted with PCPs, service users, and family members in all the projects. Written consent was sought from the participants prior to the interview. The interviewers followed a semi-structured interview guide to conduct the interviews. For the service users and family members receiving treatment in primary care centers, the questions focused on their explanatory models, help-seeking, barriers and facilitators to their treatment, and reasons for non-adherence (if they had dropped out of treatment). For PCPs, questions focused on their experience of treating mental health patients (barriers and facilitators), receiving mhGAP training, knowledge, and attitude. Similarly, for service users involved in stigma interventions [15, 21, 22], questions focused around their experiences of participating in mhGAP trainings with PCPs. All interviews were conducted in Nepali and were audio-recorded. They were then transcribed and translated by professionals.

In Tunisia, the site was chosen based on acceptance of the lead PCP to conduct the study in her primary care center, and based on placement of the primary care center in the geographical sector of the referent psychiatry department. All service users of this particular primary care center were approached to participate in the study, and two refused. In Tunisia, senior psychiatrists, who were clinician-researchers at the local hospital, conducted the qualitative interviews. They had qualitative research training, with prior experience leading key informant interviews and focus group discussions. The psychiatrists conducting the qualitative interviews was known to some of the participants including the lead PCP, the associated mental health professional, and the program manager participating in the study. But the psychiatrist conducting interviews was not known to the remainder of the study participants.

Qualitative situational analysis tool

The situational analysis tool comprised questionnaires for documentary analysis as well as topic guides for qualitative interviews and focus group discussions. The questionnaires for the documentary analysis were based on an existing case study methodology outlined for mental health projects in LMIC [23] as well as a situational analysis tool developed for a large international program of mental health interventions in primary care [24]. The topic guides were adapted from guides used for the exploration of experiences of stigma and discrimination among people with mental illness in India [25], which have also been used in other studies on stigma and discrimination in India [26].

As INDIGO-PRIMARY was carried out in culturally and socio-economically diverse sites, core topic guides comprised a list of broad questions and probes with the phrasing of questions developed locally to take the cultural and contextual factors of each site into account. The situational analysis tool is accessible online (see S1 Table).

Data analysis

Techniques of framework analysis were used for qualitative data, allowing for analysis both by site and across sites. This approach was chosen as it compares cases and codes, making analysis and dissemination digestible and relevant to policy-makers [27]. A thematic coding framework was created jointly with the research teams in all sites to structure multiple-researcher coding [28]. Inductive coding was used to identify additional codes relevant to the specific context in each site. The coding framework was charted into tables to compare various stakeholders’ experiences and views on key themes such as “attitudes of PCPs” or “experiences of stigma and discrimination in primary care” [27]. In a second step, similarities and differences between research sites were mapped out on each theme. Relevant subthemes evident in the data were identified (e.g., different types of stereotypes evident in PCP narratives across countries). Cross-country findings on each theme were summarized for a) data from PCPs (provider perspective) and b) data from service users/family members (service user perspective), with a particular focus on findings that were salient across study sites and findings that appeared to be more locally specific. In a final step, implications for interventions to address stigma and discrimination in primary care were identified.

Regarding presentation of results, we have chosen to use non-specific terminology rather than to present results as percentages or frequencies. We are aware that this issue is debated among qualitative researchers. We are following one of the recommended conventions in presenting qualitative research [29] that we feel best fits the nature of our study design and data. Neale and colleagues (2014) recommend that raw numbers be only included for features that have been assessed for all participants. Because of the semi-structured nature of these interviews, we cannot state with certainty that a particular subtheme as presented in our results was presented in manner that allowed every participant to respond with an endorsement or denial of the statement. This is better suited to a structured questionnaire, which is something that could follow from this research to establish prevalence of different attitudes or experiences in a multi-site study. In addition, Neale and colleagues recommend that reporting % be limited to studies in which the sample size is at least 50 respondents are greater. Although our total sample is 248, all sites except Nepal are below the 50-respondent threshold. Therefore, any reporting of percentages would be biased toward the Nepal results and not representative across countries. Neale and colleagues discourage using terms with specific quantitative meaning (e.g., majority, minority) with a specific justification, and instead recommend non-specific language (e.g., few, several, some, many). Ultimately, we chose this latter strategy of using non-specific terms when helpful to convey meanings, and we have avoided use of percentages, frequencies, and specific quantitative terminology. This best captures the spirit that qualitative data should be used to understand phenomena rather than quantifying frequency of themes to infer prevalence of such phenomena [29].

Ethical statement

Informed consent was obtained from all participants taking part in data collection for this study. Ethics approval was obtained from the PNM Research Ethics Subcommittee, King’s College London (Reference Number (No.) RESCMR-17/18-4109; approval date 23 February 2017) and local Ethics Committees in the participating sites: Czech Republic (National Institute of Mental Health, Klecany, Czechia. No 128/17, 19.4.2017), Hungary (Semmelweis Egyetem Regionális, Intézményi Tudományos és Kutatásetikai Bizottság, No. SE TUKEB 162/2017), India (NIMHANS Institutional Review Board, Ref No: NIMHANS/IEC(BEH.SC.DIV.)7th MEETING/2017; and permission from Directorate of Health Family Welfare Servicers, Government of Karnataka, India, No: DD/Mental Health/10/18-19 dated 26.04.2018), Italy (Comitato Etico IRCCS San Giovanni di Dio–Fatebenefratelli, Brescia, No. 102/2017; Ethics Committee of the Provinces of Verona and Rovigo, No. 44066, 13 September 2017), Lebanon (Saint Joseph’s University Beirut, No. CEHDF 1193; 3 July 2018), Nepal (Nepal Health Research Council, No. 139/2016, 28 July 2016), Tunisia (Ethics Committee of Razi Hospital La Manouba, 28 January 2017).

Description of settings

The studies were conducted in primary care settings in seven countries, with all countries having different models of primary care-based mental health services (see Tables 1 and 2). We first describe the context in the LMICs: In India, three primary healthcare centers in the Ramanagaram District, Bengaluru Division, of Karnataka State in the southern central region of the country participated in the qualitative situational analysis. In Bengaluru, there is a history of initiatives to support primary care-based mental health services going back multiple decades [30]. In Nepal, there has been a recent initiative through PRIME, through which a district mental health plan was developed that included training primary care workers on mhGAP and psychosocial skills [20]. The PRIME initiative was conducted in southern Nepal in the Chitwan district bordering India. The prior qualitative study from which data were drawn for this study included 32 primary care centers for anti-stigma study conducted in those settings [15]. In Lebanon, two primary care centers participated in the Mount Lebanon area. Some of the participating Lebanese PCPs were are also trained on mhGAP, and community mental health centers provide specialized mental health services. However, not all PCPs had mhGAP training at the facilities. This represents the situation throughout the country where only some PCPs have had exposure to mental health training. In one primary care center in the Greater Tunis area of Tunisia, some providers have been exposed to training through the national mental health program, although none had received formal mhGAP-IG training. Among those who have had training, PCPs generally have limited responsibilities in mental healthcare with the exception of basic medication management for depression and anxiety. In summary, the PCPs in the LMIC are representative from the perspective of a mix of those who have had and have not had a focused mental health training, and all recruited PCPs worked at primary care centers where some mental healthcare was being provided.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Overview of mental health and primary care in participating countries: Low- and middle-income countries (LMICs).

https://doi.org/10.1371/journal.pone.0258729.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Overview mental health and primary care in participating countries: High-income countries (HICs).

https://doi.org/10.1371/journal.pone.0258729.t002

Regarding the HICs: in the Czech Republic, three primary care centers in Prague participated in the study. Czech general practitioners receive general psychiatry training in medical school, but there is no further systematic mental health training for these doctors working in primary care. Their mental healthcare activities are limited to prescribing antidepressants and referring patients to mental health specialists. In Hungary, four primary care centers in Budapest participated in the study. Family doctors in primary care centers are responsible for referring patients to mental health specialists. Then with written permission from a specialist, family doctors can prescribe antidepressants and anxiolytics for a limited time of 6 months to 1 year. In Italy, two primary care centers in Brescia and one primary care center in Verona participated in the study. Italy has a long tradition of primary care- and community-based mental health services. PCPs detect mental illness and refer patients to specialists. PCPs are authorized to prescribe psychiatric medication, and they primarily treat common mental disorders.

To classify the types of primary care models of mental healthcare integration, we draw upon the framework adapted by van Ginneken and colleagues [30]. This framework classifies primary care mental health services into five types: 1) education and training models where PCPs are trained to independently manage mental illness; 2) consultation-liaison models in which PCPs independently manage with only educational support from specialists; 3) collaborative care models where care managers are in place to foster linkages among primary care centers and specialists; 4) referral/replacement models in which PCPs identify and refer patients to specialists; and 5) specialist-community models in which PCPs are trained within specialty programs to treat severe mental disorders in community settings. Models in all the LMIC partner sites (India, Nepal, Lebanon, and Tunisia) can be described as generally consultation-liaison. In the Czech Republic and Hungary, the model is predominantly replacement/referral. In Italy, the model is predominantly collaborative care.

Sample description

Qualitative data included 248 total participants: 64 primary care providers, 11 primary care facility managers, 111 people with mental illness, and 62 family members of people with mental illness (see Table 3). Across the sites, the majority of primary care providers were women (43 women vs. 18 men), with India, Lebanon, and Tunisia each only having one male PCP. In the other sites, only 2–3 PCPs were men. PCPs were roughly equally divided among doctors and nurses. In some sites, e.g., Nepal, PCPs were managed by health auxiliary staff who did not have a medical degree (e.g., neither MBBS nor MD credentialing). Across sites, the number of participating PCPs was equally divided between those who did vs. did not receiving mental health training. In India, Nepal, and Lebanon, the majority had prior training. For example, in Lebanon and Nepal, the primary care doctors and primary care workers, respectively, had received mhGAP trainings. In Tunisia, the Czech Republic, and Hungary, the majority did not have prior mental health training. The majority of service users for whom gender information was available were women (21 women vs. 6 men). In the study, information on the diagnoses of service users was not available for nearly half of the participants; the remainder were comparably divided between common and severe mental illness. For those service users where age was recorded, most were older: 22 service users (90%) were 40 years of age or above compared to two under 40 years of age. Family members were comparably divided between men and women, with all family members 40 years of age or older.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Qualitative sample participant demographics.

https://doi.org/10.1371/journal.pone.0258729.t003

There was considerable heterogeneity within and between study sites in the types and backgrounds of respondents. In Hungary, the study examined the work of PCPs in four primary care centers. The PCPs included four family physicians and three primary care nurses working in Budapest. The doctors in all four primary care centers are certified family physicians, having completed a specialist degree required in Hungary, and all four were lead clinicians. The medical primary care nurses all have completed education in nursing. In addition, all 4 clinics employ a receptionist, who has a high school degree. Two of the interviewed family physicians completed psychotherapeutic training, which is not compulsory for the training as general practitioner in Hungary. In India, there was a wide range of service providers from three different primary care centers, including medical officers, nurses, junior health workers, community health workers (Asha’s). Each professional category had the required qualifications as mandated by the government for their positions. For instance, medical officers were MBBS graduates, nurses had undergraduate nursing qualification, junior health workers had diploma degree qualifications, and Asha workers had the equivalent of a 10th grade education. The service providers interviewed had received very minimal training about mental health. In Nepal, the healthcare workers were ‘prescribers’ (i.e., medical officers with MBBS degrees, health assistants, and auxiliary health workers) who had prescription right as per the Government of Nepal and ‘non-prescribers’ (i.e., staff nurses and auxiliary nurse midwives) who did not have prescription rights.

Primary care providers’ perspectives

Perspectives of people with mental illness and their families

Stakeholder perspectives on “what needs to be done”

Limitations of the study

Our study findings are limited by heterogeneity in the employed study methodology across sites. As the study was carried out without external funding in most sites, participant numbers varied depending on available site resources. Despite efforts to harmonize study methodology by providing questionnaire guidelines, data analysis tools and methodological support, there were still cross-site variations in terms of research experience of interviewers, depth of probing on certain topics, and other aspects of research implementation. The samples interviewed in some sites were small, and it is possible that more experiences of discrimination would have been identified in a larger sample of service users. Due to the small sample sizes in some sites, differing levels of detail in the data and the wide range of aspects of stigma studied, there are likely other commonalities and differences that could be unveiled through more in-depth work in each site.

Although the diversity of settings covering low-income, lower-middle-income, upper-middle-income, and high-income was a strength, the diversity of settings also meant that there were differences in the backgrounds of PCPs and differences in health system structures across settings. There was also heterogeneity in relation to the mental health background of the PCPs. This would suggest that those with mental health training have more information and skills in the treatment of people living with mental illness, which can influence their attitude towards patients, potentially decreasing stigmatization. We did not adequately sample PCPs with a background in mental healthcare vs. those without a mental health background in order to reach meaning saturation of these subgroups across all sites. For example, in Hungary, all four PCPs have professional contact with the community psychiatry center of the district, which shows a more open, tolerant attitude towards patients living with mental illness, and this is not likely to be representative of PCPs across the country. There was also differential understanding of topics across sites. For example, in India, although they were aware of social stigma, there was less understanding of the concept of structural stigma.

In Lebanon, the primary care doctors were general practitioners who had received mhGAP trainings. However, general practitioners throughout the country had not universally been exposed to mhGAP training at the time of the study, therefore results cannot be generalized to PCPs without any mental health training. In Nepal, the secondary data were extracted from multiple interviews that was conducted with health workers, patients receiving mental health treatment in primary healthcare facilities, and service users and family members involved in anti-stigma interventions in Nepal.

A limitation of the Lebanon study was that qualitative interviews were limited to two urban areas with higher economic status than many other areas of the country. In the future, underprivileged areas should be investigated further and may reveal differences in the findings reported here. Similarly, in India, because the study site has more than 60 primary care centers, further studies can be conducted by including more primary care centers of the district on a larger scale, and similar studies can also be conducted in other districts of the state to enhance generalizability.

Similar to challenges with the heterogeneity within and between sites related to care providers, the service users and their family members could not capture the diversity of demographic differences within and between countries. In India, the interviewed service users and family members were from a rural background. As a result of this, we found that the service users and their family members had low levels of mental health literacy and limited knowledge of their own conditions and how treatment and attitudes may have been influenced by their diagnoses. The service users for the study were identified by psychiatric diagnoses on mental health outpatient cards. However, information on the diagnoses was often incomplete and the service users were rarely aware of the exact diagnosis for their conditions. In Nepal, as the PRIME intervention focused on 4 priority disorders (depression, alcohol use disorder, psychoses, and epilepsy), all the patient and service user participants were representative of only these disorders. In addition, all the participants interviewed were above 18 years of age and so represent only the adult population. There are a number of limitations that should be considered when interpreting results, such as the service users reporting lack of stigmatizing experiences. Because most interviews were conducted by clinician-researchers at primary care centers, desirability bias could have masked stigmatizing attitudes among PCPs as well as stigma experiences among service users and their family members. For example, in Lebanon, contact with PCPs was established through supervisors, which may bias responses to more positive attitudes or greater willingness to take on mental health services than participants may have reported if not referred by their supervisors. In India, the context under which the interviews were conducted likely influenced the level of detail and type of responses. Because of high clinical workloads and shortage of time for attending interviews, the qualitative interviews conducted in clinical sites were abbreviated.

Moreover, because in most sites service users were nominated by staff at the primary care centers or known to the study teams, they may reflect a biased sample who have more positive experiences of mental health services in primary care. For example, in Lebanon, service users were reached by contacting primary care facility focal points who then referred the service users for the qualitative interviews. This recruitment technique may have reduced the likelihood of providing comments that were critical of services received at those clinics. Another consideration is that by conducting the interviews in primary care centers in some sites as well as potentially associating interviewers with primary care staff, the service users may have been concerned that negative responses would impact their care. In the consent forms, they were told their care would not be impacted, but the context of the interviews may have not been reassuring for this point.

Similarly, the level of motivation of primary care centers to train more in mental health may be biased based on the sites participating. In some sites, the primary care centers taking part in this study had purposely been chosen as pilot sites for an intervention to reduce stigma based on their expressed interest and reliability and may therefore not have been representative of other primary care centers in their site. Another factor to consider is gender. The vast majority of PCPs interviewed (75%) were women. We do not know if men in these roles would have had similar perspectives.

Ultimately, the study provides an overview of some the types and diversity of attitudes and experiences related to stigma in primary care settings. However, given the type of data collection and methods described here, the findings should not be viewed as comprehensive and universal throughout the countries studied.

Recommendations

This study provides a number of lessons for both researchers and for practitioners interested in reducing mental illness stigma. We summarize these key recommendations below.

Research recommendations

1. For qualitative research on attitudes and experiences of service users, clinician-researchers may not be the ideal sole qualitative interviewers because of power differentials and expectations of service users about how to respond to persons they perceive as part of the healthcare system. For future research, it would be helpful to have qualitative researchers who are not part of the health system be included as interviewers. These interviews could also be conducted in non-clinical settings to help reduce the perception that the research team is directly linked to clinical care providers. Although the consent information in our study indicated that responses would be kept confidential, this may not have been the perception when interviews were conducted in clinical settings and sometimes by researchers who also worked in those facilities.
2. Collaboration with service users for qualitative data collection may contribute to reduced bias in responses of service users. Research teams could consider recruiting and training service users as qualitative researchers. Some of the sites in the study trained clinicians without prior qualitative experience to conduct interviews for this study. Those training efforts could be directed to service users. Similarly, service users could have been involved in the qualitative analysis to reflect upon whether the responses of participants were consistent with local experiences or if there may have been bias introduced by the recruitment and data collection strategies.

Stigma reduction and mental healthcare strengthening recommendations

1. Equip PCPs for taking on basic mental healthcare as suggested by the WHO [13] by providing training on identification of mental disorders, basic mental health treatments, psychosocial counselling, and communication skills as outlined in mhGAP training. For example, the statements by PCPs in Nepal and Lebanon, where mhGAP training had already been implemented, highlight that they felt much more confident and equipped to take on a role in providing mental healthcare in their site after their training—a finding which has also been documented in respective evaluation studies [36]. This suggests that implementing mhGAP training would seem to be a valuable next step in providing basic mental health training to PCPs in primary care settings where this has not yet been done. These mental health trainings could also be brought into pre-service training for health workers.
2. Prepare PCPs for being able to handle a wider range of interactions with service users with mental illness (according to local need) with empathy, skill, and confidence, including managing crisis situations. In a study of PCPs trained in mental healthcare in Liberia, Uganda, and Nepal, lower stigmatizing attitudes were associated with greater common factors of mental healthcare, such as empathy, collaboration, and promoting hope for recovery [37]. The importance of training PCPs in understanding and empathetic communication was mentioned specifically by several service users. New initiatives such as the WHO Ensuring Quality in Psychological Support (EQUIP) platform could be used to enhance empathy and communication skills [38].
3. Address universal and locally specific stereotypes, attitudinal barriers and stigmatizing behaviors among PCPs [6]. Our findings highlight the importance of addressing universally held stereotypes such as service users being dangerous or mental illness not being treatable, but also identifying any locally specific myths and addressing what matters most to service users and PCPs in different settings [39]. Based on the existing evidence in anti-stigma interventions in healthcare and particularly the findings of previous anti-stigma work done in one of our study sites, Nepal, we further suggest addressing stigmatizing stereotypes, attitudes and behaviors among PCPs through contact-based interventions employing trained service users [15, 22].
4. Train PCPs to support service users and families with the significant experiences of stigma enacted outside healthcare by providing psychoeducation, providing practical guidance and addressing internalized stigma [40]. To address internalized stigma, interventions should also cover basic counselling skills and simple interventions for self-esteem building and empowerment. One of the particularly salient findings across sites was experience of stigma enacted by family members, who were often carrying a large burden of care without adequate information or support. Training interventions for PCPs should therefore comprise delivering psychoeducation to service users and family members who often accompany service users to appointments and guidance with handling challenging situations at home. Findings from Nepal suggest that training service users for social contact interventions with PCPs should also include family members, which has the potential to also reduce family-based discrimination [19]. Similar interventions have been suggested and tried in the training of community mental health workers, e.g., in India [12] and in the International Federation of Anti-Leprosy Associations’ (ILEP) stigma reduction guidelines for health workers [41].
5. Assure that PCPs have ongoing support. Ongoing supervision and support are important for appropriate quality of care and to monitor motivation, treatment quality, and potential burnout as PCPs become more engaged in mental health services. There are many different models for this that could be adapted to the different countries; the issue is not just about training, but ongoing support for PCPs [42, 43].
6. In addition to training interventions targeted at individual health workers, interventions at the structural level are required. Primary care managers should review the work processes in their health facilities to identify any procedures that might expose service users to stigma, such as visibly marked patient files increasing fears of disclosure or consultation rooms lacking privacy. Checklists and tools should be developed or adapted to capture these aspects of structural stigma. Structural interventions should also seek to address PCP burnout and wellbeing by aiming to keep workloads manageable, creating support structures for staff, and teaching skills on self-care and burnout prevention.