Introduction
The world population is aging rapidly, particularly in low- and middle-income countries (LMICs) compared with high-income countries (HICs) (UN 2015), and while South Asia has in the past lagged behind other parts of the world in this, its fertility rate is declining and life expectancy is climbing, bringing on a rapid demographic transition (Chand, Reference Chand2018). In India, which accounts for most of the region’s population, 8.6% of the population are aged 60 and over, a proportion projected to reach 19% by 2050 (Agarwal et al., Reference Agarwal, Lubet, Mitgang, Mohanty and Bloom2016). The Lancet Global Mental Health series (2007 and 2011) highlighted the wide gap between mental health needs and provision (Eaton et al., Reference Eaton2011), but this gap is not well characterized in older people. Common mental disorders (CMDs), which encompass depression, anxiety, and related disorders, are a large part of the global disease burden, contributing substantially to morbidity, mortality, and decreased quality of life for people of all ages (Hay et al., Reference Hay2017), but particularly in older people (Guerra et al., Reference Guerra2016; Ogbo et al., Reference Ogbo, Mathsyaraja, Koti, Perz and Page2018).
There is a need to focus on older people across LMICs, but also on South Asia in particular. The majority of existing data in this area focus on HICs, with a relative lack in older people and LMICs. Initiatives such as the 10/66 International Dementia Research Collaboration (Prince et al., Reference Prince, Ferri, Acosta, Albanese, Arizaga and Dewey2007) and the Study on Global Ageing and Adult Health (SAGE) (Kowal et al., Reference Kowal2012) have greatly added to the understanding of mental disorders in older people in LMICs, however, they still leave much to learn. The work of the 10/66 Collaboration is primarily focused on dementia, and SAGE aims to provide a broad representation from different geographic regions across LMICs (Lotfaliany et al., Reference Lotfaliany, Hoare, Jacka, Kowal and Berk2019), rather than a sharply focused impression of any one region in particular. While it can be useful to examine an issue in relation to income status, for example, LMICs, this risks conflating a range of very distinct cultural contexts. South Asian countries are often considered together in large meta-analyses, citing challenges common across the region, such as lack of access to resources, stigma, and underdeveloped mental health legislation, policies, and plans (Hossain et al., Reference Hossain, Purohit, Sultana, Ma, McKyer and Ahmed2020; Ogbo et al., Reference Ogbo, Mathsyaraja, Koti, Perz and Page2018; Ranjan and Asthana, Reference Ranjan and Asthana2017). There may also be common psychosocial constructs across the region (Rama et al., Reference Rama, Béteille, Li, Mitra and Newman2014), suggesting that it is useful to consider South Asia as a whole in a review such as this.
Appropriate screening tools play an important role in the identification and characterization of mental health needs. They aid assessment in routine clinical practice, facilitate epidemiological work, which in turn, informs policymakers in allocating scarce resources appropriately according to need. There are well-known effects of culture on presentations of mental disorders and idioms of distress (Kerr, Reference Kerr2001), and evidence that mental disorders can present differently in older people (Yesavage et al., Reference Yesavage1982). It is currently not known, however, to what extent appropriate screening tools or measures for CMDs have been validated in South Asia, and where exactly the gaps in availability and validation of these measures are.
This scoping review aims to identify the extent to which screening tools for CMDs have been validated in older adults in South Asia, and to identify key research gaps and priorities for future work through performing a systematic search of the literature including validation studies, epidemiological studies, and existing literature reviews. The quality of the current validation data were assessed through the quality assessment of included validation studies using the Quality Assessment of Diagnostic Accuracy Studies – version 2 (QUADAS-2) (Whiting et al., Reference Whiting2011).
As a secondary aim, this scoping review will summarize the screening tools currently, and previously, being used in epidemiological studies of CMDs in older adults in South Asia and whether these have been previously validated and locally developed or culturally adapted.
The need for this review was identified by leading older people’s mental health clinicians and researchers from South Asia at the inaugural meeting of North East England South Asia Mental health Alliance (NEESAMA) in November 2018. Mental health of older adults is a key theme of this initiative.
Methods
Search strategy
A systematic literature search was performed in July 2019 in the following databases:
Medline, Embase, PsycInfo, Scopus, and Web of Science. The search strategy was developed iteratively in Medline and translated across the remaining databases. To capture CMDs, we combined the keywords, “depression” OR “affective” OR “anxiety” OR “GAD” OR “panic” OR “phobia*” OR “obsessive” OR “compulsive” OR “OCD” OR “trauma” OR “PTSD” with the Medline subject headings, “ANXIETY DISORDERS”/ OR “MOOD DISORDERS”/ OR “NEUROTIC DISORDERS”/ OR “TRAUMA AND STRESSOR RELATED DISORDERS”/ OR “MENTAL HEALTH”/. These were combined with thesaurus and free-text terms to capture older adults, and countries in the South Asia region.
Complete search terms for Medline are included in Appendix A1, published as supplementary material online attached to the electronic version of this paper.
Selection criteria
Studies were included if they presented data on any CMD in a community or outpatient older population sample (aged >50), from South Asia. The age of 50 was used as a cutoff rather than the current Indian standard of 60 or the HIC standard of 65 as some well-known seminal studies conducted in the 1990s used this 50+ cutoff and were considered relevant when planning this review. The South Asian countries included were Afghanistan, Bangladesh, Bhutan, Maldives, Nepal, India, Pakistan, and Sri Lanka. Only studies which used a tool, questionnaire, or instrument to identify or screen for a clinically defined CMD were included. Studies that used self-reported diagnoses considered only well-being rather than a psychiatric diagnosis of CMD, or did not specify the age range of the sample were excluded. Systematic and other reviews and opinion pieces were examined for references and the scope and major findings of reviews included for discussion and context of the current knowledge base, however, no data were extracted. There was no specifier for study design to restrict to validation studies as a pilot search indicated that this would exclude papers known to be of interest from the search results. This allowed the collection of all instances of tool use rather than just those in which the purpose was validation. The search was limited to peer-reviewed journal papers published between 1990 and August 2019 in the English language.
Search results were independently screened for eligibility by two authors (LF and CN) in a two-step process; title and abstract followed by full-text screening, as documented in the PRISMA diagram (Figure 1) (Moher et al., Reference Moher, Liberati, Tetzlaff, Altman and Group2009). Any disagreements were resolved via discussion and the input of a third reviewer (EBM). The reference lists of the 176 studies included after full-text screening were hand-searched to identify any other potentially relevant studies.
Figure 1. PRISMA flowchart.
Data extraction and synthesis
Validation studies
Studies presenting a new tool, or validating an existing tool were of primary interest and are presented in Table 1. Data extracted included: the country and region the study sample was drawn from; sample size (limited to participants in South Asia where several countries were included); setting (community dwelling, outpatient, or care home); urban or rural; CMD(s) assessed; name of tool, whether it was novel or previously developed and/or validated elsewhere and if so, where it was developed originally; language of the validated version.
Table 1. Adaptations of HIC older age depression screening tools as index tests with a reference standard
Prince et al. (Reference Prince2004) could not report sensitivity for the GMS for methodological reasons.
Abbreviations: OPD, outpatient department; GDS, Geriatric Depression Scale; HIC, high-income country; ICD, International Classification of Diseases; DCR, diagnostic criteria for research; PHQ, patient health questionnaire; CIS-R, Clinical Interview Schedule – Revised; EURO-D, EURO – Depression; CES-D, Centre for Epidemiological Studies – Depression; WHO, World Health Organization; opthal, ophthalmology; SCID, Structure Clinical Interview for DSM-IV; GMS, Geriatric Mental State + AGECAT algorithm; MADRS, Montgomery–Asberg depression rating scale.
Where the tool was compared to a reference standard, further information was extracted according to its performance as a diagnostic test (sensitivity, specificity, and cutoff value used) and gold standard assessment used for validation if relevant.
These studies were quality assessed using the QUADAS-2 (Whiting et al., Reference Whiting2011) by two authors (LF and EBM). In addition to the recommended signaling questions, one was added asking whether the approach to cultural adaptation or translation increased the risk of bias.
Epidemiological data
Data were also extracted from all epidemiological studies that used an instrument to diagnose a CMD, whether or not there was any validation described – see Table S1 published as supplementary material online attached to the electronic version of this paper. Data extracted included: demographic data on the population, participants, and study setting as above; study design; primary outcome of the study; CMD(s) assessed; tool used to identify CMD(s); type of tool (screening/diagnostic); where the tool was developed (HIC or LMIC); if the tool was validated for that context, and if not, where it was validated; any other significant features. These studies were not quality assessed other than to judge the cultural appropriateness of the tool used and no quantitative data were extracted. Their inclusion was intended to add to an understanding of how CMD screening tools are used in this context.
A proportion of the records, selected by systematic sampling, were independently reviewed to ensure the reliability of data extraction (S-MP). Due to temporary restrictions on access to library resources, 14 papers were not available. These were all epidemiological studies.
Results
Validation studies
Identification of novel tools or contextually adapted and validated tools was a central purpose of this review. A total of 15 tools aiming to meet this need were identified from the literature. One was a novel instrument developed in South Asia, and 14 were adaptations of existing instruments developed originally in HICs. Ten of these compared the adapted tool with an established gold standard. All tools which underwent a comparison with a gold standard in this way were for depression. These 10 tools were quality assessed using the QUADAS-2 (Whiting et al., Reference Whiting2011).
Novel tools
Only one entirely novel culturally specific tool was identified (Madnawat and Kachhawa, Reference Madnawat and Kachhawa2007) – aiming to specifically address death anxiety in urban community residents of Jaipur district, India. This scale was constructed de novo by local experts and so should be optimally relevant to this population. The study was not designed to assess any measures of validity, however, and so it is difficult to comment objectively on its merits compared to other, less culturally specific tools.
Adaptations of HIC tools as index tests with a reference standard
Ten studies made use of an existing screening tool developed in a HIC, translated, and/or adapted this to a particular cultural context and validated through comparison to a gold standard. In seven of these, the gold standard was an expert clinical interview, and in the others, this was a structured diagnostic tool (Geriatric Mental state [GMS], Clinical Interview Schedule – Revised [CIS-R] or Structured Clinical Interview for DSM-IV [SCID]). Eight out of these 10 studies were from India, 1 from Nepal, and 1 from Sri-Lanka. All screened for depression. They are detailed in Table 1. It can be seen here that the diagnostic accuracy of specific tools as a diagnostic test varies substantially across different settings. Studies of GDS-15 adaptions report sensitivities of 73–100%, and specificities of 48–94%, although a range of cutoffs was used. Area under the receiver-operator characteristic (AUROC) value was not commonly reported. An analysis of the specific adaptions, which mostly affected performance was beyond the scope of this study.
The QUADAS-2 tool, a quality assessment tool for diagnostic accuracy studies (Whiting et al., Reference Whiting2011) highlights quality concerns or unclear reporting in all but 2 out of the 10 studies assessed (Table 2). The most common source of bias identified was lack of blinding, that is, the index test was administered with knowledge of the results of the reference standard, or vice versa. The approach to patient selection was often incompletely reported, explaining the frequently unclear risk of bias in this category.
Table 2. QUADAS-2: quality assessment and risk of bias in diagnostic accuracy studies
✓, Low Risk; ✘, High Risk; ?, Unclear Risk.
Other measures of validity
A further four studies aimed to explore various measures of validity, but not examine performance as a diagnostic test. These were not included in the quality assessment. Two of these studies examined scales, which were subsequently or previously compared to an appropriate gold standard in an equivalent setting (Chokkanathan and Mohanty, Reference Chokkanathan and Mohanty2013; Ganguli et al., Reference Ganguli, Dube, Johnston, Pandav, Chandra and Dodge1999) and another for the same culture but in a different context (clinic vs. community) (Gautam and Houde, Reference Gautam and Houde2011). Their related validations are included above. Thus, the most significant exclusion from quality assessment is the GAD-7 Urdu translation for an urban community in Pakistan (Ahmad et al., Reference Ahmad, Hussain, Shah and Akhtar2017).
Screening tools utilized in epidemiological studies
As part of a wider aim to scope research conduct in this area, we have reviewed the use of CMD diagnostic and screening tools used in existing relevant epidemiological research. This existing work covers a broad range of settings (Table 3), and a broadly representative geographical area (Figure 2). The majority of studies (n = 144, 82%) measured only depression, and used the GDS (n = 95, 54%) (Table 3), predominantly the 15-item version. Only 57% used a tool that was culturally validated in line with WHO guidance (World Health Organisation, 2016), or with well-established cross-cultural applicability (e.g. GMS [Prince et al., Reference Prince2004], CIDI [Haro et al., Reference Haro2006]). Where a culturally validated screening tool was not used, most studies (n = 123, 70%) did not report a cultural adaptation or review of the tool used in the study methods, for example, using local translators and clinicians (see Table S1, published as supplementary material online attached to the electronic version of this paper).
Table 3. Summary characteristics of epidemiological data studies using screening and diagnostic tools
Abbreviations: GDS, Geriatric Depression Scale; CES-D, Centre for Epidemiological Studies – Depression; EURO-D, EURO – Depression; PHQ, patient health questionnaire; GMS, Geriatric Mental State, CIDI, Composite International Diagnostic Interview.
Figure 2. Map showing the distribution of all studies included in the review across South Asia.
Reviews
A total of 15 relevant reviews were identified in the search, including 7 meta-analyses, (Barua, Reference Barua2013; Barua et al., Reference Barua, Ghosh, Kar and Basilio2010; Barua et al., Reference Barua, Ghosh, Kar and Basilio2011a; Reference Barua, Ghosh, Kar and Basilio2011b; Brandão et al., Reference Brandão, Fontenelle, da Silva, Menezes and Pastor-Valero2019; Grover and Malhotra, Reference Grover and Malhotra2015; Parkar, Reference Parkar2015; Patel and Shaji, Reference Patel2010; Pilania et al., Reference Pilania2019; Prakash et al., Reference Prakash, Shaji, Bharath and Kumar2014; Qidwai and Ashfaq, Reference Qidwai and Ashfaq2011; Rao, Reference Rao1993; Sami et al., Reference Sami, Nilforooshan, Pachana and Oude Voshaar2015; Shaji et al., Reference Shaji, Kishore, Lal, Pinto and Trivedi2004; Thapa, Reference Thapa2019). Of the reviews identified, six were systematic reviews examining depression prevalence in India (Barua, Reference Barua2013; Barua et al., Reference Barua, Ghosh, Kar and Basilio2010; Barua et al., Reference Barua, Ghosh, Kar and Basilio2011a; Reference Barua, Ghosh, Kar and Basilio2011b; Grover and Malhotra, Reference Grover and Malhotra2015; Pilania et al., Reference Pilania2019). A further three systematic reviews assessed excess mortality in depression (Brandão et al., Reference Brandão, Fontenelle, da Silva, Menezes and Pastor-Valero2019), prevalence of all mental disorders in Nepal (Thapa, Reference Thapa2019), and the natural course of anxiety disorders (Sami et al., Reference Sami, Nilforooshan, Pachana and Oude Voshaar2015). The remaining six reviews were narrative reviews describing elderly mental health in India (Prakash et al., Reference Prakash, Shaji, Bharath and Kumar2014; Parkar, Reference Parkar2015; Shaji et al., Reference Shaji, Kishore, Lal, Pinto and Trivedi2004; Rao, Reference Rao1993), Pakistan (Qidwai and Ashfaq, Reference Qidwai and Ashfaq2011), and South Asia as a whole (Patel and Shaji, Reference Patel2010). There were no studies specifically on tool use. Data were not extracted systematically, however, findings relevant to this review are discussed below.
Funding
Seventy-three percent of the studies gave a statement regarding funding or conflict of interest. Funding was predominated by either local research grants or the funders of major epidemiological exercises such as the 10/66 Collaboration and the World Mental Health Survey.
Unavailable papers
Fourteen papers were not available in U.K. libraries and were unavailable for review. These were all small epidemiological studies, not anticipated to significantly impact findings.
Discussion
Validation and adaptation of tools
Reference standards
In validating a screening tool, an expert diagnostic interview remains the preferred gold standard (Ali et al., Reference Ali, Ryan and Silva2016), but this is a scarce resource in many settings and a more practical compromise must often be found. The CIS-R, for example, is a structured, standardized interview, for use by lay interviewers to detect minor psychiatric disorders such as anxiety and depression. It has been shown to be highly reliable in other LMICs, for example, Chile (Lewis et al., Reference Lewis, Pelosi, Araya and Dunn1992) which was then an LMIC, and Malaysia (Subramaniam et al., Reference Subramaniam, Krishnaswamy, Jemain, Hamid and Patel2006), although the Indian context remains unassessed. Similarly, the GMS/AGECAT algorithm, as used by the 10/66 group (Prince et al., Reference Prince, Ferri, Acosta, Albanese, Arizaga and Dewey2007) can produce a range of diagnoses. It is specifically designed for use by lay interviewers with older participants, accounts for cognitive impairment, and has performed well in LMICs including those where literacy rates may be low (Prince et al., Reference Prince2004). More widespread validation of these promising tools would provide an important resource for researchers struggling with the lack of available expert interviewers.
Adapting HIC tools for LMICs
This review has identified a tendency to use unvalidated screening tools without an adaptation process in epidemiological studies, undermining their conclusions. The WHO (2016) provides a straightforward summary of how the process of translation and cultural adaptation might be done. This and similar adaptation processes are known to improve the performance of tools as screening instruments for CMD across LMICs (Ali et al., Reference Ali, Ryan and Silva2016). The GDS, in particular, required alteration to meet the needs of different cultures in a cross-cultural assessment of geriatric depression (Mui et al., Reference Mui, Burnette and Chen2001), though it has been identified as a screening tool well suited to the Indian context (Prakash et al., Reference Prakash, Shaji, Bharath and Kumar2014) and performs well in the validation studies included in Table 1. Advocating for resource allocation, and the design of effective public policy to meet the mental health needs of older persons in South Asia requires good quality data available locally (Arokiasamy et al., Reference Arokiasamy2017; Ebrahim et al., Reference Ebrahim, Pearce, Smeeth, Casas, Jaffar and Piot2013), especially in LMICs where mental health needs are often overlooked in favor of the many other pressing concerns (Brandão et al., Reference Brandão, Fontenelle, da Silva, Menezes and Pastor-Valero2019). In omitting any cultural adaptation, the resulting data will not take account of the important influence of culture, language, and education on the presentation of mental disorders (Kerr, Reference Kerr2001; Prince et al., Reference Prince, Ferri, Acosta, Albanese, Arizaga and Dewey2007), and therefore the population’s needs may be inadequately measured and provided for.
A meta-analysis of the performance of screening tools for CMD across LMICs found that tools specifically designed for a particular setting tend to perform better than equivalent tools adapted from a HIC, however, more limited data evaluating these restricted the conclusions they were able to draw in comparing them with HIC versions in wider use (Ali et al., Reference Ali, Ryan and Silva2016). This may be because cultural factors fundamentally affect how the disorder is experienced. Simple translations of HIC tools may miss out on the emic factors, which can provide an important dimension to understanding disorders as they manifest locally. No validated examples of this were identified in this review, however, examples from elsewhere include the Shona Symptom Questionnaire for Zimbabwe (Patel et al., Reference Patel, Simunyu, Gwanzura, Lewis and Mann1997), the case description method for CMD in China (Sheng, Reference Sheng2010), the Chinese Military Mental Health Scale for CMD (Wang et al., Reference Wang, Zhang, Chen and Yao2012), and the Pakistan anxiety and depression questionnaire (Mumford et al., Reference Mumford, Ayub, Karim, Izhar, Asif and Bavington2005). These are not designed specifically for the older population however.
Reviews
Limitations of the current research base
Reviews identified through searching highlight the shortage of high-quality research available in this area. The lack of appropriate diagnostic and screening tools available to researchers contributes to this shortcoming.
Meta-analyses have demonstrated a wide range of prevalence estimates for CMD in older people, particularly in India. This may be in excess of that found worldwide (Barua, Reference Barua2013; Barua et al., Reference Barua, Ghosh, Kar and Basilio2011b; Pilania et al., Reference Pilania2019; World Health Organisation, 2001), however, these estimates are limited by methodological imprecision in the studies available for review (Barua et al., Reference Barua, Ghosh, Kar and Basilio2010; Reference Barua, Ghosh, Kar and Basilio2011b; Grover and Malhotra, Reference Grover and Malhotra2015; Pilania et al., Reference Pilania2019). This limits the extent to which these reviews can usefully inform the provision of measures to reduce the mental health gap for older people in South Asia.
A meta-analysis of the prevalence of depression in India (Pilania et al., Reference Pilania2019) found an overreliance on screening tools such as the Geriatric Depression Scale (GDS) and Centre for Epidemiological Studies – Depression Scale (CES-D) to make a diagnosis in epidemiological surveys, and noted an increased prevalence in studies relying on such instruments. Similarly, a meta-analysis in Nepal (Thapa, Reference Thapa2019) found the prevalence of mental disorders was significantly reduced when studies relying on screening tools were excluded, noting a particular reliance on the GDS.
There are therefore limited and imprecise data available to policymakers. In the data that do exist, there is an overreliance on screening tools used for diagnosis. Contributing to this imprecision is the lack of cultural validation in the screening tools that are used.
Outside of India
There was a particular lack of reviews concentrating on research outside of India. One meta-analysis found the prevalence of mental disorder in Nepal to be similar to that found in India if the analysis was limited to diagnostic interviews (Thapa, Reference Thapa2019). The one review focused on Pakistan covered older persons’ mental health broadly and did not attempt a meta-analysis (Qidwai and Ashfaq, Reference Qidwai and Ashfaq2011).
Lack of anxiety studies
This review shows that there is a clear focus on depression over other disorders such as anxiety. This is evident in the validation studies and epidemiological surveys found. A review of older persons’ mental health needs in India reported a common overlap of depression and anxiety with almost half of depressed older patients reporting significant anxiety symptoms (Parkar, Reference Parkar2015). Furthermore, a worldwide meta-analysis of longitudinal studies of anxiety disorders in older persons found a poorer prognosis of mixed anxiety and depression compared to pure anxiety or depression (Sami et al., Reference Sami, Nilforooshan, Pachana and Oude Voshaar2015). The finding that 82% of the studies included in this review considered only depression and not anxiety therefore highlights a significant research gap. There was only one example of an anxiety-specific screening tool identified in this review (Ahmad et al., Reference Ahmad, Hussain, Shah and Akhtar2017), and this was not compared to a gold standard. This lack presents a significant challenge to researchers aiming to fill this gap.
Limitations
This review was intended to provide an overview of the literature on screening tools for CMD in older persons in South Asia, and with such a broad scope, the treatment of each area lacked the depth of analysis that would be required to make stronger conclusions. Some of these areas (e.g. prevalence data or the cultural adaptation of tools) would have benefited from a review or meta-analysis in their own right. Similarly, limiting the quality assessment only to validation studies assessed against a gold standard precludes comments on data quality in other areas. Only English language papers were included due to language restrictions of the authors. Grey literature was not reviewed.
Conclusions and recommendations
A number of recommendations follow from the research gaps identified in this review; that is in the lack of culturally validated diagnostic and screening tools for use throughout South Asia, and particularly for anxiety disorders which have been relatively neglected.
The understanding of the mental health needs of older persons in South Asia still suffers from a scarcity of data, but also a lack of quality tools with which to tackle this. Validating appropriate tools should be a research priority.
There is a severe lack of attention to anxiety disorders both as a contributor to disease burden, and in terms of tools available to researchers.
The GDS appears to be the most suitable tool to adapt for screening older people for depression in this region, however, it has only been validated in a handful of settings, limiting its use.
As well as screening tools, research in resource-constrained settings would benefit from local validations of diagnostic tools to substitute for expert clinical diagnosis where this is not available.
There is a prominent lack of epidemiological studies in Bangladesh and of validation studies of any kind outside of India.
There is likely to be value in developing novel tools for specific cultural contexts as these may outperform tools adapted from HICs.
Conflicts of interest
None.
Source of funding
Supported by a grant from the British Council (Grant no. 525227071).
Description of authors' roles
LF – Database search, record screening (title and abstract), data extraction from full text, results analysis, critical appraisal, and primary author of the manuscript.
S-MP – Primary supervisor of the project, contributed to manuscript draft, review of data extraction, and project conception.
EBM – Methodology advice and supervision, final say where disagreement over inclusion, critical appraisal, and draft for methodology section of the manuscript.
CN – Record screening (title and abstract).
AL – Region-specific expertise, project conception, and input to manuscript drafts.
RW – Coordination of sharing of ideas between sites, project conception.
MV – Coordination of sharing of ideas between sites, project conception.
All authors approved the final version submitted.
Acknowledgments
We would like to acknowledge the North East England South Asia Mental health Alliance (NEESAMA), which made this collaboration possible and, in particular, to Dr Aditya Sharma and Professor Jacqui Rodgers for their support of the work of NEESAMA. We are especially grateful to Newcastle University, Cumbria, Northumberland, Tyne and Wear NHS Foundation Trust (CNTW) and the British Council for their financial support of NEESAMA.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S1041610220003804.
Open access
