Migrant health and language barriers: Uncovering macro level influences on the implementation of trained interpreters in healthcare settings

There is a knowledge translation gap between policies promoting equitable access to healthcare and person-centred care, and the use of untrained interpreters in cross-cultural consultations leading to disparities in health outcomes. An 11 member inter-sectoral working group met at four workshops to discuss and agree on levers and barriers to the provision of trained interpreters in healthcare settings in Ireland. The process was informed by Participatory Learning and Action (PLA) research to support interstakeholder dialogue and learning. Normalisation Process Theory (NPT) was used as a conceptual framework to analyse levers and barriers. The NPT analysis explored sense-making, engagement and enactment and found challenges with sense-making and engagement in senior level service planners, managers and governmental offices. This had negative impacts on other key actors, including healthcare providers, medical students and interpreters. This also meant that the enactment of interpreted consultations in practice settings was replete with barriers, most notably a lack of resources, training and supportive organisational structures. The emergent action plan focused on improving sense-making and engagement through intersectoral awareness raising, designed to stimulate a series of complementary levers for implementation. Combining PLA and NPT provided new insights into macro level influences on implementation work at the level of a national healthcare system. The approaches used in this study are applicable in other fields. © 2021 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license ( http://creativecommons.org/licenses/by/4.0/ )


Introduction
International migration has been increasing in recent decades. Current estimates suggest that there are 281 million international migrants [1] . There is no universal definition of 'migrant' [2] and in this paper we focus on migrants who are not fluent in the language of their host country irrespective of why or how they migrated.
In its Strategy and Action Plan for Refugee and Migrant Health, the World Health Organization's Regional Office for Europe recommends the adaptation of health services so that they are accessible and responsive to increasingly diverse populations [3] . This includes adaptations to manage cross-cultural consultations when healthcare providers and migrants do not share a language and cultural background. Studies have shown that language and working to a professional code of conduct and they may not be used by healthcare professionals. Common informal responses such as reliance on ad hoc supports in the form of family members and friends as untrained interpreters [12][13][14] introduce a greater chance of error into the interpreted consultation [15] . Considering the interpreted consultation as a complex intervention, it is useful to explore micro, meso and macro level factors that will promote or inhibit its implementation in daily routines in healthcare settings. At a micro level, research shows that interactional factors are important. For example, patients may prefer to use a family member or friend because they consider that this person will represent their concerns and advocate for them [14] . Healthcare providers, particularly those who have had no training, can also find the triadic nature of the interpreted consultation unusual [ 9 , 16 ]. At meso level, it is challenging to organise and incorporate an interpreted consultation into busy clinical routines, particularly in primary care, where 10-minute consultation slots are usual [17] . A recent European study, using Normalisation Process Theory (NPT) [18] emphasised the importance of paying attention to hitherto under-researched macro level factors such as policy and resources that will enable or constrain changes to organisational routines in daily practice settings [17] . NPT is designed to alert policy makers, healthcare managers, healthcare providers and researchers to macro, meso and micro level influences on implementation [18] . It offers a description of four types of implementation work that stakeholders have to engage in, individually or collectively. These are summarised in Table 1 and can be considered to have organic and fluid relationships with each other as they occur in real space and time during implementation processes.
In this paper, we report the use of NPT and participatory learning and action research to inform the work of an inter-sectoral national working group charged with analysing the provision of trained interpreters in healthcare settings in Ireland. To our knowledge, this is the first analysis in the field of implementation work conducted at the level of a national healthcare system. It offers an opportunity to learn about macro level influences, which has not been done so far. Previous analyses of levers and barriers come from patient and service provider perspectives (micro level), whereas the current analysis includes perspectives of senior healthcare system service planners and experts from related sectors (see Methodology).
The specific objectives of our paper are two-fold. First, we describe the use of participatory approaches and NPT to develop an action plan for the national health service to support the implementation of routine use of trained interpreters during healthcare consultations. Second, we use NPT to analyse the levers and barriers identified and conceptualise the emergent action plan to elucidate the complexity of macro level aspects of implementation.

Methodology
The University of Limerick Education and Health Sciences Research Ethics Committee advised that ethical approval was not required as the work constituted reflective practice and collating of experiential and published knowledge.

Setting
Ireland started to experience immigration in recent decades. Census data from 2016 show that around 17% of the Irish population are foreign born [19] . The groups with the largest increase since the 2011 census were from Romania, Brazil and Spain. The data also show that Ireland has become a multilingual country, with Polish and French being the two most commonly spoken foreign languages. Significantly, among the individuals speaking a language other than English at home, 14% indicated that they could speak English 'not well' or 'not at all'.
There have been efforts to address the linguistic and cultural barriers faced by migrants in Ireland [20] . Irish healthcare providers might be aware of the importance of using interpreters; however studies have shown that healthcare providers still 'manage' to communicate without an interpreter (muddle through) or use the services of ad hoc interpreters such as family members or friends, hospital staff or other bilingual individuals [ 13 , 21 , 22 ]. Evaluation of a free pilot interpreting service in general practice settings in the east of Ireland (2007)(2008)(2009) found that primary general practitioners (GPs) continued their consultations without interpreters or with ad hoc , informal interpreters [16] . This has become the status quo in Ireland, which raises questions about patient safety and the equity of access to healthcare for migrants with limited English.

The working group (health system stakeholders)
A working group comprising 11 individuals, including the authors, representing six stakeholders, was established by the National Social Inclusion Office at the Health Service Executive (HSE), the government entity providing public health and social care services in Ireland. Members of the working group were invited to participate on the basis of their expertise in interpreting and/or social inclusion. They had collective knowledge of service planning, training, practice and research on interpreting and included HSE senior service planners (4), educationalists (2), interpreters (2), an HSE primary care nurse (1), migrant (1), and researchers (3) who designed and facilitated the process. One participant represented the educationalist and interpreter profiles while another represented the interpreter and migrant profiles. This was a multi-stakeholder group, independent of HSE, with diverse perspectives on routine use of trained interpreters in Irish healthcare settings.

Data generation and analysis
Four workshops were held to identify levers and barriers to the implementation of routine use of interpreters in the Irish healthcare system and an action plan to address the barriers. The workshops were informed by NPT and participatory research.
Participatory research encompasses a broad family of approaches. One of these is Participatory Learning and Action (PLA) research. This was developed by Chambers in the Global South and is an adaptive strategy designed to bring diverse groups and individuals together and enable them to participate together, to learn and act on a shared goal [23] . The desired outcome is a comprehensive and workable plan for shared action. PLA has been adapted  and used in healthcare settings [24] , including in implementation research about interpreting [ 13 , 17 ]. PLA involves the use of visual, goal-oriented tools and techniques that enable stakeholders to explore shared and differential perspectives on the problems and solutions under investigation [ 23 , 24 ]. In keeping with the flexibility of PLA [24] , tailored PLA techniques were used in the workshops to stimulate discussion and analysis about levers and barriers (flexible brainstorming) and potential actions (wall of challenges and tree of solutions). Fig. 1 provides a visual summary of the questions asked and the PLA tools used.
To support the analysis of material generated during the workshops, NPT was employed as a heuristic device for 'thinking through' implementation to ensure that all forms were considered. Specifically, the academic team mapped the emergent information from workshops 1 and 2 about levers and barriers against NPT's constructs. Any gaps in discussion of implementation were then addressed. For example, after workshop 1, the NPT appraisal revealed that there had been no discussion about implementation by hospital administrators and managers. Therefore, a specific prompt was introduced in workshop 2 to explore this topic. In total, we identified 10 levers and 11 barriers to implementation and 18 potential actions ( Table 2 ).
For the purpose of this paper, we advanced the theoretical analysis further. The thematic analyses of levers, barriers and action plan ( Table 2 ) generated during the workshops were mapped against NPT constructs to clarify which kind of implementation work they related to. AM developed a coding frame and did an initial mapping. SJP then independently used the coding frame to map the findings (See Supplementary table 1 ). The analyses were compared and discussed to resolve differences. The analysis was iteratively developed and finalised with all co-authors. The levers, barriers and action plan mapped onto three NPT constructs: sense-making, engagement and enactment. There were no data relating to appraisal because no data were available about the routine use of trained interpreters over a period of time in the Irish setting. No data fell outside the conceptual framework of NPT.

Sense-making-does it makes sense to routinely use trained interpreters?
Several barriers ( Table 2: barriers 2, 4, 6 and 10) indicate that the need for the routine use of trained interpreters does not make sense to several key stakeholders. This 'starts from the top': working group members considered that there was limited political will in senior HSE management and government departments (Health and Justice) to mandate and promote the routine use of trained interpreters. This was considered important as it contributed substantially to other barriers such as lack of resources and developing training programmes for interpreters.
Limited awareness among these stakeholders and clinical and non-clinical HSE staff (e.g. managers, receptionists) regarding the clinical risks associated with not using trained interpreters was considered a major barrier. For many healthcare staff the status quo was to use patients' family members and friends as interpreters. This was seen as an 'easy' solution and lack of trained interpreters as a pity, but not a problem. This indicates lack of sufficient critical analysis or understanding of differences between these options for supporting communication with migrants during cross-cultural consultations.
Racism and ethnocentrism were discussed as a barrier that influences and is influenced by the sense-making process. While lack of trained interpreters in the healthcare system leads to disparate treatment of migrants resulting in racism and ethnocentrism, conscious and unconscious societal racism and ethnocentrism were considered as possible reasons for the lack of political will to establish routine use of trained interpreters.
Lack of awareness about clinical risks and concerns about the cost of interpreting often led to the myth that 'interpreting will cost a lot', which in turn hinders effort s to promote routine use of trained interpreters. Therefore, to counteract this myth, a review of national and international literature and data was suggested as an action to make a business case for the implementation of trained interpreters and to counteract this myth. The working group identified some existing practices and experiences as levers ( Table 2: levers 6-9) to strengthen sense-making. Healthcare providers who were trained to work with trained interpreters found the experience transformative since it allowed them to focus on their own clinical role and tasks, rather than 'muddling through' the consultation. The same was true for trained interpreters who worked with clinicians trained to work with them. The working group considered highlighting such examples as levers to help stakeholders, mainly healthcare providers and interpreters, to differentiate between the status quo and the use of trained inter- preters. Enhancing differentiation, combined with highlighting the clinical risks for patients' health of not using trained interpreters, was considered necessary to create a shared understanding about a new practice among these stakeholders.
Migrants and healthcare providers felt more trust in each other if a trained interpreter was used during their consultations. Highlighting such experiences among migrant communities was considered to facilitate the sense-making and was seen as an important lever for creating demand from the user (migrant patient) side for using trained interpreters.
The actions/solutions categorised under this construct ( Table 2 , actions 1, 2, 5-7, 9, 13, 17, 18) mainly dealt with information gathering and awareness raising. They involved reviewing documents to see what measures are being taken at government level (Health and Justice) to tackle racism and ethnocentrism in intercultural and/or migration integration strategies; reviewing professional codes of ethics for medical professionals to establish if they included the clinical risks of not using trained interpreters; and finding out if undergraduate and postgraduate training courses for healthcare staff (medicine, nursing, allied health) covered these risks. Another action was identifying sources of governmental or HSE funding to support the development of training for healthcare staff to work with trained interpreters.
Overall, there was a specific interest in advocating for a whole-of-government approach to develop coherent and comprehensive policy and practice to support the implementation of trained interpreters. This relied on actions to raise awareness among government ministers, senior HSE officials and healthcare providers. Specific actions were to highlight the issue at the launch of the second HSE intercultural health strategy and to develop infographics focusing on the clini-cal risks and patient safety problems associated with informal interpreters.

Engagement -How committed are stakeholders to implementing routine use of trained interpreters?
Due to low sense-making and the acceptance of the status quo, commitment was found to be low among healthcare staff and service planners. As a result, healthcare providers did not actively demand trained interpreters. Migrants were unaware of their limited rights and entitlements ( Table 2: barrier 7) and as a result did not demand trained interpreters. The barriers that prevented interpreters from feeling the need to get trained and improve their skills were poor working conditions (barrier 8) and the likelihood that investment in training would not result in better rates of pay.
The working group identified three levers ( Table 2: levers 2, 4, 5) that could strengthen the commitment of key senior governmental and health system stakeholders. The Irish Human Rights and Equality Commission Act 2014 [25] requires public sector organisations, including the HSE, to monitor and address inequities in service provision and could be a basis for demands for trained interpreters to reduce health inequalities. Action ( Table 2: action 8) focused on obtaining legal advice on these matters.
Another important lever was the identification of key individuals to raise awareness and mobilise migrant communities to demand trained interpreters. In parallel, the identification of key interpreters who uphold standards and advocate the routine use of trained interpreters was seen as a key lever to initiate engagement and drive it forward. To catalyse the use of trained interpreters by medical professionals, the Irish Translators' and Interpreters' Association was requested to review codes of ethics of various associations of medical professionals ( Table 2: action 10) to determine if they mentioned or encouraged the use of trained interpreters.

Enactment -What actions could be taken to implement the routine use of trained interpreters?
The lack of resources, training and organisational structure to implement and routinely use trained interpreters is evident among the barriers ( Table 2: barriers 1 ,3,5,9 and 11). Limited understanding on why and how to routinely use trained interpreters has resulted in a lack of budgetary resources. As a result, there is limited experience and understanding (see 'Sense-making' above) among stakeholders of the need to use trained interpreters. Similarly, the absence of both interpreter training and training for healthcare staff in how to work with interpreters acts as a barrier.
Increased time required by GPs for an interpreted consultation was not accounted for in service planning, making it challenging for GPs to use interpreters during clinical consultations. Lack of separate waiting areas for patients and interpreters created challenges for interpreters when patients requested their assistance with various tasks outside their professional role. Interpreters were seldom briefed about upcoming assignments, meaning they could not prepare beforehand. There was no unit/agency or person responsible at county or national level for coordination of measures to introduce and organise the routine use of trained interpreters among healthcare staff (clinical and non-clinical), interpreting companies and various authorities involved.
Measures identified to strengthen this construct ( Table 2: levers 3 and 10) were to share national and international good practices for improving the skill sets of interpreters and healthcare staff. Such training, based on previous experience and available evidence, involving the three main actors in a clinical consultation (healthcare staff, trained interpreter and migrant patient) was identified as an important lever. Another lever identified ( Table 2: lever 1) was the use of technology (telephone and video interpreting) to improve access to the few trained interpreters that are available.
Actions 5 and 16 ( Table 2 ) aimed at increasing availability of funds and other resources for training programmes, whereas actions 3, 11, 12, 14 and 15 ( Table 2 ) focused on the need to facilitate training of interpreters and all categories of healthcare staff to routinely use trained interpreters. While training is important, conditions such as organisation of consultations and regulatory framework also need to be in place for the routine use of trained interpreters to happen. Actions 4 and 14 ( Table 2 ) focused on reviewing such existing conditions with the aim of suggesting improvement.

Discussion
Implementing the routine use of trained interpreters is a major challenge in many healthcare settings. Our analysis using NPT provides new insights into implementation work at the level of a healthcare system. It elucidates that there were fundamental macro level problems with sense-making and engagement within senior level HSE and governmental offices. This had negative impacts on sense-making and engagement of other key actors, including healthcare providers, medical students and potential students of interpreting. This also meant that the enactment of interpreted consultations in practice settings is replete with barriers, most notably a lack of resources, training and supportive organisational structures. The emergent action plan focused on improving sensemaking and engagement through inter-sectoral awareness raising, designed to stimulate a series of complementary levers for implementation.
The strength of this analysis is that it is underpinned by implementation theory -NPT. This ensured a comprehensive investigation of implementation work. It offers a generalisable framework for accumulating knowledge with other NPT studies about implementing interpreted consultations. The analysis was strengthened by our use of PLA as this supported inter-stakeholder dialogue and learning. While the working group comprised 'information rich' stakeholders, there were limitations as there was minimal representation of migrants, NGOs, companies providing interpreting services, HSE service managers and department of Justice who might well have brought additional perspectives to the analysis. Further, discussions in the workshop were about the healthcare system in general. Thus, issues about interpreter provision in specific settings or with specific types of migrant, for example, provision of interpreting in emergency settings for undocumented migrants, were not examined.
This analysis of barriers to implementation of interpreted consultations, generated by a health system working group, resonates strongly with existing literature in terms of health system barriers with resources, training deficits and interactional challenges [ 4 , 12 , 14 ]. In addition, this analysis adds to the field because it highlights inter-sectoral and societal barriers, not previously documented in the literature about implementing interpreted consultations. These include gaps in educational curricula for healthcare students, challenges with professional accreditation for interpreters, lack of political will across government departments and ethnocentrism.
The analysis reveals the complex inter-play between levers, barriers and actions. For example, there is a 'supply and demand' problem, which requires multiple, parallel and/or sequential interventions. It is necessary to improve the working conditions and professional status of trained interpreters so that a supply of interpreters is available, and it is necessary to increase healthcare providers' and managers' awareness of the need to use trained interpreters so that there is a demand for interpreted consultations. One intervention without the other will be fruitless. Furthermore, this analysis reveals the limits of agency among individual working group members to drive the implementation work forward because they rely on additional layers of stakeholders in the health service and other sectors to also be engaged in the implementation work. Finally, this analysis reveals why this is hard to achieve. HSE and governmental stakeholders appear to have limited understanding of clinical risks associated with the status quo and have concerns about the cost implications of interpreting. Thus, they lack accurate, impactful information to increase political will and motivation to mobilise resources and other necessary levers among competing priorities and limited resources at various levels of the government and HSE. Taken together, this analysis elucidates the nature and complexity of implementation work within the health system to address macro level barriers to the routine use of trained interpreters, while continuing to address the barriers at micro and meso levels.
The implications of this analysis for policy and practice are twofold. First, as with other complex interventions, context matters [26] and it matters that context is fluid and adaptable [27] . Therefore, notwithstanding the importance of education and training interventions to support the implementation of trained interpreters, it is important to increase attention to the layers and dynamics of contexts that shape policies on provision of trained interpreters. Participatory approaches, such as PLA, emerge as a valuable strategy to design and support inter-stakeholder and inter-sectoral dialogues. The attention within participatory approaches to multiple perspectives leads to a comprehensive knowledge base about levers, barriers and action plans. This in turn reveals the potential for participatory approaches to support strategic navigation and negotiation for service planning and policy making within the health service and government departments.
Second, as with other complex interventions, there are a range of barriers and levers [26] . It is important to continue to document and understand these. It is also important, however, to improve knowledge about the weighting of barriers and levers and to understand how exactly we can use levers to overcome barriers and/or to understand more about the relationship between them. Is it a case of working through documented barriers one by one? If so, in which order? Or are there major interventions that can expedite implementation? It is interesting, for example, to consider what impact the introduction of a legal instrument in the Irish context will have on the status quo. The aforementioned Irish Human Rights and Equality Commission Act (2014) [25] puts a clear responsibility on the HSE to monitor and address any discriminatory practices. Therefore, this could act as a very significant lever to change in this regard and warrants careful monitoring and evaluation.
Finally, as mentioned at the outset, interpreter provision addresses only one aspect of cross-cultural communication, the language barrier [9] . Other strategies such as cultural mediation, health literacy programmes and training for health care providers are also important and need to be implemented to holistically address communication barriers faced by migrants [ 5 , 6 ].

Conclusion
In conclusion, the implementation of trained interpreters in routine healthcare delivery is shaped by complex macro factors that require inter-sectoral attention and policy-making. This is important to support healthcare adaptations for diversifying populations in line with international policies and health equity goals.

Declarations of Competing Interest
None.