The other side of “getting by”: A case study of interpreting provision decision-making and consequences for patients

Abstract Rates of provision of professional interpreting services to patients have been shown to be low in hospital emergency departments and wards. This study aimed to elicit the reasons for the context and consequences of non-provision of professional interpreting at an adult hospital and identify strategies to increase provision. This was a qualitative case study of multiple perspectives (including patients, family members, interpreters, clinical and administrative hospital staff) using in-depth interviews, participant-observation, focus group discussions, staff survey and medical record review. The data were analysed using thematic, content and systems thinking analysis to develop a theoretical framework for providers’ decision-making processes and contextual constraints. The patient and family perspectives showed that ad hoc communication negatively affected their hospital experience and patient-centred care, and highlighted errors in communication largely unknown to treating staff. Key reasons shown for low rates of professional interpreter engagement by staff were: (1) a lack of familiarity and clarity of the process of engaging interpreters combined with inadequate infrastructure, (2) low levels of trust in and confidence in working with professional interpreters and (3) little knowledge of the evidence-base (including cost-benefit) or negative consequences resulting from “getting by”. The study shows that influencing norms through (a) clarifying the pathway from identification of need to engagement of interpreters, (b) provision of clinical staff training on the evidence base and role of interpreter (c) influencing training and professionalism of interpreting are important for improving professional interpreting provision rates in a hospital setting.

The other side of "getting by": A case study of interpreting provision decision-making and consequences for patients Samantha Abbato 1 , Jennifer Ryan 2 , Chris Skelly 3 and Phillip Good 4 * Abstract: Rates of provision of professional interpreting services to patients have been shown to be low in hospital emergency departments and wards. This study aimed to elicit the reasons for the context and consequences of non-provision of professional interpreting at an adult hospital and identify strategies to increase provision. This was a qualitative case study of multiple perspectives (including patients, family members, interpreters, clinical and administrative hospital staff) using in-depth interviews, participant-observation, focus group discussions, staff survey and medical record review. The data were analysed using thematic, content and systems thinking analysis to develop a theoretical framework for providers' decision-making processes and contextual constraints. The patient and family perspectives showed that ad hoc communication negatively affected their hospital experience and patient-centred care, and highlighted errors in communication largely unknown to treating staff. Key reasons shown for low rates of professional interpreter engagement by staff were: (1) a lack of familiarity and clarity of the process of engaging interpreters combined with inadequate infrastructure, (2) low levels of trust in and confidence in working with professional interpreters and (3) Jennifer  Through their partnership, Mater Health Services have been invested in identifying and addressing the needs of their patients of cultural and linguistic diversity (CALD) since 2010. This has included: regular auditing of professional interpreting rates across a number of hospital campuses and monitoring changes in these rates; cultural and language needs analysis of a diverse range of CALD consumer groups; understanding the context in which provider decisions are made that impact CALD patients and identifying strategies for improvement in the health service response to CALD patients.

PUBLIC INTEREST STATEMENT
Rates of professional interpreting in emergency departments and wards remain relatively low. When professional interpreters are not engaged, the quality of care is decreased and patient safety compromised. Studies seeking to understand the context of healthcare communication and interpreting for patients with low English proficiency have focused on health provider decision-making. To our knowledge, this is the first study to include an examination of the patient experience of communication and interpreting and their perspective of the consequences of provider decisions to use ad hoc and no interpreting through their hospital journey. We show that from the patient perspective, failure to engage professional interpreters can be detrimental to patient care and can unfairly burden family members. This dissatisfaction with care remains largely invisible to the provider. Knowledge of the experience of "the other side" and the real-life impact of decision-making that accepts ad hoc interpreting may help shift health practice norms.
little knowledge of the evidence-base (including cost-benefit) or negative consequences resulting from "getting by". The study shows that influencing norms through (a) clarifying the pathway from identification of need to engagement of interpreters, (b) provision of clinical staff training on the evidence base and role of interpreter (c) influencing training and professionalism of interpreting are important for improving professional interpreting provision rates in a hospital setting.

Introduction
Communication between clinicians and patients is a crucial component of health care and has a significant effect on emotional wellbeing and clinical outcomes (Simpson et al., 1991). The importance of engagement of professionally accredited interpreters for limited English proficiency (LEP) patients in the hospital setting is supported by substantial evidence (Chan et al., 2010;Flores, 2005;Karliner, Jacobs, Chen, & Mutha, 2007). Numerous studies report that the quality of healthcare is compromised when LEP patients need but do not get access to trained, professional interpreters (Flores, 2005;Karliner et al., 2007). The use of untrained "ad hoc" interpreters, including family members and friends of patients, non-clinical hospital employees and hospital clinical staff with no formal interpreter training, is associated with decreased quality of care and increased patient safety risks (Flores, Abreu, Barone, Bachur, & Lin, 2012;Flores et al., 2003;Karliner et al., 2007). It has been shown that the quality of health care and health outcomes are compromised through the use of ad hoc interpreters to a similar extent as they are in the absence of any interpreting (Flores et al., 2012(Flores et al., , 2003. In addition, patient satisfaction with care is increased through the provision of professional interpreting (Karliner et al., 2007;Mahmoud, Hou, Chu, Clark, & Eley, 2014).
Despite the ample evidence supporting the importance of professional interpreter provision for effective communication between LEP patients and their health care providers, rates of interpreting provision in the emergency departments and wards of Australian, New Zealand and United States hospitals remain low (Garrett, Forero, DIckson, & Klinken, 2008;Ginde, Clark, & Camargo, 2009;Gray, Hilder, & Donaldson, 2011;Kazzi & Cooper, 2003;Lopez, Rodreguez, Huerta, Soukup, & Hicks, 2015;Ryan, Abbato, Greer, Vayne-Bossert, & Good, 2017;Schenker, Perez-Stable, Nickleach, & Karliner, 2011). Even when professional interpreters are readily available and healthcare providers perceive the benefits of using them, accredited interpreters are consistently used for less than 20% of patients with LEP (Diamond, Schenker, Curry, Bradley, & Fernandez, 2009;Ginde, Sullivan, Corel, Caceres, & Camergo, 2010;Hsieh, 2015;Schenker et al., 2011). Others have shown that the engagement of interpreters in the hospital setting is complex (Diamond et al., 2009;Hsieh, 2015). Two recent in-depth qualitative studies have examined the complexity involved in hospital providers' decision-making processes to engage professional and ad hoc interpreters for communicating with LEP patients (Diamond et al., 2009;Hsieh, 2015). Both studies focused on the clinical provider perspective and included in-depth interviews of resident physicians (Diamond et al., 2009) and a combination of in-depth interviews and focus group discussions with health care professionals (Hsieh, 2015). Major factors influencing the clinician decision to engage professional or ad hoc interpreting included: (a) time constraints, (b) therapeutic objectives, (c) alliances of care and (d) organisational-level considerations (Hsieh, 2015).
Provider perceived time constraints were shown to be a barrier to professional interpreter use in both studies with providers frequently choosing ad hoc methods of interpreting. Both studies showed that clinical providers regularly used family members or friends of patients to interpret, bilingual staff members or got by with hand gestures when waiting for an onsite professional interpreter was perceived as creating a major delay to their patient schedule. More readily available phone or video interpreters were not seen as a viable option because of poor quality and perceived lack of interpersonal closeness and rapport (Diamond et al., 2009;Hsieh, 2015).
In deciding whether or not to engage a professional interpreter, providers weigh up the perceived value of the communication for patient care against their time constraints (Diamond et al., 2009;Hsieh, 2015). For example, professional interpreters are more likely to be engaged when the procedure is considered major (e.g. discharge instruction or family meetings or consent for procedures) compared to minor (e.g. pain management) (Diamond et al., 2009;Hsieh, 2015). These studies show that providers often consider family members the best solution as interpreters in a clinical emergency, in part because of perceived lack of trust in phone interpreting (Hsieh, 2015;Hsieh, Ju, & Kong, 2010).
At the organisational level, resource limitations such as the availability of onsite interpreters, adequacy of equipment and hospital policies and procedures have been shown to influence clinician engagement of interpreters (Diamond et al., 2009). For example, if standards for communicating with patients with LEP are unclear, speaker-phones are not available and underuse of professional interpreters common, clinicians may be less likely to engage a professional interpreter even if they personally see this as an appropriate decision.
Importantly, professional interpreting engagement practices are ultimately impacted by the culture and norms of the practice environment as well as at the level of the individual clinician. At the level of the practice environment, appropriate interpreter use requires the establishment of clear norms endorsed by senior hospital leadership, in additional to structural changes (Diamond et al., 2009).
It has been argued that providers' underuse of professional interpreters should not simply be viewed as problematic and that different ways of communicating with LEP patients, including through ad hoc interpreters should be viewed as complementary (Hsieh, 2015). Hsieh (2015) asserts that providers' decision-making is calculated and includes consideration of interpersonal, organisational, therapeutic and ethical considerations. However, a major omission from this To our knowledge, this is the first explanatory case study of communication with LEP patients and interpreter engagement through the hospital journey from the emergency department through to the adult wards that incorporates the in-depth experience of patients and their families. It includes the in-depth exploration from multiple perspectives (including patients, family members, clinical, nursing, allied health, hospital administration and interpreters) of the complexity and uniqueness of interpreter engagement as a practice within a particular system in "real-life" experience and context. The objective of this study was to understand the reasons for low rates of professional interpreter engagement within the complexity of a specific hospital context incorporating multiple perspectives involved to identify key strategies for increasing appropriate interpreter engagement and improved cultural responsiveness. In addition, we wanted to elicit the lived experiences of patients and their family members resulting from the provider interpreting decisions throughout their hospital journey.

Study site
The study was conducted at a public tertiary referral adult hospital in Brisbane, Australia. Approximately 2.8% of patients admitted through the ED to the adult hospital are identified through the hospital database as requiring an interpreter (based on 2013-2014 hospital statistics), and is reflective of local population statistics (Australian Bureau of Statistics, 2013). Hospital statistics identified a total of 64 distinct languages spoken by patients (2013)(2014). Interpreter services were available through two full-time interpreters providing Cantonese, Mandarin and Vietnamese patient interpreting services during daytime hours, a pool of 35 contracted interpreters providing services for 22 languages and 4 different interpreting agencies for provision of onsite and phone.
Approval to undertake this qualitative study was granted by the Mater Health Services Human Research Ethics Committee (MHS HREC).

Study design
An explanatory case study approach was used to understand the real-life system of communication with patients of LEP in a hospital emergency and adult ward environment (Yin, 2014). The case study approach is appropriate for seeking evidence for "how" and "why" something occurs (Simons, 2009). The purpose of this case study was to explain "why" a low rate of professional interpreter engagement occurs at the ED and adult wards when the hospital policy states that professional interpreting is to be provided to all patients who need it. The focus was on understanding the decision-making for engaging interpreters within the specific hospital setting. The approach also enabled an understanding of "how" provider communication and interpreting decisions impact patients with LEP and their families.
A mixed-methods approach was used. Specific methods included: interviews with patients, family members and providers, focus group discussions with providers and interpreters, observations in the adult ward, surveys of providers, analysis of hospital documents, patient health records and hospital databases, including interpreting records. Rigour of method was maximised through triangulation of data source (patient, family member, health record, database, staff) and of interviewer (three researchers conducted interviews, analysed and compared and contrasted findings).
A systems thinking approach (Meadows, 2008;Williams & Hummelbrunner, 2011) was incorporated into the case-study analysis for communicating the complexity of provider decision-making and identifying potential areas for intervention for successful change within this specific hospital context.

Sampling and recruitment
Participant recruitment for the case study took place over a period of 2 months (November-December 2015).

Patients with LEP and their family members
For a period of 5 weeks, all patients admitted to the adult wards from the ED and requiring an interpreter were identified via the hospital database. A targeted sampling frame that included gender, age and language group was used as a guide for participant recruitment to facilitate a sample of patients representing a diverse range of patient demographics (See Table 1). For patients identified as LEP and fitting the targeted sampling frame, the Cultural Diversity Coordinator (CDC) consulted ward nursing staff to establish the patient's discharge date and ascertain the capacity of the patient to participate in the research. Of a total of 34 patients identified as eligible to participate from the database, 17 were already discharged or preparing for discharge and 2 were assessed by nursing staff as too unwell to participate. Of the 15 patients invited to participate by nursing staff or the CDC with the aid of translated or interpreted participant information sheets and consent forms, 2 refused consent. One of the final 13 recruited patients was unable to participate directly as a result of speech difficulties but one of his family members (son) participated in an interview on his behalf. The final sample comprised 6 male and 7 female patients. A total of 12 patients and 8 family members were interviewed. Two patients did not have family members who accompanied or visited them at hospital and family members of three patients did not participate because consent was not provided through the patient or directly from the family member prior to patient discharge. Table 2 shows the final sample of patients and family members participating.  Hospital admission and interpreting provision data was examined and all health records were extracted and reviewed for the 13 patients participating in the study.

Hospital staff and interpreters
All staff of the ED and wards were invited through email invitation and flyers to participate in focus group discussions, in-depth interviews and an online or hardcopy survey about their general communication and interpreting experienced including perceived barriers and enablers to engaging professional interpreters. Based on staff records, it is estimated that 616 staff received an internal email invitation to the online or hardcopy survey. The final response rate to the staff survey was 122 or 19.8% of staff. Table 3 summarises the response rate to the staff survey by staff role. An additional 39 staff participated in one of four focus group discussions and 7 in-depth interviews (see Table 3).

Patients with LEP and their family members
A semi-structured interview guide was used to explore the recent communication and interpreting experience of patients from their first encounter with the hospital at ED through to their experience as an inpatient. Topics included: experience of communication with hospital staff and of being provided an interpreter throughout their recent hospital journey, and their satisfaction with this experience. Interviews with patients were of 30-60 min duration, conducted by one of three researchers and took place either at their bedside or in a private room in the ward. Onsite interpreters were engaged for the majority (n = 11) of patient interviews and phone interpreting provided for the remaining patient. Interviews with family members conducted either in a private room at the hospital or over the phone, were guided by a semi-structured interview guide with the aim of eliciting the family member's experience and their perspective of the patient's communication and interpreting experience from ED through to their inpatient stay. Family member interviews lasted between 20 and 40 minutes. Interpreting services were offered for all family member interviews but were only required for 1 of the 11 family member participants.
The health records of all 13 patients participating were reviewed for information about: identification of communication difficulties, language and interpreting needs and comparison with the patient and family member's recollection of their hospital journey. The hospital interpreting records were accessed to provide a complete official record of interpreting provided for the 13 patients and this information was compared with patient and family interviews and health records. Additional observations made by the researchers when interviewing patients at the wards were recorded.

Hospital staff and interpreters
All 4 staff focus group discussions were conducted in a hospital meeting room, were facilitated by the lead researcher (first author) and notes and observations taken by an assistant researcher. Clinician interviews were conducted by phone (n = 6) and 1 nurse interview was conducted face-to-face in a private room at the hospital. All staff focus groups and interviews were guided by a themes list and questions including: (1) Components of effective communication with patients of LEP, (2) How decision is Staff surveys were completed online or hardcopy where they were distributed and collected after completion during staff meetings. The survey took between 10 and 15 min to complete. The survey included 20 questions on practice of working with interpreters, personal competence rating, appropriateness and satisfaction with different types of interpreting, including professional, ad hoc and no interpreting. Participants were invited to share recent (past 6 months) positive and negative experiences of working with professional interpreters.

Data analysis
The staff survey data were exported into Excel and analysed. All interviews and focus group discussions were audio-recorded and fully transcribed verbatim by the trained research assistants and reviewed by the lead researcher for accuracy.
All qualitative data collected from patients, family members, staff and hospital records, including responses to the open-ended "experience" questions of the staff survey, were combined for analysis of major themes and content and to develop the explanatory case study. All three researchers participated in the analysis and identification of major themes. An interim and final meeting were held where the researchers came with their identification of major themes and compared and contrasted findings from the qualitative data. There was a high degree of concurrence of themes identified and only difference in wording of themes was found.
Patient stories of their hospital journey with a focus on their communication and language needs and the extent to which these needs were met were developed based on the qualitative data. Additionally, the consequences and experience of ad hoc interpreting and other communication methods were elicited from the analysis. The individual stories were central to understanding the "lived experience" resulting from the communication decisions made by hospital staff, the involvement of family members in these decisions and the consequences of these actions for the patient and family members.
A theory generating approach incorporating an understanding of interrelationships, a commitment to multiple perspectives and an awareness of boundaries (Simons, 2009) was used to facilitate a holistic examination of the decision-making process for interpreter engagement in the context of the ED and ward setting and factors associated with this process at different levels (e.g. individual as well as hospital systems and processes) and from different perspectives (e.g. clinical staff, patients, family members and interpreters). A systems-thinking causal loop diagram (Williams & Hummelbrunner, 2011) was developed to link together the connected components of the system of the hospital ED and ward and the variables elicited from the case study. Based on the case study data, the causal loop diagram tells the story of communication and interpreting for with patients of LEP. The different types of feedback loops highlight potential areas of intervention for improving professional interpreting provision within the particular system of the hospital.

Staff survey
79 (64.8%) and 41 (33.6%) staff, respectively, rated the importance of engaging an interpreter for patients of LEP as essential or very important (n = 122). Only 17 (13.9%) of the 122 staff members responding to the survey had participated in training in working with professional interpreters. 40 (32.7%) rated their skills as substantial and 68 (55.7%) said they had "some" skills in engaging interpreters.
Although 86.1% of staff members had never received training and the majority rate their skills as less than substantial, only 26.6% said they wanted training in working with interpreters.

Major themes of the qualitative components of the study
The major themes of the case study based on qualitative thematic and content analysis are summarised below.
(1) The majority of patient and family member narratives reveal dissatisfaction and distress with communication as a result from little or no professional interpreting provision.
(2) The patient and family interviews show several major miscommunication errors resulting from ad hoc interpreting that negatively impact patients and family members but are largely unknown to providers.
(3) Decision-making around timing of engaging professional interpreting by providers is about "optimal" timing to maximise use of professional interpreting results in professional interpreting intended and higher perceived rates of provision but in the majority of cases, professional interpreters are not accessed.
(4) The complex nature of the interpreter's role and pressure on their role boundaries from hospital staff, patients and their own position within their community and their variability in knowledge of medical terminology, affects the provider's trust and control in the patientinterpreter-clinician interaction.
(5) Norms of using ad hoc or no interpreters, structural constraints (e.g. no speaker phones) and a lack of evidence-base of the harms of this practice for the patient and benefits of engaging professional interpreters, support decisions to use ad hoc interpreters in favour of engaging professional interpreters.

Interpreting decision-making and its impact on clinician-patient communication from the patient and family perspective
More than 90% of staff rated the engagement of professional interpreting as important for communication about diagnosis, prognosis and procedures and 70% (n = 85) rated engagement as important for determining patient needs and informing them of hospital routines. 70% (n = 71) of clinical staff responding to the survey said that professional interpreters were provided at the hospital when needed either most of or all of the time. An ED nurse participating in an FGD echoed this perception: "I don't think we would have anyone (with LEP) leave the ED without interpreting." However, only 4 of the 13 patients (30.8%) participating were provided an interpreter at any time in ED and only 3 of the patients (23.1%) were provided an interpreter at the ward prior to the study interview.
Acknowledging the normal pattern of several short interactions with patients during the course of their ED stay, clinical staff "get by" with visual assessment and "ad hoc" interpreting from family members and bilingual staff until later in the ED journey where they can get "the most" out of having an interpreter present for one interaction. At the ward with unpredictable timing of patientclinician interactions and the reserving of interpreting engagement for important "events", nurses, doctors and allied health staff also regularly "get by" using other means of communicating. As this ward nurse explains: 'We'd never (engage an interpreter) just for a nurse. We'd always just make a family meeting. Just say if there was consent or they needed to explain a diagnosis, they will usually organise something almost like a family meeting with the social worker there, and the doctors and OT, or try to make it a group thing so then they can do it all in one. But we'll never get it for a nurse, I've never seen that before.' The patient narratives of their experience highlight the usually silent consequences of the decision to "get by" with communication in the ED and wards.
A 33-year-old pregnant Vietnamese-speaking woman, An (not her real name), and her husband explain that they arrived in the ED late on a Sunday night with An experiencing severe pain in her belly and requesting an interpreter. An recalls: 'Like my English is just for communication and not modern day, very basic one, so I couldn't explain that I had already been to my GP who gave me the saline water through the vein. . .I got so distressed and so frustrated because at that time I only wished I had an interpreter there to help me or at least to support me in those moments. . .I think that it was very important that they understood me when I explained my symptoms and my health problem. I was so much in pain I couldn't talk and I couldn't hear very well, and I got already distressed. . .I got all the worries all over me and I think it is very important that they knew what was happening with me that I had to tell them. . .I don't know the reason why they didn't arrange for an interpreter to come but when I first came in, after a few words I kept saying "I need an interpreter". . .' An and her husband, explain that when the doctor described the required procedure to them in the early hours of the Monday morning, An heard the doctor say, "lose your baby" and became extremely distressed as she misunderstood the doctor to mean that "she may lose her baby regardless of having the surgery or not." It was not until interpreting was provided that An and her husband understood that if she didn't have the operation to remove her appendix, she might lose the baby. An's health records concurred that she repeatedly asked for a professional interpreter, "Patient speaking in full sentences. . .Patient states has pain, unwilling to elaborate without an interpreter". However, there was no sign of provider awareness that miscommunication had occurred as a result of the provider decision to "get by" without one.
A Mandarin-speaking man in his late fifties arriving in the ED with heart problems, accompanied by his son, explains his concern that although his son's English is quite good, he was not able to describe his symptoms in the specific detail required for an accurate diagnosis. He recalls: 'This time my real problem is not really aching in the heart but pressure in my heart. Sometimes it is more like a needle-pain in my heart. . .The doctors would ask really specific feelings like this. It is difficult for my son to be able to explain these feelings. My son was not able to describe the needle pain.' Nursing and allied staff talk about their repeated attempts to communicate with patients using a combination of English and body language. The patient narratives provide several examples of the communication strategy between nurses and patients failing. For example, a Mandarin speaking man in his eighties says: 'When taking the medications (the nursing staff) provided, a couple of different pills, some pills of large size, some are just small pills, they wanted me to swallow all of them at once. But it was difficult for me and especially for the large ones, I couldn't swallow all of them together and I had to separate them and take them one by one. I couldn't quite understand that. . .I chose to take them one by one! Otherwise they would be stuck in my throat. . .I just try my own way, by taking the pills one by one rather than taking them all together!' Some of the patients interviewed had communicated through a bilingual staff member at some point during their current hospital admission. This Mandarin speaking patient, a man in his eighties explains how this attempt to communicate through a staff member failed: 'Staff couldn't understand me so they grabbed a staff who could speak Mandarin to help with our communication. However, as most people who have lived here for a long time actually cannot speak Mandarin or understand it very well, therefore he didn't grab the key point of what I said. So basically the whole communication was broken.' English-proficient family members raise the issue of the necessary burden of being available to interpret for their family member on the ward and their concerns for the communication when they are not available to interpret as a result of other commitments. One family member interviewed recalls the emotional impact of being asked to interpret a terminal prognosis directly to her father from the treating doctor without any prior warning about what she would be communicating. Nurses in a focus group discussed a recent incident where they engaged the husband to interpret for a woman who was a long-term patient over a period of several days until they realised the couple were separated and the patient was uncomfortable with him knowing the details of her health situation.

Beyond "cost and time" concerns-uncovering the contextual barriers to interpreter engagement
Staff talked about cost and time concerns when engaging professional interpreters. However, a lack of familiarity and clarity of the process of engaging interpreters combined with inadequate infrastructure, low levels of trust in and confidence in working with professional interpreters and little knowledge of the evidence-base (including cost-benefit) or negative consequences resulting from "getting by" were the key contextual barriers to professional interpreter engagement uncovered through the surveys, FGDs and interviews with staff.
Without adequate infrastructure (e.g. speaker phones), a clear pathway to appropriate interpreter engagement and familiarity with the procedure, the decision to engage an interpreter is perceived as a time-consuming burden that results in the availability of "ad hoc" options being associated with "relief". As this allied health staff member says with almost unanimous agreement within the group: 'It is almost a relief when they say the family will interpret. It is easier for me, more accessible. It's easier to treat this person, rather than have to formally set up for an interpreter and come back when the interpreter is here.' Lack of familiarity about the process of accessing a phone interpreter is compounded by low confidence in working with interpreters, in many cases resulting in resistance to the engagement of phone interpreters where rapport building is more challenging. Negative interpreter engagement experiences recalled by staff members show that rapport, trust and control are important and often missing for staff in working within the staff-interpreter-patient triad. The findings showed that it was common for clinicians to lack confidence that what they communicated was interpreted to the patient. Clinicians' concerns are not just about inaccuracies with interpreting but include the interpreter's trust in them as a medical professional and can even result in interpreters at times interfering with the communication of medical information to the patient: 'Interpreter told clinician she felt the information she was giving the patient was 'wrong' because her family member had a similar condition and was told different information.' The lack of medical knowledge of interpreters is another problem raised by medical staff affecting their confidence in the clinician-interpreter-patient interaction. As this doctor states: 'There is nothing that can replace having someone who is fluent in both languages and has training in being an interpreter, particularly in regards to medical procedures, Medical English being another language in itself.'

A systems analysis using a causal loop diagram
Less than 3 in 10 staff members indicating through the survey that engagement of ad hoc interpreters is inappropriate and only around the same proportion wanting training in working with professional interpreters support a set of norms of a status quo for communication with patients of LEP.
These results suggest a level of complexity exists around the interpreting process within the hospital setting. The interpreting process itself is a fairly straight forward and logical process as shown in the linear component in the centre of Figure 2. This process consists of: (1) Effective interpreter engagement (an input) (2) Effective interpreting (the activity) (3) Effective communication (the output) (4) Good health outcome (the outcome) However, this logical process operates within a more complex system, as has been suggested from the qualitative data, which we have mapped into a casual loop diagram (CLD) (Figure 2). We have found that the process of interpretation has a number of positive feedbacks from the delivery of effective communication and good health outcomes (bottom half of the CLD). Positive feedback creates "reinforcing loops". Reinforcing loops are "virtuous" when things are going well, e.g. "increasing effective communication causes empathy to increase" and are "vicious" when things are not going well, e.g. "decreasing effective communication causes empathy to decrease". Our results suggest the feedback loops created by the response of individuals within the system creates a system of overwhelmingly reinforcing loops.
We have identified a second distinct system component associated with effective interpretation within the hospital (top half of the CLD). This component is centred around the interpreter, who in the majority of cases is contracted by the hospital for each interpreting occasion. The interpreter's impact on effective interpreting involves a "balancing loop". This means that for the organisation to change the effectiveness of interpreting process they need to add resources that lead to training interventions for interpreters that in turn improve both their professionalism and their medical knowledge. Negative feedback is the opposite of positive feedback, e.g. the decrease in Figure 2. This causal loop diagram (CLD) provides a map of the interpreting process in hospital as elicited through the case study. This diagram identifies three distinct components in this complex system of feedbacks: (1) the actual interpretation process (left-right across the middle), (2) the positive feedback loops that effectively reward (virtuous) positive change or punish (vicious) negative change (bottom loops) and (3) the organisational feedback loops that require resources to improve interpreting, which can be made harder through the misguided use of family interpreters (top feedback loops) effective interpreting will increase the training intervention required. When the effectiveness of the interpreting process is improving, the requirement for interpreter training will decrease. To impact improvement in interpreting in this "balancing" loop involves repeated and continued intervention. This is in contrast to the "reinforcing" loop of hospital provider training and shifts in structure, environment and workplace norms.

Discussion
To our knowledge, this is the first examination of decision-making about interpreter engagement that is based on an explanatory case study method incorporating multiple perspectives and experiences, including the focus on the patient journey from admission at ED through to their stay at hospital wards. Consistent with previous studies (Diamond et al., 2009;Hsieh, 2015), we found that time constraints and provider schedules, judgement of the value of communication in clinical and urgency of decision-making, and environmental constraints such as the availability of resources and equipment influenced the decision to engage professional interpreting services.
Our results, however, highlight the flaws in and negative impact of calculated use of professional and ad hoc interpreting on the patient experience in the absence of feedback from the other side of the communication dyad, the LEP patient themselves. The calculated use of interpreters was thought to be high, but the reality is that there were low rates of professional interpreting provision as shown in a recent audit of the hospital-19.8% for ED and 26.1% for ward (Abbato, Greer, Ryan, Vayne-Bossert, & Good, 2018) We show that other ways of communicating with patients including "getting by" with English and charades and engaging family members and bilingual staff to interpret are far from complementary to professional interpreting (Hsieh, 2015), can be detrimental to patient care and unfairly burden family members. The patient and family member narratives of their experience of provider decision-making shed light on the feedback on the decision-making that has been missing from assessment that the choice of family members or bilingual staff over professional interpreters is preferable or acceptable depending on the task. It lends support to the anxiety and feeling of danger of clinical participants when making do without interpreting for LEP patients in other studies (Parsons, Baker, Smith-Gorvie, & Hudak, 2013). Our study shows that family members can be unfairly burdened in communicating when emotional support is needed, major errors in LEP patient communication in English or through family members interpreting on their behalf can go unnoticed by providers and that untrained bilingual medical staff may make errors in interpreting symptoms, diagnosis or prognosis (Flores et al., 2012(Flores et al., , 2003. Furthermore, in the absence of LEP patient and family member feedback and a clear pathway for professional interpreter engagement, providers are often unaware of the negative effects of "getting by" with "ad hoc" or no interpreting. Delay in provision until the "optimal occasion" in many cases results in no professional interpreting engagement at any time of the ED or ward stay. Moving on from the judgement that "communication" is adequate or "good enough" based on only the provider side of the communication dyad (Hsieh, 2015), this study supports the importance of putting into action existing protocols that are not dependent on flawed decision-making that is more about "getting by" than ensuring communication has occurred for both parties. The findings support the establishment of workplace culture informed by evidence that includes engaging a professional interpreter as early as practical at both the ED and wards. This timing of interpreter provision as early as possible is both consistent with patient centred care, patient safety and satisfaction and it removes the complexity of the decision-making of the "optimal time" of interpreter engagement, which has no evidence-base and has been shown to simply delay engagement. It also facilitates a pattern of provision based on need where an interpreting engagement may not be limited by "once-only" provision.
These internal influences on provider decision-making are likely to be difficult to change. Our causal loop diagram (CLD), a visual representation of our understanding of what we were told about the interpreting process, illustrates that two different types of feedback are operating on the interpretation process ( Figure 2).
First, there is a series of positive-individual level feedbacks-in response to effective communication and good health outcomes from patients for whom professional interpretation is provided. When effective communication and good outcomes result from effective interpreting, a virtuous "reinforcing loop" is setup and a process of self-sustaining improvement is possible. However, when effective communication and good health outcomes are not evident or are in decline then a vicious "reinforcing loop" makes it very difficult for the hospital to turn the situation around. The net result is that the hospital environment, its culture, the existing routines are all negatively influencing a decision to engage professional interpreters for patients of LEP, when just the opposite is required.
Second, when the organisation realises that it needs to raise the level of effective interpretation it faces two challenges: (1) resources and sustained effort to change the environment, train staff and shift established norms (2) the often easier and more expedient path of using patient relatives to interpret makes it more difficult to initiate and sustain the required training initiatives.
Therefore, a systems approach taken in mapping of the interpretation process shows the benefits of a focus on infrastructure, training and workplace norms despite the deceptively easier path of using relatives, until they create a virtuous feedback cycle to sustain the training intervention. It is important that organisations take the time to understand the complexity around their interpretation process and then use the shared understanding that a systems map or CLD provides to identify the most potentially successful pathway for intervention and to create the culture change required to implement and sustain system improvements.
Part of the solution requires that the environments in which health care providers work are well equipped with adequate speaker telephones at bedsides in both the ED and the ward (Parsons et al., 2013). Reducing barriers of perceived costs of interpreting through communication of actual costs and cost-benefits such as reduced length of stay associated with interpreter provision  and potential costs associated with medical errors (Flores et al., 2012(Flores et al., , 2003 is recommended. Training of clinicians in the procedures for accessing and working with professional interpreters and the evidence-base of the dangers of increased errors associated of non-provision is also important (Cowden, Thompson, Elizey, & Artman, 2012;Flores et al., 2012Flores et al., , 2003Parsons et al., 2013). Our findings highlight the lack of clarity about the role interpreters have in the hospital setting and are consistent with the ambiguity of the interpreter role in health care settings internationally (Fatahi, Hellstrom, Skott, & Mattsson, 2008;Sleptsova, Hofer, Morina, & Langewitz, 2014). Negative interpreter engagement experiences recalled by clinicians show that rapport, trust and control are important and often missing for staff working within the staff-interpreter-patient triad. Many of these negative experiences can be linked to unclear role boundaries and insufficient medical knowledge of professional interpreters. Although staff members, particularly doctors and allied health staff emphasise that facilitating patient-clinical rapport and interpreting accurately without adding personal advice or opinions are key components of effective interpreter engagement, the actions of both staff and patients can put pressure on the interpreter role boundaries (e.g. clinician finding out more of the patient history known to interpreter from previous interpreting for client and the patient needing a friend/advisor/advocate). Training clinical staff in understanding the interpreter role and how to work confidently and effectively with professional interpreters is critical. Improving the quality of professional interpreting, including increasing medical knowledge, through advocacy and feedback is also important. Influencing the culture of the ED and wards through the establishment of clear norms that are endorsed by senior hospital leadership and viewed as patient centred, patient safety and essential quality improvement initiatives is critical.

Funding
This work was supported by the Mater Foundation.