Learning to account for the social determinants of health affecting homeless persons

Authors


Ryan McNeil, BC Centre for Excellence in HIV/AIDS, 608–1081 Burrard Street, Vancouver, British Columbia V6T 1Z1, Canada. Tel: 00 1 604 682 2344(ext. 66520); E-mail: rmcneil@cfenet.ubc.ca

Abstract

Context  Intersecting social determinants of health constrain access to care and treatment adherence among homeless populations. Because clinicians seldom receive training in the social determinants of health, they may be unprepared to account for or address these factors when developing treatment strategies for homeless individuals.

Objectives  This study explored: (i) clinicians’ preparedness to provide care responsive to the social determinants of health in homeless populations, and (ii) the steps taken by clinicians to overcome shortcomings in their clinical training in regard to the social determinants of health.

Methods  Qualitative interviews were conducted with doctors (n = 6) and nurses (n = 18) in six Canadian cities. Participants had at least 2 years of experience in providing care to homeless populations. Interview transcripts were analysed using methods of constant comparison.

Results  Participants highlighted how, when first providing care to this population, they were unprepared to account for or address social determinants shaping the health of homeless persons. However, participants recognised the necessity of addressing these factors to situate care within the social and structural contexts of homelessness. Participants’ accounts illustrated that experiential learning was critical to increasing capacity to provide care responsive to the social determinants of health. Experiential learning was a continuous process that involved: (i) engaging with homeless persons in multiple settings and contexts to inform treatment strategies; (ii) evaluating the efficacy of treatment strategies through continued observation and critical reflection, and (iii) adjusting clinical practice to reflect observations and new knowledge.

Conclusions  This study underscores the need for greater emphasis on the social determinants of health in medical education in the context of homelessness. These insights may help to inform the development and design of service-learning initiatives that integrate understandings of the social determinants of health, and thus potentially improve the readiness of clinicians to address the complex factors that shape the health of homeless populations.

Introduction

Around the globe, more than 100 million people experience homelessness every year1 and recent research has reported a steady increase in the number of people experiencing or at risk for homelessness in many countries.2–6 Homelessness is a growing public health concern.7,8 Homeless populations suffer from disproportionately high rates of chronic and infectious diseases, such as human immunodeficiency virus (HIV) infection and acquired immune-deficiency syndrome (AIDS),9–11 hepatic diseases,11–13 cardiovascular disease14 and diabetes.15 As a consequence, homeless populations experience higher mortality rates and lower life expectancies than the housed population.16–20

Over the past two decades, increased attention to the social determinants of health has led to wider acknowledgement that health inequalities are produced by social and structural factors.21–24 Health inequalities may be understood to be the outcome of social and structural inequities embedded in our society, including but not limited to the distribution of social and material resources, social and physical environments, and gendered and racialised power relations.25 Housing has long been identified as an important determinant of health, and a growing body of research has underscored how housing influences, and is influenced by, other social determinants of health.26,27 Accordingly, researchers have drawn attention to how homelessness is closely intertwined with other social and structural factors, such as entrenched poverty, drug criminalisation and food insecurity, which combine to produce adverse health outcomes among homeless populations.28–32

There has been growing acknowledgment that the health care system is an important social determinant of health and that access to and experiences with health care services are shaped by social and structural factors.33 In this context, homeless populations have been consistently demonstrated to have poor access to health care services and to show suboptimal treatment adherence,34–36 even in contexts in which publicly funded health care insurance is universally available.36 Accordingly, researchers have turned their attention to why this is the case and articulated how the interaction of proximal factors (e.g. substance use, mental illness, etc.) and distal factors (e.g. anti-drug policies, discrimination, inadequate availability of health services, etc.) constrain access to care among homeless populations.37–40 For instance, one study found that homeless persons experience discrimination in health care settings and subsequent to these experiences are less likely to seek care.37

Certainly, there is an urgent need for more substantial social reforms (e.g. increased investment in social housing) to address social and structural factors that negatively impact the health and health care access of homeless populations. However, given that social change is almost always incremental, action aimed at addressing the social determinants of health must be accompanied by strategies that seek to lessen the impact of these factors on homeless (and similarly marginalised) populations. A critical component to this is ensuring that clinicians have sufficient training to prepare them to account for and address these social determinants of health when caring for homeless populations.

To date, several medical and nursing schools have undertaken efforts to improve the preparedness of students to care for homeless populations by offering community rotations or clinical placements in homeless health care settings. Although these programmes have not been well described or comprehensively evaluated, and most published research has been based on small sample sizes and reported in brief,41–43 preliminary findings have shown promise. For example, Buchanan et al.41 reported that students (n = 18) completing a clinical elective that included lectures, journaling and rotations in settings providing health care to homeless people (e.g. emergency shelters, drop-in centres, etc.) had improved attitudes towards homeless populations. Another evaluation found that primary care interns (n = 13) who completed a service-learning programme during their first postgraduate year, which included a 15-hour placement in a health clinic for homeless persons, were highly likely to volunteer in settings of homeless population health care in subsequent years.43 However, it is unclear how well these programmes prepared students to provide care to homeless populations or whether they situated care within the context of the social determinants of health. Given that clinicians generally do not receive training in the social determinants of health,44 this is a potential shortcoming that might limit their ability to take these factors into account when providing care and treatment. Furthermore, the limited distribution of these types of service-learning initiatives means that many students do not receive adequate exposure to homeless populations during their clinical training.

This qualitative study explores the preparedness of clinicians (i.e. doctors, nurse practitioners and nurses) to provide care to homeless persons and investigates how these clinicians overcome any shortcomings in their clinical training with respect to caring for this population. In particular, we sought to identify how clinicians learned to adjust clinical strategies to account for and address social determinants of health that shape the health of this population. This study represents an opportunity to foster a meaningful discussion on how clinical training can be updated to improve the readiness to care for one of the world’s most disadvantaged populations by situating training within the broader context of the social determinants of health.

Methods

Study design

This research was conducted as part of a larger investigation into the individual, social and structural factors impacting the delivery of health care services to homeless persons in six Canadian cities (Halifax, Ottawa, Toronto, Hamilton, Thunder Bay and Winnipeg). We undertook qualitative interviews with doctors and nurses between February 2007 and August 2008 to explore their experiences in entering into and adapting to providing care to homeless populations. We chose to conduct qualitative interviews with clinicians to facilitate in-depth discussion of their experiences and subsequently generate data allowing the analysis of social processes,45 which in this case referred to preparedness and learning among clinicians who provide care to homeless populations. In this regard, our research was situated within an interpretivist paradigm insofar as we were concerned with exploring how these experiences are understood by those who have them.46

Participants & recruitment

We sought to interview doctors, nurse practitioners and nurses with at least 2 years of experience in providing clinical care to homeless persons, and in particular those who had finished their clinical training (e.g. residency, etc.). Specifically, we aimed to recruit participants who worked primarily in settings mandated to provide care to homeless persons (e.g. community health clinics, emergency shelter-based health services, etc.) to ensure that they had sufficient experience in serving this population to have identified, and potentially addressed, any learning gaps. We consulted with an advisory committee made up of researchers and regional experts (e.g. senior health and social services administrators who provide services to homeless persons) to identify potential participants. Additional potential participants were identified through a scoping review of health services for homeless persons.

A total of 35 clinicians were sent letters or e-mails outlining the study and the study methods, and inviting them to participate. Of these, 24 individuals (68.6%) agreed to participate in this study; these included doctors (n = 6), nurse practitioners (n = 2) and nurses (n = 16). In several Canadian provinces, nurse practitioners have expanded responsibilities (e.g. limited diagnostic and prescribing powers) upon the completion of supplementary clinical training. The majority of participants were female (n = 18) and White (n = 23). All participants had completed their clinical training (e.g. residency, etc.) at least 2 years prior to the interview and the majority may be characterised as being mid-career. Furthermore, all but two participants reported that their primary clinical responsibility over at least the previous 2 years had been to provide care to homeless populations and the vast majority of participants indicated that they planned to remain in this field for the foreseeable future. Although several participants (n = 6) worked in both hospital and community settings, they reported that they primarily served homeless persons in community settings.

Data collection

We conducted semi-structured, qualitative interviews with participants in their work settings or in alternative locations of their choosing (e.g. public parks, coffee shops, etc.). Interviews were facilitated through the use of an interview topic guide that was developed based on a review of the existing literature on the delivery of health care services to homeless persons and our observations during previous studies involving clinicians who provide care to homeless populations. The subsequent interview guide (Table 1) included sections exploring: (i) how prior medical training prepared participants to provide care to this population, and (ii) how participants addressed any shortcomings in clinical training with respect to caring for homeless populations. All interviews were conducted by the lead author, an experienced qualitative health researcher. Interviews were audio-recorded. Each interview lasted 45–120 minutes, although most were approximately 60 minutes in length. Interviews were transcribed verbatim in separate files by research assistants. A research team member reviewed the transcripts while listening to the audio files to verify their accuracy and make any necessary corrections.

Table 1. Selected interview questions
QuestionsProbes
Topic area: training and education
 Did your clinical training prepare you to provide care to homeless persons in community settings?Why or why not?
 Could you describe any issues that you felt unprepared to deal with in community settings?Did you feel unprepared to deal with:
 Addictions?
 Mental illness?
 Housing issues?
 Transportation issues?
 Food insecurity?
 Why do you feel that your clinical training did not adequately prepare you to deal with these issues?Do you feel that your clinical training took into account:
 Factors that impact health?
 Factors that impact treatment?
Topic area: continuous learning
 How did you learn to overcome these issues when you began working in community settings?Did you learn to overcome them by:
 Learning from clients?
 Learning from your peers?
 Taking classes or courses?
 Learning from your experiences?
 Could you share an example of how you learned to overcome these issues?Could you share with me what you did?
Who or what did you draw on?
 Have you made any changes to your clinical practices as a result of what you have learned?How have you made these changes?

Analysis

We used an inductive and iterative process to analyse data, with an emphasis on the relationship between preparedness and subsequent learning. We imported the interview transcripts into the qualitative data analysis software NVivo Version 8 (QSR International Pty Ltd, Doncaster, Vic, Australia) to assist with coding. We developed a preliminary coding framework to facilitate analysis by extracting three broad categories from the interview topic guide and field notes (i.e. preparedness to work in community settings, learning in the field, changes to clinical practice) that corresponded to our larger research objectives and emergent themes identified during data collection. Two research team members concurrently coded the transcripts by drawing upon methods of constant comparative analysis, whereby emerging categories and sub-categories were identified and explicated through constant comparison with the data.47,48 We discussed the categories we identified during subsequent team meetings and revised the coding framework accordingly using a consensus-based approach. Once the final thematic categories had been established, two research team members independently re-coded sections of the transcripts to ensure the credibility of these categories.

Ethics

This study was approved by the institutional research ethics boards at the University of British Columbia and Saint Paul University. We obtained informed consent prior to the interviews using an approved consent protocol. We answered any questions that participants had about the study at that time. Participants retained duplicate copies of the signed consent protocol for their records. Where they are quoted directly, we identify participants by profession (i.e. doctor, nurse practitioner or nurse) and numerically in the order in which they were interviewed to preserve anonymity.

Results

Participants reported generally limited exposure to homeless persons during their clinical education and a lack of training in the social determinants of health. We explore how participants learned to provide care responsive to the social and structural contexts of homelessness. Our first theme explores how participants were initially unprepared to account for and address the social determinants of health that influence health and health behaviours among homeless persons. We then present an experiential learning framework that illustrates how participants learned to adapt their clinical practice to address these factors as they gained experience in this field.

Need to account for social and structural factors

Participants highlighted challenges to providing care to homeless persons, noting that they could not rely on assumptions embedded in their clinical training and reinforced in hospital settings (i.e. clients are appropriately housed, have proper nutrition and access to transportation, and are able to pay for care and treatment). Participant accounts indicated that this set of assumptions decontextualises clinical care in that they overlook social and structural forces that shape the health of this population. Participants strongly felt that homeless persons were adversely impacted by a myriad of factors, such as inadequate housing or discrimination, and yet reported limited training in and therefore understanding of these social determinants of health. However, participant accounts indicated that they soon recognised that these social and structural factors greatly increase the complexity of clinical care and, in particular, constrain access to care and treatment adherence:

‘[Homeless clients] have so many little things involved. When you’re in the hospital, you don’t think about food. You don’t think about clothing. You don’t think about shelter… You know they’re safe. You know everything’s taken care of. You know they’ll take their medication. You don’t even think about it. With these guys, just getting a prescription filled is 2 days of work.’

(Nurse, 6)

‘I was running a psychiatric programme where people came in during the day and went home at night. All of those people were housed… They were well enough to get on the bus and come to a programme. If you’re connected up to [the psychiatric programme] and miss three appointments, they discharge you. They don’t bother asking if the person has bus tickets or are they well enough to come to get on a bus… People don’t have bus tickets… They’re not going to buy bus tickets and, nine times out of ten, they’re not well enough to get to the hospital.’

(Nurse practitioner, 1)

Although participants acknowledged that changing these social and structural factors was beyond the scope of their training and professional practice, they reported that it was critical to account for these factors when providing care and developing treatment plans. However, with reference to the lack of emphasis on the social determinants of health in their clinical training, participants expressed their feelings of being ill prepared to cope with these challenges:

‘We have a certain set of skills we can contribute but certainly not enough to deal with the breadth of needs of this population… The housing, the income piece, the social supports… We do that [try to address these social care needs] but we’re just not particularly trained in that. We’re not particularly qualified.’

(Doctor, 2)

Experiential learning framework

This experiential learning framework represents the process through which participants acquired knowledge to account for and address social and structural factors impacting the health of this population and integrated this knowledge into their clinical practice. The process appeared to be composed of three stages: (i) acquiring experiential knowledge by being present; (ii) evaluating through observation and reflection, and (iii) continuously adjusting clinical practice to reflect new knowledge. Figure 1 depicts this experiential learning framework and emphasises the nature of the process: it was a continuous and ongoing process through which participants responded to their practice environment.

Figure 1.

 Experiential learning framework

Acquiring experiential knowledge

Participants articulated how they acquired experiential knowledge to address the health needs of homeless persons by being present and identifying ways to respond to these needs within their social and structural contexts. Being present entailed actively and deeply engaging with this population outside of traditional health care environments in order to foster understanding of how factors exogenous to the individual influence health and health behaviours. In this regard, all participants reported that they regularly met with homeless patients in diverse settings in the course of their work, such as emergency shelters and street-based settings, and that providing care in these settings helped them to gain insights into how homelessness and poverty impact on health. Participant accounts indicated that being present subsequently helped them to identify specific factors they needed to take into consideration when developing care and treatment strategies for homeless individuals. For example, one nurse reported that, in order to meet the challenges that homelessness poses to follow-up (i.e. individuals are generally difficult to reach), she learned the daily patterns of her clients (i.e. where they spent time and when) in order to mediate treatment:

‘The guys come to me asking for this [medication]. I do it [arrange for them to obtain the medication]. They’ll get it when they find me again or when I find them again. You get to learn their routines so well. Oh, he’s here on this day. He’s there this afternoon. In the mornings, he’s there… I have one guy that I knew, every time I drove by, where he is sitting at eight o’clock. He’d be sitting on the bus bench. I knew something was wrong the morning he wasn’t there. You just get to know their routine and work around them.’

(Nurse, 3)

Participants noted that, because they had only a limited capacity to visit many settings, they often supplemented direct observation and interactions by liaising with colleagues and community partners. In this context, participants reported that regular contact with community health and social service agencies (e.g. emergency shelters, syringe exchange programmes, etc.) was critical to keeping abreast of the health status and needs of individual clients, as well as of homeless persons generally. Participants articulated how doing so allowed them to extend their presence into these settings and consequently increase the responsiveness of clinical care to clients’ circumstances:

‘What I try to do is identify all the problems, recognising that the major problems may not come out immediately. It may be only over time that they come out... I always try to get as much information as possible and a lot of time I spend on the phone calling other health care providers trying to get information rather than trying to reinvent the wheel and delve out the whole history myself.’

(Doctor, 3)

Evaluating effectiveness through observation and critical reflection

Participants reported that they evaluated the effectiveness of clinical care and treatment on an ongoing basis through direct observation and critical reflection. Participant accounts indicated that they continued to engage with clients in multiple settings (e.g. community clinics, emergency shelters, etc.) to evaluate a range of treatment and health outcomes (e.g. treatment adherence, substance use, etc.) in the clients’ natural environments. Participants reported that doing so allowed them to better evaluate any social and structural factors that constrained the efficacy of care and treatment:

‘We were there. We watched people on an ongoing basis. Some of it we learned on the job... It’s just, on a day-to-day basis, being with the clients, being flexible, and living with the consequences of what you’ve done. For example, if you’re upping their narcotics then, that’s okay, as long as you’re watching very carefully what the effect is.’

(Doctor, 1)

‘I can say, let’s talk about the time or day that you can take your medication. When is it best for you? Some of our TB medication is twice a week. What about Monday and Thursdays or Tuesdays and Fridays? Pick a date, but it has to be twice a week... It is working with a client and being honest and letting the client be honest with you.’

(Nurse, 7)

Furthermore, the majority of participants articulated how they critically reflected on these experiences of providing care to homeless persons in order to evaluate the clinical care, and emphasised the importance of questioning and analysing their experiences and assumptions. Whereas all participants reported that they reflected on the job, several participants reported that they kept journals. Participant accounts illustrated that, whether through reflection or journaling, they carefully considered how social and structural factors impacted the delivery of health care services to this population. In this regard, participant accounts indicated that critical reflection involved thoughtful deliberation on their experiences in the field (i.e. observed barriers to care or treatment adherence), as well as any feedback they received from clients. One participant noted:

‘You just keep going in a circle to see if it works. It is self-reflection and asking questions… I take every opportunity to learn and to say: “How are you doing? What can we do better? What more do we need to do?”’

(Nurse, 9)

Adjusting clinical strategies

Participants articulated how they had learned to adjust their clinical practice to integrate strategies they felt successfully accounted for social and structural factors that impact on clinical care. Participants emphasised that they made these adjustments on an ongoing basis to reflect their experiences in the field and noted that they applied lessons learned to similar situations. Accordingly, as one participant observed, these adjustments to clinical practice were part of an ongoing process:

‘[Clinical care] is trying something, finding out it doesn’t work, and then reflecting on why it didn’t work and readjusting it. Or, it is trying something and finding it works reasonably well. Then, it’s transferable to this problem over here or this problem over here.’

(Nurse, 12)

Participants identified a wide range of changes they had made to their clinical practice that took into account social and structural factors that shape the health of homeless persons, including but not limited to prescribing medication that can be taken without food, dispensing medication via emergency shelters or clinics, and escorting or arranging for someone to escort clients to appointments for X-rays, laboratory tests and specialist consultations. Furthermore, they reported that they had learned to expand the scope of their clinical practice to include advocating for improvements in the material conditions of homeless persons as a means to improving the overall health of this population:

‘We can work with them to get a drug card. We can dispense medications more easily... We provide a lot of administrative services in terms of getting drug cards and disability pensions and things like that. A lot of that is above and beyond the usual in diagnosing pneumonia.’

(Doctor, 1)

In addition, some participants actively engaged in research to promote action on the social determinants of health and others engaged directly in activism (e.g. protests, lobbying of governments, etc.) aimed at ending homelessness.

Discussion

This study explored the preparedness of clinicians to provide care that is responsive to the social and structural contexts of homelessness, and investigated how these clinicians learned to adapt to providing care to homeless populations. Although participants noted that multiple social and structural factors (e.g. inadequate housing, food insecurity and geographic isolation) increased the complexity of care and treatment, they reported that they were ill prepared to respond to these factors when they began to work in this field. Within this context, participants learned to provide care that accounts for and addresses these social and structural factors by gaining experiential knowledge and integrating it into their clinical practice. We found that this was a continuous process through which participants: (i) acquired knowledge by engaging with clients in multiple settings and contexts; (ii) evaluated and reflected on the effectiveness of care strategies, and (iii) adjusted their clinical practice to increase its responsiveness to the social determinants of health.

Increased attention to the social determinants of health has led to greater recognition that health inequities are shaped by social and structural factors (e.g. housing, poverty, etc.). However, given that medical education continues to emphasise the notion that individuals are primarily responsible for accessing care and adhering to treatment,44,49–51 clinicians generally lack training in this area. Our findings illustrate how clinicians are subsequently poorly prepared to account for these social determinants of health when developing care strategies for homeless persons. This deficit is highly problematic given the multiple social and structural factors that constrain access to care and adherence to treatment in this population (e.g. inadequate housing, lack of food, etc.). Our study demonstrates that participants believed that accounting for these social determinants of health was necessary when providing care to this disadvantaged population and lies within the scope of clinical practice.

Our findings suggest that accounting for the social determinants of health within the context of clinical practice was a two-fold process that involved: (i) learning to situate clinical care within its social and structural contexts, and (ii) engaging directly in advocacy and activism to promote social change (e.g. ending homelessness). Within this context, it was evident that participants believed that, although it is necessary to situate care within its social and structural contexts, this is not enough and larger action is needed to ameliorate social suffering. This finding has important implications because it underscores the diverse ways in which clinicians take action to decrease health and social inequities. Accordingly, our study raises important questions on the training needs of clinicians, which should be used to guide future research. Notably, should we be better training clinicians to advocate for their clients when such advocacy may be critical to advancing health equity?

Whereas researchers have elsewhere highlighted the increasingly important role that experiential learning plays in medical education,52–55 they have yet to specifically study how experiential learning unfolds among clinicians serving homeless populations. Previous research has drawn largely on Kolb’s experiential learning cycle,56 which consists of four stages: (i) having a new experience (i.e. concrete experience); (ii) reflecting on that experience (i.e. reflective observation); (iii) conceptualising a model or understanding of the experience (i.e. abstract conceptualisation), and (iv) planning to and ultimately making changes to practice (i.e. active experimentation). Although we would be remiss if we did not acknowledge several continuities between this experiential learning cycle and our findings (e.g. both emphasise observation and reflection), our experiential learning framework has been generated from the experiences of clinicians who care for homeless persons, and thus more closely corresponds to how they learned to provide care responsive to the needs of this population. In this regard, our experiential learning framework represents an advance in medical education as it relates to homeless populations and should be used to inform several changes in medical curricula to increase the readiness of clinicians to care for homeless populations, as well as other populations impacted by social determinants of health.

Service-learning initiatives with homeless populations (e.g. clinical placements or training rotations in environments in which health care is provided to homeless people) delivered as part of a curriculum emphasising the social determinants of health have considerable potential to improve care for homeless populations by situating care within its social and structural contexts. Such service-learning initiatives have been launched by a variety of medical and nursing schools around the globe and have shown promise in increasing understanding of how social and structural factors impact clinical care.41–43 These should be considered as important components in training clinicians to provide care to homeless populations and should be integrated into initial clinical education. Furthermore, given that clinicians may not have had these learning opportunities, it is advisable that continuing education opportunities be offered to those who have only begun to work in homeless health care environments and those who wish to strengthen their capacity to work with this population in hospital settings.

Our experiential learning framework presents ways to optimise these service-learning initiatives, as well as to inform the development of future initiatives, by situating the curriculum in relation to the experiential learning processes of clinicians in this field. Firstly, service-learning initiatives should provide opportunities for students to be present– that is, to engage with homeless persons in multiple settings and contexts (e.g. emergency shelters, street settings, etc.) – and thereby gain insights into how social and structural factors shape clinical care. Secondly, critical reflection should be integrated into these service-learning initiatives to prompt greater attention to how clinical strategies can be improved. Finally, service-learning initiatives need to be of sufficient length to provide students with opportunities to adjust strategies to account for the social and structural factors affecting care.

This study has several limitations that should be taken into consideration. Firstly, whereas the present participants chose to provide care for this population, many clinicians might be hesitant to work with this population and subsequently less willing to immerse themselves in environments that provide health care to homeless persons. Secondly, we lacked resources to conduct follow-up interviews with participants, which limited opportunities for joint reflection. However, we achieved theoretical saturation in that no new categories were identified during the final analysis and re-coding.47,48 Finally, although our study does not seek to be generalisable in the same sense as quantitative research, our findings may be understood to be ‘transferable’57 in that they may be relevant to other settings and situations in which clinicians learn to account for and address health inequalities.

Despite these limitations, this study has generated important insights that may inform curriculum development and future research. It underscores the importance of increasing awareness of the social and structural factors that influence health as a means of preparing clinicians to provide care to disadvantaged populations. Furthermore, our experiential learning framework suggests that service-learning initiatives are of equal importance and should be developed to reflect the learning processes of clinicians who care for homeless populations. Finally, further research, ideally applying mixed-methods approaches, might test the generalisability of our experiential learning framework for caring for homeless populations and test its application to populations similarly affected by social determinants of health.

Contributors:  RM, MG-Y, SWH and JT contributed to the conceptualisation of this study. RM conducted the interviews. RM, MG-Y and LBD conducted the analysis. RM and LBD wrote the first draft of the manuscript. All authors contributed to the critical revision of the paper and approved the final manuscript for publication.

Acknowledgements:  We thank the study participants for sharing their experiences with us. We also thank Peggy Cooke, Natalie Dupuis, and Arash Kameli for providing research and administrative support.

Funding:  This study was funded by the Social Sciences and Humanities Research Council (SSHRC) (Grant # 410-2006-2050). RM is supported by a doctoral award from SSHRC.

Conflict of interest:  none declared.

Ethical approval:  this study was approved by the institutional research ethics boards at the University of British Columbia and Saint Paul University.

Ancillary