'We are always in some form of contact': friendships among homeless drug and alcohol users living in hostels.

Homeless drug and alcohol users are one of the most marginalised groups in society. They frequently have complex needs and limited social support. In this paper, we explore the role of friendship in the lives of homeless drug and alcohol users living in hostels, using the concepts of 'social capital' and 'recovery capital' to frame the analyses. The study was undertaken in three hostels, each in a different English city, during 2013-2014. Audio recorded semi-structured interviews were conducted with 30 residents (9 females; 21 males) who self-reported drink and/or drug problems; follow-up interviews were completed 4-6 weeks later with 22 participants (6 females; 16 males). Data were transcribed verbatim, coded using the software package MAXQDA, and analysed using Framework. Only 21 participants reported current friends at interview 1, and friendship networks were small and changeable. Despite this, participants desired friendships that were culturally normative. Eight categories of friend emerged from the data: family-like friends; using friends; homeless friends; childhood friends; online-only friends; drug treatment friends; work friends; and mutual interest friends. Routine and regular contact was highly valued, with family-like friends appearing to offer the most constant practical and emotional support. The use of information and communication technologies (ICTs) was central to many participants' friendships, keeping them connected to social support and recovery capital outside homelessness and substance-using worlds. We conclude that those working with homeless drug and alcohol users - and potentially other marginalised populations - could beneficially encourage their clients to identify and build upon their most positive and reliable relationships. Additionally, they might explore ways of promoting the use of ICTs to combat loneliness and isolation. Texting, emailing, online mutual aid meetings, chatrooms, Internet penpals, skyping and other social media all offer potentially valuable opportunities for building friendships that can bolster otherwise limited social and recovery capital.

What is known about this topic

Homeless drug and alcohol users have complex needs and experience multiple forms of social exclusion.

Very little is known about friendship among this population.

Promoting positive social networks among people with drink and drug problems can increase treatment initiation, improve treatment outcomes and reduce relapse.

What this paper adds

Homeless drug and alcohol users have small friendship networks that are often undermined by arguments, geographical mobility and imprisonment.

Homeless drug and alcohol users desire culturally normative friendships, underpinned by routine and regular contact.

Information and communication technologies are central to the friendships of many homeless drug and alcohol users, keeping them connected to sources of social support and recovery capital outside homelessness and substance‐using worlds.

Introduction

This paper explores the role of friendship in the lives of homeless drug and alcohol users living in hostels. A key objective is to understand how individuals who often have complex needs and experience multiple forms of social exclusion build and sustain relationships that may be instrumental in enabling them to address their addictive behaviours and establish more settled lives. A related objective is to consider whether and, if so, how service providers and therapists might better support the development and maintenance of meaningful friendships among homeless drug and alcohol users and other similarly marginalised populations. The concepts of ‘social capital’ and ‘recovery capital’, both of which emphasise the importance of positive relationships to human health and well‐being, are deployed to frame the analyses.

The anthropologist Raymond Firth noted that, ‘Friendship is of a very diverse and complex, even ambiguous, nature’ (Firth 1999, p. xiv). The term ‘friend’ lacks a shared and stable meaning, but it is generally assumed that friendship involves a degree of choice and commitment. Furthermore, friends tend to be defined as people who have something in common, enjoy each other's company, share activities or a common history, like each other and can relax together. They may also offer each other practical help or emotional support and confide in each other (Spencer & Pahl 2006). Indeed, the principle of homophily assumes that people are attracted to, and make friends with, others who share similar characteristics, such as socioeconomic status, values, beliefs or attitudes.

‘Social capital’ is likewise a difficult concept to define, but essentially refers to the benefits that individuals gain by participating in social groups and networks (Bourdieu 1985, 1986, Coleman 1988, Bourdieu & Wacquant 1992, Putnam 1995, Portes 1998). Friendship networks are an important source of social capital, along with relationships between family members, neighbours, work colleagues, members of community, interest or religious organisations, etc. (Bourdieu 1993, Putnam 1995). Relationships do not, however, in and of themselves produce social capital. Rather, social capital is generated where network members share norms and values, and trust and assist each other. Accrual of social capital also depends on individuals being able to ‘claim access’ to the resources that other members of a network have and the ‘amount and quality’ of those resources (Portes 1998, Barker 2012).

Homeless drug and alcohol users have been identified as a very marginalised group in society (Neale 2001, 2008, Coumans & Spreen 2003, Pleace 2008). They report disrupted family lives, relationship breakdowns, lack of social support and interpersonal relationships that are undermined by mistrust, broken confidences and dishonesty (Neale & Stevenson 2014b). Likewise, they frequently have complex needs relating to offending and imprisonment, histories of abuse, low educational attainment and poor health (Zlotnick et al. 1998, Neale 2001). In consequence, their social capital tends to be limited (Stevenson & Neale 2012, Neale & Stevenson 2014b). As Whiteford (2010) has noted, the social exclusion experienced by homeless people has both material and relational consequences. Additionally, the stigma, discrimination and prejudice associated with substance use can exclude individuals, particularly when they are shunned by services and demonised in the press (Buchanan 2004).

For many years, hostels have provided an important source of accommodation and support for homeless drug and alcohol users, routinely offering them food, companionship and help with health, addiction and other problems. Yet, homeless hostels often only provide shared bedrooms and communal living areas (Edgar & Meert 2006, Busch‐Geertsema & Sahlin 2007) and operate strict rules and policies, such as bans on visitors or curfews (Stevenson 2013). Such arrangements may be deemed necessary by hostel providers – for example, because of the limited physical and financial resources available or to protect the general safety and well‐being of residents and staff. Nonetheless, the lack of privacy can undermine relationships, and particularly intimate relationships (Stevenson & Neale 2012, Neale & Stevenson 2014b). Gender imbalances, commercial sexual activities and drug taking within hostel settings also create tensions and mistrust between individuals, so further eroding any social capital that homeless hostel residents do possess (Stevenson & Neale 2012, Stevenson 2013).

Although friendship among homeless drug and alcohol users has per se received little research attention, there is a small but growing literature on the wider relationships of this population (Nyamathi et al. 1999, Trumbetta et al. 1999, Blais et al. 2012, Stevenson & Neale 2012, Stevenson 2013). For example, a study of 130 homeless people diagnosed with substance abuse and severe mental illness found that social networks tended to be small, smaller networks predicted alcohol use over time and substance dependence remitted when individuals had fewer substance users in their baseline networks (Trumbetta et al. 1999). Research has also shown that drug use can hinder the formation of intimate relationships among homeless people as selling and using drugs are often prioritised over everything else (Blais et al. 2012). However, having an intimate partner can reduce feelings of anxiety and isolation and increase feelings of safety, self‐esteem and well‐being. This can, in turn, enable homeless drug and alcohol users to better manage their addiction and move away from a street‐based lifestyle (Nyamathi et al. 1999, Stevenson & Neale 2012).

Research focusing more explicitly on homelessness has shown that homeless people frequently lack trust in other people (Barker 2012) and experience loneliness (Rokach 2005). Furthermore, the family and friends of homeless people can make demands on, create conflict with or be abusive to them (Savage & Russell 2005). Consequently, homeless people may turn to friends rather than to family when they need help (Whitbeck et al. 1999, Ravenhill 2008). Indeed, homeless people, especially those living on the streets, can form dense social networks where most members know everyone else (Hawkins & Abrams 2007, Ravenhill 2008, Cloke et al. 2010, Mostowska 2013). Nonetheless, these relationships can be extremely complex: on the one hand, offering intense friendship, reciprocity, and a sense of belonging, security and solidarity; on the other hand, reinforcing a hierarchy and pecking order where violence and fighting are ‘the norm’ (Ravenhill 2008, Cloke et al. 2010, Gowan 2010).

Within addiction research, problematic drug and alcohol use has been associated with criminal, violent, abusive and exploitative relationships (Browne & Bassuk 1997, Farris & Fenaughty 2002, Neale 2002). Yet, studies have also shown that substance users can have supportive family and friends who provide financial, emotional and practical assistance, discourage drug use and enable better management of addictions (Alverson et al. 2000, Laudet et al. 2000, Neale 2002, Simmons & Singer 2006, Neale et al. 2012). Furthermore, there is evidence that therapies which actively promote positive social networks among people with drink and drug problems can increase treatment initiation, improve treatment outcomes and reduce the likelihood of relapse (Marlatt & Gordon 1985, O'Farrell et al. 1985, McCrady et al. 1986, Stout et al. 1987, Barber & Crisp 1995, Copello et al. 2002, 2006).

Associations between positive relationships and reduced addictive behaviours seem consistent with the concept of ‘recovery capital’. Recovery capital has been directly adapted from the sociological literature on social capital and is the term increasingly used to assess the resources that an individual can draw upon to initiate and sustain processes of recovery from substance dependence (Cloud & Granfield 2001, 2008). Social capital (in the form of relationships) comprises one of four key components of recovery capital; the others being ‘physical capital’ (income, savings, investments, property), ‘cultural capital’ (values, beliefs and attitudes that promote social norms) and ‘human capital’ (education, knowledge, skills, hopes, health and heredity). While the concept of recovery capital has been critiqued as poorly specified and incomplete (Neale & Stevenson 2014a, Neale et al. 2014), it is generally accepted that people who have access to recovery capital are better placed to overcome their substance misuse‐related problems than those who do not have such access (Cloud & Granfield 2008).

Methods

The data presented were collected as part of a larger study designed to increase our understanding of hostel residents' relationships in order to inform the development of social network‐focused therapeutic interventions that might assist homeless drug users and drinkers. Ethical approval was granted from a university research ethics committee, and data were collected during 2013 and 2014 from three hostels, each in a different English city. Hostel A provided 56 beds in a medium‐sized city; hostel B provided 57 beds in a large city; and hostel C provided 17 beds in a small city. All hostels catered for males and females (although hostel C had no female residents at the time of data collection). All hostels operated no visitor policies and all prohibited illicit drugs on the premises (see Table 1 for further details).

Table 1

Hostel characteristics

	Hostel A	Hostel B	Hostel C
Number of beds	56	57	17
Location	Medium city	Large city	Small city
Opening times	24 hours	24 hours	6 pm to 8 am
Institutional ethos	Secular	Secular	Faith‐based
Wet facilities	One communal wet‐room; no alcohol in bedrooms	No alcohol in communal areas	No alcohol at all
Length of stay	1 night to 6 months	3 months to 2 years	Up to 3 months but exceptions granted

In each of the three hostels, an experienced qualitative researcher conducted one‐to‐one interviews with 10 residents who self‐reported drink and/or drug problems (N = 30). The figure of 10 participants per hostel was chosen for pragmatic reasons. Specifically, we wanted an equal number of participants from each case study hostel and felt that we would be unlikely to be able to recruit more than 10 residents in the smallest hostel. Furthermore, we believed that a total of 30 participants would enable meaningful analyses, including some subgroup analyses, for the study as a whole.

The researcher recruited participants by reviewing bed lists and then randomly selecting individuals to approach, while trying to ensure a mix of men and women and participants from different ethnic groups. On initial contact, the researcher explained the nature of the research and asked individuals whether or not they currently had a problem with drugs or alcohol. If the response was affirmative, and the individual was interested in participating in the study, a time to conduct the interview was agreed. After 4–6 weeks, the researcher attempted to contact all participants again, either through the hostel or via telephone, to conduct a follow‐up interview. Written informed consent was obtained by the researcher prior to each interview, and participants were told that they were free to withdraw from the study at any time without any impact on their access to services.

Initial interviews followed a semi‐structured topic guide that covered demographic information; current and previous housing circumstances; current and previous drink and drug use; and current and previous relationships, both inside and outside the hostel. Follow‐up interviews covered similar topics, but included any salient issues emerging from the first interviews and any changes in relationships occurring between interviews 1 and 2. Both initial and follow‐up interviews lasted 45–60 minutes and were audio recorded. All interviews were conducted in privacy in one of the hostel rooms and participants were offered a £10 voucher in compensation for their time.

Audio recordings of all interviews were transcribed verbatim and entered into the software package MAXQDA for systematic coding. All coded data were then analysed using an approach known as Framework (Ritchie & Spencer 1994). For this paper, the coded data relating to ‘friends’ and ‘friendship’ were exported from the coding frame into Microsoft Office Word documents. These Word documents were then reviewed line by line to explore the nature of the relationships described, including any changes over time. The findings from this process were further reviewed with a focus on reasons for friendships, amount of contact with friends over the study period and the relevance of drink or drugs to the friendships. This produced eight relatively discrete friendship types. Finally, all analyses were linked back to the framing concepts of social and recovery capital.

Participants

At interview 1, the 30 participants included 9 females and 21 males, of whom 22 (6 females; 16 males) were re‐interviewed. Our analyses are therefore based on 52 semi‐structured interviews. Ages at first interview ranged from 21 to 54 years (mean 38 years). Twenty‐six participants were white British (2 black West Indian; 1 mixed race; 1 white Irish traveller). Ten had left school before the age of 16 and only seven had any formal qualifications. Twenty‐nine were receiving welfare benefits, and one had no income at all. None was in work. Thirteen participants said that they had hepatitis C and two said that they were HIV positive. Twenty‐four participants self‐reported mental health problems and 22 stated that they had been in prison. Length of homelessness at first interview varied from a few days to 20 years, with many individuals reporting intermittent periods of being housed or being in prison.

In terms of current (last month) substance use at first interview, most participants reported poly‐drug use: heroin and crack cocaine (n = 16), heroin and alcohol (n = 6), and heroin, crack cocaine and alcohol (n = 6). Other drugs used in the last month were cannabis, ketamine, prescription medicines and MDMA (3,4‐methylenedioxy‐methamphetamine). In addition, twelve participants reported current drug injection and a further four participants reported previous drug injection. At the follow‐up interview, three participants stated that they were no longer using their main drug stated at their initial interview and eight participants stated that they were using less of their main drug. In contrast, eight participants said that their substance use had escalated.

Findings

At interview 1, 21 of 30 participants (15 males; 6 females) spoke of current friends either inside or outside the hostel. Of these 21 participants, one male reported no friends at interview 2 and one female without friends at interview 1 reported a friend at interview 2. That said, changes within friendship networks were evident over the study period. At interview 2, a small number of individuals reported less or no contact with friends identified at interview 1. Reasons for this included arguments, friends moving away or going to prison, and, in one case, a friend establishing a new relationship with a boyfriend whom the participant did not like. In addition, a small number of participants reported new friends at interview 2. These included old friends who had moved back into the area or re‐established contact, new people met socially outside the hostel and new neighbours or housemates where individuals had moved on from the hostel.

Overall, friendship networks were relatively small; between one and six people at interview 1 (mode = 1). While a small number of individuals stated that they would like more friends, others reported that friendships did not interest them. Patrick explained:

I worked out at school … that … I find it too draining if I have too many people … I think it is getting involved in their lives and all that. I just find it really complicated. (Patrick, aged 46, hostel C)

When asked to describe the most important characteristics of friendship, participants most often referred to trust, honesty and loyalty. They also emphasised that friends were people who had things in common, looked out for each other, were good company and shared things, including secrets. Friends would additionally listen to each other, be non‐judgemental, forgive each other and be supportive. Many added that they wanted friends who did not use drugs or alcohol. This, they said, was because having other substance users or dealers around tempted them to drink or use drugs or encouraged them to commit crimes such as shoplifting. Equally, they reported that it was stressful when others constantly asked them for drugs or drug paraphernalia.

In the following sections, we describe each of the eight friendship types emerging from our data. Most participants reported friends from just one of the friendship categories, although five reported friends from two categories, two reported friends from three categories, one reported friends from four categories and one reported friends from five categories.

Family‐like friends

Eight participants (five males; three females) identified friends who could be categorised as family‐like; that is, the boundary between friendship and family member was blurred because of the perceived strength and depth of the relationship and a belief that the relationship was unconditional and unbreakable. On balance, family‐like friends seemed less likely than other types of friend to drink or use drugs problematically, more likely to have regular contact with the participant (in person, by phone, text or Skype) across both interviews 1 and 2, and more likely to offer a range of support. This included ad hoc daily telephone calls to say ‘hello’, looking out for the participant, having serious conversations to talk through problems and offering participants meals or a bed to sleep in. Very occasionally, family‐like friends might go out together for a social drink or share drugs:

They are like my close, close friends. I mean they are like my family…. These are the kinds of friends that if I haven't got nothing to eat, I will go there and can just eat anything out of their fridge without even having to ask…. (Lauren, aged 21, hostel B)

Using friends

Eight participants (five males; three females) also discussed having friendships that were primarily founded on current or previous drinking or drug use. Often, these friends were described as ‘old friends’, ‘drinking partners’ or ‘friends from clubbing days’. Sometimes participants had daily contact with using friends, and sometimes they only had very irregular contact by phone or Facebook, usually because the friend had moved away. Using friends were not consistently reported across the two interviews and lack of regular contact meant that participants did not necessarily know whether or not these individuals were still using drugs or drinking.

Although any time physically spent with using friends often involved substance use, participants reported that using friends still sometimes tried to be supportive. Brian, for example, had a friend with whom he had used drugs in his early twenties. The friend had since stopped using drugs and settled down with a family. Although Brian and his friend had not seen each other for several years, they maintained regular contact on Facebook and Brian considered the friend to be reliable and helpful. In fact, the friend had recently invited Brian to visit and Brian had wanted to go, but decided against it as he was concerned that the friend's wife and child would not appreciate him staying:

He [friend] has invited me down there [place where friend now lives]. He said if I needed to get away at all, he has invited me down there. But I wouldn't go down there while I am not 100% anyway. He has got a Mrs [wife] and kid…. (Brian, aged 34, hostel C)

Homeless friends

A further category of friend, also identified by eight participants (again five males; three females), was the homeless friend; that is, individuals whom our participants had met in hostels or living on the streets. Some, but not all, homeless friends currently used or had previously used drugs or alcohol problematically, but they seldom used drugs or drank with the participant now. Homeless friends were nearly always seen in person most days or several times a week and our participants often identified them as being particularly important and supportive. In this regard, participants stated that they spent time together, talked to each other about problems, helped each other out by giving or loaning each other money or small material items, looked out for each other and encouraged each other to address their addictions or attend 12‐step meetings. Furthermore, they shared the experience of homelessness and the difficulties of trying to secure their own accommodation. Despite this, homeless friends identified at interview 1 were often not discussed again at interview 2:

I see a guy, he lives out on the street…. I like him and I care about him…. He is a really funny decent man who is an alcoholic … and I always beeline towards him because he is a nice bloke and we have a laugh…. He enhances my life by the very fact that he is there…. He is supportive of me … always hassling me to go to AA meetings. (Rick, aged 51, hostel B)

Childhood friends

In total, four participants (three males; one female) discussed childhood friends, although three reported that they had very little contact with these individuals any more. Two childhood friends were described as generally caring or supportive even though they themselves often used drink and drugs and one was described as unsupportive as he condoned and encouraged the participant's drug use. Only one female participant reported having regular bi‐weekly contact with a female childhood friend who did not use drugs or drink problematically and who would always be there to telephone and help in an emergency. Apart from this one female friend, childhood friends identified at interview 1 were not discussed again at interview 2:

She is a friend from childhood that I don't have an immense amount in common with apart from the fact that we have been friends for life since school, and she has always been there for me in situations. Like, if I haven't spoken to her for ages, I can still pick up the phone if there is like an emergency. She is that kind of friend. (Helen, aged 38, hostel B)

Online‐only friends

Three participants (all male) identified friends whom they only ever contacted via the Internet. One participant reported that he had many Facebook friends, most of whom he did not know. However, he had an old male friend who had moved away about 5 years ago and they still contacted each other twice a week via Facebook. A second male participant also stayed in contact, again by Facebook, with an old friend who had moved abroad, and a third male had a female friend in Canada whom he had never met, but with whom he spoke daily by Facebook or Skype. Participants were not sure of the current drinking and drug use behaviours of these online‐only friends, but none was thought to be using illicit drugs, although one was believed to be drinking. All three males described their online friends as supportive, and the participant who contacted his female friend daily reported that this contact was invaluable, particularly during periods when he had been living on his own:

I have actually got a very good friend, this is the one in Canada, who I have been speaking to on Facebook for about three and a half years, pretty much every night. We Skype, and although she is in another country and I have never actually met her, I would say she has been a very good, very good close friend…. When I have lived on my own and there has been no one to chat to, it has been really nice to sort of sit down and just, you know, ‘how has your day been?’ (Greg, aged 36, hostel C)

Drug treatment friends

Only two participants identified friends made in drug treatment. One female participant had met a female friend at a local drug project and they were seeing each other daily at interview 1, although they were both still using drugs. By interview 2, the two women had lost contact as the friend had moved away. In addition, one man had met a male friend in residential rehabilitation treatment and the friend (who was abstinent) telephoned the participant (who was still using drugs) every couple of weeks at both interviews 1 and 2. The participant greatly appreciated this call, which was essentially to check that he was alright:

He's a friend who I went through [rehabilitation service] with…. He phones me up and asks me how I'm getting on … every 2 weeks. (Mark, aged 35, hostel A)

Work friends

Again, only two participants discussed friends they had made while working. One male identified a male work colleague whom he saw fortnightly at interview 1 but did not mention at interview 2. The friend was a heavy daily drinker. One female identified several female friends who all had professional jobs, at least one of whom she saw daily at both interviews 1 and 2. The participant believed that this particular friend would be helpful when she was ready to return to work:

[Alice] is a fellow journalist…. I see her about six times a week. Well I have contact with her six times a week…. She is very important [to me]…. She is going to be useful with me getting back into work. (Helen, aged 38, hostel B)

Mutual interest friends

Only one participant (male) identified anyone who might be termed a mutual interest friend; that is, the two friends shared a passion that was not drug‐ or drink‐related. This participant had a male friend whom he had met at a philosophy class and they saw each other at least weekly at both interviews 1 and 2. The friend did not have a problem with either drugs or drink. This same participant had also met a female friend, again a non‐drug user, at a music festival shortly before his second interview. They had only met and spoken by telephone a few times by interview 2, but were getting on well:

[Sophie] is just a friend, someone I met at an event. We are doing this thing [fundraising event] together so, in the last couple of weeks, I have seen her twice and we sort of contact each other on the phone. (Rick, aged 51, hostel B)

Discussion

Participants in the study reported poor physical and mental health, low levels of education, training and employment, and small friendship networks (cf. Neale 2001, 2002). Indeed, nearly a third of all participants reported no friends at all and a small number denied that they even wanted friends. Furthermore, extant friendships were undermined by arguments, geographical mobility and imprisonment (for further information on the negative aspects of our participants' relationships, see Neale & Stevenson 2014b). These findings suggest that homeless drug and alcohol users comprise a population that is low not only in social capital but also physical, human and cultural capital (Cloud & Granfield 2008). Accordingly, opportunities and choices for building and sustaining friendships are likely to be constrained by limited access to individual, interpersonal and institutional resources. Despite this, just over two‐thirds of participants had friends and most wanted friends. Moreover, the kinds of friendships they desired were culturally normative – that is, friendships based on trust, honesty and loyalty, that involved sharing belongings and confidences, supporting each other and being non‐judgemental (Spencer & Pahl 2006).

In one of the earliest writings on friendship, the Greek philosopher Aristotle argued that friends could be divided into ‘friends of utility’ (individuals who help each other and provide practical support); ‘friends of pleasure’ (individuals who share activities and enjoy each other's company); and ‘friends of virtue’ (individuals who are intimate, trust each other and really know and understand each other) (Urmson 1988, Spencer & Pahl 2006). Our data revealed examples of all three of Aristotle's friendship types. Thus, ‘friends of utility’ provided food, accommodation, gifts or loans, were accessible in an emergency and could potentially help with finding work; ‘friends of pleasure’ spent time together, shared interests and sometimes drank or used drugs with each other socially; and ‘friends of virtue’ talked to each other about their problems and felt connected by shared experiences of homelessness and addiction. Sometimes friends fulfilled more than one of these functions.

Importantly, however, friendship in our study had an additional role: that of providing regular contact. Indeed, routine communication by those whom we might term ‘friends of contact’ seemed to be valued over and above all other friendship benefits. Interaction might be established in person, by telephone call, by text or via the Internet, with the mode of communication seemingly less relevant than its frequency and dependability and the fact that someone had thought of them and recognised their existence, value and worth. Furthermore, the content of the communication could be minimal and generally did not involve any particular discussion of the participant's drinking, drug use or other problems. In fact, it was not actually necessary for the two friends to have even met in person or for the other person involved to derive any obvious personal benefit from the interaction. Rather, it was the act of texting, calling, emailing, facebooking or skyping to say ‘hello’ and ‘how are you?’ that was valued, especially when individuals were feeling lonely.

In presenting our analyses, we have identified eight categories of friendship emerging from the data. We do not claim that our typology is exhaustive as further data collection might have identified additional categories. Equally, we appreciate that some friends could potentially be ascribed to more than one friendship category or might move between categories over time – for example, a childhood friend who subsequently became a using friend or an online‐only friend who later became a family‐like friend. We also recognise that there are other existing friendship categorisations that we might have applied to our own data rather than creating a typology of our own (cf. the writings of Aristotle, but also Spencer & Pahl 2006). Despite this, the qualitative data collected provided a novel opportunity to explore the concept of friendship inductively using the accounts of homeless drug and alcohol users themselves. Moreover, the eightfold categorisation created offered a useful heuristic for considering how different types of friend may support – or undermine – social and recovery capital.

In this regard, we note that family‐like friends seemed to offer the most constant forms of both practical and emotional support (or social capital). This is in contrast to using or drug treatment friends, whose contact was unpredictable and whose willingness or capacity for support seemed limited or nominal. Homeless friends were often physically present on a daily or near‐daily basis, thus facilitating talking and the sharing of experiences. Nonetheless, homeless friends often moved on or away and thus tended not to offer more long‐term recovery resources. On balance, childhood friends did not make regular contact, were not very supportive and could undermine recovery capital by ongoing drinking and drug use; yet, there was evidence of exceptions. Meanwhile, only few participants reported online‐only friends, work friends and mutual interest friends. However, these could still provide important recovery resources; for example, helping to combat loneliness (online‐only friends), assisting with a return to paid work (work friends) and encouraging non‐drug or drink‐related activities (mutual interest friends).

Conclusion

Homeless drug and alcohol users who live in hostels tend to have few friends. Nonetheless, they can still build and sustain friendships that may help them to address their addictive behaviours and establish more settled lives. The friendships they aspire to are culturally normative; that is, based on supporting each other, spending time together and sharing values such as trust, honesty and loyalty. Routine and regular contact – even if this just a text or brief telephone call to say ‘hello’ – is highly valued. In contrast, drinking and drug use, or even talking about drinking and drugs, is generally not associated with supportive friendships. Significantly, the use of information and communication technologies (ICTs) – mobile phones, the Internet and new social media – emerged as central to many friendships, keeping participants connected to sources of social support and recovery capital outside homelessness and substance‐using worlds.

Homeless drug and alcohol users' friendships can take a number of forms, with some offering more support than others. From this, we conclude that those working with people who are homeless and use drugs – and potentially other marginalised populations – could beneficially encourage their clients to identify and build upon their most positive and reliable relationships, perhaps beginning with any friends whom they consider to be like family. Additionally, service providers and therapists might explore ways of promoting the use of ICTs to combat loneliness and isolation, just as ICTs can enable independent living and increase social contact among older people and people with disabilities (Curry et al. 2002, Jaeger & Xie 2009, Cotten et al. 2012). We recognise that this may require resources to provide better access to computers, and training and support to ensure that clients are able to use the available technologies (Neale & Stevenson 2014a). Yet, texting, emailing, online mutual aid meetings, chatrooms, Internet penpals, skyping and other forms of social media all seem to offer valuable opportunities for building friendships that can in turn bolster social and recovery capital.

Conflicts of interest

The authors have no conflicts of interest.