Methodology

To achieve the aim of this research, the researcher adopted a constructivist grounded theory approach. Grounded theory can be used to investigate areas that have not been extensively researched. It enables the discovery of concepts grounded in the data that are relevant to the specific research area, rather than attempting to test a specific hypothesis or verify an existing theory (Glaser & Strauss, Reference Glaser and Strauss1968). The intent of grounded theory is to move beyond simple description of the experience for a group of individuals to generation of a proposed theory that is representative of processes captured in the experience of the group of individuals (Creswell, Reference Creswell2007). Through the use of grounded theory, a process could be developed from the bottom up. That is, the process would be grounded in the social experiences that friends shared with the researcher. Grounding insights firmly within the experiential data of friends was particularly important, given the limited existing understanding of friendships post TBI.

Participants

Four participants with TBI were recruited to the study. These participants were past or current clients with the Mid Western Brain Injury Rehabilitation Program (MWBIRP), located in a regional city in New South Wales (NSW), Australia. This program is part of NSW Health and is attached to a major regional hospital. The program provides outpatient multidisciplinary care to people with TBI. Inclusion criteria for participants with TBI were:

1. 1. Have sustained a moderate-to-severe TBI as defined by a period of post traumatic amnesia (PTA) of at least 1 to 24 h or more and/or a Glasgow Coma Scale (GCS) of 12 or less;

2. 2. Have sustained the TBI at least 6 months prior to recruitment;

3. 3. Have been discharged from hospital;

4. 4. Have documented social communication impairments as identified in an initial discourse sample analyzed using the Prutting Pragmatic Protocol (Prutting & Kirchner, Reference Prutting and Kirchner1987). This was important because it is well documented that changes to social communication is one impairment that has significant implications for maintenance and development of friendship following TBI;

5. 5. Be able to identify friends (nonfamily) to participate; and

6. 6. Have sufficient English proficiency to engage in conversation in English.

The exclusion criteria for participants with TBI were:

1. 1. Aphasia which impacts on the ability to participate equally in conversation;

2. 2. Dysarthria of a severity which would impact on an unfamiliar listener’s ability to understand output;

3. 3. Current drug or alcohol addiction or active psychosis determined through a file audit and discussion with their clinical team; and

4. 4. Current diagnoses of clinical depression, determined through a file audit and discussion with their clinical team.

These criteria were chosen, as it was believed that their presence may have influenced the results. All of these diagnoses would have significant effects on communication abilities, different to the effects of social communication impairments. It was important to exclude these diagnosis to assist with developing saturation in the data, due to experiences being more similar among participants.

The demographics of these participants are outlined in Table 1. None of the participants were currently engaged in rehabilitation programs, but all continued to have contact with case managers.

Table 1.

Demographics of participants with TBI

These four participants identified one-to-two friends each. Nine friends participated in the study. The friendship dyads are described in Table 2.

Table 2.

Demographics of participants with TBI and their friends, plus friendship characteristics

* Age range only provided.

The eligibility criteria for friends were for them to:

• Be selected by the person with TBI;

• Not be a family member of the person with TBI; and

• Be able to participate in an interview.

Recruitment and sampling

In this study the participants are the friends of people with TBI. However, the person with TBI needed to be included to enable the recruitment of friends. Rehabilitation coordinators provided information about the study to people with TBI and sought their consent to be contacted by the researcher. The researcher contacted and visited those who had indicated a willingness to participate and completed the informed consent procedures. The person with TBI nominated friends who they thought may be willing to participate in the study. The person with TBI provided the contact details of the friends. The friends were then contacted by the researcher. Friends were told that their contact details had been provided by the person with TBI. Interviews were completed with those who agreed to participate. Once one friend had been interviewed, a second or third friend was contacted and the process was repeated. There was one occasion when the person with TBI provided the contact details of two friends and both friends chose not to participate. This person with TBI was excluded from the study.

The researcher commenced by recruiting from a group of participants who were considered to have expertise or experience surrounding the subject matter. In this study, this broad group comprised friends of participants who had sustained a TBI. Later in the research process, sampling was guided by theoretical sampling. Through the constant comparison of the data, characteristics that might play a role in shaping the friendship became apparent. These characteristics included the duration of the friendship, the age of friends, and whether friendships commenced prior to the injury being sustained or after the injury was sustained. Theoretical sampling was used to guide further recruitment of participants that allowed each of the characteristics to be represented in the study. This procedure provided the researcher with opportunities to explore whether codes developed from the data were specifically related to particular friendships characteristics (Coyne, Reference Coyne1997).

To assist with determining when to cease recruitment, saturation in relation to grounded theory was considered. While repetition did become evident within much of the data analysis, true theoretical saturation could not be reached due to the limitations imposed by the time constraints of postgraduate study.

Data collection

Semi-structured interviews, memos, and client files were used to collect data.

The first author conducted the interviews. Initially, a broad interview protocol was developed based on existing literature discussing friendships following acquired brain injury. Interviews began with an open invitation to share information: I’m interested to hear about your friendship with xx, whatever comes to mind, how long you’ve know each other, the things you like to do together, and the things that you enjoy about xx and even the things that might frustrate you or get on your nerves …… When participants responded to this opening segment the researchers considered the unfolding story and paid attention to the information that was provided, so that relevant follow-up questions could be asked (Galletta, Reference Galletta2013).

Constant comparison occurred between each interview. Hence, supplementary ideas were added to the protocol that was grounded in the data provided by previously interviewed participants. These ideas pertained to:

• How friends supported each other;

• Talking together;

• The accident;

• The hospital stay;

• Stages of recovery;

• Effects of the accident or TBI on the friend;

• How did the friend get information about the accident;

• Thoughts about the right thing to do in relation to the accident;

• Anything friends chose not to include themselves in, in regards to the accident;

• Changes in relation to communication and how the friend managed these changes;

• What would have helped the friend;

• Who helped the friend; and

• Any experiences that the friend drew on that helped manage the changes that occurred as a result of the accident.

The interviews in this study lasted between 50 and 90 min. All interviews were audio and video recorded. Interviews were conducted face-to-face with eight participants, at a convenient location for the participant. One participant lived interstate to the researcher. This interview was conducted over the phone and audio recorded. All interviews were transcribed in full and uploaded to NVivo software (Castleberry, Reference Castleberry2014). Transcription was conducted according to the guidelines used by Hines (Reference Hines2009).

Throughout data collection constant comparison was used to examine data provided by the same participant. When conducting initial coding the researcher noted ideas and topics that were raised by the participants and these were added to the interview protocol. In this way, the second participant was asked about these ideas or topics raised by the first participant. This process continued throughout the course of all nine interviews. While saturation was apparent in most categories, it should be noted that negative case sampling was not used to confirm that this was the case for all categories.

Data analysis

All the interviews were transcribed and read independently by two researchers. Initial coding was conducted on the data provided by the first participant. On most occasions each sentence or utterance was coded individually; however, narratives were coded in their entirety to avoid altering the intended meaning.

Following this, focused coding was completed. This involved using codes that occurred more frequently among initial codes or had greater significance. At this stage the researcher considered what the initial codes implied as well as what they initially revealed (Charmaz, Reference Charmaz2014). That is, some participants may have been explicit about the same idea that other participants implicitly acknowledged.

Finally, theoretical coding was completed. Three theories were referenced at this stage of coding. Social network theory, a theory of how people interact with others inside their network (Kadushin, Reference Kadushin2012). Social exchange theory, describing the way that social relationships are based on exchanges (Uehara, Reference Uehara1990). Finally, symbolic interactionism, a theoretical perspective that views the individual and the context in which the individual exists as inseparable (Charon, Reference Charon2007). Each of these theories were used when engaging in constant comparison to enhance the proposed theory at this level of coding.

Establishing trustworthiness and rigor

Several strategies were used to enhance the trustworthiness and rigor of this research. Member checking was used with all but one participant, who could not be contacted. After initial coding, the researcher produced a written summary of the information that the participant had provided, including any interpretation that had been made to achieve the initial coding. This summary was sent to participants. Participants were then contacted, providing an opportunity to further discuss the summary, and for participants to highlight any areas where they felt the information may have been misinterpreted. There were no occasions of participants highlighting information that had been misconstrued.

Given that the first author worked clinically for an extensive period of time with people who had sustained a TBI, memo writing was important to establish reflexivity. The memos included assumptions made by the researcher about factors that may have contributed to social isolation, the impact of family members on friendships, ways that clinicians could engage with friends, and possible ways to assist with the development and maintenance of friendships. This information did not come directly from participants, therefore it was important to capture this, acknowledging how these findings emerged and therefore establishing reflexivity.

A methodological journal was also kept. This journal captured topics such as previous experiences, assumptions, and hypotheses that the researcher held, identifying gaps in the data (which assisted with theoretical sampling), defining preliminary codes, highlighting subtle differences between initial codes and thoughts and perceptions surrounding the experience of conducting semi-structured interviews.