• No results found

3.3 Statistical analysis

5.2.2 Data generation

There are a number of ways in which qualitative data can be generated such as focus groups or individual/paired interviews. Focus groups are a valuable means of gaining insight into participants’ perceptions and experiences.183 Focus groups stimulate interaction and by guiding participants through a set of topics allows the opportunity to observe how issues are conceptualised, worked out and negotiated.184 185 It was anticipated that the implementation of this intervention would be associated with processes that people do collectively therefore focus groups would provide a means to unravel these activities. Examples of such processes include shared decision making regarding a patient’s potential for

mobilisation or staff working together to assist a patient to mobilise for the first time. As VEM is likely to involve a number of different members of the stroke team, each focus group, where possible, consisted of a mix of nurses, therapists and doctors, to capitalise on peoples’ different views within a group setting. The group size recommended for a successful focus group ranges from four to 12 individuals to eight to 12 individuals. This study aimed to have focus groups consisting of between six to eight individuals to allow ample speaking time.

However, due to the difficulty in releasing staff from the ward to participate in the research, the number of participants in each focus group was actually between three and four. Similarly, difficulties in releasing nursing staff from ward responsibilities to attend a focus group at some hospitals resulted in paired interviews with therapy staff being conducted. Paired interviews proved

beneficial in providing more space for thinking and allowing the participants to complement each other’s responses and stories.

As a result of the sampling strategy the focus groups/paired interviews consisted of pre-existing groups/pairs which are seen as advantageous in setting a more comfortable scene allowing participants more freedom and confidence to raise sensitive issues or opposing opinions. On the other hand this may also limit groups or individuals who disagree in the workplace with respect to feeling inhibited in the group discussion.

Each consenting participant was provided with written confirmation of the date and location of the focus groups. A reminder letter was sent to the participant by mail or email seven days prior to the focus group. A one page demographic questionnaire was enclosed to be completed prior to attendance. Participants were also provided with a scenario questionnaire which required staff to rate the appropriateness in turn of three different mobilisation strategies for both haemorrhagic and ischaemic stroke. This approach is based on a consensus

method known as the ‘appropriateness criteria’ where different scenarios known to affect decision making around implementing a certain technology are

presented. The expected time of completion of these two activities was 15 minutes. It was planned that participants would be provided time at the beginning of the session to complete these questionnaires if need be. This occurred in the majority of cases, with the participants forgetting or having a

“lack of time” to complete the questionnaires. The questionnaire proved to be difficult for HCPs to complete as a stand-alone exercise prior to attending the focus group. Instead, the questionnaire was used as a tool for discussion during the focus groups.

It has often been viewed that involving professionals from different disciplines may potentially inhibit those that are in the company of more dominant

professions. Whilst based on the stroke units for the observational study it became obvious that there were differences in practices and opinions between professional disciplines. Additionally, it is recognised that doctors are not actually involved in the conduct of mobilisation, so may have different beliefs from those staff that mobilised patients on a daily basis. Therefore, to ensure professionals felt comfortable during their participation in the study semi-structured interviews were held with doctors as opposed to inviting them to the focus group. There was an opportunity to have informal discussions with nursing assistants about participating in the focus groups. Nursing assistants decided against participation, believing that they would not have anything to offer and that they would feel “uncomfortable” in the presence of trained staff. Due to time-constraints it was not an option to hold separate focus groups for nursing assistants. Nursing assistants were still encouraged to take part and it was emphasised to the nurse manager that these staff members were not excluded from taking part.

An interview schedule was used to ensure that topics were covered in a consistent manner yet flexibility was still allowed. The interview schedule previously used in the Stroke Care Outcomes: Providing Effective Services study was used to inform the questions in this schedule.186 The interview schedule (Appendix 13) included three main sections:

1. Current stroke service to capture information about the organisation and context of the ward

2. Mobilisation practices to gauge perceptions of very early mobilisation in relation to associated benefits, risks and value

3. Changes that have occurred in stroke care to assess the factors that may influence the implementation of very early mobilisation

A pilot study consisting of one focus group (one nurse and two physiotherapists) was conducted on the 8th October 2010. This hospital (ID 7), was later excluded from the main study. The fluency of the interview schedule was tested,

ambiguous questions were identified and the quality of recording equipment was checked. The interview was transcribed verbatim and the responses examined to ensure that the data collected during the pilot met the research aims. The pilot study did not highlight any major concerns regarding the conduct of the

interview. Changes were made to the ordering and the wording of some of the questions and the addition of the following question: “Imagine that next week in the unit you had to start mobilising nearly every patient within 24 hours post-stroke what would you need to do this?”

Focus groups were conducted with nurses and therapists between December 2010 and May 2011, in a pre-booked room within or near to the ASU at the staff member’s place of work to maximise attendance and situate the participants with familiar and convenient settings. The focus groups were recorded using a digital recorder. An introduction was provided at the beginning of each group detailing the background, purpose and confidentiality of the research. The aim of the study was explained to participants and that this was an opportunity for them to discuss the process of care offered within their units in particular their mobilisation practices. Discussion was facilitated to encourage involvement from all the participants and to probe any responses where appropriate. The focus groups lasted for one and half hours.

Semi-structured interviews were conducted with doctors between December 2010 and May 2011. Interviews were conducted using the same schedule as the focus groups and were semi-structured in nature so as to remain open to discussion beyond the specific interview questions yet maintain focus on the topic.

5.2.3 Data analysis

There are a number of methods that exist to analyse qualitative data such as content and grounded theory which are based on the epistemological (nature of knowledge and how it can be acquired) approach used to address the research question. Content analysis examines both the content and context of data with themes being linked to external factors such as age and gender.187 Grounded theory develops analytical categories and identifies relationships between them with this process continuing until categories and relationships are ‘saturated’, and new data no longer contributes to theory under development.188 Thematic analysis, the choice of analysis in this study, “is a method for identifying, analysing, and reporting patterns (themes) within data”.189 Thematic analysis was seen as an appropriate method to both reflect current stroke practices yet unpack ‘the surface of reality’.189 Framework analysis is a matrix based analysis which allows transparent data management to ensure that all the stages of data development can be systematically constructed.190

Immersion in the data was achieved by listening to the interview recordings and by reading the full set of transcripts in entirety and repeatedly. The first three focus groups were transcribed verbatim with the remainder of the recordings being professionally transcribed verbatim. Transcribing provided the opportunity to take an early analytical mind to the data, improve interview style in terms of fluency, to seek definitions and ensure the effective use of prompts in future interviews. Transcripts were checked against the original audio recordings for accuracy. Separate field notes were made throughout the coding process about recurring themes, impressions of the data and questions to follow-up on. Where participants used a particular tone or placed emphasis, annotations were

inserted in the transcripts.

Codes were used to categorise similar text together. Thematic coding is a multi-step procedure. Concepts or codes were assigned to the empirical data codes and were initially formulated as close to the text as possible and then became increasingly more abstract.191 The codes were than categorised into generic concepts and relationships between the categories were developed.191 There were no restrictions to the number of times that the extracts were coded. The

second coder (NB) read and coded a subset of transcripts. The coding was discussed to define and identify any overlap of codes using the initial version of the coding framework. Thereafter, a second version of the coding framework was devised whereby NB applied this to a different subset of transcripts to test the interpretation of the codes. Barriers and facilitators were identified both directly from the relevant sections of the transcripts i.e. participants responses to question 7 of the interview schedule, and by adopting a more implicit

approach. For example, staff when discussing current stroke processes (in response to question 1 of the interview schedule), highlighted how delays in discharge plans impacted on bed availability and thus admission of new patients to the ASU, this was then implicitly coded as a barrier to VEM (‘delayed

admission to the ASU’). Prevalence data to gauge the importance of each of the barriers and facilitators were not provided (see discussion).

Themes were subsequently developed from patterns in the data such as

“conversation topics, vocabulary, recurring activities, meanings, feelings, or folk sayings and proverbs”.192 The transcripts were initially read through to identify emerging and recurring themes. This was an iterative process of arranging codes into broader interpretations, discussing emerging themes and writing up of ideas. Analysis moved from the specific (detailed analysis of each transcript) to the more general (comparing patterns across all the transcripts). The barriers and facilitators that had been identified and coded were interrelated (a

facilitator was often the reverse of a barrier) and were more general contextual factors. Therefore, the barriers and facilitators were interpreted together and more broadly within the themes. To represent the keyness of a theme the

following terms were used: ‘the majority of staff’, ‘many staff’, or ‘a number of staff’. The analysis, as well as identifying the themes, also detected differences between professional groups and experiences of delivering the intervention within a trial setting.

The focus groups were analysed in the same way as the interviews, and with particular attention given to the additional aspects that need to be considered when analysing data from group discussion as opposed to individual interviews.

These aspects include group dynamics, interactions and the influence of other views.190 Group dynamics is describing how certain events may affect the way in

which the topic of interest is discussed such as what and how it is said.

Interactions are exchanges that occur between group participants to affirm or disagree. The influence of other views is where participants have the

opportunity to listen and engage in different viewpoints from others. This allowed the identification of challenges or consensus between participants.

The literature was interspersed with the findings to provide background and to develop theories about the data. The final themes derived from the data were linked together using three headings (specification of the intervention,

organisational characteristics and provider characteristics) to achieve a coherent and detailed narrative based on the experiences of the participants. Themes were described under the relevant heading with extracts which best illustrate the theme provided. Descriptors for each extract are provided in parentheses.

These include participant ID, profession and experience of VEM (‘non-VEM’ refers to participants with no experience of delivering the trial intervention and ‘VEM’

refers to participants with experience of delivering the trial intervention). The

‘experience of VEM’ descriptor was not applicable for quotes from doctors or from specialist nurses. Where extracts were conversations from paired

interviews or focus groups the interview ID (as featured in Table 5-2) was provided under the quote. The term ‘staff’ is used interchangeably with ‘HCPs’

and includes all participants. The term ‘therapist’ refers to physiotherapists and occupational therapists. Nvivo software (version 9) was used to code the data.

5.2.4 Ethics and management approval

Ethical approval was granted by the West of Scotland Research Committee on 8th October, 2010. Research and Development Management approvals were granted on 21st October 2010, 17th January 2011, and 28th January 2011 for each of the three health board areas. Two substantial amendments were subsequently approved; the first amendment was to approve the changes of the interview schedule after the pilot study and the second to approve the use of semi-structured and paired interviews in the study design.