«Citation: Dalkin, Sonia (2014) The Realist Evaluation of a Palliative Integrated Care Pathway in Primary Care: What Works, For Whom and in What ...»
Programme theory 4 The literature highlighted that communication about death and dying can be difficult (Buckman 1984, Ptacek and Eberhardt 1996, Ptacek and Ptacek 2001, The National Council for Palliative Care 2009). Literature in psychology suggests that in health related risk situations individuals utilise distinctive attentional processing styles which allows them to be classified as a monitor or blunter (Miller 1995). It could be that matched coping styles could help examine why some palliative care consultations lead to more proactive discussions than others. Thus, the following programme theory was
The innate coping style of the GP and patient (context) facilitates a consultation if matched (mechanism), making a preference discussion and use of the locality advance care plan more likely to occur (outcome).
- Are matched coping style consultations more successful (in terms of producing outcomes such as preference discussions and advance care plans)?
- Can matched coping styles facilitate practice?
Programme theory 5 Using a patient centred framework based on shared decision making is advocated in policy and in translational tools used in practice. In the literature chapter it was identified that Hockley, Watson et al. (2010) and Bower, Roderick et al. (2010) found using ACP (amongst other resources), which is based on shared decision making, resulted in a decrease in hospital deaths and an increase in home deaths, respectively.
This knowledge and that gained from integration into the locality resulted in the
following programme theory:
The ICP can facilitate preferred place of death (outcome) and prevent emergency admissions (outcomes) through identifying patient preferences (context) and using ACP (mechanism).
- Are home deaths increasing?
- Are care home deaths increasing?
Gaps in knowledge led to the formulation of the research question. Knowledge of the quantitative ICP outcomes available, reading the literature and access to contextual knowledge then enabled the formulation of programme theories and of questions to be asked of the data. The literature in Chapter 1 (p.6) highlighted gaps in knowledge which the programme theories address. For example Smith (2012) identified an increase in home deaths due to implementation of a palliative care heart failure service, however the findings do not highlight what contexts and mechanisms were necessary for this outcome to occur; programme theory 5 aims to identify not only if home deaths are increasing in the locality but how and why, through investigation of patient preferences and ACP. The programme theories above recognise and illustrate key components of the programme (ICP) and include its strategies, functions or activities. Programme theories also map out the outcomes the ICP intends to generate and the specific components that contribute to particular outcomes.
Data Collection and Analysis It is possible to use and integrate different combinations of data sources, methods and types of data as part of a study (Plowright 2011). Contributions of different data types and analyses were not confined to specific programme theories but some data analysis did feature more prominently in specific programme theories. Realist evaluation values mixed methods and states that data type should be selected on the basis of how informative it will be to the study. For more than a century, the supporters of quantitative and qualitative research paradigms have engaged in dispute (Johnson and Onwuegbuzie 2004). From these disputes, purists have emerged on both sides (Campbell and Stanley 1963, Lincoln and Guba 1985). Quantitative purists (Ayer 1959, Popper 1959, Schrag 1992, Maxwell and Delaney 2004) express assumptions consistent with a positivist philosophy and maintain that social science inquiry should be objective (Johnson and Onwuegbuzie 2004). Thus they pursue time and context free generalizations (Nagel 1986). Qualitative purists (Smith 1983, Smith 1984, Guba and Lincoln 1989, Schwandt 2000) reject positivism and are also known as constructionists or interpretivists (Johnson and Onwuegbuzie 2004). They state that time and context free generalisations are neither possible nor desirable, and that research is value bound.
They also contend that it is impossible to differentiate causes and effects fully. Both quantitative and qualitative purists view their paradigm as the ideal for research meaning these paradigms and associated methods should not be mixed (Johnson and Onwuegbuzie 2004). However, to include only quantitative or qualitative methods falls short of the major approaches being used in the social and human sciences (Creswell 2003). Currently, research practice is often not purely quantitative or qualitative but on a continuum between the two, using mixed methods (Creswell 2003). Mixed methods is a third research paradigm which recognises that both quantitative and qualitative research are important and useful and thus uses a combination of at least one quantitative and one qualitative component in a single research study (Bergman 2008).
The goal of mixed methods research is to draw from the strengths of both quantitative and qualitative research whilst minimising the weaknesses in single research studies and across research studies. It employs strategies of inquiry that involve collecting data either simultaneously or sequentially to best understand research problems (Creswell 2003). Thus by using mixed methods numbers can be used to add precision to words, pictures, and narrative and a broader and more complete range of research questions can be answered (Johnson and Onwuegbuzie 2004). Furthermore, the use of mixed methods produces more complete knowledge that is necessary to inform theory and practice which is essential in this study (Johnson and Onwuegbuzie 2004). However, mixed methods can be more time consuming and expensive and some of the details of mixed methods research remain to be worked out fully (Johnson and Onwuegbuzie 2004).
Framework There are five types of data collection within this project. This is in order to understand and evaluate the ICP in a variety of ways, find evidence to support the programme theories and allow generation of multifaceted CMOCs. Each data collection form and analysis is explained further below and in Table 1.
Fourteen of the fifteen practices implementing the ICP took part in the study. GP practices were lettered to maintain anonymity. All fourteen practices provided MIQUEST and GP practice data. Health care professionals from all fourteen practices were invited to take part in the focus groups, along with members from other key organisations such as the ambulance service, social services, and OOH services. Three of the GP practices were chosen as sites to collect data from patients and bereaved relatives. Practice H collected questionnaire data, practice D acted as gatekeepers and referred bereaved relatives and patients for interviews, and practice E acted as gatekeepers and recorded consultations with palliative care patients. Data collection was sited to avoid over burdening health care professionals, palliative care patients and bereaved family, friends and carers of palliative care patients. Practice D is in a large town (urban), practice E is in a smaller town (semi-rural) and practice H is in a village in a sparsely populated area (rural).
All strands of data collection took place simultaneously. This allowed the generation of CMOCs across data sources, and allowed for further exploration of one factor through another data collection method. Furthermore, all different data collection forms were sited at different practices (Practices D, E and H) in order to only invite
theories 1, 2, 3 and 5) Recruitment: No recruitment was needed for the GP practice data as it was already available from NHS North of Tyne PCT. Permission to use the data was sought from the GP practices using an invitation letter (Appendix 3), information sheet (Appendix 4), and consent form (Appendix 5). Permission for its use was also given by the Research and Development (R&D) manager at NHS North of Tyne PCT.
Data Collection: Quantitative data available from the fourteen pilot sites is available in the form of the Death Audit and Morbidity Information Query and Export Syntax (MIQUEST). Both sets of data are routinely collected by practices: Death Audit data retrospectively since 2007 and MIQUEST searches have been run routinely since 2009. Whilst some of the data overlap, other data, such as that on locality advance care plans and sudden deaths, do not. This data is mostly about identifying outcomes of the ICP and the structure of the ICP implementation so that essential mechanisms that lead to these outcomes can be explored, identified and tested.
Data Analysis: Statistical analysis was performed on MIQUEST and Death Audit data using IBM SPSS 17.0. The statistical tests used were: descriptive statistics, correlation, multiple regression, repeated measures t-test and cluster analysis. These tests investigated the tools the ICP uses, the good death outcomes the ICP generates and the embeddedness of the ICP. The embeddedness of the ICP refers to how many ICP interventions a GP practice uses; the more interventions that are used, the more it can be assumed the ICP is becoming routine in that GP practice. Practice comparisons were viable using statistics and contributed to essential contextual evidence. All the analyses have been undertaken in the knowledge that intervention recording rates can vary. However, since realist approaches seek to develop understanding of how, for whom and in what circumstances interventions work (as opposed to seeking absolute proof that they do), this was not considered an issue.
2. The Quality of Dying and Death Measure (contributes to programme theories 3 and 5) Recruitment: Nurses from GP Practice H identified bereaved relatives systematically by looking at patients who died six months before data collection began (01/02/2012). Bereaved relatives were sent the invitation letter (Appendix 6), information sheet (Appendix 7) and consent form (Appendix 8) in the post by the GP practice (H) on headed paper. If they responded, the response was sent to me at Northumbria University. The bereaved relative was then sent the QDDM and an email was sent to the GP practice nurse, who asked the relevant GP (who had previously had a significant involvement with the bereaved relatives loved one) to also fill in the same questionnaire. The GP was also provided with an invitation letter (Appendix 9), information sheet (Appendix 10) and consent form (Appendix 11). This was then emailed to me. Questionnaires were then matched and analysed.
All bereaved relatives recruited to the study were between four and six months post bereavement; this was an inclusion criteria stated by the study and followed by the GP practices. All potential participants were discussed with the founder of the ICP who is a retired GP with an interest in palliative care, to ensure that they were suitable and deemed fit to participate in the study.
Data Collection: There are several ‘good death’ measures available in the literature;
however a recent systematic review found the QDDM (Curtis, Patrick et al. 2002) to be the most widely studied and best validated (Hales, Zimmerman et al. 2010) (Appendix 12 includes statistics about the psychometric properties of the measure).
The QDDM was used to determine whether individuals who experience the ICP had a good death. This measure was distributed to bereaved families and key health care workers involved in a patient’s care. This section of data collection concerns the experiential and perceived outcomes of GPs and bereaved relatives. Experiential and perceived outcomes are extremely important, as the quantitative data lacks this rich quality.
Data Analysis: The questions were used to identify areas within palliative care that GPs and bereaved relatives perceive similarly and differently to support CMOC.
3. Focus groups with health care professionals and key organisation workers (contributes to all programme theories) Recruitment: Consent to contact staff who were eligible to take part in the three focus groups was obtained through R&D approval (Appendix 13) and letters to relevant organisations (Appendix 14). The date and time of the focus group was arranged and emailed to all the relevant potential participants (GPs, community matrons, social care workers, OOH staff, ambulance service, urgent care staff, voluntary organisations and relevant hospices including the locality palliative care unit) along with an invitation letter (Appendix 15), information sheet (Appendix 16) and consent form (Appendix 17). Participants were asked to read the study information provided, reply to the email to confirm their attendance and to bring their completed consent form with them to the focus group. Spare copies of the consent form were available at the focus group for those who forgot to bring it.
Focus Group 1 (FG1) aimed to highlight how the ICP functioned in practice on a regular basis. It took place in May 2012 and had twelve attendees (three GPs, two community matrons, two ward staff nurses, one social care team lead, one junior doctor, one palliative care unit project manager, one palliative care unit deputy manager, one team leader for the OOH nursing service). Focus Group 2 (FG2) took place in October 2012 and aimed to get constructive feedback on primitive CMOC;
it had eight attendees (two GPs, one MacMillan nurse, one ward staff nurse, one community matron, one social care team lead, one district nurse, one palliative care unit deputy manager). Focus Group 3 (FG3) had six attendees and took place in October 2013 (four GPs, one care home manager, one social care team lead). This focus group aimed to refine final CMOCs. FG1 and FG3 did not require participants to do pre-focus group work (materials to review). FG2 required participants to study Soft Systems Methodology (SSM) maps that had been created based on the ICP (example provided, see appendix 37). This allowed participants to become familiar with the maps, generate ideas prior to the focus group and encouraged them to be confident in the discussion of them. FG2 participants were sent the SSM maps to review before they participated in the focus group.