«Citation: Dalkin, Sonia (2014) The Realist Evaluation of a Palliative Integrated Care Pathway in Primary Care: What Works, For Whom and in What ...»
GP2 (FG2): “Yes I think there’s, there’s, before you’re about to have it there’s a stimulus, after you’ve had it because of the feedback there’s a stimulus and I know I’ve, in the past our statistics improved because it stimulated us.” Community Matron 2 (FG1): “There’s something about, which I’ve thought about since the beginning of this meeting, is that there’s been a great leadership with this, these visits, it was the same with others, like diabetes and heart quality visits that there was, exactly what you said, before we’ve got to get it ready, we’ve got to get it right and then you do sort of pick up on the points and then it’s just human nature everything drops off again and I think if the leadership doesn’t keep going it’s sort of keeping that, gotta keep it going, so there is something about that (inaudible) tha t’s lost.” This highlights how CQI is itself modifying the delivery of the programme. The knowledge of an upcoming PCQV visit (resource) is enough to make staff feel they need to adhere to the programme (ICP) specification more rigidly. However, the negative facet to this is that post PCQV, after initial motivation and enthusiasm from the visit has peaked and then diminished, health care professionals may begin to lose sight of the ICP goals and ethos and make less use of the associated interventions.
In summary, the health care professionals’ reasoning is changed, as they improve their performance and use of ICP interventions, due to the knowledge of an upcoming PCQV or in reaction to feedback from a PCQV (resource). This then leads to the more frequent use of interventions advocated by the ICP, as described in the outcome section to follow.
Context: Proactive and patient centred care Many of the ICP interventions are focused on proactive care (palliative care registration, preference discussion, ACP, anticipatory medication), meaning that in the use of the interventions GPs are implicitly being proactive. Thus the medical and political drive for proactive care is a factor that will affect the use of ICP interventions. Patient centred care is another philosophy that is widely advocated in health care (Little, Everitt et al. 2001, McCormack, Treiman et al. 2011, Mulley, Trimble et al. 2012) and can also be considered as a context that affects the use of the ICP interventions and consideration of place of death. Patient centred care is the process of providing care that is responsive to individual personal preferences, values and needs and ensures that patient values are central to all clinical decisions.
Some of the ICP interventions, such as preference discussions and ACP, explicitly present operationalisations of patient centeredness. All health care professionals in the focus groups were very aware of the notion of patient centred care and highlighted it in FG2 as the most important context in regard to several interventions that the ICP uses: the ICP overall, palliative care registration, ACP, MDT meetings and the traffic light system. CQI also aims to improve services with a view to providing patients with a good death (being patient centred).
FG3 highlighted that proactive and patient centred care were the crucial context in this CMOC, with the NHS drive for CQI acknowledged but not pivotal. The NHS drive for improvement is actually incorporated in the social programme and forms part of the mechanism resource.
GP3 (FG3): “I would suggest that proactive and patient centred care is the most important context… Well of the two, whilst the NHS drive for CQI is there in the background I don’t see it as something that drove us.” GP4 (FG3): “No I mean I, I agree, the NHS drive for the CQI is there in a lot of domains but I would say that locally, the end-of-life care is one of the things that we did really progress, and I think that’s probably come from the feedback from (palliative) care quality visits.
Erm that was one of the reasons I think that we were more enticed into that role than all the other stuff that was sent to us really” Despite the important role of CQI as a mechanism, it was not considered to be a contextual factor that drove the use of ICP interventions.
Figure 10: CMOC3, CQI and intervention use The drive to provide patient centred and proactive care in the locality (context) means that the education and feedback PCQVs provide (resources) result in motivation and enthusiasm from the health care professionals, both prior to and post PCQV. This makes them engage with the ICP more (reasoning). This mechanism results in increased use of interventions (outcome): palliative care registration, preference discussions, LCP. However, locality advance care plans and anticipatory medication use have not significantly increased. Explanation of the non-significant increase in locality advance care plans will be discussed elsewhere in the thesis (Chapter 6: Preference discussions and locality advance care plans, p.199).
Anticipatory medication may not have significantly increased because GP practices can be very variable in their recording of anticipatory medication. This CMOC demonstrates the importance of the PCQV in continually improving the ICP and increasing intervention use and recording through motivating and educating the health care professionals who deliver it. CMOC2 (Figure 10) illustrates the relationship between the contexts, mechanisms and outcomes in this configuration.
CMOC4 (I) – Peer support, leadership and embeddedness of the ICP Outcome: Embeddedness - three high performing practices Cluster analysis can be used to discover structures in data but does not explain why they exist. Thus in this analysis it will be used to identify relatively homogeneous groups of GP practices based on selected characteristics (such as ICP interventions or place of death outcomes), input by the analyst, which will be stated below in each cluster analysis. Two separate cluster analyses were conducted - one focused on outcomes (place of death, number of interventions administered and number of palliative care registered deaths), the other focused on a number of intermediate outcomes from ICP interventions (palliative care registration, the locality advanced care plan, OOH notification, DS1500 forms for monetary benefits, anticipatory medication and LCP). Two clusters were selected from the data collected from fourteen GP practices. Hierarchical clustering was used, which involves factors being organised into a hierarchical structure as part of the procedure. This was done with a between groups linkage that clusters objects based on the distance between them.
Cluster Analyses I The first cluster analysis completed showed two clusters when using four factors from 2011/12 data: patient death in their own home (MIQUEST), patient presumed preferred place of death including home, care home and palliative care unit/hospice (MIQUEST), ratio of interventions administered to total deaths (MIQUEST) and ratio of palliative care registered deaths to total deaths (MIQUEST). These four factors were selected as they are outcomes of the ICP and offer explanatory potential however they are not outcomes of intervention use in their primary form (as in Cluster Analyses II). The analysis identified two clusters, cluster 1: GP Practices A, B, C, F, G, H, J, K, L, M and N, and cluster 2: GP Practices D, E and I.
T-tests were performed to find out which factors significantly contributed to the clusters. Normality of data was checked for all outcomes the t-tests were performed on; normality pertained in all instances. There was a significant difference between clusters for the numbers of home deaths patients achieved (t(12) = -3.07, p.05).
Cluster 2 (practices D, E and I) had significantly more patients who died in their own homes (M = 10.33, SD = 4.51) than cluster 1 (M = 4.36, SD = 2.58). There was also a significant difference between clusters for presumed preferred place of death (MIQUEST) (t(12) = -6.29, p.001). The presumed preferred place of death is a combination of home, care home and palliative care unit/hospice deaths. It is assumed that these are preferred places of death for patients as a considerable effort would be needed in order for a patient to die in this location, and hospital is reported as the least preferred place to die in both locality and regional data (Gomes, Calanzani et al. 2011). Significantly more patients died in the presumed preferred place of death in cluster 2 (M = 29.33, SD = 6.81) than in cluster 1 (M = 8.45, SD = 4.68). The ratio of interventions administered (t (12) = -1.7, p.05) and the ratio of palliative care registrations to total deaths (t(12) = -1.29, p.05) were both not significantly different between cluster 1 and cluster 2.
To summarise, cluster 2, which consisted of practices D, E and I helped patients to achieve a home or presumed preferred place of death significantly more than cluster 1 (all other practices from the locality).
Cluster Analyses II The second cluster analyses focused on interventions and formal outcomes of the ICP. This cluster organised the GP practices in the same way as cluster I; cluster I consisted of GP practices A, B, C, F, G, H, J, K, L, M and N, and cluster 2 consisted of GP Practices D, E and I. The factors entered into the cluster analysis were palliative care registration, locality advance care plan, OOH notification, DS1500 forms, anticipatory medication and LCP. All data was from 2011/12.
T-tests were performed to find out what significant differences contributed to the clusters. Normality of data was checked for all interventions t-tests were performed on; normality pertained in all instances. There was a significant difference between clusters for the number of palliative care registrations made (t(12) = -4.63, p.001). Cluster 2 (practices D, E and I) registered significantly more patients on the palliative care register (M = 36, SD = 12.12) than cluster 1 (M = 13.27 SD = 6.23).
There was not a significant difference for use of the locality advance care plan between the two clusters, although the p value was near significance (t(12) = -1.94, p = 0.76) and the means indicated that cluster 2 (M = 9.33, SD = 3.22) completed more advance care plans than cluster 1 (M = 4.64, SD = 3.8).
OOH notifications were significantly different between the two clusters (t(12) = p.01), with cluster 2 completing significantly more OOH notifications (M = 15.33, SD = 5.77) than cluster 1 (M = 6.27, SD = 1.153).
A t-test showed that cluster 1 and 2 had a significant difference for DS1500 forms completed (t(12) = -2.45, p.05), with cluster 2 administering more DS1500 forms (M = 8.33, SD = 6.66) than cluster 1 (M = 3.09, SD = 2.02).
Clusters 1 and 2 showed a significant difference on the amount of anticipatory medication prescribed to patients (t(12) = -4.76, p.001). Cluster 2 prescribed significantly more anticipatory medication (M = 14.67, SD = 3.22) than cluster 1 (M = 5.18, SD = 3.03).
There was a significant difference in the amount the LCP was used between cluster 1 and cluster 2 (t(12) = -7.26, p.001). Cluster 2 used the LCP significantly more (M = 26.67, SD = 6.51) than cluster 1 (M = 6.45, SD = 3.67).
To summarise, cluster 2, consisting of practices D, E and I, administered or completed significantly more formal interventions from the ICP than cluster 1 on all formal interventions apart from use of the locality advance care plan, where the outcome was not significant, but it was extremely close to significance. They also had significantly more home or presumed preferred place of death (home, care home or palliative care unit). Since it has now been highlighted by this quantitative analysis that practices D, E and I can be considered as ‘high performers’ a comparison can be made; between the high performing practices and the other practices in the locality.
Mechanism: Opinion leaders and a champion Peer support and leadership are both very important when implementing new initiatives (Greenhalgh, Robert et al. 2004) and contribute to embeddedness.
Practice D, E and I all have peer support and leadership in the form of opinion leads or a champion. Practice I hosts the GP Locality Lead for Research who is very research active and a data active GP who ensures the recording, analysis and feedback of data to the practice is of a high standard. This GP and the GP Locality Lead for Research could be considered as peer opinion leaders (Greenhalgh, Robert et al. 2004). These two peer opinion leaders at practice I had a positive effect on ICP diffusion in their practice, they adopted the ICP promptly after sign up due to their active roles and their desire for constant progress through research, feedback, and innovation and this had an effect on others in the practice (I).