6.2.1 STUDY I
It is a difficult task to change the culture and attitudes of staff to improve end-of-life care outcomes at resident level. In a previous systematic review it was noted that interventions that have narrow aims and low complexity are more likely to be successful in changing care practices compared with those with more ambitious aims and high complexity (Low et al. 2015). In the two
78
largest trials included in the systematic review, complex interventions were implemented. An early pilot study of the INTERACT intervention had shown significant reductions in burdensome hospitalizations (Ouslander et al. 2009), but more rigorous replications did not prove its effectiveness (Kane et al.
2017). Similarly, the large European multicenter collaboration PACE, with its well outlined and long-lasting intervention, did not affect EOL care (Van den Block et al. 2019). Ambitious and complex interventions may be hindered by common barriers observed in previous studies, such as high staff turnover, a high training attrition rate, low organizational support, lack of resources or commitment, as well as opposing cultures and attitudes (Low et al. 2015;
Collingridge Moore et al. 2020). The use of complex interventions has natural appeal, as it is compelling to believe that two possibly beneficial interventions together would make for a more effective intervention. This is reflected in the preference for complexity in the trials presented in Study I. However, individual components of the interventions are mostly not based on solid evidence, and expert opinion is prone to be changeable. For example, the expert panel of the European Association of Palliative Care found that some resources such as written documents were not considered important anymore, even if they were widely used in earlier initiatives such as the INTERACT project (Froggatt et al. 2020).
Many of the trials provided participants with possibilities for reflection and integration of new competencies into everyday practice. These have been suggested to be vital in effective adult learning (Knowles 1980; Kolb 1984;
Schön 1987; Dolmans et al. 2005). Additionally, addressing professionals’ own feelings related to fear of death has been suggested to be important in affecting their attitudes and culture in caring for those at the end of life (Linn et al.
1983).
A variety of training interventions was used in these trials. They included traditional behavioristic interventions such as lectures and also activating learning methods, such as case-based learning and role-plays. None of the trials justified their choice of training methodology on the basis of learning theories or adult education traditions. Similarly, descriptions of assessment of learning needs and motivation of staff members were absent. However, such assessments could be important, considering that in a post-hoc analysis, differences in motivation were found to determine the differences between facilities in the INTERACT project (Huckfeldt et al. 2018). The train-the-trainer approach, where one or several staff members are trained to disseminate and implement intervention in their own facilities was used in nine studies and in most of the complex interventions. It is likely to be efficient in distributing interventions and in engaging facilities by making the interventions originate from “inside” the facility, with the added possibility to adjust the intervention to suit the practices in each facility. On the other hand, nurses or nurse practitioners who were most often recruited to be facilitators, trained other staff members without any apparent pedagogical training and would be more susceptible to pressure from other staff members or managers than an outside expert might be. Also, the training backgrounds of these facilitators was rarely studied.
79
Implementing attitudes, skills and tools acquired in training into practice is a critical step towards resident-level results (Collingridge Moore et al.
2020). In several trials the use of the new skills or practices after initial training were evaluated – giving feedback to trainees or even re-training staff members. However, this effort did not seem to translate into better resident outcomes. It remains impossible to determine whether low effectiveness was due to insufficiencies in training or in implementation.
6.2.2 STUDIES II–IV
Findings from the trial reported in Studies II–IV are in line with those in previous trials. Resident characteristics concerning hospitalizations at baseline according to MDS were similar, but we were not prepared for the drastically differing caring cultures between the facilities. In our trial this heterogeneity of practices was reflected in the fact that one ALF was many times more likely to admit residents to hospitals. We noted in our training sessions that the staff in this particular ALF were reluctant to take responsibility in acute situations. In the small-group discussions, staff from this facility emphasized the fact that they could not take responsibility for a resident dying in the ALF. This was in contrast to the attitudes of most staff members in other study sites, who were more likely to view death as a natural part of their residents’ life courses. Within the limited scope of the training sessions, these beliefs and attitudes were not sufficiently changed to affect resident outcomes. It is important to note that previous research suggests that staff members are more likely to favor a residents’ hospitalization if they have inadequate possibilities for physician consultation or when fear of legal consequences drives actions (Laging et al. 2015).
Trials in Study I and others imply activity in ACP as an effective means to reduce hospitalizations. In Study I we recognized two successful ways to promote ACP in LTCFs: employing a nurse especially for this purpose (Molloy et al. 2000), or using an ACP video decision-aide together with formalized conversion instructions to staff (Hanson et al. 2017). In our trial, ACP discussion activity was high in both our intervention and control arms, and it markedly increased during the follow-up year. However, this did not have an effect on hospital use. Also, we did not gather information about the results of the ACP discussions such as ADs, making it difficult to judge the usefulness of these discussions. Effective ACP interventions may need to include families in the decision-making process, as suggested by the findings in Study I.
In Study III, a significant difference was observed between groups in ESAS total scores at six months, implying that in the intervention group symptoms decreased compared with the control group. Regarding the ESAS instrument, a minimal clinically significant change for the total score has been approximated to be 3–5 points for improvement (Hui et al. 2016). Thus, the observed difference of around three points between groups at six months can be considered clinically significant. The difference was not present at the later 12-month assessment, making it difficult to evaluate the overall significance of the finding. Historically, pain has been undertreated in LTCFs, with an increased likelihood of undertreatment with declining cognition (Hunnicutt et
80
al. 2017). Systematic evaluation of pain with tools such as PAINAD has been suggested as a means of improving pain treatment. However, the results of implementation trials remain inconclusive (Liu and Lai 2017; Rostad et al.
2018), and even interventions with analgesics show mixed results (Jordan et al. 2011; Lukas et al. 2019). The observed low baseline PAINAD and ESAS scores suggest the possibility of a floor effect in these instruments.
Proxies’ satisfaction with care is an important measure of care quality in LTCFs. Satisfaction with care measured by using SWC-EOLD instrument has been seen to reflect the level of consensus about treatment choices between staff and proxies (van der Steen et al. 2012). In a recent study it was noted that Finnish LTCF residents’ proxies gave the lowest satisfaction with care scores when six European countries were compared (ten Koppel et al. 2019a). In line with the results of previous trials described in Study I, in Study III we noted that SWC-EOLD scores were not easily affected. Current interventions may be insufficient in affecting the many aspects they reflect, such as facility culture, communication and capabilities for joint decision-making.