Evaluation of an initiative to improve advance care planning for a home-based primary care service

Advance care planning (ACP) refers to “the process of discussing and recording preferences concerning goals of care for patients who may lose capacity or communication ability in the future” [1]. ACP is associated with improved symptom management [2], increased caregiver satisfaction [3, 4], and reduced hospital use at the end of life [5]. Nevertheless, studies show that providers lack the confidence and competency to initiate ACP conversations [6], especially for frail populations [7]. Advance care preferences arise from the ACP process. These include an expressed preference in the event of an acute health crisis, for inclusion or avoidance of: life-extending procedures such as cardio-pulmonary resuscitation and intubation; transfer to hospital; home-based therapeutic interventions; and/or comfort care and palliation. We call these preferences rather than directives because the ultimate decision for a preferred intervention rests with the most responsible provider based on weighing the medical appropriateness in various circumstances.

Frailty is a “multidimensional syndrome of loss of reserves (e.g., energy, physical, ability, cognition, health) that gives rise to vulnerability” (p. 489 [8]). Frailty, more than age, predicts poor surgical outcomes [9], longer hospital stays, institutionalization [10], and death [10]. Failure to adequately consider these outcomes with frailty during ACP has the potential to favour life prolonging measures over symptom control and quality of life [11]. Despite these important considerations, providers report a disproportionate hesitancy to engage in ACP with frail older patients compared to those with cancer [12].

The Palliative and Therapeutic Harmonization (PATH) model was developed by two geriatricians in Nova Scotia, Canada [13] to increase engagement in ACP with frail older adult populations and their families in inpatient and outpatient settings. This study aims to evaluate the implementation of a PATH approach to ACP in a home-based primary care program for frail homebound older adults in Vancouver, Canada.

Interest for implementing the PATH model originated amongst primary care clinicians, nurses, and allied health providers working in a home-based primary care program run by the Vancouver Coastal Health Authority. Primary care providers identified the need for improved ACP after attending presentations on the PATH approach. They noted its high relevance to their practice population, which focuses on the care of frail older adults. The staff sought and successfully obtained donor funding to support the training and implementation of a PATH approach to ACP.

A total of eight physicians and three nurse practitioners participated in the full PATH training. Data were abstracted from 314 charts, 200 before implementation and 114 following implementation. The mean (standard deviation) age of patients was 88.1 (± 7.2) years. A little over one third (35.4%) were male and for a similar proportion (109/314, 34.7%) English was not their first language. Approximately 50.3% of patients had documented dementia and 19.4% were severely or very severely frail (Table 3).

EMR documentation of frailty stage was 27.0% before implementation compared to 74.6% after (p < .0001). There was an increase in documentation of a substitution decision-maker(s) (63.5 to 71.9%, p = 0.128) that was not statistically significant. Likewise, there was increased documentation about whether to pursue hospital-based care (61.5 to 100%, p < .0001). In contrast, following the intervention, documentation of resuscitation code status decreased from 79.5 to 67.5% (p = 0.018). The mean (standard deviation) composite documentation score (i.e., the sum of the four measures, ranging from 0 to 4), was 2.32 (1.16) before versus 3.14 (1.11) after the intervention, p < .0001 (Table 4).

Expressed preference for DNR was not statistically different before and after implementation of the PATH process (90.6% versus 88.3%, p = 0.591), whereas expressed preference for DNH increased (23.6% before and 51.8% following the initiative, p < 0.0001) (data not shown). Separate multiple regression models were run for an expressed DNR preference and an expressed DNH preference. There was no significant effect on expressed preference for DNR after the intervention, and age and osteoporosis remained significant in the final model. There was a significant effect on expressed preference for DNH after the intervention even after adjusting for age, male gender, and relevant comorbidities (adjusted odds ratio 3.14 (95% confidence interval 1.78, 5.55)) (Table 5).

This study evaluated the implementation of a PATH approach to ACP in a home-based primary care program for frail older adults. It measured the proportion of patients with chart documentation of key surrogate measures of ACP and expressed decision-making preferences for DNR and DNH, before and after the intervention.

Compared to before the PATH intervention, there was significant improvement in the EMR documentation of frailty stage, and documentation of a hospitalization decision in the after period. Although not statistically significant, the proportion of patients having a documented SDM listed on their EMR increased from 63.5 to 71.9%. While EMR documentation of these variables does not necessarily reflect the quality of ACP communication, the variables are reasonable surrogate indicators of ACP, and their increase following PATH training suggests more proactive ACP occurred as a result of the training.

One exception to the above findings is the observed paradoxical decline in the documentation of a resuscitation decision in the post-implementation period compared to baseline. This finding may relate to the relative focus that PATH communication places on educating families and patients regarding hospital interventions rather than cardio-pulmonary resuscitation, which is a relatively rare event even when individuals are frail and homebound. This may also reflect a presumed DNR for patients who opted for a DNH. In addition, documentation of a resuscitation decision was high at baseline (79.5%) and was recorded in over two thirds of cases post implementation, despite the observed decline.

This analysis demonstrated a significant increase in the proportion of patient/ SDM(s) dyads expressing a preference to avoid hospital (DNH) in favour of home-based care interventions for future health crises, which increased from 23.6% before to 51.8% after implementation. Although the actual rate of hospital transfer before and after implementation was not measured, patients’ recorded preferences to avoid cardiopulmonary resuscitation and hospitalization have been shown to result in lower rates of hospital admission [1], an outcome that is desirable in frail populations if symptoms can be managed in community settings.

Beyond the empirical results, some lessons about the importance of context and institutional support for practice change were learned through informal discussions with participating clinicians, course instructors, program administrators and practice coaches over the course of the project, and from a qualitative evaluation report on ACP discussions at HomeViVE [15]. Context enablers of the initiative included the enthusiasm of the provider team and access to on-call after-hours care. The ongoing support and advice of the PATH geriatricians during the implementation period, and the deployment of a practice coach who provided clinicians with feedback data on their ACP metrics in real-time were additional enablers.

A contextual barrier to PATH ACP implementation in the home-based primary care setting was the fee-for-service remuneration method for family physicians. Family physicians, who comprised the majority of the primary care providers, and fee-for-service remuneration of these practitioners likely discourages comprehensive geriatric assessment and ACP communication, which require extensive time commitments. Another identified barrier was the requirement that the PATH model, as it was originally conceived, delegate a substantial portion of the assessment for staging frailty to trained allied health providers. Despite an administrative agreement to implement the PATH ACP initiative in the home-based primary care program, reassigning nursing or allied heath staff to this role was not supported by the program administrator, leaving the burden of ACP activities to the most responsible primary care physician or nurse practitioner. The PATH ACP approach, therefore, had to be carried out longitudinally using shorter, cumulative assessments by the most responsible primary care provider, instead of having other members of the team complete the CoCGA in more extended visits. Additional systemic barriers include the dual charting system, for home-based clients, between the HomeViVE and home care team, and lack of inter-operability between these systems.

As with all studies there were some limitations. First, data was not extracted from a reference population of patients over the same time period. It is, therefore, possible that the observed changes may be a result of other ACP interventions, unrelated to the PATH ACP initiative, taking place within the health region at that time. In addition, other key elements of the PATH ACP approach, such as the use of the PATH form that guides decision-making about future health crises and documentation of preferred place of death, were not measured. This study did not measure hospital utilization, the patient and family experience, clinical satisfaction with PATH intervention, or the cost-effectiveness and sustainability of the program. All of these measures could be important areas for future research to evaluate ACP practice improvement interventions.

Despite these limitations, this study had a number of strengths including the fact that it was implemented in the community, whereas most published research evaluating ACP initiatives take place in nursing home or hospital settings [1]. A second strength is the sample size, which was relatively large compared to other evaluative studies of ACP interventions [1]. Furthermore, in a systematic review [16] of ACP evaluation studies, none of the studies discussed implementation challenges nor explored how the intervention may have differed in practice compared to what was originally envisaged. As such, the description of contextual elements discussed here present new insights. Finally, the results and findings of this study are generalizable to other longitudinal home-based primary care programs for frail older adults operating in Canada [17, 18] and the US [19, 20].

In conclusion, evaluation of the PATH ACP initiative suggests a promising approach to ACP in the setting of home-based primary care of frail older adult populations. As with all complex adaptive systems, further implementation and scaling of the approach needs to consider key contextual enablers and barriers.