Exploring the Uptake of Advance Care Planning in Older Adults: An Integrative Review

A search strategy was developed in collaboration with a coauthor medical librarian (R.L.W.). The search included MeSH terms: “advance care planning,” OR “advance directives,” OR “living wills,” AND “patient preferences,” OR “consumer participation,” OR “patient participation,” OR “personal autonomy” in multiple combinations. A filter of age 65 years and older was placed on each search. A comprehensive search was conducted through December 2019 in PubMed, CINAHL, PsycINFO, and Embase (Figure 1).

Studies were included if they met the following criteria: 1) addressed three factors: a) components of ACP (advance directives, identification of healthcare proxy, discussions with patients, families, and health-care providers), b) ACP uptake, and c) older adults; 2) peer-reviewed; 3) published in English; 4) featured primary research (data collected directly by investigators); and 5) encompassed serious illness, including frailty. Exclusion criteria included 1) published in a non-English language; 2) opinion articles, study protocols, case studies, and conference abstracts; 3) systematic reviews; and 4) focused on ACP for specific disease entities such as cancer, dementia, or congestive heart failure. The decision to exclude specific diseases was intentional and prognostication-related as ACP approaches differ between older adults with clear terminal conditions (i.e., heart failure, cancer) and those who are on trajectories of frailty and facing the end of life with greater uncertainty. We were most interested in ACP uptake in cases of uncertainty. Seventy-seven studies were preliminarily included in the review, with one additional study added from a subsequent hand search of references.

An integrative design and synthesis approach was used to evaluate studies in an inductive stepwise process, including 1) quality appraisal of studies using critical appraisal tools, 2) extraction of descriptive content and rating of evidence (levels) with the research hierarchy, 3) analysis of applicable statistical results (effect sizes) of factors (variables) that influence ACP as a dependent variable, and 4) content analysis of results to identify categories, defined themes, and descriptors within studies. Our approach allowed for the study findings to move beyond a summary used in a narrative review to an approach used to generate new insights and understanding of ACP uptake in a broad, yet systematic, manner.

Quality appraisal of studies was conducted using Joanna Briggs Institute Critical Appraisal tools for analytical cross-sectional, quasi-experimental, randomized controlled trials (RCTs) and qualitative studies. Studies were examined to determine the extent to which each met appraisal criteria, noting limitations that might influence the accuracy of study findings (Appendix I). Data were reviewed by two independent researchers (E.F. and C.A.M.), and disagreements were resolved through discussion. In the second step, data were extracted from each study, and evidence tables were developed which included study objective(s), level of evidence, setting and location, sample and participant characteristics, study design and data collection methods, ACP uptake features/factors, and ACP uptake outcome (Appendix II). Results from studies comprising quantitative analyses from which effect sizes could be generated were categorized into key sets including 1) demographic variables, 2) psychosocial variables, 3) variables related to physical function/disability, and 4) other variables (i.e., targeted interventions, process facilitators). To extract categories and themes from studies, results sections of all studies containing a qualitative analysis were line-by-line coded for factors contributing, negatively or positively, to uptake of ACP. In the fourth step, categories were developed; themes from all 78 studies were defined and organized under each category, and descriptors from each study were placed in tables (Appendix III). From the tables of categories, themes and descriptors, an integrated conceptual model was developed (Figure 2) that provides a broad visual depiction of the major categories and relationships.

Seventy-eight studies (1996–2019) met inclusion criteria. The review included 23 qualitative, five mixed-methods and 50 quantitative studies (including 12 RCTs). Studies were conducted in 13 countries: United States n = 40, Canada n = 10, Australia n = 7, Hong Kong n = 3, Taiwan n = 1, United Kingdom n = 4, Norway n = 2, Germany n = 2, Belgium n = 3, South Korea n = 1, Japan n = 1, Sweden n = 1, and Netherlands n = 2. One study represented 11 countries as a whole, but results were not reported by country. Sample sizes for the 78 studies included qualitative (range: 7 to 503), mixed-methods (range: 21 to 289), and quantitative (range: 18-25,550). The 12 RCT sample sizes ranged from 14 to 2294. Details of age range, gender, and race/ethnicity of participants are shown in Appendix II.

Demographic variables analyzed for their effects on ACP included age, gender, race/ethnicity, marital status, education, and place of residence. Most studies reported that the likelihood of engaging in ACP increased with increasing age (ORs = 1.01 to 2.22).^{,  ,}  Gender demonstrated generally mixed effects (ORs = 0.60 to 1.14)^{,  ,  ,  ,  ,}  ; however, being married increased likelihood of ACP as compared to not being married (OR = 1.07 to 1.84).^,, Higher levels of education were most commonly associated with increased likelihood of ACP (OR = 1.25 to 3.90),^,, and one study reported the effect after adjusting for other demographic variables (OR = 1.63 and 2.40). Living in a private setting, compared to an institutional setting, dramatically reduced the likelihood of ACP (OR = 0.53 to 0.66) (Table 1).

Results from studies of psychosocial variable effects on ACP uptake are summarized in Table 2. Those variables have included living situation, social support, perceived health, quality of life, nature of marital relationship, general family functioning, number of children, relationship with children, religiosity/spirituality, presence of depressive symptoms, and personal preferences. Social support, described as having someone to listen and provide emotional support, was associated with increased likelihood of ACP (someone to listen, OR = 6.93; someone to provide emotional support, OR = 2.83). If grouped into categories (good/better/excellent vs. poor), the likelihood of ACP generally was reported to increase with better perceptions of health (OR = 1.01 to 1.90).^, Higher family functioning, or the degree to which families function as a unit, was associated with increased likelihood of ACP (OR = 1.69 and OR = 1.99), as was increased emotional support from adult children (OR = 1.10, 1.39), yet the likelihood of ACP tended to decrease as the number of children increased (OR = 0.58, 0.88). Other psychosocial variables including religiosity/spirituality were associated with increased likelihood of ACP even when adjusted for religious affiliation, degree of religiosity or spirituality, beliefs, values, sociodemographic, and health status (ORs = 1.32, 2.25). Decreased or absent depressive symptoms were consistently associated with increased likelihood of ACP (OR = 1.24 to 2.31).^,, A couple of studies looked at the effects of personal preferences as precursor variables (decision control preferences and wish to be informed of diagnosis of terminal disease) on the likelihood of ACP. Generally, higher decision control preferences increased the likelihood of ACP (OR = 0.93 to 2.00), while the wish to be informed increased the likelihood after adjusting for age and gender (OR = 9.19) (Table 2).

Disability and functioning variables included physical functioning, frailty, cognitive functioning, comorbidity, stroke, and the use of health-care services. As shown in Table 3, associations of physical functioning with ACP were mixed, yet most studies reported an increased likelihood of ACP with better functioning (OR = 1.79, 1.88, OR = 1.20–1.26).

Increased cognitive functioning (higher Mini-Mental State Examination scores) was generally associated with increased likelihood of ACP (OR = 1.57; OR = 1.60). After adjusting for age, sociodemographics, and other pertinent variables, studies generally found that the likelihood of ACP increased with increasing comorbidities (OR = 1.30, 1.82)^, (Table 3).

Several studies, as shown in Table 4, investigated the effect of specific programs/variables on engagement in ACP using implementation and clinical trial designs. Generally targeted interventions increased engagement outcomes (multifaceted, OR = 2.17; electronic motivational prompts, OR = 3.20; defined ACP program [Respecting Choices], OR = 23.27), as did those that focused on process facilitators (decision aids, OR = 22.0, OR = 2.61; standardized templates, OR = 3.82; and online ACP program [PREPARE], OR = 17.14 [documentation], OR = 5.6 [discussions]) (Table 4).

The category of enhanced communication processes emerged as the central requisite addressed in some way in every study for successful ACP and improved outcomes. The term “enhanced” conveys the necessity of additional factors/traits required to move beyond the status quo. These requirements are represented by themes of honest/frank discussion,^{,,  ,  ,}  building personal relationships,^{,,  ,}  shared decision-making,^,,, stages of change in terms of readiness,^,, framing,^{,,  ,}  empathy,^,, and technique/style.

Five studies^{,,  ,  ,}  reported the importance of honest/frank discussion in ACP with a focus on the delivery of truthful information. The concept of “conditional candour” in one study described a preference for frank information while also assessing readiness, and another highlighted the necessity for physicians to communicate through an honest and straightforward approach while continuing to be attentive to patients' informational and emotional needs. Another theme to emerge was that of building relationship and forming a connection or bond between two or more parties.^{,,  ,}  The importance of building a relationship extended to both patients and surrogate decision-makers, highlighting engagement in ACP as a dyadic activity, with the surrogate being the one who knew the older adult best and would ensure their treatment was in accordance with values. The importance of being “known” by a provider was noted as critical to the ease of ACP discussions and honoring preferences.

Shared decision-making is reported as a way to support and maintain individual patient autonomy while requiring skilled providers to seek permission and respect patient communication and patient engagement.^,,, Another approach focused on the assessment of readiness for change in participants as a factor for modifying/adapting individual approaches.^,, Recognition of ACP as specific behaviors in which individuals progress through stages of change or readiness has been explored by Fried et al. and grounded in the Transtheoretical Model of behavior change.

The role of the provider in approaching conversations with patients about the end of life is key, as is recognition that conversations are often emotionally charged and require preparation. Health-care providers have a responsibility to assist in the decision-making process through education about ACP and communication skills that impart compassion and empathy. The specific roles and necessary preparation of health-care providers is critical for the uptake of ACP and the timing of ACP to occur within a framework that emphasizes responsiveness to patient and family emotions, while also focusing on overall goals of a patient's care. The category of provider role and preparation is supported by four themes, including interdisciplinary roles,^{,,,,,,  ,  ,  ,}  provider awareness of readiness,^{,,,,,,  ,}  trained facilitators,^,,, and a long-standing relation with a family physician. Ten studies reported the use of interdisciplinary team members as part of a study protocol to deliver ACP education, assess ACP knowledge, lead ACP meetings, and complete care planning documents.^{,,,,,,  ,  ,  ,}  Four studies used a social worker to conduct ACP during home visits or an office visit before the scheduled provider visit,^,,, noting that the social workers' professional role of counseling and facilitative communication is a natural fit for ACP. Nurses' role in ACP is represented in five studies^,,,, through a nurse facilitator to guide participants toward sharing and expressing their EOL wishes to a nurse-led interactive educational workshop and advance practice registered nurses' leading conversations following a Respecting Choices model.