key: cord-0834565-gv0zs6dh
authors: Garland, Allan; Keller, Heather; Quail, Patrick; Boscart, Veronique; Heyer, Michelle; Ramsey, Clare; Vucea, Vanessa; Choi, Nora; Bains, Ikdip; King, Seema; Oshchepkova, Tatiana; Kalashnikova, Tatiana; Kroetsch, Brittany; Steer, Jessica; Heckman, George
title: BABEL (Better tArgeting, Better outcomes for frail ELderly patients) advance care planning: a comprehensive approach to advance care planning in nursing homes: a cluster randomised trial
date: 2022-03-23
journal: Age Ageing
DOI: 10.1093/ageing/afac049
sha: a771d71072e1c44d7b87ec9b61caabc2938fa0b4
doc_id: 834565
cord_uid: gv0zs6dh

BACKGROUND: Nursing home (NH) residents should have the opportunity to consider, discuss and document their healthcare wishes. However, such advance care planning (ACP) is frequently suboptimal. OBJECTIVE: Assess a comprehensive, person-centred ACP approach. DESIGN: Unblinded, cluster randomised trial. SETTING: Fourteen control and 15 intervention NHs in three Canadian provinces, 2018–2020. SUBJECTS: 713 residents (442 control, 271 intervention) aged ≥65 years, with elevated mortality risk. METHODS: The intervention was a structured, [Formula: see text] 60-min discussion between a resident, substitute decision-maker (SDM) and nursing home staff to: (i) confirm SDMs’ identities and role; (ii) prepare SDMs for medical emergencies; (iii) explain residents’ clinical condition and prognosis; (iv) ascertain residents’ preferred philosophy to guide decision-making and (v) identify residents’ preferred options for specific medical emergencies. Control NHs continued their usual ACP processes. Co-primary outcomes were: (a) comprehensiveness of advance care planning, assessed using the Audit of Advance Care Planning, and (b) Comfort Assessment in Dying. Ten secondary outcomes were assessed. P-values were adjusted for all 12 outcomes using the false discovery rate method. RESULTS: The intervention resulted in 5.21-fold higher odds of respondents rating ACP comprehensiveness as being better (95% confidence interval [CI] 3.53, 7.61). Comfort in dying did not differ (difference = −0.61; 95% CI −2.2, 1.0). Among the secondary outcomes, antimicrobial use was significantly lower in intervention homes (rate ratio = 0.79, 95% CI 0.66, 0.94). CONCLUSIONS: Superior comprehensiveness of the BABEL approach to ACP underscores the importance of allowing adequate time to address all important aspects of ACP and may reduce unwanted interventions towards the end of life.

In 2016, U.K. nursing homes (NHs) housed 4% of people 65 years and over, and 15% of those 85 and older [1] . In Canada, comparable figures are 6.3% over 65 years, and 11.0% over 75 years [2] and numbers are similar elsewhere in the industrialised world [3, 4] . Most NH residents are frail, live with dementia and need assistance with activities of daily living [4, 5] . The ageing population will drive further growth of NH populations [6, 7] . With ∼20% of all deaths occurring in NHs [8] [9] [10] and median survival of 2 years [11] [12] [13] , NH residents must have the opportunity to consider, discuss and document their healthcare wishes, a process known as advance care planning (ACP) [14] . The disproportionate mortality experienced by NH residents from the COVID-19 pandemic underscores the importance of ACP in NHs [15] .

There are important problems with ACP in NHs. Most people have mediocre knowledge of, and engagement in, ACP [16, 17] . An American study of families of NH residents reported little or no ACP discussion before medical emergencies [18] . Through stakeholder engagement, we identified four 'weak links' in ACP: (1) uncertain identification of substitute decision-makers (SDMs), (2) poor understanding of the appropriate role of SDMs, (3) insufficient context provided for clients to understand the available medical choices and (4) failure to prepare in advance for health crises [19] . In 38 Canadian NHs, we identified low participation of physicians and residents in ACP discussions, infrequent discussion of artificial life support, inconsistent adherence to prior ACP decisions and incomplete communication of ACP decisions upon transfer of residents to an emergency department or hospital [20] .

To address these deficiencies, we developed BABEL ACP (Better tArgeting, Better outcomes for frail ELderly patients), a comprehensive, person-centred approach based on best practices [21] . Here we describe the results of the BABEL ACP Trial in Canadian NHs [22] . We hypothesised that compared with prevalent ACP practices, BABEL ACP would improve comprehensiveness of ACP, and resident comfort at the end of life.

This unblinded, parallel-group, cluster randomised study included a convenience sample of 29 NHs in three Canadian regions: Calgary, Alberta; Winnipeg, Manitoba; and Kitchener-Waterloo, Ontario. See Appendix 1 for further information about securing participating NHs. Participants were recruited and data collected from 28 August 2018 to 9 August 2020. Trial registration: ClinicalTrials.gov (NCT03649191). This study was approved by the Research Ethics Boards of the Universities of Manitoba (#HS20669), Calgary (REB17-1688), Waterloo (#31782) and Conestoga College (#CC256).

Data were derived from NHs, emergency departments, hospitals, and physicians' offices; see Appendix 1 for further details. Surveys were paper or email-based. Medical records were reviewed manually by study personnel. All data were entered into the University of Manitoba's Research Electronic Data Capture system and analysed using Stata 16 (StatCorp, College Station, TX). Resident inclusion criteria were: age ≥ 65 years, informed consent from competent residents or SDMs and presence of any of four risk factors that confer an elevated risk of dying within the next 6-12 months (Appendix 1). We excluded residents admitted for respite care or transferred from a different participating NH.

NHs were randomised after recruitment in a 1:1 ratio by the principal investigator to control and intervention groups using a random number generator and enrolled by the site investigators or research assistants. Control homes continued their usual ACP processes [20] , without alterations during the study. Intervention homes used BABEL ACP for residents who consented to participate and agreed to deliver the Full BABEL discussion (the core of the intervention) within 6 weeks of informed consent. The BABEL ACP intervention comprises [21] : (i) orientation and training for all NH staff, (ii) a workbook to guide ACP discussions (online materials), (iii) training tools for its use and (iv) stakeholder knowledge tools. The first three components support evidence-based best practices, and the third supports participants engaged in ACP. All participating NHs repeat ACP discussions yearly and when a resident's clinical status changes, intervention homes used the Full BABEL discussion for these repeat discussions in study participants. A cluster randomised trial was chosen to avoid application of BABEL components to control residents in homes with both control and intervention participants.

Screening for study eligibility utilised the Resident Assessment Instrument-Minimum Data Set (MDS) version 2.0 in each NH. MDS is a validated, widely used tool to characterise the needs and support care planning for NH residents [23, 24] . Repeated quarterly in all participating NHs, individuals who did not initially qualify were rescreened after subsequent MDS reassessments. In Alberta and Manitoba research, coordinators had direct access to MDS information on all screened residents, but due to privacy rules in Ontario, such information was only available for residents entered into the study. Screening data were maintained on a custom-designed electronic screening log.

Study personnel sought informed consent for eligible residents. SDMs consent was also sought for completion of surveys regarding ACP and their perceptions of the care provided to their loved ones.

Twelve individual-level outcomes were assessed, of which two were designated a priori as co-primary. The first (Supplementary Table 1 ) was the Audit of ACP, asking which of seven ACP elements had been discussed since NH admission; [25] the score for this instrument ranges from 0 to 7, with higher values indicating more comprehensive ACP. This survey was provided 6 weeks after study entry to capable residents, otherwise to the SDM. Analysis was by ordinal logistic regression. The second co-primary outcome (Supplementary Table 1 ) was Comfort Assessment in Dying [26, 27] , a questionnaire completed post-mortem by a nurse caring for the resident during the final week of life. The frequency of 14 symptoms was rated 1-3 as: a lot, somewhat or not at all, summed to a 14-42 point scale where higher values indicated fewer adverse symptoms. Analysis used ordinary least squares regression. Ten secondary outcomes, (Supplementary Table 2 ) included five measures of medical resource use. Qualitative methods were used to explore staff perspectives on BABEL ACP (Appendix 2).

Main analyses used mixed-effects regression modelling on complete records, at the level of individuals, with clustering within NHs, adjusting for baseline covariates [28] . Resident covariates were: age at study consent, sex, interval between NH admission and consent, marital status, pre-NH location, province, whether resident consent was obtained from the resident or SDM, presence of each of the four study inclusion factors indicating high risk (Appendix 1), whether the resident was receiving hospice care and six parameters obtained from the most recently completed MDS. These latter were: Cognitive Performance Scale [29] , Activities of Daily Living Scale [30] , Aggressive Behaviour Scale [31] , Depression Scale [32] , Pain Scale [23, 33] and a modified Charlson Comorbidity Index (CCI) [34] , as derived from the comorbid conditions available in the MDS. NH covariates were number of beds and public versus private ownership. As both co-primary outcomes were missing for ∼one-quarter of participants, sensitivity analyses used fixed effects regression with multiple imputation (10 full datasets) for missing dependent variables [35, 36] .

Main analysis was by intention-to-treat (ITT). As 21% of intervention home residents never received the Full BABEL discussion, we also performed per-protocol (PP) analyses of medical resource use outcomes, considering three groups of residents: (a) control, (b) intervention who never received the Full BABEL discussion and (c) intervention who received the Full BABEL discussion and only including resource use occurring after that discussion occurred.

We adjusted for multiple comparisons among all 12 outcomes in ITT analyses, using Simes' false discovery rate step-up procedure [37, 38] , considering two-sided, adjusted P-values <.05 to be statistically significant.

Sample size analysis utilised PASS 2008 (NCSS, LLC, Kaysville, Utah) for two independent means with a cluster randomised design applied to the Audit of ACP outcome. Assumptions were: mean scores of 5.0 in control homes and 5.6 in intervention homes, both with standard deviations of 1.5, 5% Type I error rate, 80% power, NHs of 150 beds with 43% of residents qualifying for the study, refusal rate of 25% among those eligible and an intraclass correlation coefficient of 0.20. This calculation indicated the need for 12 NHs in each group, for a total of 1,164 participants.

The study included 713 residents in 29 NHs (15 intervention, 14 control). Recruitment difficulties (Appendix 3) resulted in failure to reach the target sample size.

Due to later average study initiation of intervention homes (Table 1) , fewer intervention residents were screened ( Figure 1 ). With slightly lower rates of eligibility and consenting for intervention residents, there were 271 residents in the intervention group and 442 in the control group. Among intervention subjects, 57 (21.0%) never received a Full BABEL discussion; of the 214 who did, 153 (71.4%) received it within 8 weeks of consenting, and 12 (5.6%) received it >6 months after consenting. Physicians participated in only 40 (18.7%) of full BABEL discussions.

NH characteristics were similar between the two groups (Table 1) . However, there were relevant differences in subject characteristics (Table 1) , including control subjects spending an average of 29 days longer in the study than intervention residents in the ITT analysis; this grew to 115 days longer in the PP analysis.

Comprehensiveness of ACP was significantly higher in intervention participants (Table 2, Figure 2, Supplementary  Table 3 and 4). The overall odds of intervention home respondents rating ACP comprehensiveness as being better was 5.21-fold higher (95% confidence interval [CI] 3.53, 7.61) than for control home respondents; this remained consistent through sensitivity analyses (Supplementary Table 5 ). Both with and without adjustment, comfort assessment in dying was similar between decedents in the two groups ( Table 2, Supplementary Table 3 , 5 and 6).

Among medical resource use outcomes (Table 3, Supplementary Table 7) , after model-based and multiple comparison adjustments, the only significant one was lower antimicrobial use in intervention NHs (incidence rate ratio [IRR] = 0.79, P = 0.048). In PP analyses, not adjusted for multiple comparisons, intervention residents who received the Full BABEL discussion had lower rates of transfer to Table 7 ). After adjustment via mixedeffects, multivariable, grouped Poisson regression, IRR for death among intervention compared with control participants was 0.85 (95% CI 0.69, 1.03), with a P-value of 0.08 before further adjustment for multiple comparisons and 0.24 after further adjustment ( Table 2 ). The comparable results for entry into palliative care were IRR = 0.75 (0.54, 1.03) with model-based P-value of 0.08 before and 0.24 after further adjustment.

Seventy-five residents were competent and able to complete the Resident ACP Self-efficacy survey (Supplementary Table 2 ). Scores were similar between control and intervention residents (Supplementary Table 8 ).

Among the 272 residents who died, 217 SDMs completed post-mortem satisfaction surveys provided 1 month after death. Responses were similar between groups for both Satisfaction with Care at End of Life (Supplementary Table 9 ), and whether the care plan chosen in the NH was actually followed (Supplementary Table10).

Discordance between residents' care wishes and the care they received could only be assessed for the 214 residents who received the Full BABEL discussion, as it was the source of details about their wishes. Among the five types of care assessed, five individuals received care that they had not desired (Supplementary Table 11 ), of which two received it after their Full BABEL discussion-one who was transferred to hospital, and one who received antimicrobials in the NH.

Semi-structured interviews of NH staff showed that BABEL ACP improved their knowledge and confidence A. Garland et al. in ACP, leading to better discussions and more informed residents and SDMs (Appendix 4).

The BABEL approach to ACP resulted in: (1) more comprehensive ACP in NHs and (2) decreased use of antimicrobials. PP analysis (Table 3) was consistent with the Full BABEL discussion being the influential element of the intervention and may indicate inadequate sample size as the reason that other measures of medical resource group were not significantly lower in ITT analysis. Likely contributors to this latter include that the study was not powered for those outcomes, and the recruitment difficulties described. The qualitative portion of the study indicated that the BABEL intervention improved ACP discussions, NH staff knowledge and comfort in conducting discussions, and health record documentation of ACP decisions. The major barrier to its implementation was the time required to deliver the Full BABEL Discussion, generally estimated at ∼60 min.

Despite our expectation that the intervention would lead to fewer residents receiving aggressive, potentially uncomfortable care at the end of life, it did not increase the outcome of comfort assessment in dying, or rate of entry into formal palliative care. Similar null results using the same outcome scale were observed in a trial of a palliative care program in European NHs [39] . A possible contributor to this null finding was generally high comfort in dying of both groups (Supplementary Table 3 ); this suggests that problems with end-of-life care are not because NH personnel do not Supplementary Table 4. know how to provide it once a decision has been made to implement it, but because of trouble eliciting with residents when to provide it.

Our finding that the BABEL approach improved the comprehensiveness of ACP underscores the importance of allowing adequate time for these discussions; shortcuts that do not compromise the quality of ACP are unlikely to be found. This is consistent with pre-study observations by NH staff that time was a barrier to ACP, and with studies of outpatient care demonstrating that longer visits are associated with more shared decision-making [40, 41] , better communication [42, 43] , higher quality of care [44] and more attention to complex issues [42, 43, 45] . Although time is at a premium in NHs, is this time being utilised wisely? An anecdote from one of our control NHs is particularly informative (Mira Djokic, personal communication). A resident became acutely ill, was sent to hospital, stayed several days and died hours after NH repatriation. The time spent by NH personnel on such reactive care, making the resident ready for transport and resettled upon repatriation far exceeded 60 min for proactive planning required for a full BABEL discussion.

Conducting ACP 'upstream' from NH admission is attractive, but problematic. ACP requires trained personnel, but who would take on this role in the community? Average primary care outpatient visits of 20 min [46] are insufficient to conduct comprehensive ACP. Remuneration is a disincentive; in the United States, Medicare pays $170 for 60 min of ACP versus $370 for seeing four established patients for 15 min each [47] . Conducting ACP in homecare or assisted living settings is impeded by even lower staffing ratios than in NHs [48] ; also, half or less of NH admissions come from such settings [22, 49] . Thus, for comprehensive ACP to occur in NHs, adequate staffing with sufficient expertise and training is needed; likely we must reconfigure NH care processes.

Addressing growing nursing shortages in NHs [50, 51] and increasing physician presence and engagement in NHs [20, 52] are essential to support fulsome discussion of medical issues such as prognosis and treatment options that are essential to comprehensive ACP. Professional societies have defined core competencies for NH nurses [53] and physicians [54] ; however, actual competency is highly variable [52] , with few having geriatrics training [50, 55] . Katz et al. have noted the difficulty in convincing attending physicians to adhere to specific practice standards [56] . High quality NH care, including comprehensive ACP, requires considerably more on-site physician presence than is commonly provided [54, 57] . And it has been suggested to make NH medicine a specialty with required training, board examinations and closed-model, full-time, salaried, on-site positions [54, 56] .

Interventional studies of ACP in NHs are uncommon. A 2015 systematic review identified only 13 [58] , none of A. Garland et al. which assessed the breadth of relevant outcomes of our study, or adjusted their analyses for multiple comparisons. Hospitalisation rate was the most common outcome assessed, with reductions ranging 9-26%, compared with our 23% reduction in the adjusted incidence rate of transfer to emergency department or hospital. The two studies assessing mortality found that their ACP interventions did not alter mortality rates, in agreement with our finding. A more recent cluster randomised study in six Australian NHs [59] introduced a goal of care form incorporating a nuanced delineation of ACP wishes, a feature shared by BABEL ACP. This study reported a nonsignificant reduction in the incidence rate of transfer to emergency or hospital of 26%, similar to our 23% reduction. This similarity might indicate that a vital component of ACP is establishing the nuanced spectrum of options and outcomes, which is counter to practice and regulations in many jurisdictions that mandate goals of care designations with few gradations [60, 61] . Finally, a cluster randomised study in 28 American NHs found no effect of an intervention aimed at reducing antibiotic use in residents with severe dementia and suspected infection [62] . This negative result might reflect an intervention aimed at NH staff and limited to a single clinical issue, whereas the significant reduction in antibiotic use in BABEL may reflect it being an interactive intervention aimed at engaging residents and their proxies in a larger discussion about appropriate goals of care.

The main strength of this study is that it evaluated an ACP intervention that is more person-centred and comprehensive than any other of which we are aware, and that provides NH-tailored, evidence-informed, stakeholderbased resources. Second, we collaborated with NH stakeholders to design the intervention in a way that best practices were embedded into usual care processes. Other strengths are a wide-ranging set of both quantitative and qualitative person-centred outcomes, cluster randomisation and analysis that rigorously adjusted for multiple comparisons.

Our study also has limitations. The primary limitation was failure to reach the estimated sample size, which raises the probability of making Type II errors in inference. Second, as it was performed in 29 NHs in three Canadian provinces, additional studies are necessary to demonstrate generalisability of our approach, especially to other countries where ACP and end-of-life practices in NHs differ. Another limitation is related to the known problem of suboptimal adherence to prior ACP decisions during medical emergencies [18, 20] . Although aspects of BABEL ACP were designed to address this issue, our study did not assess such adherence, and thus we are unable to speak to the possibility that augmenting these features might improve the benefits of BABEL ACP. Also, as elaborated in Appendix 5, the study refinement that was part of the original study plan led to modification of outcomes after the original ethics submission, but before the study began.

Placed in the context of prior studies, our findings indicate that ACP needs to be recognised as a core expectation and competency of NH personnel. ACP is fundamental to NH care. We must rethink existing training, structures, policies and care processes in NHs to effectively provide the comprehensive ACP that this large, and growing, population needs.

Supplementary Data: Supplementary data mentioned in the text are available to subscribers in Age and Ageing online.

Declaration of Sources of Funding: Canadian Frailty Network (TG2015-15); Research Manitoba. The funders had no role in conception, design, data acquisition, data analysis, interpretation, drafting or approval of the manuscript.

Beyond the 'Iron Lungs of Gerontology': Using Evidence to Shape the Future of Nursing Homes in Canada

An international perspective on long term care: focus on nursing homes

Nursing Home Care. Centers for Disease Control

Prevalence of frailty in nursing homes: a systematic review and meta-analysis

United Nations

Sizing Up the Challenge: Meeting the Demand for Long-Term Care in Canada. Ottawa: The Conference Board of Canada

Percentage Distribution of Deaths, by Place of Death -United States

Factors associated with site of death: a national study of where people die

Doctors' learning experiences in end-of-life care -a focus group study from nursing homes

A census-based longitudinal study of variations in survival amongst residents of nursing and residential homes in Northern Ireland

Mortality and hospitalization at the end of life in newly admitted nursing home residents with and without dementia

Mortality in nursing home residents: a longitudinal study over three years

Defining advance care planning for adults: a consensus definition from a multidisciplinary Delphi panel

Mortality associated with COVID-19 outbreaks in care homes: early international evidence. London: international Long term care policy

What do Canadians think of advanced care planning? Findings from an online opinion poll

Understanding cardiopulmonary resuscitation decision making: perspectives of seriously ill hospitalized patients and family members

Provider perspectives on the influence of family on nursing home resident transfers to the emergency department: crises at the end of life

Applying the knowledge-to-action framework to engage stakeholders and solve shared challenges with person-centered advance care planning in long-term care homes

Problems with advance care planning processes and practices in nursing homes

Institute for Aging

CCRS Profile of Residents in Continuing Care Facilities

Instruments: an overview of the interRAI Suite

The development and validation of a questionnaire to audit advance care planning

Scales for evaluation of end-oflife care in dementia

The validity and reliability of scales for the evaluation of end-of-life care in advanced dementia

The Role of Covariable Adjustment in the Analysis of Clinical Trials

MDS cognitive performance scale

Scaling ADLs within the MDS

The aggressive behavior scale: a new scale to measure aggression based on the minimum data set

Development of a minimum data set-based depression rating scale for use in nursing homes

Minimum Data Set (MDS) 2.0 Canadian Version. inter-RAI

A new method of classifying prognostic comorbidity in longitudinal studies: development and validation

Missing data

Don't be Fancy. Impute Your Dependent Variables!

Frequentist q-values for multiple-test procedures

Adaptive linear stepup procedures that control the false discovery rate

Evaluation of a palliative care program for nursing homes in 7 countries: the PACE cluster-randomized clinical trial

Exploring the relationships between participatory decision-making, visit duration, and general practitioners' provision of argumentation to support their medical advice: results from a content analysis

Physicians' shared decision-making behaviors in depression care

Physician-patient communication in HIV disease: the importance of patient, physician, and visit characteristics

Duration of a well-child visit: association with content, family-centeredness, and satisfaction

Primary care visit duration and quality: does good care take longer?

Relationship of depression screening and physician office visit duration in a national sample

National Trends in primary care visit use and practice capabilities

Long-term care providers and services users in the United States

Minimum Data Set (2.0) Frequency Report: Calendary Year Trends

The Long-Term Care Workforce: Can the Crisis be Fixed?

Who Will Care For Us? Long-Term Care and the Long-Term Workforce: Long-Term Care and the Long-Term Workforce

Physician visits at the nursing home: does more mean better care?

Nurse competencies for person-directed care in nursing homes

Medical care delivery in US nursing homes: current and future practice

Retooling the RN workforce in long-term care: nursing certification as a pathway to quality improvement

An international perspective on improving physician care in nursing homes

The nursing home physician workforce

The effects of advance care planning interventions on nursing home residents: a systematic review

Introducing goals of patient Care in Residential Aged Care Facilities to decrease hospitalization: a cluster randomized controlled trial

Advance Care Planning: Have a Say in Your Health Care. Winnipeg: Winnipeg Regional Health Authority

The trial to reduce antimicrobial use in nursing home residents with Alzheimer disease and other dementias (TRAIN-AD): a cluster randomized clinical trial