Effect of a Video-assisted Advance Care Planning Intervention on End-of-Life Health Care Transitions among Long-Stay Nursing Home Residents

Abstract

Objective:

To determine the relationship between an advance care planning (ACP) video intervention, Pragmatic Trial of Video Education in Nursing Homes (PROVEN), and end-of-life health care transitions among long-stay nursing home residents with advanced illness.

Design:

Pragmatic cluster randomized clinical trial. Five ACP videos were available on tablets or online at intervention facilities. PROVEN champions employed by nursing homes (usually social workers) were directed to offer residents (or their proxies) ≥1 video under certain circumstances. Control facilities employed usual ACP practices.

Setting and participants:

PROVEN occurred from February 2016 to May 2019 in 360 nursing homes (119 intervention, 241 control) owned by two healthcare systems. This post hoc study of PROVEN data analyzed long-stay residents ≥65 years who died during the trial who had either advanced dementia or cardiopulmonary disease (advanced illness). We required an observation time ≥90 days before death. The analytic sample included 923 and 1,925 advanced illness decedents in intervention and control arms; respectively.

Methods:

Outcomes included the proportion of residents with 1 or more hospital transfer (i.e., hospitalization, emergency room use or observation stay), multiple (≥3) hospital transfers during the last 90 days of life, and late transitions (i.e., hospital transfer during the last 3 days or hospice admission on the last day of life).

Results:

Hospital transfers in the last 90 days of life among decedents with advanced illness were significantly lower in the intervention versus control arm, (proportion difference =−1.7%, 95% CI −3.2% to −0.1%). The proportion of decedents with multiple hospital transfers and late transitions did not differ between the trial arms.

Conclusions and implications:

Video-assisted ACP was modestly associated with reduced hospital transfers in the last 90 days of life among nursing home residents with advanced illness. The intervention was not significantly associated with late health care transitions and multiple hospital transfers.

Brief summary:

Video-assisted advance care planning significantly reduced the prevalence of hospital transfers in the last 90 days of life. Multiple hospital transfers in the last 90 days and transitions within 3 days of death were not affected.

Introduction

Health care transitions in the last months of life can fragment care, burden patients and their families, and may confer limited clinical benefit. Nursing home (NH) residents, who experience terminal illness trajectories characterized by multiple cognitive and functional limitations are particularly susceptible to receiving aggressive interventions (e.g., hospitalizations) that may be inconsistent with their preferences.^1,2 Prior research suggests that transfers in the last 3 days of life, and multiple hospitalizations in the last 90 days of life are potentially burdensome care among terminally-ill NH residents.^3,4 Such transitions are common among NH decedents and have been associated with unmet needs and poor quality in end-of-life care.^3,5 In one study, nearly 1 in 5 decedents with advanced cognitive impairment, and up to 37.5% in some states, had a burdensome transition.³

Advance care planning (ACP) enables individuals to state their future preferences for medical treatment.⁶ Consistent ACP and creating a plan to honor those wishes could improve care quality by ensuring health care transitions align with patient goals and care preferences. ACP may include formulating advance directives such as designating a health care proxy, medical orders for life-sustaining treatment, and preparation for hospice care.⁷ ACP has been associated with better palliative care outcomes in NHs,^8–10 and has the potential to clarify patient preferences to avoid unnecessary health care transitions. However, in usual practice, ACP is often inadequate and advance directives are commonly poorly documented and not routinely ascertained.¹¹

Video ACP tools have been developed and tested for efficacy in several small randomized clinical trials (RCTs) in an effort to address inconsistent ACP counseling and improve patient-provider communication around goals of care.^12–17 While video ACP support tools were introduced to clinical care as early as 2012,¹⁸ evaluations of their real-world effectiveness are limited.¹⁹ The Pragmatic trial of Video Education in Nursing Homes (PROVEN), a pragmatic cluster randomized clinical trial, was designed and conducted to evaluate the effectiveness of ACP videos in practice. The main results indicated that an ACP video intervention did not significantly affect primary outcomes of hospital transfers, burdensome treatment use, or hospice enrollment among long or short stay NH residents.^19,20 Nonetheless, it is possible that the intervention impacts end-of-life care, a period over which aggressive treatments may be inconsistent with patient goals and care preferences. The objective of this analysis is to determine whether ACP videos were associated with reduction in potentially burdensome health care transitions at end-of-life among long-stay NH residents with advanced illness who died.

Methods

The Institutional Review Board approved the pragmatic clinical trial and waived informed consent. The study was registered in ClinicalTrials.gov (NCT02612688). The trial protocol design was previously published.¹⁹ This report is a post hoc analysis of the subgroup who died.

Nursing home randomization

PROVEN was conducted between February 2016 and May 2019 in 360 NHs owned by 2 for-profit NH chains across 32 states. To be eligible for randomization NHs were required to have >50 beds and admit both post-acute and long-term residents. NHs which corporate leaders deemed to have serious organizational problems or inability to transfer electronic health records (EHR) were excluded from randomization. NHs were stratified by chain and then by the distribution of the primary outcome (i.e., hospitalizations per 1000 person-days alive among residents with advanced illness). NHs in each stratum were randomized into the intervention and control groups in a 1:2 ratio.

Data sources

We used Medicare claims, Minimum Data Set (MDS) assessments, and EHR data. We obtained baseline resident characteristics (e.g., demographics, cognition, physical function) from the MDS assessment closest to the start of the PROVEN trial implementation period or based on the first assessment in which residents met criteria for advanced illness. Additionally, we used MDS items linked with Medicare claims to capture outcome variables defined below. We ascertained hospice enrollment from Medicare claims.

Study sample

In this analysis we included NH residents who were: 1) ≥65 years of age, 2) died during the trial period, 3) were long stay (>100 days), 4) enrolled in Medicare fee-for-service, and 5) had ≥1 quarterly or comprehensive MDS assessment within 90 days before death. We included participants with advanced illness, defined as having either advanced dementia or cardiopulmonary disease (chronic obstructive pulmonary disease [COPD] or congestive heart failure) and functional impairment based on MDS assessments.¹⁹ Advanced dementia was defined by the presence of Alzheimer’s disease or other dementia, advanced cognitive impairment (cognitive function scale score of 3 or 4) and requiring extensive or total assistance for eating and transferring.

Intervention

The intervention consisted of five previously created videos (6 to 10 minutes long) offered in English or Spanish. Video topics covered were: 1) general goals of care, 2) goals of care for advanced dementia, 3) hospice, 4) hospitalization, and 5) ACP for healthy patients.^17,21–23 Each NH identified 1-2 ACP video program champions, typically social workers, who were tasked with offering and showing the videos to residents or their proxies. Program champions received training prior to participant enrollment, and were instructed to offer videos 1) within 7 days of admission or readmission, 2) every 6 months for long-stay residents, 3) when decision-making arose on a topic for which there was a specific video, 4) upon a significant change in clinical status, and 5) special circumstances when goals of care are considered (e.g., family visiting). Champions were asked to record when a video was offered and whether the resident or proxy watched the video. More information about the intervention is available elsewhere.^19,20 Control facilities conducted their usual ACP practices.

Outcomes

We descriptively assessed burdensome treatment use defined as experiencing any of tube feeding, parenteral therapy (for medication delivery or hydration), admission to an intensive care unit, and invasive mechanical ventilation.¹⁹ We considered diverse measures of health care transitions including those relevant at end-of-life.^3–5 We analyzed the proportion of residents with any hospital transfers in the last 90 days of life. A hospital transfer included admissions, emergency room visits without admission and observation stays in the last 90 days. We also examined the proportion of decedents with multiple (≥3) hospital transfers in the last 90 days of life. Additionally, we analyzed the proportion with “late” transitions defined as either a hospital transfer in the last 3 days of life or hospice admission in the last calendar day before death.⁵

Statistical analysis

We described baseline characteristics using proportions for categorical variables and means and standard deviations (SDs) for continuous variables. Using hierarchical logistic regression models accounting for facility-level clustering, we estimated the marginal differences in proportions and 95% confidence intervals (CIs). We used a two-sided test of the difference in proportions to examine the null hypothesis. Standard errors (SEs) were calculated via bootstrapping.²⁴

Results

Of 360 NHs, 119 were intervention and 241 were control facilities. The CONSORT diagram of NHs assessed for eligibility and those included in the study has been published previously.¹⁹ We analyzed a sample of 2,848 decedents among long-stay residents with advanced illness (Table 1). There were 923 and 1,925 decedents in the intervention and control arms, respectively. The distributions of baseline demographic and clinical characteristics were similar for decedents in the intervention and control arms. The mean follow-up time prior to death was 214 days in both arms.

Table 1:

Characteristics of long-stay nursing home decedents with advanced illness

Characteristics, n (%)	Intervention N=923	Control N=1925	P-value*
Baseline characteristics
Age in years, mean (SD)	85.4 (8.6)	85.2 (8.5)	0.66
Female	640 (69.3)	1316 (68.4)	0.60
Race/ethnicity			0.24
Non-Hispanic White	778 (84.4)	1625 (84.4)
Non-Hispanic Black	116 (12.6)	247 (12.8)
Other	29 (3.1)	53 (2.7)
Advanced dementia	661 (71.6)	1393 (72.4)	0.68
Advanced CHF/COPD	304 (32.9)	597 (31.0)	0.30
CFS score, mean (SD)	3.2 (0.8)	3.2 (0.8)	0.80
ADL score, mean (SD)	22.3 (3.9)	22.5 (4.0)	0.31
Follow-up characteristics
Follow-up time in days, mean (SD)	214 (80)	214 (78)	0.97
Hospice care in the last 90 days of life	283 (30.7)	597 (31.0)	0.85
Potentially burdensome treatment in the last 90 days of life
≥1 burdensome treatment use	200 (21.7)	494 (25.7)	0.02
Intensive care unit admission	136 (14.7)	351 (18.2)	0.02
Tube feeding	58 (6.3)	156 (8.1)	0.09
Invasive mechanical ventilation	n<11	17 (0.88)	0.52
Parenteral therapy	128 (13.9)	265 (13.8)	0.94
Intervention fidelity
Offered video	472 (51.1)	--
Shown video, among those offered**	115 (24.4)	--

CFS, cognitive function scale where higher values indicate higher impairment; ADL, activities of daily living where higher scale values indicate greater functional limitations and more extensive dependency

^*Chi-square tests (categorical variables) and t-tests (continuous variables) were used to calculate P-values

^**Using a denominator of n=472 decedents who were offered the opportunity to watch an advance care planning video

During follow-up, almost one-third of decedents enrolled in hospice in the last 90 days of life in the intervention (30.7%) and control (31.0%) arms (Table 1). Receipt of 1 or more potentially burdensome treatments was less common in the intervention compared with the control arm (21.7% vs. 25.7%, P=0.02) largely due to differences in intensive care unit (ICU) use (intervention:14.7% vs. control:18.2%). Among decedents, 3.8% (intervention) and 5.7% (control) experienced any hospital transfer and 1.8% (intervention) and 3.0% (control) had multiple hospital transfers within the last 90 days of life (Figure 1). The proportion of decedents with a hospital transfer within the last 3 days of life was 4.6% in the intervention and 5.9% in the control arm. Transfers to hospice in the last day of life occurred among 3.9% and 4.5% decedents in intervention and control arms; respectively.

Figure 1:

Prevalence of health care transitions and treatment use within the last 90 days of life among long-stay nursing home residents

* Late transition includes transfer to hospice in the last day of life or hospitalization/emergency room use/observation stay in the last 3 days

** Burdensome treatments include tube feeding, parenteral therapy (for medication delivery or hydration), admission to an intensive care unit, and invasive mechanical ventilation

In regression models we found a statistically significant reduction in the prevalence of 90-day hospital transfers among decedents with advanced illness (Table 2; marginal proportion differences [MPD]: −1.7%, 95% CI −3.2% to −0.1%). Multiple hospital transfers in the last 90 days of life did not significantly differ (MPD: −0.8%, 95% CI −1.7% to 0.1%) between trial arms. There were no significant differences in hospital transfers in the last 3 days (MPD: −1.1%, 95% CI −3.5% to 1.0%) or hospice transfer in the last day of life (MPD: −0.6%, 95% CI −1.8% to 1.0%).

Table 2:

Health care transitions in the last 90 days of life among long-stay nursing home residents with advanced illness

Outcomes	Proportion Difference, % (95% CI)
Any hospital transfer in the last 90 days life	−1.71 (−3.21, −0.09)
≥3 hospital transfers in the last 90 days of life	−0.83 (−1.71, 0.14)
≥1 late transition*	−2.22 (−5.29, 1.26)
Acute hospitalization in the last 3 days of life	−1.09 (−3.48, 1.03)
Transfer to hospice in the last day of life	−0.59 (−1.78, 1.00)

Bold indicates a statistically significant difference at alpha = 0.05

^*hospitalization/emergency room use/observation stay in the last 3 days or transfer to hospice in the last day of life

Discussion

This study found that an ACP video intervention was associated with a modest but significant reduction in the proportion of hospital transfers in the last 90 days of life among NH residents with advanced illness. The intervention was not significantly associated with health care transitions in the last 3 days or multiple hospital transfers in the last 90 days before death.

ACP is consistently rated by patients, families, and clinicians as important, especially by those who have experienced decision-making for serious illness.²⁵ Prior research has shown mixed findings of the impact of ACP; however, studies show an overall benefit for patient, family, and clinician outcomes.²⁵ Heterogenous trial designs, settings, disease states, and outcomes likely contribute to the mixed findings. Although ACP has been identified as particularly relevant for NH residents, there are knowledge gaps regarding circumstances under which ACP is effective and how best ACP can be implemented in NH settings.²

In conducting this post hoc analysis in a decedent subgroup, we hypothesized the intervention would reduce health care transitions among those who are terminally ill. Our findings indicate little to no impact of the ACP video intervention on health care transitions at end-of-life among NH residents with advanced illness. However, the descriptive results on burdensome treatment use indicate the intervention was significantly associated with ICU admissions. This result provides a signal that ICU utilization could be an outcome that is sensitive to ACP interventions among NH residents with advanced illness. Further exploration is needed.

The present findings combined with earlier results which found the PROVEN intervention ineffective at reducing hospital transfers among (long-stay or short-stay) NH residents in the PROVEN trial raise questions about the degree to which factors outside the intervention could explain nonsignificant findings.^19,20 Only one-quarter (24.4%) of decedents with advanced illness in the intervention group were shown an ACP video among those offered the opportunity to watch a video. This low intervention fidelity underscores the difficulty of implementing pragmatic RCTs in NHs. Embedding large scale trials into the real world needs to address workflow integration, highly variable quality of care, limited resources, and competing priorities.^26,27 Important opportunities remain to improve ACP uptake at end-of-life among NH residents, and the utility of video decision support and other communication enhancing strategies deserve continued evaluation.^28,29

This study has limitations. First, this analysis was not defined as part of primary or secondary outcomes of the PROVEN trial. Our analysis could be underpowered and differences may exist in the decedent cohort that are not protected by randomization if mortality is influenced by treatment arms within the cohort. Second, there was a secular trend of decreased hospital transfer rates across nursing in the U.S. during the period when PROVEN was conducted.³⁰ However, secular changes outside the trial should similarly affect the intervention and control arms. Third, we lacked context about resident or family preferences regarding whether or not to forego hospital transfers or other transitions at end-of-life. Advance directives were not included because they are not collected in the MDS 3.0 and were not systematically documented in EHRs of all NHs participating in the PROVEN trial.

Conclusions and Implications

Although video-assisted ACP was not associated with late transitions and multiple hospital transfers, it was modestly associated with reduction in the prevalence of hospital transfers in a long-stay decedent cohort with advanced illness. Given time and resource constraints in NHs, this analysis suggests that targeting long-stay residents with advanced illness could be a promising strategy to reduce potentially burdensome care at end-of-life.