The Role of Regional and State Initiatives in Nursing Home Advance Care Planning Policies

Abstract

Objective:

Antibiotic use at the end of life (EoL) may introduce physiological as well as psychological stress and be incongruent with patients’ goals of care. Advance care planning (ACP) related to antibiotic use at the EoL helps improve goal-concordant care. Many nursing home (NH) residents are seriously ill. Therefore, we aimed to examine whether state and regional ACP initiatives play a role in the presence of “do not administer antibiotics” orders for NH residents at the EoL.

Methods:

We surveyed a random, representative national sample of 810 U.S. NHs (weighted n=13,983). The survey included items on “do not administer antibiotics” orders in place and participation in infection prevention collaboratives. The survey was linked to state Physician Orders for Life-Sustaining Treatment (POLST) adoption status and resident, facility, and county characteristics data. We conducted multivariable regression models with state fixed effects, stratified by state POLST designation.

Results:

NHs in mature POLST states reported higher rates of “do not administer antibiotics” orders compared to developing POLST states (10.1% vs. 4.6%, respectively, p=0.004). In mature POLST states, participation in regional collaboratives and smaller NH facilities (<100 beds) were associated with having “do not administer antibiotics” orders for seriously ill residents (β = 0.11, p=0.006 and □ = 0.12, p=0.003, respectively).

Discussion:

NHs in states with mature POLST adoption that participated in infection control collaboratives were more likely to have “do not administer antibiotics” orders. State ACP initiatives combined with regional antibiotic stewardship initiatives may improve inappropriate antibiotic use at the EoL for NH residents.

BACKGROUND

Advance care planning (ACP), a process in which patients’ wishes, goals and priorities related to healthcare services are elicited and documented, is an important tool to help achieve goal-concordant care.¹ ACP is particularly relevant for the approximately 1.3 million Americans living in nursing homes (NHs) annually, as eighty-five percent of these long-stay residents are over the age of 65, and many are seriously ill. ^2-5

The use of ACP has been shown to improve quality of life in seriously ill adults by aiding end-of-life (EoL) decision-making.⁶ ACP facilitates palliative care, which focuses on symptom relief and aims to improve quality of life for individuals with serious illness.⁷

Suspected infections are common in NH residents and are often treated with antibiotics, although evidence of the benefits of antibiotic treatment in residents nearing EoL is inconclusive,^8,9 and older, frail adults, such as NH residents, are at higher risk of adverse events.⁵ In a study of NH residents with advanced dementia, only 44% of residents who were treated with antibiotics met minimum clinical criteria for initiating treatment.⁹ The inappropriate use of antibiotics increases the risks of multidrug-resistant organisms and C. difficile.^9-11 Researchers have found that patients without specific “do not administer antibiotics” ACP orders have increased incidence of resistant bacteria and high antibiotic usage by time of death.^10,12 Importantly, for NH residents at the EoL, administering antibiotics may not align with their goals of care, as these treatments may be burdensome and increase stress for residents and their loved ones.^13,14 However, a national survey found that few NHs had formalized ACP orders, and even fewer had “do not administer antibiotics” orders.¹⁵ Integrating palliative care and infection management is a key strategy to reduce risks associated with inappropriate use of antibiotics at the EoL.^5,16,17 This integration includes NH antibiotic stewardship, aimed at improving the use of antibiotics through mindful prescription in EoL care that is aligned with the resident’s goals of care.

Improving patient-centered, goal-concordant antibiotic use in NH residents at EoL requires a shift to a culture change model of care and strong antibiotic stewardship programs, both of which are associated with national and local policies.^18-21 The Physician Orders for Life-Sustaining Treatment (POLST) is a state-driven tool that facilitates ACP discussions, elicits patient preferences, and formalizes them in order to deliver care that reflects individuals’ goals.⁵ The National Academy of Medicine has recommended nationwide implementation of POLST, particularly for frail individuals with serious illness.⁵ POLST usage has resulted in high consistency between treatment and patient preferences for resuscitation, hospitalization, and antibiotics administration,^22,23,24 POLST has resulted in state variation in content and rates of adoption.^25,26 For example, out of 51 POLST state forms, only 32 were found to have asked about antibiotic preferences.²⁷

To strengthen overall healthcare quality including infection control programs in NHs, the Centers for Medicare & Medicaid Services (CMS) funded 14 regional collaboratives known as Quality Innovation Network-Quality Improvement Organizations (QIN-QIOs).^28,29 QIN-QIOs provide various resources to NHs,³⁰ including online training modules and help in tracking healthcare-associated infections through enrollment and participation in the National Healthcare Safety Network (NHSN).^31-34 In 2017, over 78% of NHs participated in QIN-QIO regional collaboratives.³⁵ Notably, participation in QIN-QIO initiatives has been associated with more comprehensive NH antibiotic stewardship programs.³⁶ Also, collaboration at the federal and state levels increased NH ACP use when California NHs began collecting information about POLST in the CMS Long-Term Care Minimum Data Set (MDS), which is the federally mandated clinical assessment process for all CMS-certified NHs.^37,38

This study aimed to examine if state and regional level initiatives impacted ACP policies at the NH level. More specifically, we analyzed whether the policy environment created by the adoption of state POLST programs and facility participation in regional collaboratives had downstream effects on NH ACP policies regarding “do not administer antibiotics” orders.

METHODS

Survey Sample

In 2017-2018, a national survey of U.S. NHs was conducted and completed by NH Directors of Nursing. The sample included nonspecialized, free-standing NHs with ≥30 beds, and a Certification and Survey Provider Enhanced Report (CASPER) assessment dated 2014 or later,³⁹ stratified by QIN-QIO region, NHSN enrollment,³³ and participation in our previous study.¹⁵ Probability weights were computed based on sampling strata and nonresponse predictors. Details of the survey process have been published elsewhere.¹⁵ Authors obtained approval of Columbia University Medical Center Institutional Review Board prior to beginning this study.

Data Sources and Measures

Survey data were linked to state-level POLST data.²⁷ State POLST status was categorized by POLST program designation, as nonconforming or developing and mature or endorsed; the classifications used were based on information dated April 2018, contemporaneous with our survey data collection.^27,40,41 Although there are other methods to categorize POLST status use, POLST program designation focuses on national policies and governance with implementation, which is aligned with our analysis of policy initiatives. Mature or endorsed adoption state status indicates that POLST is a part of standard care. Nonconforming or developing status indicates that either the state POLST program was developed but failed to comply with requirements or that the state has not taken any action regarding POLST implementation.⁴⁰ Table 1 provides an overview of POLST program designation by state. The survey data were also linked to administrative datasets, including: facility-level measures from CASPER data including NH quality;³⁹ MDS resident-level assessments from 2017 (quarterly or annual assessments for long-term residents 65 and older);³⁸ and county-level data from the Area Health Resources Files (AHRF).⁴²

Table 1.

POLST Program Designation by State

POLST Program Designation	N	States (initials)
Mature	3	CA, OR, WV
Endorsed	20	CO, GA, HI, ID, IN, IA, KS, LA, ME, MO, MT, NH, NY, NC, PA, TN, UT, VA, WA, WI
Developing	24	AL, AK, AZ, AR, CT, DE, FL, IL, KY, MI, MN, MS, NV, NJ, NM, ND, OH, OK, RI, SC, SD, TX, WY, WASHINGTON DC
Nonconforming	4	MD, MA, NE, VT

The ACP policy for antibiotic use was measured by the likelihood of “do not administer antibiotics” orders for a hypothetical patient with Alzheimer’s near the EoL with a suspected infection. Respondents indicated the likelihood that a “do not administer antibiotics” order would be in place on a 3-point Likert scale (not at all likely, somewhat likely and very likely), which we collapsed to “very likely” or “not”. We inquired about participation in regional QIN-QIO infection prevention initiatives. Other survey facility characteristics included staff turnover in the previous 3 years of key personnel (infection preventionists, directors of nursing, and administrators) and on-site access to an advanced practice registered nurse.

The CASPER facility measures included: the proportion of hospice use, calculated as the resident hospice count divided by the total resident count, then categorized by quartiles based on frequency distribution; facility size; percent occupancy; type of ownership (for-profit, non-profit, or government); payer status (Medicare/Medicaid or other); membership in a multi-facility organization; and location in a rural area. The CMS five-star quality rating was categorized by quartiles based on frequency distribution.

Resident characteristics from MDS data were calculated as facility proportions by calendar quarter, averaged across the year and normalized by their standard deviations. Characteristics included demographics (i.e., sex and race), functional status, and select comorbidities. Functional status measures included proportions of residents’ dependent for their activities of daily living (bed mobility, transferring, locomotion, dressing, eating, toileting, and hygiene). Comorbidities included the proportion with active Alzheimer’s and other dementias diagnoses, the proportion with severe cognitive impairment based on Cognitive Function Scale scores,⁴³ and the proportion with cardiovascular advanced illness.⁴⁴ In addition, AHRF data were used to extract county-level characteristics including the percent of the population that was 65 and older, the percent of persons 65 and older below poverty, the percent of the population in NHs, and the number of hospices in the county.

Analyses

We compared “do not administer antibiotic” orders and facility, resident, and county characteristics of NHs in developing or nonconforming POLST states to those in mature or endorsed states, using χ² and one-way analysis of variance (ANOVA) tests, as appropriate. Multivariable linear regression models, stratified by POLST status, were developed to evaluate the association between NH facility characteristics with the likelihood of EoL patients having “do not administer antibiotics” orders, controlling for resident population and county characteristics. Models also included state fixed effects to adjust for state-level variation.

RESULTS

A total of 892 facilities completed surveys, a 49% response rate. Of those, 810, representing 13,983 NHs nationally, had complete data for the variables of interest and were included in the regression analyses. Just over half, 53.9%, were located in states with a developing or nonconforming POLST, and 46.1% were in states designated as having a mature or endorsed POLST.

Table 2 shows NH characteristics by their state POLST designation. Facilities in mature or endorsed states had higher rates of being very likely to have “do not administer antibiotics” orders (10.1% versus 1.6%, p = 0.004). NHs in mature or endorsed POLST states differed from developing/nonconforming POLST states by the following characteristics: smaller-sized NHs (51.7% versus 41.2%); a lower number of non-profits (20.1% versus 27.6%); lower proportions of Hispanic residents (3.7% versus 4.4%); higher proportions of residents dependent for transferring (18.6% versus 16.6%), locomotion (23.6% versus 20.1%), eating (7.9% versus 6.5%) and toileting (16.4% versus 14.3%); lower proportions of residents with active dementia diagnoses (39.7% versus 42.7%) and cardiovascular advanced illness (4.7% versus 6.1%); and more hospices per county (13.6% versus 6.1%).

Table 2:

Nursing home facility, resident and county characteristics by POLST maturity level

	All	Developing, Nonconforming POLST	Mature, Endorsed POLST
	n=13983	n=7538	n=6445
	% (SE)			P
Do not administer antibiotics orders	7.1 (0.93)	4.6 (1.10)	10.1 (1.54)	0.004
Survey Facility Characteristics
Participated in QIN-QIO infection prevention initiatives	35.6 (1.83)	36.1 (2.54)	35.0 (2.63)	0.8
APRN on-site	63.3 (1.89)	65.6 (2.59)	60.7 (2.74)	0.2
Staff turnover in last 3 years	Mean (SE)			P
Infection preventionist	2.0 (0.05)	2.0 (0.07)	2.0 (0.06)	0.9
Director of nursing	2.2 (0.05)	2.2 (0.08)	2.1 (0.07)	0.2
Administrator	1.9 (0.06)	1.9 (0.06)	1.9 (0.09)	1.0
General Facility Characteristics	% (SE)			P
Small facility (30-99 beds)	46.0 (1.94)	41.2 (2.65)	51.7 (2.78)	0.007
Ownership type
Government	6.4 (0.86)	5.2 (1.05)	7.8 (1.40)	0.1
Non-profit	24.2 (1.56)	27.6 (2.30)	20.1 (2.01)	0.014
For-profit	69.4 (1.70)	67.2 (2.43)	72.1 (2.33)	0.1
Part of multi-facility (chain) organization	57.0 (1.92)	54.1 (2.70)	60.4 (2.71)	0.1
Located in a rural area	28.6 (1.72)	27.4 (2.39)	29.9 (2.48)	0.5
	Mean (SE)			P
Percent occupancy	80.0 (0.62)	79.6 (0.91)	80.5 (0.81)	0.4
Percent Medicare as primary payer	13.6 (0.50)	14.0 (0.74)	13.2 (0.65)	0.4
Percent Medicaid as primary payer	58.7 (0.86)	58.4 (1.21)	59.0 (1.22)	0.7
Percent other primary payer	27.7 (0.73)	27.7 (0.94)	27.8 (1.13)	1.0
Resident Characteristics	% (SD)			P
Demographics
Female	63.0 (0.39)	63.4 (0.54)	62.5 (0.56)	0.2
White	83.5 (0.84)	83.1 (1.18)	84.0 (1.19)	0.5
African American	8.1 (0.55)	8.4 (0.75)	7.7 (0.81)	0.5
Hispanic	3.7 (0.44)	4.4 (0.73)	3.0 (0.42)	0.044
Other (Alaskan native/Am Indian, Asian, native Hawaiian/Pac Islander, multirace)	4.7 (0.41)	4.1 (0.55)	5.3 (0.63)	0.1
Functional status	Mean (SD)			P
Bed mobility dependent	7.7 (38.96)	7.3 (37.00)	8.1 (40.78)	0.2
Transfer dependent	17.5 (47.70)	16.6 (45.25)	18.6 (49.71)	0.013
Locomotion dependent	21.7 (56.34)	20.1 (51.26)	23.6 (60.14)	<.001
Dressing dependent	10.6 (49.15)	10.0 (45.45)	11.4 (52.44)	0.1
Eating dependent	7.2 (36.69)	6.5 (24.48)	7.9 (45.51)	0.018
Toileting dependent	15.3 (61.68)	14.3 (56.35)	16.4 (66.31)	0.038
Hygiene dependent	11.9 (50.62)	11.1 (48.46)	12.7 (52.54)	0.1
Comorbidities
Alzheimer’s active diagnosis	9.3 (40.91)	9.2 (42.38)	9.4 (39.44)	0.7
Other dementias active diagnosis	41.3 (55.39)	42.7 (57.27)	39.6 (52.74)	<.001
Severe cognitive impairment (CFS=4)	12.4 (37.20)	12.3 (39.58)	12.5 (34.73)	0.9
Cardiovascular advanced illness	5.5 (32.67)	6.1 (35.87)	4.7 (28.87)	0.008
County Characteristics	Mean (SE)			P
Percent population 65+, 2016	16.7 (0.17)	16.9 (0.26)	16.4 (0.21)	0.1
Percent persons 65+ below poverty, 2012-16	1.4 (0.02)	1.4 (0.03)	1.4 (0.04)	0.9
Percent population in nursing facility, 2010	0.7 (0.01)	0.7 (0.02)	0.6 (0.02)	0.5
Number of hospices, 2016	10.0 (1.78)	7.0 (0.80)	13.6 (3.72)	0.019

Note: All frequencies and percentages weighted.

APRN, advanced practice registered nurse; CFS, Cognitive Function Scale; EOL, end-of-life; QIN-QIO, Quality Innovation Network-Quality Improvement Organization.

Table 3 shows the stratified multivariable analyses. In NHs in mature/endorsed states, independent predictors of very likely to have a “do not administer antibiotics” orders included: participating in QIN-QIO infection initiatives (β = 0.11, p = 0.006); and, smaller sized facilities (β = 0.12, p = 0.003). There were no independent predictors for NHs in developing/non-conforming states.

Table 3:

Multivariable regression estimates of predictors of “do not administer antibiotics” orders in nursing homes, stratified by POLST maturity level

	Developing, Nonconforming POLST		Mature, Endorsed POLST
Facility Characteristics	Estimate (SE)	P	Estimate (SE)	P
Participated in QIN-QIO infection prevention initiatives	0.02 (0.20)	0.8	0.11 (0.04)	0.006
APRN on-site	0.03 (0.30)	0.8	0.03 (0.04)	0.5
Staff turnover in last 3 years
Infection preventionist	0.01 (−0.04)	1.0	0.00 (0.02)	0.8
Director of nursing	0.01 (0.09)	0.9	0.00 (0.02)	0.8
Administrator	0.01 (−1.03)	0.3	0.00 (0.01)	0.8
Proportion of hospice use by quartiles
Quartile 1	reference		reference
Quartile 2	0.02 (−0.06)	0.9	−0.01 (0.05)	0.9
Quartile 3	0.04 (−0.15)	0.9	−0.06 (0.05)	0.2
Quartile 4	0.03 (0.74)	0.5	−0.07 (0.05)	0.2
CMS Five-Star Quality Rating by quartiles
Quartile 1	reference		reference
Quartile 2	0.02 (−0.40)	0.7	−0.02 (0.05)	0.7
Quartile 3	0.04 (−0.09)	0.9	−0.06 (0.05)	0.2
Quartile 4	0.04 (0.70)	0.5	−0.06 (0.05)	0.2
Small facility (30-99 beds)	0.02 (−0.23)	0.8	0.12 (0.04)	0.003
Ownership type
Government	0.04 (−0.28)	0.8	0.07 (0.07)	0.3
Non-profit	0.02 (−0.30)	0.8	0.00 (0.04)	1.0
For-profit	reference		reference
Part of multi-facility (chain) organization	0.02 (1.15)	0.2	0.01 (0.04)	0.7
Located in a rural area	0.03 (−1.21)	0.2	0.03 (0.04)	0.5
Percent occupancy	0.00 (0.30)	0.8	0.00 (0.00)	0.6
Insurance Type
Percent Medicare as primary payer	0.00 (0.62)	0.5	0.00 (0.00)	0.9
Percent Medicaid as primary payer	0.00 (0.58)	0.6	0.00 (0.00)	0.4
Percent other primary payer	reference		reference

Note: Adjusted for resident and county characteristics with state fixed effects. Abbrev: QIN-QIO, Quality Innovation Network-Quality Improvement Organization; APRN, advanced practice registered nurse; SE, standard error.

DISCUSSION

In this study, after controlling for resident, facility, and county characteristics, we found that NHs participating in regional QIN-QIO infection prevention initiatives and located in states with mature or endorsed POLST adoption status were more likely to have “do not administer antibiotics” orders in place for EoL residents. These findings suggest that state and regional level policy initiatives, such as POLST and the QIN-QIO shared-learning infection prevention collaboratives, have a positive impact on NH ACP antibiotic use.

Our results suggest a need to support efforts encouraging antibiotic stewardship in NHs. In an analysis of palliative care policies in our national NH survey, we found that on average, NH were least likely (63.24% of the time) to have “do not administer antibiotics” orders in place for residents, and over half (52.8%) of NHs reported administering antibiotic treatments near the EoL.⁴⁵ Similarly, a study of NH residents with advanced dementia found that ACP related to antibiotics were the least commonly completed (only 4% of forms).⁴⁶ ACP is a multi-faceted and ongoing process. These results indicate the need for state and regional educational initiatives directed at NHs in the integration of palliative care and infection management, and greater educational resources in antibiotic stewardship are needed for patients and families.

Regional QIN-QIO programs offer NHs training through shared learning collaboratives on antibiotic stewardship and appropriate use to increase patient safety, improve quality of care, and support infection control programs.^28,30,47,48 While antibiotic stewardship guidance through QIN-QIOs is largely focused on healthcare-associated infections and preventing drug resistance, our findings suggest a possible spillover effect from generalized antibiotic stewardship into palliative and EoL care. Our findings also suggest a compound beneficial effect of having both QIN-QIO and mature/endorsed POLST participation. A featured webinar through a QIN-QIO, “The Role of POLST in Advance Care Planning,” indicates a potential crossover of resources between QIN-QIO and POLST.⁴⁹ Increased collaboration could yield positive outcomes for goal-concordant EoL care for NH residents.

Additionally, it is possible that such initiatives have the potential to mitigate inappropriate antibiotic use at the EoL to ensure that goals of care are aligned with patient and family wishes and priorities.¹ A study by Manu et al., which outlined care goals in comfort-oriented or palliative care-driven advanced directives, found that older adult NH residents with multiple chronic conditions and indwelling catheters were less likely to have antibiotic use compared to those without similar advance directives.⁵⁰ Together, these findings highlight the importance of the adoption of advance directive initiatives in NHs (i.e. POLST or other paradigms) that can guide clinicians in providing optimal patient-oriented care at the EoL. These findings highlight how imperative it is for NHs to incorporate policies that prioritize the timely completion of comprehensive advance directives to ensure improved quality of life and reduced treatment burden, particularly for seriously ill older adults.⁶

This study was limited by the nature of the survey data, which were retrospective, facility level and subject to self-reporting biases. Possible response bias was analyzed, and respondents only differed from non-respondents on a few characteristics.¹⁵ It is possible that hospice providers primarily conduct ACP for EoL patients in NHs, which suggests there could be more coordination between hospice providers and NH facilities in goals of care discussions and ACP policies. Investigators have found that NH staff may have poor communication with hospice providers, which may impact ACP for residents.^51-53 The facility level survey made it impossible to capture individual level data on preferences for care, which is a limitation.

CONCLUSION

State and regional initiatives together may impact NH policies on care planning related to antibiotic use for residents approaching the EoL. Incorporating antibiotic stewardship into palliative care at EoL may enhance ACP and infection management procedures. Future research on the coordination and communication between hospice providers and NHs is needed in general for ACP and specific to goals of care and antibiotic use.