Positive Beliefs and the Likelihood of Successful Community Discharge From Skilled Nursing Facilities

Abstract

Objectives:

To examine the association of patient and direct-care staff beliefs about patients’ capability to increase independence with activities of daily living (ADL) and the probability of successful discharge to the community after a skilled nursing facility (SNF) stay.

Design:

Retrospective cohort study of SNF patients using 100% Medicare inpatient claims and Minimum Data Set resident assessment data. Linear probability models were used to estimate the probability of successful discharge based on patient and staff beliefs about the patient’s ability to improve in function, as well as patient and staff beliefs together. Estimates were adjusted for demographics, health status, functional characteristics, and SNF fixed effects.

Participants:

Fee-for-service Medicare beneficiaries (N=526,432) aged 66 years or older who were discharged to an SNF after hospitalization for stroke, hip fracture, or traumatic brain injury.

Interventions:

Not applicable.

Main Outcome Measures:

Successful community discharge (discharged alive within 90d of SNF admission and remaining in the community for ≥30d without dying or health care facility readmission).

Results:

Patients with positive beliefs about their capability to increase independence with ADLs had a higher adjusted probability of successful discharge than patients with negative beliefs (positive, 63.8%; negative, 57.8%; difference, 6.0%, 95% confidence interval [CI], 5.4–6.6). This remained true regardless of staff beliefs, but the difference in successful discharge probability between patients with positive and negative beliefs was larger when staff had positive beliefs. Conversely, the association between staff beliefs and successful discharge varied based on patient beliefs. If patients had positive beliefs, the difference in the probability of successful discharge between positive and negative staff beliefs was 2.5% (95% CI, 1.0–4.0). If patients had negative beliefs, the difference between positive and negative staff beliefs was −4.6% (95% CI, −6.0 to −3.2).

Conclusions:

Patients’ beliefs have a significant association with the probability of successful discharge. Understanding patients’ beliefs is critical to appropriate goal-setting, discharge planning, and quality SNF care.

Patient-centered care requires that patients are a part of their health care team and that their preferences, needs, and values guide clinical decision-making.^1,2 Patient-centered approaches improve patient satisfaction, quality of care, and functional outcomes among patients with chronic and complex medical conditions, and are critical components of quality health care.^2–6 The Centers for Medicare and Medicaid Services (CMS) has recently prioritized making health care quality measurement more patient-centered through function-related quality measures and linking quality measurement to reimbursement.^7–11

Patients, families, and health care providers recognize that increasing independence with activities of daily living (ADL) and self-care are important rehabilitation goals.^12–15 Setting goals is an integral part of patient-centered practice as it requires patients, families, and health care providers to establish mutually agreed-upon objectives that guide the care plan.^1,7,16 Expectations about recovery, positive affect, and beliefs about the ability to do what is required to recover (self-efficacy) may help patients to set and achieve their goals.^17–23 Through interactions with their patients, health care providers can affect patients’ self-efficacy, expectations, and functional outcomes,^24–27 whereas not involving patients in decisions can create discordance between patients and their health care teams.^28–31 Despite the recognized benefits, financial and organizational pressures,^32,33 doubts about patients’ capacity, and practitioners’ difficulty eliciting goals from patients are barriers to patient involvement in clinical decisions.^{12,28–31,33–38}

Our objective was to examine the association between beliefs about a patient’s capability to increase independence with ADLs and successful discharge to the community. Discharging to the community, and remaining in the community for 31 days is a skilled nursing facility (SNF) quality measure, designed to reflect the goal shared by most patients and families of the patient returning home and remaining home after a postacute rehabilitative stay.³⁹ Specifically, we examined whether the probability of successful discharge after an SNF stay was associated with: (1) a patient’s beliefs about their capability to increase independence in ADLs; (2) SNF staff beliefs about a patient’s capability to increase independence in ADLs; and (3) the interaction between patient and staff beliefs (eg, the association between patient’s beliefs and successful discharge conditional on staffs’ beliefs and vice versa). Our analyses focused on patients admitted to SNFs with stroke, hip fracture, and traumatic brain injury (TBI) because a high proportion of these individuals use inpatient postacute care services,^40–42 and because their functional limitations with self-care and mobility are the primary reasons for postacute care referral and need for institutionalization after postacute care.^43–52 Based on evidence supporting the relationship between independence with ADLs and discharge disposition^45,53 and the role of self-efficacy^17,54,55 and positive affect^22,23 in goal achievement, we hypothesized that positive beliefs held by patients and direct-care staff would be associated with a higher probability of successful discharge.

Methods

Data source

This study is a retrospective cohort analysis using 100% Medicare administrative data for the years 2011 to 2016. We combined the following sources of individual-level data: the Medicare Master Beneficiary Summary File (Medicare enrollment data [ie, demographics and managed care participation]), the Medicare Provider Analysis and Review files (inpatient hospital and SNF encounters), hospice and home health claims, and Minimum Data Set 3.0 (MDS), a federally mandated clinical assessment completed for all residents in CMS-certified nursing homes. The Brown University Institutional Review Board approved this study.

Sample

We combined claims and enrollment data to identify fee-for-service Medicare beneficiaries discharged to a SNF after acute hospitalization for hip fracture, stroke, or TBI between January 1, 2011, and December 31, 2016. Patients were identified based on the International Classification of Diseases code contained in the primary diagnosis field of their hospital claim (see supplemental appendix S1, available online only at http://www.archives-pmr.org/, for specific International Classification of Diseases codes). We identified patients aged 66 and older who had not used SNF or hospice services in the previous year. We excluded patients with missing successful discharge information, a single MDS assessment (ie, those with very short nursing home stays), and incomplete or “unable to determine” patient or staff belief data (primary explanatory variable). The application of the inclusion and exclusion criteria to arrive at the final sample is described in figure 1. Differences between patients who were included versus those who were excluded owing to MDS assessments or missing belief data are detailed in supplemental table S1 (available online only at http://www.archives-pmr.org/).

Fig 1.

Flowchart of Medicare beneficiaries who met the inclusion and exclusion criteria. *A total of 88% of those with incomplete successful discharge information were censored owing to admission after September 1, 2016.

Study variables

Study outcome

The primary outcome was successful discharge from the SNF within 90 days of admission. Successful discharge was identified using residential history file methodology⁵⁶ and defined as discharge home after an SNF stay and remaining home without admission to an inpatient health care facility (ie, acute hospital, long-term care hospital, SNF, or inpatient rehabilitation hospital), hospice use, or death for 30 continuous days.⁵⁷ Patients who were readmitted or who died during the SNF stay were not considered successfully discharged. The outcome aligns with the CMS claims-based quality measure, which reports the percentage of patients who return to home and remain alive without readmission to an acute care or long-term care hospital for 31 days.³⁹

Primary explanatory variables

The primary predictors were the rating of the patients’ “Functional Rehabilitation Potential” from the MDS, which indicates whether the resident or the direct-care staff believe the patient is “capable of increased independence in at least some ADLs.” The MDS manual states that a resident’s beliefs should be based on a conversation with the resident and should reflect the resident’s perceptions regardless of how accurate the staff believes those perceptions to be. The rating of staff beliefs should be based on an interdisciplinary team meeting involving staff who routinely work with the resident.⁵⁸ Responses of “yes” were coded as positive beliefs, responses of “no” were coded as negative beliefs, and patients were excluded if beliefs were “unable to determine,” which was more often among those with significant cognitive deficits.

Covariates

Sociodemographic variables included age, race, and dual Medicare-Medicaid eligibility during the month of admission. The Elixhauser Comorbidity index was calculated based on the set of hospital discharge diagnoses.⁵⁹ Marital status was categorized into currently married versus other categories. Depression was based on the Patient Health Questionnaire-9 or the Patient Health Questionnaire-9 Observational Version and coded as none to minimal depression (score, <5), mild depression (score, 5–9), or moderate to severe depression (score, >10).⁶⁰ Body mass index (BMI) was categorized into underweight (<18.5), normal or overweight (18.5–30), or obese (>30).⁶¹ Cognition was characterized using the Cognitive Function Scale score, which ranges from 1 (intact) to 4 (severe impairment).⁶² Because patient beliefs may be reported differently among those with more significant cognitive impairment, we included a dummy indicator for each Cognitive Function Scale score value in all models to limit this source of bias. The ADL-long form score was used to assess independence with ADLs. Scores range from 0 to 28, with higher scores indicating less independence.^63,64

Analyses

The characteristics of patients with positive and negative beliefs were summarized using standard descriptive measures, including means, standard deviations, frequencies, and proportions. Bivariate comparisons were made using t tests for continuous variables and chi-square tests for categorical variables. To examine the association of patient beliefs with successful discharge, we used linear probability models controlling for patients’ sociodemographic, health status, and functional status characteristics, and SNF fixed effects. Separate models were used for the full sample and each diagnosis group. By including SNF fixed effects in each model, we estimated the average within-SNF association between patient beliefs and successful discharge (ie, differences between patients with and without positive beliefs pooled across SNFs). Thus, relevant facility-level attributes such as quality were effectively controlled for.

We then examined the association of staff beliefs with successful discharge in using the same methodology but replacing patient beliefs with the staff beliefs as the primary explanatory variable. In a third specification, we included both patient and staff beliefs and an interaction term allowing us to examine the association of patient beliefs with successful discharge conditional on the staffs’ beliefs and vice versa. Point estimates from each model are presented as adjusted probabilities of successful discharge based on either patient or staff beliefs, or their combination, which we derived using the marginal standardization form of predictive margins. All analyses were completed with STATA MP 16.0.^a Additional information about the data and methods used for these analyses can be found at https://repository.library.brown.edu/studio/item/bdr:1117604/.

Results

Our final sample included 526,432 Medicare fee-for-service beneficiaries who were discharged to an SNF after hospitalization for hip fracture (n=341,540), stroke (n=142,200), or TBI (n=42,692) between 2011 and 2016. Patient characteristics are described in table 1. The majority (63.4%) of patients were successfully discharged home. Of the 192,834 (36.6%), patients who were not successfully discharged, 47,926 (24.9%) remained in an SNF, 19,427 (10.1%) died before SNF discharge, 28,018 (14.5%) died within 30 days of SNF discharge, and 97,463 (50.5%) entered an inpatient health care facility within 30 days of SNF discharge.

Table 1.

Patient characteristics by patient and staff beliefs

Characteristics		Patient Beliefs			Staff Beliefs
	All Subjects	Positive	Negative	P Value	Positive	Negative	P Value
n (%)	526,432	487,236 (93)	39,196 (7)		489,403 (93)	37,029 (7)
Age, mean ± SD	84.1±7.6	84.1±7.6	84.2±7.7	.01	84.1±7.6	84.0±7.7	.26
Female, %	70.8	70.9	70.5	.1	70.9	69.9	.01
Currently married, %	33.3	33.4	31.5	.01	33.4	31.8	.01
Race, %				.01			.01
White	90.2	90.4	87.4		90.4	87.7
Black	5.6	5.4	8.3		5.4	8.3
Other	4.2	4.2	4.3		4.3	4.0
Medicaid eligible, %	12.5	12.2	15.9	.01	12.3	15.4	.01
Elixhauser Comorbidity Index,* mean ± SD	15.5±13.6	15.4±13.6	16.0±13.6	<.01	15.4±13.6	15.8±13.6	<.01
Obesity,† %				<.01			<.01
Underweight	7.5	7.4	9.0		7.4	8.8
Normal/overweight	74.8	74.9	73.5		74.9	73.6
Obese	15.4	15.4	14.7		15.4	14.7
Depression,‡ %				<.01			<.01
None to minimal	73.2	74.0	63.3		73.8	65.5
Mild	15.3	15.3	15.0		15.3	15.1
Moderate to severe	5.0	5.0	5.8		5.0	5.5
Cognitive function,§ %				<.01			<.01
1	52.5	53.8	36.6		53.5	39.5
2	23.8	24.0	22.6		23.9	23.1
3	20.3	19.5	30.0		19.7	28.0
4 (severely impaired)	2.8	2.3	10.1		2.4	8.6
Admission ADL score‖	18.4±3.6	18.3±3.5	19.5±4.6	<.01	18.4±3.5	19.1±4.6	<.01
mean ± SD
Index hospitalization length of stay, d, mean ± SD	6.9±4.4	6.8±4.2	7.8±6.1	P<.01	6.9±4.3	7.6±6.1	P<.01
ICU admission, %	27.7	27.4	31.5	P<.01	27.4	31.2	P<.01

NOTE. Comparisons were made using t tests for continuous variables and chi-square tests for categorical variables.

^*Elixhauser Comorbidity Index (sample range, −4 to 98, with higher scores indicating more comorbidity burden).

^†Derived from the resident’s BMI.

^‡According to the Patient Health Questionnaire or Patient Health Questionnaire Observational Version.

^§According to the Cognitive Function Scale (score range, 1–4, with higher scores indicating more cognitive impairment).

^‖According to the Morris ADL scale (score range, 0–28, with higher scores indicating worse ADL function).

A higher proportion of patients with positive beliefs were married, white, had a normal or overweight BMI, and had none to minimal reported depression or cognitive impairment compared with patients with negative beliefs. Patients with positive beliefs also had better ADL function and lower comorbidity burden at SNF admission than those with negative beliefs. Compared with patients with positive beliefs, those with negative beliefs had longer hospital lengths of stay, and a higher proportion were black, underweight, eligible for Medicaid, admitted to the intensive care unit (ICU) during hospitalization, and had significant cognitive impairments. Similarly, among staff with positive beliefs, a higher proportion of patients were white, women, had a normal or overweight BMI, none to minimal reported depression or cognitive impairment, and had better ADL function and lower comorbidity burden compared with when staff had negative beliefs. Among staff with negative beliefs, patients had longer hospital lengths of stay, and a higher proportion were black, underweight, eligible for Medicaid, admitted to the ICU during hospitalization, and had impaired cognition. Similar patterns were observed in each diagnosis group, and are detailed in supplemental table S2 (patient beliefs), supplemental table S3 (staff beliefs), and supplemental table S4 (patient and staff beliefs) (available online only at http://www.archives-pmr.org/).

Patients with positive beliefs about their capability for increased independence with ADLs had a higher probability of successful discharge than patients with negative beliefs after adjusting for sociodemographic, health, and cognitive characteristics. In the full sample, the adjusted probability of successful discharge was 63.8% (95% confidence interval [CI], 63.8–63.9) among patients with positive beliefs and 57.8% (95% CI, 57.3–58.4) among those with negative beliefs. Thus, positive beliefs conferred a 6.0% (95% CI, 5.4–6.6) increase in the probability of successful discharge. Similarly, positive staff beliefs conferred a 4.4% (95% CI, 3.8–5.0) increase in the probability of successful discharge. Similar patterns were observed across each diagnosis group (table 2).

Table 2.

Adjusted probability of successful discharge based on patient and staff beliefs

	All Patients	Hip Fracture	Stroke	TBI
	Estimate (95% CI)	Estimate (95% CI)	Estimate (95% CI)	Estimate (95% CI)
Patient beliefs*
Positive	63.8 (63.8–63.9)	68.0 (67.9–68.0)	54.5 (54.4–54.6)	60.9 (60.7–61.1)
Negative	57.8 (57.3–58.4)	63.1 (62.3–63.9)	48.0 (47.0–49.0)	54.4 (52.2–56.6)
Difference	6.0 (5.4–6.6)	4.9 (4.1–5.7)	6.6 (5.5–7.6)	6.5 (4.1–8.9)
Staff beliefs†
Positive	63.7 (63.6–63.7)	67.9 (67.8–67.9)	54.4 (54.3–54.5)	60.8 (60.6–61.0)
Negative	59.3 (58.7–59.9)	64.6 (63.8–65.4)	49.3 (48.3–50.3)	55.8 (53.4–58.1)
Difference	4.4 (3.8–5.0)	3.2 (2.4–4.1)	5.1 (3.9–6.2)	5.0 (2.5–7.6)

NOTE. Estimates reflect adjusted probability (95% CI).

^*Probability estimates were based on models that included patient beliefs and the covariates of age, age², sex, race, marital status, depression, cognitive function, hospital length of stay, ICU use, obesity, and Medicaid eligibility.

^†Probability estimates were based on models that included staff beliefs and the covariates listed above.

The probabilities of successful discharge based on the interaction of patient and staff beliefs overall and in each patient group are described in table 3. The effect of positive patient beliefs on successful discharge in the context of staff beliefs and the effect of positive staff beliefs on successful discharge in the context of patient beliefs are reported in table 3.

Table 3.

Adjusted probability of successful discharge based on the interaction of patient and staff beliefs

	All Patients	Hip Fracture	Stroke	TBI
	Estimate (95% CI)	Estimate (95% CI)	Estimate (95% CI)	Estimate (95% CI)
Patient/staff beliefs
Patient positive/staff positive	63.8 (63.8–63.9)	68.0 (67.9–68.0)	54.6 (54.5–54.7)	61.0 (60.8–61.2)
Patient positive/staff negative	61.3 (59.9–62.8)	65.8 (63.9–67.8)	51.9 (49.1–54.7)	57.7 (52.1–63.3)
Patient negative/staff positive	54.1 (52.9–55.4)	58.7 (57.0–60.4)	44.6 (42.4–46.8)	51.5 (46.8–56.1)
Patient negative/staff negative	58.7 (58.1–59.4)	64.2 (63.4–65.0)	48.7 (47.6–49.8)	55.1 (52.6–57.6)
Difference in probability of successful discharge associated with patient positive beliefs
If staff were positive	9.7 (8.4–11.0)	9.3 (7.5–11.0)	9.9 (7.7–12.2)	9.5 (4.8–14.2)
If staff were negative	2.6 (1.0–4.2)	1.6 (−0.5 to 3.7)	3.2 (0.3–6.1)	2.6 (−3.5 to 8.6)
Difference in probability of successful discharge associated with staff positive beliefs
If patients were positive	2.5 (1.0–4.0)	2.2 (0.2–4.1)	2.7 (−0.1 to 5.5)	3.3 (−2.5 to 9.0)
If patients were negative	−4.6 (−6.0 to −3.2)	−5.5 (−7.4 to −3.6)	−4.1 (−6.5 to −1.6)	−3.7 (−8.9 to 1.6)

NOTE. Estimates reflect adjusted probability (95% CI). Probability estimates were based on the interaction of patient and staff beliefs in a model adjusted for age, age², sex, race, marital status, depression, cognitive function, hospital length of stay, ICU use, obesity, and Medicaid eligibility.

In the full sample, positive patient beliefs conferred a 9.7 percentage point (95% CI, 8.4–11.0) increase in the probability of successful discharge if staff beliefs were positive, and at 2.6 percentage point (95% CI, 1.0–4.2) if staff beliefs were negative. Positive patient beliefs conferred a significant increase in the probability of successful discharge if staff beliefs were positive in each diagnosis group. If staff beliefs were negative, positive patient beliefs were associated with a significant increase in the probability of successful discharge among patients with stroke. The direction of the effect was consistent among patients with hip fracture and TBI, although the results were not significant (see table 3).

Staff beliefs were also associated with the probability of successful discharge after adjusting for sociodemographic, health, and cognitive characteristics. However, the direction of the association varied based on patient beliefs. In the full sample, positive staff beliefs were associated with a 2.5 percentage point (95% CI, 1.0–4.0) increase in the probability of successful discharge if the patient had positive beliefs. However, if patient beliefs were negative, the difference in the probability of successful discharge was −4.6 percentage points (95% CI, −6.0 to −3.2), suggesting that when patients had negative beliefs, staff’s positive beliefs were associated with a lower probability of successful discharge. Similar patterns were observed in diagnosis-specific analyses (see table 3).

Discussion

In this study of older adults admitted to SNFs after hip fracture, stroke, or TBI, we examined the association of patient and staff beliefs about the patients’ capability to increase independence with ADLs and the probability of successful discharge. We found that patients with positive beliefs had a higher probability of successful discharge after an SNF stay compared with patients with negative beliefs after adjusting for a broad set of individual-level covariates, including depression and cognitive status, as well as SNF fixed effects. We also found that positive patient beliefs were associated with successful discharge regardless of the direction of staff beliefs. We found that staffs’ positive beliefs about patients’ capability to increase independence with ADLs were associated with successfully discharging to the community, but only if patients also had positive beliefs. Taken together, our findings suggest that patients’ beliefs about the potential to increase independence may be more important to discharge outcomes than the beliefs of direct-care staff.

Our findings that patient beliefs are associated with desirable outcomes is supported by previous literature suggesting that patients’ expectations about recovery and self-efficacy may be associated with positive outcomes.^21,26,65 Drawing on the self-efficacy framework, it is possible that patients who believe they will improve put more effort into rehabilitation and are ultimately more likely to improve and achieve desired outcomes.^54,66 Patients with positive expectations may also be viewed as more motivated and easier to work with than those with more pessimistic outlooks.⁶⁷ Clinicians themselves may be motivated by patients’ perceived motivation and use different approaches to encouragement and care that affect recovery and long-term outcomes.^67,68 Our results also suggest that patients may have good insight into their recovery potential, contradicting the perception that patients’ capacity or knowledge is a barrier to patient-centered practices.^28,33,38

Contrary to our hypothesis, staffs’ positive beliefs were only associated with successful discharge if patients also had positive beliefs. If patients had negative beliefs, the probability of successful discharge was greater if staff shared negative beliefs. We hypothesize that this finding reflects the importance of patients’ beliefs and the benefits of shared recognition of a patient’s poor prognosis. Patient and staff agreement about expectations for recovery, even negative expectations, may allow for early coordination of services and caregiver training, which allows patients to remain at home in the presence of functional impairment.^69,70

Functional improvement, accurate goal setting, and helping patients return to their home environment have always been a goal of inpatient postacute rehabilitative care. However, after the passage of the Improving Post-Acute Care Transformation Act of 2014 and the corresponding CMS “Meaningful Measures” initiative, there has been increased attention on engaging patients and families as partners in their care.¹⁰ SNF quality measures now incorporate functional goals, functional change, and the proportion of patients who are discharged to the community and remain in the community for 31 days. The quality measures are publicly reported through the CMS’ Nursing Home Compare website and must be reported to avoid a 2% payment reduction penalty.^11,71 The results of this work suggest that patients’ beliefs regarding the capability to increase independence with ADLs may have a measurable effect on the probability of achieving desired quality outcomes and that, despite the barriers, understanding and incorporating patient beliefs may help facilitate accurate goal-setting and effective discharge planning.

In this investigation, we focused on the relationship between patient and staff beliefs on successful community discharge. However, we did not examine the relationship between patient and staff beliefs on ADL improvement during an SNF stay. Future work examining the associations between beliefs and functional improvement will provide additional insight into the role of beliefs on postacute discharge outcomes. Additionally, further examination of concordance and discordance between patients and staff and strategies for shared decision-making in cases of discordance may enhance the integration of beliefs and expectations into effective care planning.

Study limitations

Our findings should be considered in light of several limitations. First, there may be residual confounding from factors related to beliefs and successful discharge that we are unable to obtain from the available data. For example, details of direct-care staffs’ training; use of patient-centered approaches and encouragement strategies; and patients’ pre-injury functional status, family support, living situation, and other social supports may influence the probability of successful discharge, but were unavailable and not included in this analysis. Additionally, we did not include information about health care utilization (eg, primary, specialty, mental health, or home health care) after the SNF stay that may be associated with successful discharge. Information about awareness of deficits was also unavailable, and although we control for cognitive function, differences in patient report of beliefs among those with cognitive impairments may be a source of bias. Additionally, the included data sources contain limited information about injury severity among the patients with stroke or TBI. We included ICU use and cognitive function in our models, which may be associated with injury severity, but additional investigation into the role of injury characteristics on the relationship between perceptions and discharge outcomes is warranted. Finally, our sample included patients with 3 primary diagnoses who were homogenous in terms of experience with postacute care. Therefore, our findings may not be generalized to individuals admitted to SNFs for conditions we did not examine, non-Medicare beneficiaries, and those enrolled in Medicare Advantage.

Conclusions

Patients’ positive beliefs about their ability to increase independence with ADLs are associated with an increased probability of successful discharge from SNF after a hip fracture, stroke, or TBI. Our findings suggest that patient beliefs about the potential for functional recovery have a measurable effect on desirable outcomes. Despite the barriers, integrating patients into care decisions is critical to achieving outcomes that are important to patients, clinicians, and facilities. Direct-care staff who interact with patients need the resources, support, and training to be able to engage with patients; understand their beliefs, goals, and expectations; and integrate patients into care planning.

List of abbreviations:

ADL

activities of daily living

BMI

body mass index

CI

confidence interval

CMS

Centers for Medicare and Medicaid Services

ICU

intensive care unit

MDS

Minimum Data Set 3.0

SNF

skilled nursing facility

TBI

traumatic brain injury