Rural/Urban Differences in Discharge from Rehabilitation in Older Adults with Traumatic Brain Injury

Abstract

Background/Objectives

Rates of traumatic brain injury (TBI) among older adults and treatment of this population in nursing homes are increasing. The objective of this study is to examine differences in the quality of care and outcomes of older adults with TBI in rural and urban settings by 1) comparing the rates of successful community discharge; and 2) reasons for not achieving successful discharge among patients in rural and urban environments.

Design

Retrospective national cohort study of SNF patients using Medicare inpatient claims linked with Minimum Data Set assessments. Demographic, health, and facility characteristics were compared between rural and urban settings using descriptive statistics. Logistic regression with state random effects was used to identify characteristics that predicted successful discharge.

Setting

U.S. skilled nursing facilities (n=11,771).

Participants

Medicare beneficiaries aged 66+ discharged to a SNF following hospitalization for TBI between 2011 and 2015 (n = 61,021).

Measurements

Successful community discharge defined as discharge from SNF within 100 days of admission and remaining in the community for ≥ 30 days without dying or admission to an inpatient healthcare facility.

Results

Unadjusted rates of successful discharge were significantly lower for patients in rural settings compared to patients in urban settings (52.1% vs. 58.5%, p<0.01). Patients in rural settings had lower adjusted odds (OR 0.84, 95%CI=0.80-0.89) of successful discharge. Reasons for not discharging successfully differed between rural and urban settings with rural patients less likely to discharge from SNF within 100 days though also less likely to be re-hospitalized within 30 days of SNF discharge.

Conclusion

Given the low overall rate of successful community discharge and worse outcomes among rural patients, further research to explore interventions to improve SNF care and discharge planning in this population is warranted.

Introduction

Traumatic brain injury (TBI) is a major cause of death and disability in the United States across all age groups. Older adults have the highest rates of hospitalization and death following TBI of any age group in the US, with rates of trauma center admissions for older adults increasing faster than population growth.^1,2 Following hospitalization for TBI, older adults are less likely to be discharged home compared to their younger counterparts; the proportion of older adults with TBI transferring from the hospital to a skilled nursing facility (SNF) increases sharply with age.^1,2 The goals of post-acute care, including SNF care, are to address ongoing medical needs, maximize independence, and facilitate safe return to the community.³

The quality and type of healthcare, including post-acute care services, vary based on where a patient lives.⁴ Patients discharged from rural hospitals more often receive post-acute care in SNFs than via home health services,⁵ and receive care in nursing homes with lower quality indicators.⁶ Among TBI patients of all ages, mortality rates are higher for patients in rural settings.⁷ However, it is unclear how the rural-urban differences observed in prior research relate to the growing number of older adults with TBI receiving post-acute care in SNFs and their likelihood of successful discharge from SNF.

The objective of this study is to examine differences in the quality of care and outcomes of older adults with TBI in rural and urban settings by 1) comparing the rates of successful community discharge; and 2) reasons for not achieving successful discharge among patients in rural and urban environments. Given prior research on post-acute care use and health outcomes among patients in urban and rural environments, we hypothesized the odds of successful discharge would be lower among older patients with TBI residing in rural settings.

Methods

Data Sources

Data sources included the 2011-2015 Centers for Medicare and Medicaid Services (CMS) Medicare Master Beneficiary Summary File (demographics and insurance status), the Medicare Provider Analysis and Review files (inpatient claims, diagnosis codes and length of stay), and the Minimum Data Set 3.0 (MDS), a federally mandated assessment of patients’ health and function at all CMS–certified nursing homes. The MDS assessment is completed on each resident, often by a designated coordinator, and with input from interdisciplinary staff with knowledge of the resident. Medicare enrollment, claims (inpatient and hospice), and MDS data were combined using the residential history methodology to create a longitudinal record of Medicare utilization to identify patients who successfully discharge to the community and elucidate the reasons for not successfully discharging among those who did not successfully return to the community.⁸ Hospital characteristics were obtained from the American Hospital Association Annual Survey. SNF characteristics were obtained from the Online Survey Certification and Reporting (OSCAR) from 2011-2012 and from the Certification and Survey Provider Enhanced Reporting (CASPER) from 2013-2015. Quality ratings were obtained from Nursing Home Compare.

Population

We identified Medicare beneficiaries aged 66 years and older enrolled in traditional fee-for-service Medicare, who were admitted to a SNF after being hospitalized with a TBI between January 2011 and September 2015 (n=136,387). Patients with an International Classification of Diseases, Ninth Revision (ICD-9) code indicating TBI from the hospital claim (800–801.99, 803–804.99, 850–854.99, 959.01) and those with an MDS assessment associated with the post-hospitalization SNF admission were included. Patients were required to be enrolled in fee-for-service Medicare for the year before the index hospitalization and for four months after. We excluded patients with an inpatient (acute hospital, SNF/nursing home, or inpatient rehabilitation) or hospice admission in the year before the index hospitalization as a proxy for poor pre-injury health (n=74,285). Patients with incomplete rural/urban designation (n=282) or SNF characteristics data (n=799) were also excluded. The final sample included 61,021 patients.

Study Variables

Study Outcome.

The primary outcome was successful community discharge, a claims-based SNF quality measure designed to reflect a meaningful patient outcome.⁹ Successful community discharge is defined as discharge from SNF to the community within 100 days of SNF admission and remaining in the community for at least 30 days without dying or admission to an inpatient healthcare facility.⁹ We also report reasons for not successfully discharging to the community. These included remaining in a skilled nursing facility or nursing home for more than 100 days, discharging directly to a hospital, dying while a resident in the skilled nursing facility, or admission to a hospital, hospice, or death within 30 days of discharge from the SNF to home.

Explanatory Variables.

Our primary explanatory variable was a binary variable indicating rural/urban designation of the patient’s home county using the 2013 National Center for Health Statistics Urban-Rural Classification Scheme.¹⁰ Demographic characteristics included age, sex, race, current marital status, and dual Medicare and Medicaid enrollment. Health characteristics included hospital length of stay, intensive care unit (ICU) admission, Charlson Comorbidity Index^11,12 score at hospital discharge, and functional status, cognition, and mortality risk categorized with data from the MDS. Functional status was characterized using the ADL long-form score, which reflects the level of assistance required to perform seven ADL tasks on a 0-to-4-point scale. Scores range from 0 to 28 points, with higher scores indicating worse function.¹³ Patients who scored 23 or greater were categorized as having severe functional impairment.¹⁴ Cognitive status was characterized using the MDS-derived Cognitive Function Scale (CFS).¹⁵ The CFS ranges from 1 to 4 with higher score indicating more cognitive impairment.¹⁵ Patients with scores of 2 to 4 on the Cognitive Function Scale (CFS) were categorized as having cognitive impairment.¹⁶ Health status was characterized using the Mortality Risk Score (MRS3), which uses personal, functional, and health characteristics to assess mortality risk on a 1-to-39-point scale.¹⁷ Scores of 8 or higher on the MRS3 are associated with a 5.4-fold increase in the odds of 30-day mortality, and patients with scores of 8 or higher were characterized as having high mortality risk.¹⁷ Hospital characteristics included bed size, teaching status, and trauma center level. SNF characteristics included bed size, occupancy rate, for-profit and chain status, and CMS overall Five-Star rating.¹⁸

Statistical Analysis

Patient and facility characteristics were summarized using means, standard deviations, frequencies, and proportions. Bivariate comparisons between patients from rural and urban environments were made using t-tests and chi-square tests. Cohen’s D and Cohen’s W were calculated to examine the strength of the bivariate associations. The odds of successful discharge were calculated using multivariable logistic regression model adjusting for demographic, health, and facility characteristics that were significant in the bivariate analysis, and U.S. state random-effects. In the main analysis, missing values from the MDS assessment or SNF Five-Star rating were dummy coded. We also performed a sensitivity analysis comparing the results including and excluding those with missing MDS or SNF star-rating values. Lastly, we compared the proportion of reasons for not achieving a successful discharge between patients in rural and urban settings. The Institutional Review Board at Brown University approved this study. All analyses were completed with STATA 14.0 (StataCorp, College Station, Texas). Additional information about the data and methods used for these analyses can be found at https://repository.library.brown.edu/.

Results

Unadjusted rates of successful discharge were significantly lower for patients in rural areas compared to those in urban areas (52.1% vs. 58.5%, p<0.01). We identified small but significant differences in demographic, health status, and facility characteristics between patients from rural and urban areas, detailed in Table 1. Results from the multivariable logistic model indicated that, after adjustment, patients from rural areas have lower odds of successful discharge compared with patients from urban areas (OR 0.84, 95%CI=0.80-0.88; see Table 2). The sensitivity analysis indicated similar odds of successful discharge among patients in rural areas if those with missing MDS assessment values or SNF star-rating were excluded (OR 0.85, 95%CI=0.80-0.89).

Table 1.

Characteristics of Older Adults with TBI in SNF in Rural and Urban Settings, 2011-Q3 2015

N (%)	Urban 50,891 (83.4)	Rural 10,130 (16.6)	p Value	Effect Size
Successful discharge to community, n (%)	29,749 (58.5)	5,277 (52.1)	<0.01	w=0.05
Demographic Characteristics
Age	84.3 (7.4)	83.3 (7.5)	<0.01	d=0.13
Female, %	63.5	61.9	<0.01	w=0.01
Race, %
White	90.2	95.3
Black	4.5	2.3	<0.01	w=0.07
Other/unknown	5.4	2.4
Currently Married, %	33.5	35.6	<0.01	w=0.02
Medicaid Eligible, %	13.3	17.6	<0.01	w=0.05
Health Characteristics
Hospital length of stay, days	6.7 (6.3)	6.9 (6.3)	<0.01	d=0.04
Intensive Care Unit admission, %	53.1	51.8	0.02	w=0.01
Severe Functional Impairment,1 %	9.7	10.4	0.02	w=0.01
Cognitive Impairment,2 %	53.0	55.2	<0.01	w=0.02
High Mortality Risk, %	14.7	17.5	<0.01	w=0.03
Charlson Comorbidity Index Score	1.54	1.50	0.04	d=0.02
Hospital Characteristics
Level 1 trauma center, %	28.5	30.3	<0.01	w=0.02
Teaching hospital, %	73.1	61.8	<0.01	w=0.09
Hospital bed size
Mean (SD)	462 (324)	392 (305)	<0.01	d=0.22
Median (IQR)	384 (254-607)	338 (147-566)
SNF Characteristics
For profit, %	68.5	67.1	<0.01	w=0.01
Chain, %	57.2	59.9	<0.01	w=0.02
SNF bed size
Mean (SD)	131 (77)	105 (48)	<0.01	d=0.36
Median (IQR)	120 (88-159)	101 (72-125)
Occupancy rate, % (SD)	84.4 (13.6)	81.5 (14.7)	<0.01	d=0.20
Star rating, %
5-star	27.3	20.6
4-star	28.1	29.7
3-star	18.4	20.3	<0.01	w=0.06
2-star	16.6	17.6
1-star	8.3	10.8

Note: Values reflect mean (SD), unless otherwise indicated. Effect Size interpretation d=0.2-small, d=0.5-medium, d=0.8-large, w=0.1-small, w=0.3=medium, w=0.5-large.

¹Activity of Daily living long-form score ≥ 23,

²Cognitive Function Scale ≥ 2,

³Mortality Risk Score ≥ 8.

Abbreviations: SD: Standard Deviation, IQR: Interquartile range, SNF: Skilled Nursing Facility.

Table 2.

Factors Predicting Successful Community Discharge from SNF among Older Adults with TBI, 2011-Q3 2015

	Adjusted Odds Ratio	95% CI	p value
Rural setting	0.84	0.80-0.88	<0.01
Demographic Characteristics
Age	0.97	0.97-0.97	<0.01
Female	1.29	1.24-1.34	<0.01
Race
White	Ref.
Black	1.04	0.95-1.14	0.36
Other	1.21	1.10-1.32	<0.01
Currently married	1.16	1.11-1.21	<0.01
Medicaid eligible	0.55	0.52-0.58	<0.01
Health Characteristics
Length of hospital stay (days)	0.97	0.96-0.97	<0.01
Intensive Care Unit admission	1.02	0.98-1.06	0.28
Severe Functional Impairment1	0.37	0.34-0.40	<0.01
Cognitive Impairment2	0.37	0.36-0.38	<0.01
High Mortality Risk3	0.55	0.52-0.58	<0.01
Charlson Comorbidity Index Score	0.89	0.88-0.90	<0.01
Hospital Characteristics
Level 1 trauma center	1.09	1.04-1.15	<0.01
Teaching hospital	1.05	1.01-1.10	0.03
Hospital bed size
>Median Number of beds	0.97	0.92-1.01	0.12
SNF Characteristics
For profit	0.98	0.94-1.02	0.28
Chain	1.07	1.03-1.11	<0.01
Occupancy rate	0.67	0.58-0.78	<0.01
SNF bed size
>Median Number of beds	0.97	0.93-1.01	0.10
Star rating
5-star	Reference
4-star	0.91	0.87-0.95	<0.01
3-star	0.85	0.81-0.90	<0.01
2-star	0.80	0.76-0.85	<0.01
1-star	0.71	0.66-0.77	<0.01

Note:

¹Activity of Daily living long-form score ≥ 23,

²Cognitive Function Scale ≥ 2,

³Mortality Risk Score ≥ 8.

Abbreviations: CI= Confidence Interval; SNF= Skilled Nursing Facility.

Reasons for not discharging successfully differed between rural and urban settings (see Figure 1). Patients in rural areas were less likely to discharge from SNF within 100 days and more likely to die in SNF than their urban counterparts. However, patients in rural settings were also less likely to return to the hospital directly from SNF or to be re-hospitalized within 30 days of SNF discharge.

Figure 1. Reason for successful community discharge failure among older adults with TBI admitted to SNF, by rural and urban setting.

Note: * indicates significant difference in column proportions at p<0.05. NH= Nursing Home; SNF = Skilled Nursing Facility.

Discussion

This study has three main findings: First, the overall rate of successful community discharge in this study population was less than 60%, indicating that a large number of older adults treated in SNFs in both rural and urban settings struggle to return to the community following TBI. Second, rural TBI patients have lower rates of successful community discharge than their urban counterparts after adjusting for differences in demographic, health status, and facility characteristics. Finally, reasons for not having a successful discharge varied between the two groups, with those in rural areas more likely to remain or die in a SNF and less likely to be hospitalized during or within 30 days of a SNF stay.

Our findings align with previous work demonstrating worse outcomes, specifically higher mortality rates among trauma patients with TBI, in rural settings compared with urban settings.⁷ The unique contribution of our study is that we examine differences in discharge outcomes following post-acute care, an important patient-centered and policy-relevant outcome that can directly influence discharge planning processes.

The identified differences in the reasons for not achieving successful discharge in rural and urban environments warrant further investigation. For example, in our sample, patients in rural areas were more likely to transition to long-term care compared to patients in urban areas. It is possible that the barriers to use of home and community-based services (e.g., home health) observed in rural settings encourage patients to remain in inpatient post-acute care settings longer.^19,20 We also found a higher proportion of patients in rural settings died while in a SNF compared to those in urban settings. This finding corresponds with evidence of higher mortality rates in rural settings among patients receiving post-acute care²¹ and patients with TBI.⁷ Yet, we found patients in rural areas were less likely to be re-hospitalized during their SNF stay and in the 30 days after SNF discharge, potentially aligning with decreased access to hospital care that has been previously been documented in rural environments.²²

There are several limitations to this study. There may be residual confounding factors that were not available in our data but may impact successful discharge including details of injury severity, type and timing of pre-hospital trauma care, and access to post-injury social supports. Additionally, we did not examine patients with Medicare Advantage insurance, non-Medicare enrollees, and those with a healthcare facility or hospice admission in the year prior to injury, thus our results should not be generalized to those patient groups. Finally, we cannot address differential selection that might occur in who is discharged to SNF among patients in rural and urban settings.

Future research examining access to healthcare services and community resources among patients with TBI from rural and urban environments may yield insight into the observed differences in successful discharge rates and provide potential targets for intervention. Additionally, studies examining successful discharge among the excluded patient groups (i.e. those with prior healthcare utilization), and that include details of local area resources, personal supports, and injury characteristics will improve understanding of the post-acute care experiences of older adults with TBI receiving care in SNFs.

Conclusion

In this analysis, we examined differences between patients from rural and urban settings in terms of demographic and health status characteristics, the facilities where they receive care, and their rates of successful discharge to the community. We found that patients from rural areas have a lower rate of successful discharge compared to those from urban areas after adjusting for demographic, health status, and facility characteristics that differ between the two environments. Given the low overall rate of successful community discharge and worse outcomes among patients in rural settings, further research is needed to explore interventions that improve the outcomes of older adults with TBI receiving SNF care in rural settings and to address challenges to discharge planning specific to this population.

Key Points.

• Less than 60% of Medicare beneficiaries, recently hospitalized with traumatic brain injury (TBI), successfully discharge home following a skilled nursing facility (SNF) stay.

• Rates of successful discharge among older adults with TBI in SNFs are lower among patients in rural areas compared to urban areas.

• Older adults with TBI in rural areas are more likely to remain in a SNF for more than 100 days or die while admitted to a SNF compared to patients in urban environments, whereas those in urban areas are more likely to be admitted to a hospital directly from the SNF or within 30 days of SNF discharge.

Why does this paper matter?

This paper reports a difference in the rate of successful discharge from SNF between older adults with TBI in rural versus urban settings. The findings are foundational for exploring potential disparities in care and developing interventions that may improve the outcomes of older adults with TBI.