Care Patterns and Predictors of Community Residence among Older Patients after Hospital Discharge for Traumatic Brain Injury

Abstract

BACKGROUND:

An increasing number of older adults with traumatic brain injury (TBI) require hospitalization, but it is unknown whether they return to their community following discharge. We examined community residence following acute hospital discharge for TBI in Texas and identified factors associated with 90-day community residence and readmission.

METHODS:

We conducted a retrospective cohort study using 100% Texas Medicare claims data of patients older than 65 years hospitalized for a TBI from January 1, 2014, through December 31, 2017, and followed for 20 weeks after discharge. Discharges to short-term and long-term acute hospital, inpatient rehabilitation facility (IRF), skilled nursing facility (SNF), long-term nursing home (NH), and hospice were identified. The primary outcome was 90-day community residence. Our secondary outcome was 90-day, all-cause readmission.

RESULTS:

In Texas, 26,985 Medicare fee-for-service patients were hospitalized for TBI [Racial and ethnic minorities: 21.1%; Females 57.3%]. At 90 days and 20 weeks following discharge, 80% and 84% were living in the community, respectively. Female sex (OR=1.16[1.08-1.25]), Hispanic ethnicity (OR=2.01[1.80-2.25]), “other” race (OR=2.19[1.73-2.77], and prior primary care provider (PCP; OR=1.51[1.40-1.62]) were associated with increased likelihood of 90-day community residence. Patients aged 75+, prior NH residence, dual eligibility, prior TBI diagnosis, and moderate-to-severe injury severity were associated with decreased likelihood of 90-day community residence. Being non-Hispanic Black (HR=1.33[1.20-1.46]), discharge to SNF (HR=1.56[1.48-1.65]) or IRF (HR=1.49[1.40-1.59]), having prior PCP (HR=1.23[1.17-1.30]), dual eligibility (HR=1.11[1.04-1.18]), and prior TBI diagnosis (HR=1.05[1.01-1.10]) were associated with increased risk of 90-day readmission. Female sex and “other” race were associated with decreased risk of 90-day readmission.

CONCLUSIONS:

Most older adults with TBI return to the community following hospital discharge. Disparities exist in returning to the community and in risk of 90-day readmission following hospital discharge. Future studies should explore how having a PCP influences post-hospital outcomes in chronic care management of older patients with TBI.

INTRODUCTION

A sizable portion of traumatic brain injury (TBI) related emergency department visits, hospitalizations, and deaths in the US occur in older adults.^1,2 Accidental falls are the most common cause of TBI among older adults, followed by motor-vehicle accidents and being struck by or against an object.³ Following TBI, cognitive, physical, and psychological symptoms or limitations may last for several years.^4–6 Compared to younger adults, older age is associated with longer-lasting symptoms and poorer outcomes.⁷ Older adults are also at increased risk of mortality following TBI.⁸

Following acute hospitalization, patients with TBI may be discharged (a) directly home with or without services, or receiving additional medical services in an (b) inpatient rehabilitation facility (IRF), (c) skilled nursing facility (SNF), (d) short-stay hospital, (e) long-term nursing home (NH), or (f) hospice facility.⁹ In the US, older adults with TBI are more likely than younger patients to be transferred to a post-acute care facility following the initial hospitalization.^10–12 Older patients with TBI who are socioeconomically disadvantaged, have functional impairments, or live in rural areas are less likely to return to the community following the acute hospitalization and subsequent SNF care.^13,14 Older adults with TBI are more likely to be readmitted following acute care hospitalization,^15,16 especially those transferred to IRFs^17,18 or SNFs.^13,17 Although they experience slower functional gains over shorter lengths of stay than younger adults,^3,19 older adults with TBI benefit from receiving rehabilitation and care supporting their return to home.

Following hospitalization, most patients desire to return to the security of their home, which is often viewed as a milestone or achievement.^20–22 Accordingly, it is important to explore whether patients can return home at some point following their hospitalization. Therefore, the objectives of the study were to: a) describe residential status of older patients in the first 20 weeks following TBI, b) identify predictors of community residence at 90 days (primary outcome), and c) identify predictors of readmission at 90 days (secondary outcome).

METHODS

The Institutional Review Board of the University of Texas Medical Branch at Galveston approved this study with a waiver of informed consent. The Centers for Medicare & Medicaid Services (CMS) approved our Data Use Agreement to obtain the Medicare files. The methods and results followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines.²³

Data Source

We used 100% Texas Medicare fee-for-service claims data files from January 1, 2013 through December 31, 2018 to capture 12 months of baseline data prior to admission and follow-up data for patients admitted between 1/1/2014 and 12/31/2017. We used the Master Beneficiary Summary Files to obtain enrollment information and demographics. We also used the Medicare Provider Analysis Review (MedPAR) files to identify admissions to short-term and long-term acute care hospitals, IRFs, and SNFs facilities, and discharges to hospice. Outpatient Standard Analytic files and Carrier files were used to identify any TBI diagnosis in outpatient settings the year prior to hospital admission. The Minimum Data Set v3.0 (MDS) was used to in conjunction with MedPAR claims to identify long-term NH stays.²⁴

Cohort Selection

We identified 64,277 Medicare fee-for-service beneficiaries with a primary or secondary TBI discharge diagnosis in January 1, 2014 through December 31, 2017 (Figure 1). TBI was defined using International Classification of Diseases, Ninth and Tenth Revisions, Clinical Modification codes (ICD-CM-10: S01, S02.0, S02.1, S02.3, S02.7-S02.9, S04.0, S06, S07.0, S07.1, S07.8, S07.9, S09.7-S09.9, T01.0, T02.0, T04.0, T06.0, F078.1 or ICD-CM-9: 800-801, 803-804, 850-854, 950.00-950.3, 959.01). Index hospitalization was defined as the first TBI-related hospitalization during the study period (n=53,934). Patients aged 66 years old or older at admission, with continuous part A and B enrollment and without Medicare Advantage (HMO) for at least 12 months prior to admission and 90 days after discharge were included (n=30,485).We used a one-year baseline period before the index hospitalization to identify comorbid conditions, a primary care provider (PCP),²⁵ and TBI outpatient diagnoses. Since our cohort only included TBI hospitalizations, we cannot determine how many patients had an outpatient TBI diagnosis without a TBI hospitalization. We excluded patients hospitalized with a TBI in the 12 months prior to the hospital admission, those who died during hospitalization, and those discharged to hospice, resulting in 26,985 patients in the final analytical sample.

Figure 1.

Flowchart of Cohort Selection.

Residential Status, Community Residence, and Readmission Outcomes after Discharge

Residential status was determined daily for a period of 20 weeks after hospital discharge, which is when the trajectory of residential status reached a plateau. For each patient, we identified all acute hospital admission and discharge dates during the follow-up period using inpatient claims. Using these inpatient admission and discharge dates, we then created time intervals in days for each inpatient episode of care within the follow-up period. For each time interval, we determined the residential status based on the claim type (e.g., hospital, IRF, SNF). The MDS was used to ascertain long-term NH stays.²⁴ Over the 20-week follow-up period (i.e., 140 days post-index event acute care hospital discharge), we assumed that, at any point, if the patients were not in any inpatient or NH facility, this meant they resided in the community. With this method, “community” included assisted living and similar unlicensed facilities. To strengthen this assumption, patients were followed for up to 12 months after discharge or as long as they had Medicare Part A and B coverage without HMO. The follow-up was censored if Medicare coverage ended, the patient died, or if discharged to hospice. For the primary analysis, we then explored whether the patient with TBI resided in the community at 90 days following hospital discharge. While the secondary analysis explored 90-day all-cause hospital readmission.

Demographic and Clinical Factors

The following demographic and clinical factors were included in the analysis: Age (66-69, 70-74, 75-79, 80-84, 85-89, 90+ years), Sex (male, female), and Race and ethnicity defined by the Research Triangle Institute (RTI) race code (non-Hispanic white, Black or African-American, Hispanic, Asian-Pacific Islander, American Indian/Alaska Native, Other, Unknown),²⁶ year of admission, residence prior to admission (community, NH), having a PCP prior to admission,²⁵ Medicaid eligibility (yes, no), original reason for original Medicare entitlement [age or disability/end stage renal disease (ESRD)], prior TBI outpatient diagnosis (yes, no), TBI as the primary admission diagnosis (yes, no), length of stay (days, by quartile), Elixhauser weighted comorbidity score (quartile), and the Injury Severity Score (RISS) (quartile). The Injury Severity Score is a trauma injury severity score calculated using the three most severely injured regions out of the six body regions (head or neck, face, chest, abdomen and pelvic contents, extremities, pelvic bones, and general).²⁷ Each region is scored from 0 (minor) to 6 (untreatable/death), then squared and summed. If any region is a 6, the total score automatically is 75. The total score ranges from 0 to 75. The RISS is calculated based on ICD-9-CM and ICD-10-CM injury diagnosis codes and R statistical software (ICDPIC-R).²⁸ RISS scores can be interpreted as mild (<9), moderate (9-15), severe (16-24), and profound (≥25).²⁹ Initial discharge destination was categorized as community, SNF, IRF, or other. The other discharge destinations included transfer to either a short-stay or long-term acute care hospital.

Statistical Analysis

We calculated descriptive statistics of the sample’s demographic and clinical factors. We then tested the association of discharge destination with sample characteristics using X² analyses. We estimated unadjusted and adjusted odds ratios (aORs) of residing in the community 90 days following discharge using binary logistic regression models. Multivariable logistic regression models were adjusted for the demographic and clinical factors to determine the predictors of residing in the community 90 days following discharge, except for comorbidities. We decided to use injury severity scores rather than the Elixhauser comorbidity score, since the former is more predictive of post-hospital outcomes among older adults with TBI.³⁰ For this analysis, we excluded patients if they lost Medicare coverage, died, or were discharged to hospice at any time prior to 90 days following hospital discharge.

Fine and Gray’s competing risk models were used to determine the factors associated with risk of all-cause acute hospital readmission within 90 days from hospital discharge.³¹ Death and discharge to hospice were treated as competing events. Patients were censored if they lost Medicare coverage, died, were discharged to hospice, or reached the end of the 90-day follow-up period after discharge. For this analysis and following CMS all-cause unplanned readmission measures (i.e., short-term acute care hospital), we excluded patients who were transferred to another acute care hospital or long-term NH upon discharge. All analyses were performed using SAS software (version 9.4., SAS Institute, Cary, NC).

RESULTS

Demographics and Clinical Characteristics

The final sample included 26,985 Medicare fee-for-service patients aged 66 and older who were admitted to an acute care hospital for TBI in Texas from January 1, 2014 to December 31, 2017 (Figure 1). The demographic and clinical characteristics are described in Table 1. Most subjects were ≥ 75 years of age (73.3%), female (57.3%), Medicare-eligible because of age (89.5%), not dual eligible (79.2%), had a PCP prior to admission (66.4%) and lived in the community prior to admission (95.6%). Patients classified as Hispanic, Black/African American, and “other” race made up 21% of the sample. The RTI “other” race category included Asian/Pacific Islander, American Indian/Alaska Native, other races, and unknown race. Half of the sample had a primary admission diagnosis of TBI (51.8%); 41% had a prior TBI outpatient diagnosis in the year prior to admission. Most patients stayed in the acute hospital 1-7 days (76.9%), median 4 days. Nearly half of the patients had a moderate-to-severe trauma injury severity score (48.7%).

Table 1.

Descriptive statistics.

Characteristic		Sample, N (%)
Total		26,985 (100)
Age group	66-69	2,889 (10.7)
	70-74	4,328 (16.0)
	75-79	4,908 (18.2)
	80-84	5,699 (21.1)
	85-89	5,233 (19.4)
	90+	3,928 (14.6)
Sex	Female	15,454 (57.3)
	Male	11,531 (42.7)
RTI Race *	Black	1,206 ( 4.5)
	Hispanic	3,787 (14.0)
	White	21,278 (78.9)
	Other	714 ( 2.6)
	Asian/Pacific Islander	467 (1.7)
	American Indian/ Alaska Native	57 (0.2)
	Other	113 (0.4)
	Unknown	77 (0.3)
Year of admission	2014	5,009 (18.6)
	2015	5,790 (21.5)
	2016	7,885 (29.2)
	2017	8,301 (30.8)
Prior Residence	Community	25,792 (95.6)
	Nursing Home	1,193 ( 4.4)
Prior PCP	No	9,062 (33.6)
	Yes	17,923 (66.4)
Medicaid	No	21,359 (79.2)
	Yes	5,626 (20.8)
Original Entitlement	Age	24,148 (89.5)
	Other	2,837 (10.5)
Prior TBI DX (outpatient)	No	15,915 (59.0)
	Yes	11,070 (41.0)
TBI primary DX	No	13,020 (48.2)
	Yes	13,965 (51.8)
LOS, quartiles (days)	1st (1, 2)	7,119 (26.4)
	2nd (3, 4)	8,115 (30.1)
	3rd (5, 7)	5,512 (20.4)
	4th (8, 128)	6,239 (23.1)
Elixhauser weighted comorbidity score, quartiles	1st (−14, 0)	8,108 (30.0)
	2nd (1, 6)	6,284 (23.3)
	3rd (7, 14)	5,990 (22.2)
	4th (15, 67)	6,603 (24.5)
Injury Severity Score (RISS), quartiles	1st (0, 3)	8,087 (30.0)
	2nd (4, 9)	5,751 (21.3)
	3rd (10, 15)	7,014 (26.0)
	4th (16, 66)	6,133 (22.7)

Note. DX: Diagnosis; LOS: Length of Stay; PCP: Primary care provider; TBI: Traumatic Brain Injury; RISS: Injury Severity Score using R software. Elixhauser weighted comorbidity score and RISS quartiles increase signify more comorbidities and more severe bodily injury, respectively.

^*RTI Race code includes both race and ethnicity.

Discharge Destination

Of eligible patients, 1,885 died during hospitalization and 1,531 were discharged to hospice. Of the 26,985 hospitalized patients who survived, 12,387 (45.9%) were discharged directly to the community, 8,199 (30.4%) were discharged to a SNF, 4,547 (16.9%) were discharged to an IRF, and 1,852 (6.9%) were discharged to another hospital or long-term care NH. All demographic and clinical characteristics differed significantly by discharge destination (i.e., community, SNF, IRF, other; Supplementary Table S1). A greater proportion of younger patients aged 66-69 (60%), Hispanic patients (48.7%), patients of “other” races (51.8%), males (48.2%), those with milder injury (56.2%), and those with fewer comorbidities (52.2%) were discharged to the community.

Community Residence after Hospital Discharge

Figure 2 shows where the patients resided after acute hospital discharge. Slightly fewer than half of the patients were living in the community immediately following discharge. By week 3, roughly 60% were living in the community. By 90 days following hospital discharge, 80% resided in the community. By week 20 (i.e., 140 days post-discharge), 84% were living in the community. The percentage of patients in a SNF steadily declined, from roughly 30% at the time of index hospital discharge to 5% at 20 weeks following discharge. About 15% of older adults with TBI were in an IRF immediately after the index hospitalization, which declined to 2-3% by week 3 (day 21 post-discharge). Lastly, there was a steady increase in the percentage of patients in a long-term care NH, which comprised roughly 10% of the patients during weeks 15 to 20. Supplementary Tables S2 and S3 include the residential status at 90 days and 140 days, respectively, by the residence type prior to admission.

Figure 2. Residential Status for the First 140 days after Discharge.

The denominator for the calculation of percentages included patients who were alive, had Medicare coverage, and were not discharged to hospice at each time-point. A total of 641 patients were excluded from the denominator because of loss of coverage by the end of follow-up.

Table 2 presents the unadjusted rates and aOR of patients with TBI living in the community at 90 days after hospital discharge, by patient characteristics. Female sex was associated with higher adjusted odds of residing in the community (aOR = 1.16; 95% Confidence Interval [CI], 1.08-1.25), as well as Hispanic ethnicity (aOR = 2.01; 95% CI, 1.80-2.25), and being classified as “other” race (aOR = 2.19; 95% CI, 1.73-2.77). Patients who had a PCP prior to hospitalization were more likely to reside in the community at 90 days after hospital discharge (aOR = 1.51; 95% CI, 1.40-1.62). Patients aged 75-79 (aOR = 0.65; 95% CI, 0.56-0.76), 80-84 (aOR = 0.57; 95% CI, 0.49-0.65), 85-89 (aOR = 0.47; 95% CI, 0.40-0.54), and 90+ years (aOR = 0.36; 95% CI, 0.31-0.41) had decreased odds of community residence compared to those aged 66-69. Patients who previously resided in a NH (aOR = 0.09; 95% CI, 0.07-0.10), were dual eligible (aOR = 0.16; 95% CI, 0.15-0.18), were previously diagnosed with a TBI in an outpatient setting (aOR = 0.87; 95% CI, 0.81-0.94), or who had a more severe trauma injury scores (RISS 16-66; aOR = 0.82; 95% CI, 0.75-0.91) had decreased odds of residing in the community at 90 days following discharge.

Table 2.

Adjusted Odds Ratios for Residing in the Community 90 days after Discharge.

Characteristic	Sample N (%)	Unadjusted Rate (%)	Adjusted OR (95% CI)
Age group
66-69	2,605 (11.3)	2,214 (85.0)	REF
70-74	3,842 (16.7)	3,250 (84.6)	0.86 (0.74 - 1.01)
75-79	4,340 (18.9)	3,524 (81.2)	0.65 (0.56 - 0.76)
80-84	4,858 (21.1)	3,843 (79.1)	0.57 (0.49 - 0.65)
85-89	4,302 (18.7)	3,320 (77.2)	0.47 (0.40 - 0.54)
90+	3,038 (13.2)	2,161 (71.1)	0.36 (0.31 - 0.41)
Sex
Male	9,545 (41.5)	7,668 (80.3)	REF
Female	13,440 (58.5)	10,644 (79.2)	1.16 (1.08 - 1.25)
RTI Race*
White	18,113 (78.8)	14,583 (80.5)	REF
Black	1,007 ( 4.4)	712 (70.7)	1.12 (0.95 - 1.33)
Hispanic	3,241 (14.1)	2,509 (77.4)	2.01 (1.80 - 2.25)
Other†	624 ( 2.7)	508 (81.4)	2.19 (1.73 - 2.77)
Year of admission
2014	4,250 (18.5)	3,288 (77.4)	REF
2015	4,871 (21.2)	3,832 (78.7)	0.99 (0.88 - 1.10)
2016	6,743 (29.3)	5,392 (80.0)	1.04 (0.93 - 1.15)
2017	7,121 (31.0)	5,800 (81.4)	1.15 (1.03 - 1.28)
Prior Residence
Community	22,112 (96.2)	18,163 (82.1)	REF
Nursing Home	873 ( 3.8)	149 (17.1)	0.09 (0.07 - 0.10)
Prior PCP
No	7,534 (32.8)	5,464 (72.5)	REF
Yes	15,451 (67.2)	12,848 (83.2)	1.51 (1.40 - 1.62)
Medicaid
No	18,273 (79.5)	15,794 (86.4)	REF
Yes	4,712 (20.5)	2,518 (53.4)	0.16 (0.15 - 0.18)
Prior TBI DX (outpatient)
No	13,614 (59.2)	11,062 (81.3)	REF
Yes	9,371 (40.8)	7,250 (77.4)	0.87 (0.81 - 0.94)
Injury Severity Score (RISS)
1st (0, 3)	6,933 (30.2)	5,553 (80.1)	REF
2nd (4, 9)	4,927 (21.4)	3,938 (79.9)	0.99 (0.89 - 1.09)
3rd (10, 15)	5,907 (25.7)	4,688 (79.4)	0.93 (0.85 - 1.03)
4th (16, 66)	5,218 (22.7)	4,133 (79.2)	0.82 (0.75 - 0.91)

Note. The analysis included patients who were alive, had Medicare coverage and were not discharged to Hospice 90 days after discharge (N = 22,985).

OR: Odds Ratio; CI: Confidence Interval; DX: Diagnosis; PCP: Primary care provider; TBI: Traumatic Brain Injury; RISS: Injury Severity Score using R software.

^*RTI Race code includes both race and ethnicity.

^†The RTI “other” race category included patients classified as Asian/Pacific Islander race, American Indian/Alaska Native race, other races, and unknown race.

90-day Hospital Readmission following Discharge

Patients were censored at follow-up due to Medicare coverage ending (n=523), death (n=2,501), or discharged to hospice (n=976). A total of 7,582 patients (29.3%) were readmitted within 90 days of discharge (Supplementary Figure S1). The unadjusted rates and adjusted hazard ratios for readmission are found in Table 3. Non-Hispanic Black patients had a 37.7% 90-day readmission rate, compared to 28.9% for non-Hispanic white patients (HR=1.33; 95% CI, 1.20-1.46). Those discharged to a SNF or IRF had a 34.5% and 33.7% 90-day readmission rate, respectively, compared to those discharged directly home (24.4%); HR 1.56; 95% CI, 1.48-1.65 and HR 1.49; 95% CI, 1.40-1.59, respectively. Patients who were dual eligible had a 90-day readmission rate of 31.9% compared to 28.6% for those who had Medicare only (HR = 1.11; 95% CI, 1.04-1.18). In addition, those previously diagnosed with a TBI in an outpatient setting had a greater rate of 90-day readmission compared to those without (HR = 1.05; 95% CI, 1.01-1.10). Patients with RISS scores of 4-9 (HR = 0.90; 95% CI, 0.84-0.96) and 16-66 (HR = 0.92; 95% CI, 0.86-0.98) were less likely to be readmitted compared to those with milder injury RISS scores (0-3). Those of female sex (HR = 0.80; 95% CI, 0.76-0.83) and “other” race (HR = 0.85; 95% CI, 0.73-0.99) were less likely to be readmitted at 90 days compared to males and non-Hispanic white patients, respectively.

Table 3.

Adjusted Hazard Ratios Derived from Modeling Time to First Hospital Readmission within 90 days after Discharge.

Characteristic	Sample (n=25,905)	Unadjusted rate for 90-day readmission	Adjusted HR (95% CI)
Age group
66-69	2,625 (11.6)	763 (27.9)	REF
70-74	3,843 (17.0)	1,178 (28.7)	1.01 (0.92 - 1.11)
75-79	4,207 (18.6)	1,433 (30.7)	1.06 (0.97 - 1.15)
80-84	4,757 (21.1)	1,617 (29.6)	0.99 (0.91 - 1.08)
85-89	4,230 (18.7)	1,499 (29.6)	0.98 (0.90 - 1.07)
90+	2,923 (12.9)	1,092 (28.2)	0.93 (0.85 - 1.03)
Sex
Male	9,558 (42.3)	3,492 (32.0)	REF
Female	13,027 (57.7)	4,090 (27.3)	0.80 (0.76 - 0.83)
RTI Race *
White	17,885 (79.2)	5,920 (28.9)	REF
Black	942 ( 4.2)	429 (37.7)	1.33 (1.20 - 1.46)
Hispanic	3,147 (13.9)	1,060 (29.4)	0.99 (0.92 - 1.06)
Other†	611 ( 2.7)	173 (25.7)	0.85 (0.73 - 0.99)
Year of admission
2014	4,081 (18.1)	1,405 (29.3)	REF
2015	4,752 (21.0)	1,629 (29.5)	0.99 (0.92 - 1.07)
2016	6,663 (29.5)	2,194 (29.0)	0.96 (0.90 - 1.03)
2017	7,089 (31.4)	2,354 (29.3)	0.99 (0.92 - 1.05)
Prior Residence
Community	22,218 (98.4)	7,187 (29.0)	REF
Nursing Home	367 ( 1.6)	395 (34.4)	1.12 (1.00 - 1.25)
Destination
Home	12,387 (47.8)	3,019 (24.4)	REF
SNF	8,199 (31.7)	2,831 (34.5)	1.56 (1.48 - 1.65)
IRF	4,547 (17.6)	1,531 (33.7)	1.49 (1.40 - 1.59)
Nursing Home	772 ( 3.0)	201 (26.0)	1.06 (0.91 - 1.24)
Prior PCP
No	6,938 (30.7)	2,291 (26.4)	REF
Yes	15,647 (69.3)	5,291 (30.7)	1.23 (1.17 - 1.30)
Medicaid
No	19,015 (84.2)	5,873 (28.6)	REF
Yes	3,570 (15.8)	1,709 (31.9)	1.11 (1.04 - 1.18)
Prior TBI DX (outpatient)
No	13,490 (59.7)	4,346 (28.6)	REF
Yes	9,095 (40.3)	3,236 (30.2)	1.05 (1.01 - 1.10)
Injury Severity Score (RISS)
1st (0, 3)	6,929 (30.7)	2,329 (29.8)	REF
2nd (4, 9)	4,809 (21.3)	1,557 (27.9)	0.90 (0.84 - 0.96)
3rd (10, 15)	5,799 (25.7)	2,049 (30.3)	0.98 (0.92 - 1.04)
4th (16, 66)	5,048 (22.4)	1,647 (28.6)	0.92 (0.86 - 0.98)

Note. Patients were excluded from the analysis if they were transferred to another hospital upon discharge (n=1,080).

HR: Hazard Ratio; CI: Confidence Interval; DX: Diagnosis; LOS: Length of Stay; PCP: Primary care provider; RISS: Injury Severity Score using R software; TBI: Traumatic Brain Injury.

^*RTI Race code includes both race and ethnicity.

^†The RTI “other” race category included patients classified as Asian/Pacific Islander race, American Indian/Alaska Native race, other races, and unknown race.

DISCUSSION

In this retrospective cohort study using 100% Texas Medicare claims, we provide the clinical course of TBI following hospital discharge in older adults. The majority of older adults with TBI in Texas who survived hospitalization were able to return to the community at 90 days following hospital discharge. We identified a number of social determinants of health (e.g., age, sex, ethnicity, race, socioeconomic status, and access to primary care) that were associated with both community residence and all-cause readmission at 90 days following discharge. Older patients with TBI who received post-acute care prior to community residence had greater readmission rates than those discharged directly home. Overall, these findings expand our understanding of post-hospitalization outcomes among older adults and aligns with the bio-psycho-socio-ecological model on improving TBI care.³² Our findings may be used to guide chronic care management and the development of high-quality transitional care programs to improve outcomes of older adults with TBI.

We found that institutional care patterns varied over the 20-week follow-up period, but was consistent with average length of stay and Medicare payment policies. For example, the observed drop in SNF utilization by week three was likely associated with Medicare’s coinsurance policy where costs to beneficiaries increase from $0 to $194 coinsurance per day on day 21 of SNF care. By 90 days and 20 weeks after discharge, we also found that 7% and 10% of patients, respectively, were institutionalized in a long-term care NH. This rate is within the 7-39% rate for stroke patients.³³ Ultimate placement in a long-term care NH after hospitalization is influenced by both the quality of the hospital and the quality of post-acute care services,^34,35 and other post-TBI risk factors, such as older age, living alone, fall risk or safety concerns, Black race, less education, motor-vehicular injuries, baseline cognitive and functional impairments.^36–38

Like previous studies,³⁹ racial and ethnic disparities in post-injury outcomes were identified. Hispanic patients were more likely to reside in the community compared to non-Hispanic white patients, but did not differ based on readmission rates. Regardless of age, Hispanic patients with TBI are less likely to receive post-injury rehabilitation services compared to other ethnic and racial groups.^40–42 While patient preference and cultural values may influence post-acute care use, other individual (e.g., language barriers), extended familial support, geographical (e.g., limited post-acute care provider supply), or systemic factors (e.g., racism/discrimination) may be considered.^40,42

Black Medicare patients with TBI had greater 90-day readmission rates compared to non-Hispanic white patients, which is consistent with other studies.^43,44 Our findings suggest the need to better address the medical needs of older Black patients with TBI to reduce their risk of readmission. Patients of “other” races were more likely to reside in the community and had lower 90-day all-cause readmission rates compared to non-Hispanic white patients. The “other” race category was small (2.6% of the total) and was comprised mostly of Asian/Pacific Islander patients. More research is needed to explore the post-TBI outcomes of older Asian/Pacific Islander and American Indian/Alaska Native adults.

Patients aged 75 years and older were less likely to reside in the community compared to those aged 66-69, but differences in readmission rates by age were not identified. However, we did identify disparities by sex. Female patients were more likely to reside in the community and had lower readmission rates than male patients at 90 days following hospital discharge. Other studies suggest that males with TBI are more likely to be discharged to the community than females with TBI.⁹ Some studies suggest females do better and others suggest poorer outcomes compared to males with TBI.⁴⁵ Future work should further examine the sex differences in readmission rates after TBI among older adults.

Access to primary care and health status appear to influence post-TBI health outcomes. Having a PCP prior to hospitalization was associated with increased likelihood of community residence and readmission. Older patients who regularly saw a PCP prior to hospitalization, received post-acute care, prior TBI diagnosis, and dual eligible had higher readmission rates, consistent with other studies.^15,16 Older adults, particularly those who are vulnerable and socio-economically disadvantaged, may have greater illness, inadequate treatment of coexisting medical conditions, recurrent TBI, inadequate medication management, or lack of or poor follow-up and continuity of care that negatively influence outcomes.^46,47 Future work should explore the role of follow-up care and continuity of care by PCPs with reducing adverse outcomes in older adults who need transitional care. Efforts to improve transitional care could help reduce readmission rates, particularly among those who are sicker and functionally dependent. Patients with greater injury severity compared to those with milder injuries had lower rates of readmission, which could be influenced by the increased likelihood of mortality among those with greater injury severity scores.

Limitations and Strengths

Limitations of our findings must be noted. Although we were able to account for injury severity (i.e., RISS scores), administrative claims data do not account for functional status,⁴⁸ presence of a caregiver, and other social determinants of health associated with post-discharge outcomes. We defined TBI broadly using ICD-9 and ICD-10 codes representing a wide spectrum of injury severities, which may underreport or misdiagnose TBI. We cautiously included S09 (8.1% as primary diagnosis), which are other and unspecified head injuries. Exclusion of S09 may result in missed cases, while its inclusion may result in false positives.⁴⁹ We also could not determine if patients without inpatient or outpatient Medicare claims (n=149) moved out of Texas during the follow-up period; we assumed that they resided in the community. We also do not have information on those with Medicare Advantage, which represent an increasing number of eligible Medicare patients. In addition, our data are limited to Texas and the findings may not be the same for other states or nationally. For example, among states that reported data to CMS, Texas reported spending 61% of their Managed care Long-term Support Services (MLTSS) on institutional and home and community-based services, which must be considered when interpreting the results. Future studies should explore whether percentage of community residence in Texas differs from states with lower MLTSS spending.

Despite these considerations, this study contributes to our understanding of care trajectories and quality of care (i.e., community residence and readmission) following TBI among older adults. A unique feature of this study is that we were able to follow longitudinally a population-based sample of older adult patients with TBI over 20 weeks following hospitalization, as well as identify factors associated with community residence following discharge. Previous studies have either examined residential status at longer time points (e.g., 1-, 2-, or 5-years following injury), focused on working adults, or following inpatient rehabilitation.^37,50 In addition, we were able to examine primary care utilization and comorbidities prior to the hospitalization, an analysis which was absent from the TBI literature.

CONCLUSION

Older adults receive a variety of types of institutional care following acute hospitalization for TBI. Most patients with TBI who survive following discharge reside in the community, but some older adults with TBI experience continued institutionalization. Access to primary care by older adults with TBI was associated with post-hospitalization outcomes. Future studies should examine the role of continuity of care in primary care to explore its role in chronic care management and influence on post-hospital outcomes of older patients with TBI.

Supplementary Material

Supplementary File

Supplementary Figure S1. Cumulative Incidence of First Hospital Admission after Discharge (n = 25,905). 1,080 subjects were excluded from the analysis because they were transferred to another hospital upon discharge. Subjects without any hospital readmission were censored at the time of loss of Medicare coverage (n=523), discharged to hospice (n=976), death (n=2,501), or at the end of the 90-day follow-up period.

IMPACT Statement:

We certify that this work is novel. This is the first longitudinal cohort study to explore factors associated with being alive and residing in the community at 90 days after hospital discharge for traumatic brain injury among older adult Medicare beneficiaries.

Key Points.

• 80% of Medicare fee-for-service beneficiaries with traumatic brain injury (TBI) discharged alive from Texas hospitals resided in the community at 90 days after hospital discharge, while 7% of the beneficiaries were in a long-term care nursing home and 12.5% of the beneficiaries had died.

• Of those who survived, female sex, Hispanic ethnicity, other race and ethnicity, and having a primary care provider were associated with increased likelihood of residing in the community at 90 days after hospital discharge for TBI.

Why Does this Paper Matter?

Older adults with TBI is a growing population, and the ability to return to the community following hospital discharge is a quality indicator and patient-centered goal.