Postacute Discharge Destination and Major Adverse Cardiovascular Events Among Patients With Intracerebral Hemorrhage

Abstract

BACKGROUND:

Studies evaluating health system factors associated with major adverse cardiovascular events (MACE) among intracerebral hemorrhage (ICH) survivors are lacking. We evaluate differences in MACE incidence across postacute ICH care settings—inpatient rehabilitation facilities (IRF), home, or skilled nursing facilities (SNF).

METHODS:

Using data from Florida, New York, Maryland, Washington, and Georgia, we identified adult ICH survivors discharged to home, IRF, or SNF (April 2016–December 2018). Multivariable logistic models, adjusted for sociodemographic factors, treatment intensity, comorbidities, and frailty, estimated adjusted odds ratios (aORs) and 95% CI for the association between discharge disposition (IRF versus home; IRF versus SNF) and MACE (a composite of acute stroke, acute myocardial infarction, systemic embolism, and vascular death), recurrent ICH, acute ischemic stroke, acute myocardial infarction, vascular death, and all-cause mortality within 1 year. Cardiovascular outcomes were ascertained using International Classification of Diseases, Tenth Revision codes. We assessed interaction between age and discharge disposition, performing stratified analyses for patients <65 and ≥65 years when the interaction was significant.

RESULTS:

Among 58 591 patients with ICH (mean age [SD], 68.1 [16.0] years; 47.1% female), 17 647 ICH survivors discharged home (46.4%), to IRF (25.5%), or to SNF (28.1%) were included. Within 1 year, 1302 (7.4%) patients experienced MACE, with rates for recurrent ICH, acute ischemic stroke, acute myocardial infarction, vascular death, and mortality at 2.5%, 3.2%, 0.6%, 1.3%, and 3.5%, respectively. In fully adjusted models, patients discharged to IRF had significantly lower odds of MACE (versus home: aOR, 0.84 [CI, 0.71–0.98]; versus SNF: aOR, 0.79 [CI, 0.67–0.93]), with a significant discharge disposition-age interaction (P=0.047). In stratified analysis, IRF discharge (versus home) was only significantly associated with MACE in patients aged <65 years (aOR, 0.70 [CI, 0.54–0.92]), not in those aged ≥65 years (aOR, 0.94 [CI, 0.77–1.15]). Patients discharged to IRF had significantly lower odds of recurrent ICH (versus SNF: aOR, 0.60 [CI, 0.45–0.80]), vascular death (versus SNF: aOR, 0.70 [CI, 0.49–0.99]), and all-cause mortality (versus SNF: aOR, 0.63 [CI, 0.50–0.79]).

CONCLUSIONS:

IRF care (versus SNF and home) was associated with lower odds of MACE. Further research is needed to determine specific components of IRF care contributing to better outcomes.

Spontaneous intracerebral hemorrhage (ICH) accounts for 10% to 15% of all stroke cases in the United States but is associated with mortality rates ranging from 40% to 50% at 30 days to 54% at 1 year.^1,2 Owing to advances in acute neurological and critical care, ICH case fatality has declined in the United States during the past decade.^3,4 However, up to 80% of ICH survivors are left with moderate to severe disability, and there is emerging evidence that patients with ICH are at a high risk of subsequent major adverse cardiovascular events (MACE), including recurrent strokes, acute myocardial infarction (AMI), systemic embolism, and death.^5,6 Therefore, developing effective postacute transitional care systems that promote secondary prevention of MACE and maximize recovery in patients with ICH remains critically important.^7,8

The American Heart Association/American Stroke Association’s guidelines recommend discharge to an inpatient rehabilitation facility (IRF) for all eligible stroke survivors.^9,10 With the overall goal of improving functional outcomes and reintegration into the community, patients under IRF care receive intensive rehabilitation from a multidisciplinary team with specialized multimodal rehabilitation expertise, typically around 3 hours per day for at least 5 days a week. Bundled together, IRF care interventions could improve the functional independence of the patients with ICH, enhance their ability to commit to a physically active and healthy lifestyle, help them reintegrate into the community, and build their self-reliance. These interventions, therefore, may provide a framework to reduce the short and long-term risk of experiencing MACE by promoting better overall cardiovascular health, improving compliance with prescribed treatments, and encouraging lifestyle changes that mitigate the factors contributing to MACE.

Previous observational studies have shown that ischemic stroke survivors who receive postacute IRF care experience greater improvements in functional independence compared with patients who are discharged to skilled nursing facilities (SNF) and that poststroke functional independence is associated with a lower risk of stroke recurrence.^11–14 Also, prior studies, which are largely observational and conducted among patients with ischemic stroke, have demonstrated that IRF treatment and a higher intensity of rehabilitation are associated with a lower likelihood of readmission and stroke recurrence.^12,15,16 Furthermore, the initiation of secondary stroke prevention measures during acute care admission has yielded high treatment adherence rates after discharge.¹⁷ However, large population-based studies evaluating the association of IRF treatment with the likelihood of experiencing MACE among ICH survivors are lacking. In this study, we utilize large and racially diverse data from 5 US states to evaluate the association between IRF treatment and experiencing MACE among ICH survivors, compared with home or SNF discharge.

Methods

Per institutional review board policy, this study does not constitute human subject research, as all data sets used in the project are deidentified and publicly available from the Healthcare Cost and Utilization Project¹⁸ upon completion of requisite training and data use agreements. Therefore, this study is exempt from institutional review board review and informed consent requirements. We followed the STROBE guidelines (Strengthening the Reporting of Observational Studies in Epidemiology).¹⁹

Data Source, Study Design, and Case Identification

In this longitudinal observational cohort of prospectively collected claims data, we utilized the State Emergency Department Databases and State Inpatient Databases (SID) for Florida, New York, Maryland, Washington (SID only), and Georgia to identify all patients discharged with ICH as the principal diagnosis from April 2016 to December 2018.¹⁸ ICH was identified using the International Classification of Diseases, Tenth Revision code I61, which has 89.5% sensitivity and 97.7% positive predictive value.²⁰ The SID and State Emergency Department Databases administrative databases are maintained by the Agency for Healthcare Research and Quality under the Healthcare Cost and Utilization Project.¹⁸ The SID captures all statewide inpatient discharges, while the State Emergency Department Databases represents all patients who present to a hospital-affiliated emergency department across the state. The included states, spanning 3 distinct US census regions, were selected due to their diverse large patient discharge volumes and because they provide unique state-specific linkage variables assigned to patients based on encrypted person number, date of birth, and sex.²¹ If any of this information is missing, the linkage variables are set as missing. These linkage variables facilitate longitudinal tracking of patients over time, allowing for the identification of repeated visits to the emergency department and inpatient care, thereby enabling the identification of patients who experience the outcome measures over time. Patients aged <18 years and those with missing age and linkage variables were excluded. In addition, patients with missing length of stay were excluded to enable the identification of same-day events (outcome events registered within 24 hours of a previous discharge). In our analysis, same-day events were considered a continuation of the previous admission and, therefore, not regarded as new events. Likewise, patients with a prior ICH diagnosis were excluded to restrict our analytic sample to first-time occurrence of ICH, consistent with prior studies.^6,22 Furthermore, patients with a concurrent diagnosis of head trauma, arteriovenous malformation, intracranial aneurysm, or brain malignancy were also excluded to restrict our analysis to primary ICH cases. In addition, patients who died during the index ICH hospitalization were excluded from this analysis. Patient data from January to March 2016 was used to ensure a minimum of a 3-month look-back period to identify patients with a prior ICH diagnosis; SID and State Emergency Department Databases data from 2019 were utilized to allow for a full year of follow-up for all patients. Patients were followed up from the date of index ICH admission to the date of the first outcome event or until they were censored (date of death or 365 days after their index ICH).

Outcomes

The primary outcome is experiencing MACE (defined as any stroke, AMI, systemic embolism, or vascular death) within 1 year of the initial ICH event. Vascular deaths, which comprise deaths related to acute vascular events such as systemic embolism, extra-axial hemorrhage, revascularization procedures, thrombolysis, mesenteric ischemia, sudden cardiac death, and extracranial hemorrhage,^6,22 were only classified as 1-year MACE events if the acute vascular event occurred within 1-year post-ICH and resulted in mortality within 30 days of admission, consistent with prior studies.^6,22 Secondary outcomes include recurrent ICH, acute ischemic stroke (AIS), AMI, vascular death, and all-cause mortality within 1 year of initial ICH admission. In this study, only outcome events that resulted in hospitalization or emergency department visits were included, as the acute clinical nature of these events makes it highly likely that they would necessitate hospital-level care and thus be captured in administrative records.^23–25 The International Classification of Diseases, Tenth Revision codes used to identify the study’s outcomes are provided in Table S1. To improve accuracy, only principal discharge diagnosis codes were considered for all study outcomes.^20,26,27 Only the first event was considered in our analysis for patients experiencing the same event type more than once. Furthermore, patient migration is not tracked within Healthcare Cost and Utilization Project data. As a result, we are unable to ascertain outcomes for patients who relocate outside the study area. Consequently, patients who relocate to other states are effectively lost to follow-up.

Independent Variables and Covariates

The primary independent variable was patient discharge disposition, coded based on the discharge status data element entered on the UB-04 claim form. Discharge disposition was classified as home discharge (including patients discharged home under the care of an organized home health service organization in anticipation of receiving covered skilled care), discharge to IRF, and discharge to SNF. All other discharge types, including hospice, left against medical advice, and discharge to long-term and other short-term healthcare facilities, were excluded from the analysis. The frequency distribution of the excluded discharge destinations is provided in Table S2. Race and ethnicity data are provided in Healthcare Cost and Utilization Project databases as a single variable, with ethnicity given precedence over race. Accordingly, patients with Hispanic ethnicity were categorized as Hispanic regardless of race. Consequently, race and ethnicity were categorized as non-Hispanic White, non-Hispanic Black, Hispanic, Asian American and Pacific Islanders, and others (Native Americans and others). Insurance status was classified as Medicare, Medicaid, Private Insurance, and other insurance (including self-pay). Patient location was classified based on the urban-rural classification scheme for US counties developed by the National Center for Health Statistics. Counties in metropolitan areas with a population ≥1 million were classified as large metropolitan counties. In contrast, metropolitan counties of 50 000 to 999 999 population were classified as small metropolitan counties, and other counties not satisfying these preceding criteria were classified as nonmetropolitan/micropolitan (ie, rural counties). Income status was categorized based on the income quartile of the patient’s zip code. Patients with a comorbid history of AIS, AMI, or peripheral vascular diseases are considered to have a history of occlusive vascular diseases. Patients’ comorbidity burden was further assessed using a version of the Charlson Comorbidity Index that has been validated among patients with ICH.²⁸ Patient frailty was assessed using the hospital frailty risk score.²⁹

Statistical Analysis

We analyzed baseline sociodemographic and clinical characteristics as proportions for categorical variables and medians (interquartile range) for continuous variables. We conducted bivariate comparisons of the demographic characteristics of patients discharged to IRF, home, and SNF using the Kruskal-Wallis test for continuous variables and the χ² test for categorical variables. We report the unadjusted (crude) 1-year cumulative incidence rates (CIR) and 95% CI of each outcome.³⁰

We fit multivariable logistic regression models to estimate the adjusted odds ratio (aOR) and 95% CI of the associations of IRF discharge compared with either home discharge or SNF discharge (ie, IRF versus home, IRF versus SNF) for each study outcome (MACE, ICH recurrence, AIS, AMI, vascular death, and all-cause mortality). First, we fitted univariable logistic regression models with discharge disposition as the only independent variable. The univariable models were subsequently adjusted for age, sex, and race (minimally adjusted models). Finally, we developed fully adjusted models, which included a large range of clinical and sociodemographic variables known to be associated with postdischarge outcomes, including age, sex, race, insurance type, patient location, income quartile of patients’ zip code, anticoagulant use, past or current history of smoking, invasive treatments (invasive ventilation, tracheostomy, percutaneous endoscopic gastrostomy), comorbidities (hypertension, diabetes, hyperlipidemia, obesity, atrial fibrillation, and history of occlusive vascular diseases), length of stay in acute care, Charlson Comorbidity Index, and patient frailty.^6,22,29,31 We included all covariates regardless of statistical significance; thus, no variable selection procedures were applied. This approach aligns with methodological literature, which advocates for including all literature-identified, scientifically plausible covariates, particularly in exploratory regression analyses.³² The variance inflation factor was used to assess potential multicollinearity among covariates. We performed an additional analysis that included only patients whose ICH location was not classified as unspecified (I61.9). Furthermore, in a separate analysis, we compared patients discharged home with home health services and those discharged to IRF. The overall fitness of each model was evaluated using the Hosmer-Lemeshow and Pearson χ² goodness-of-fit tests. Observations with missing data for any of the included covariates accounted for <2% of the total analytic sample. We, therefore, used complete case analysis for multivariable modeling, consistent with recommended practices when missingness is low and unlikely to bias results.^33,34 All analyses were conducted in Stata version 18.³⁵

Stratified Analyses

Using the final main effects logistic models, we also assessed for potential interactions between age and discharge disposition and performed stratified analysis by age (<65 years, ≥65 years) if the interaction was statistically significant (P<0.05). The threshold of 65 years was selected to represent the age of Medicare eligibility. Furthermore, a stratified analysis was conducted among patients at or below the median age (<70 years) and those above the median age (≥70 years).

Results

Sociodemographic and Clinical Characteristics of the Study Cohort

Overall, we identified 67 715 total ICH discharges representing 58 591 unique patients with ICH. Among these, 17 647 individual patients with ICH (median age [interquartile range], 69 [57–79] years; 45.2% female; 55.5% non-Hispanic White; 23.2% non-Hispanic Black; 10.4% Hispanic; 4.9% Asian American and Pacific Islanders) who were discharged to IRF, home, or SNF were eligible for inclusion in this study (Figure 1). The major reasons for exclusion include discharge to destinations other than home, IRF, or SNF (n=9103; see details in Table S2), missing discharge destination (n=1322), in-hospital mortality during index ICH hospitalization (n=8096), presence of diagnosis codes for other causes of intracranial bleeding: intracranial aneurysm, head trauma, brain malignancy, and arteriovenous malformation (n=2661), prior history of ICH (n=1094), missing age and age <18 years (n=193), and missing length of stay (n=65; Figure 1). Table S3 provides details of the distribution of missing values for the analysis variables.

Figure 1.

Patient selection chart. FL indicates Florida; GA, Georgia; ICH, intracerebral hemorrhage; IRF, inpatient rehabilitation facilities; MD, Maryland; NY, New York; SID, State Inpatient Databases; SNF, skilled nursing facilities; and WA, Washington.

Among the eligible ICH survivors, 8194 (46.4%) were discharged home (with or without home health), 4506 (25.5%) were discharged to IRF, and 4947 (28.1%) were discharged to SNF. Compared with those discharged home, patients with ICH discharged to IRF and SNF were significantly older (Table 1). Furthermore, a greater proportion of patients with ICH discharged to IRF (versus home) were non-Hispanic White, Medicare-insured, and were more likely to have hypertension, diabetes, hyperlipidemia, obesity, atrial fibrillation, and a history of anticoagulant use. In addition, patients discharged to IRF (versus home) are significantly more likely to have received invasive treatments: invasive ventilation (18% versus 4.7%); tracheostomy (3.8% versus 0.7%); percutaneous endoscopic gastrostomy (9.5% versus 1.6%; Table 1).

Table 1.

Descriptive Sociodemographic and Clinical Characteristics of the Overall ICH Cohort and Bivariate Comparative Characteristics by Discharge Disposition

ICH survivors <65 years of age (versus those ≥65 years) were more predominantly male (62.8% versus 49.4%), non-Hispanic Black (35.6% versus 14.8%), non-Medicare–insured (87.5% versus 10.5%), and more likely to be discharged home (56.2% versus 39.8%; Table S4).

MACEs (Primary Outcome)

Overall, 1302 ICH survivors experienced MACE within 1 year of follow-up (CIR, 7.4%, [95% CI, 7.0–7.8]). The CIR (95% CI) among patients discharged to home, IRF, and SNF are 7.4% (95% CI, 6.9–8.0), 6.2% (95% CI, 5.5–6.9), and 8.4% (95% CI, 7.6–9.2), respectively (Figure 2). In the fully adjusted model, patients discharged to IRF had significantly lower odds of experiencing MACE compared with patients discharged home (aOR, 0.84 [95% CI, 0.71–0.98]) or to SNF (aOR, 0.79 [95% CI, 0.67–0.93]; Table 2; Table S5). However, there was a significant interaction between discharge disposition and age (P value for home versus IRF=0.047; Table S6), and thus, we repeated the statistical models after stratifying by age. Among the 7145 subjects aged <65 years, IRF discharge (versus home) is associated with significantly lower adjusted odds of experiencing MACE (aOR, 0.70 [95% CI, 0.54–0.92]), whereas among the 10 502 subjects ≥65 years, there is no significant difference in the odds of experiencing MACE between patients discharged to IRF and those discharged home (aOR, 0.94 [95% CI, 0.77–1.15]; Table S6; Figure S1). Similar findings were observed when a threshold value based on median age (<70 years versus ≥70 years) was used (Table S7). The overall mean variance inflation factor was 1.37, indicating minimal overall multicollinearity. In addition, all individual variables exhibited low variance inflation factors, with the highest being 1.94 for age, suggesting no severe collinearity for any single variable.

Figure 2.

Incidence rates with 95% CIs of study outcomes. AIS indicates acute ischemic stroke; AMI, acute myocardial infarction; ICH, intracerebral hemorrhage; IRF, inpatient rehabilitation facilities; MACE, major adverse cardiovascular events; and SNF, skilled nursing facilities.

Table 2.

Association Between Discharge Disposition and the Odds of Experiencing Study Outcomes Within 1 Year of Follow-Up

Secondary Outcomes

ICH Recurrence

A total of 435 (2.5% [95% CI, 2.2–2.7]) ICH survivors experienced ICH recurrence within 1 year of follow-up. The CIR (95% CI) among patients discharged to home, IRF, and SNF are 2.7% (95% CI, 2.2–3.0), 1.8% (95% CI, 1.4–2.2), and 2.8% (95% CI, 2.3–3.4), respectively (Figure 2). In the fully adjusted model, ICH recurrence is significantly lower among patients discharged to IRF, relative to patients discharged to SNF (0.60 [95% CI, 0.45–0.80]; Table 2). However, there was no significant interaction between age and discharge disposition.

AIS and AMI

Overall, 568 (3.2% [95% CI, 3.0–3.5]) ICH survivors experienced AIS within 1 year of follow-up. Among patients discharged to home, IRF, and SNF, the unadjusted CIRs (95% CI) of AIS are 3.2% (95% CI, 2.9–3.6), 3.0% (95% CI, 2.6–3.6), and 3.4% (95% CI, 2.9–3.9), respectively (Figure 2). There is no statistically significant difference in experiencing AIS between ICH survivors discharged to IRF and those discharged to home (aOR, 0.96 [95% CI, 0.77–1.21]) or to SNF (aOR, 1.03 [95% CI, 0.81–1.32]; Table 2). However, relative to recurrent ICH and AIS, fewer ICH survivors (n=114) experienced AMI (CIR, 0.7% [95% CI, 0.5–0.8]). The unadjusted CIR (95% CI) among patients discharged to home, IRF, and SNF are 0.8% (95% CI, 0.6%–1.0%), 0.5% (95% CI, 0.3%–0.7%), and 0.6% (95% CI, 0.4%–0.8%), respectively (Figure 2). There is no significant difference in experiencing AMI between patients discharged to IRF and those discharged to home (0.63 [95% CI, 0.37–1.06]) or to SNF (0.87 [95% CI, 0.49–1.55]; Table 2). Notably, there was no significant interaction between age and discharge disposition for both AIS and AMI outcomes.

Vascular Death and 1-Year All-Cause Mortality

Overall, 232 (1.3% [95% CI, 1.2–1.5]) experienced vascular death within 1 year. The rate among patients discharged to home, IRF, and SNF are 0.87% (0.7%–1.1%), 1.2% (0.9%–1.5%), and 2.2% (1.8%–2.7%), respectively (Figure 2). Discharge to IRF (versus SNF) was significantly associated with lower odds of vascular death within 1 year (aOR, 0.70 [95% CI, 0.49–0.99]; Table 2).

A total of 619 (3.5% [95% CI, 3.2–3.8]) patients with ICH discharged alive died within 1 year of follow-up. The 1-year all-cause mortality rates (95% CI) among patients discharged to home, IRF, and SNF were 2.4% (95% CI, 2.1%–2.8%), 2.66% (95% CI, 2.21%–3.18%), and 6.1% (95% CI, 5.5%–6.8%), respectively (Figure 2). Although there was no significant difference in all-cause mortality between patients discharged home and those discharged to IRF in a fully adjusted model, patients discharged to IRF (versus SNF) had significantly lower odds of 1-year mortality (aOR, 0.63 [95% CI, 0.50–0.79]; Table 2). However, the interaction between age and discharge disposition was not statistically significant.

Model estimates from the unadjusted, minimally adjusted, and fully adjusted multivariable models, as well as model goodness-of-fit statistics, are provided in Table S5; Tables S8 through S13. Tables S14 and S15 present sensitivity analyses incorporating ICH location as a covariate (Table S14) and a restricted analysis comparing IRF discharge to home discharge with home health services (Table S15).

Discussion

In this analysis, we utilized emergency department and hospitalization data of 17 647 ICH cases across a contemporary 4-year time period from 5 large and ethno-racially diverse US states to demonstrate that about 1 in 14 ICH survivors experienced MACE within 1 year of follow-up. Furthermore, patients with spontaneous ICH discharged to IRF (versus home or SNF) had significantly lower odds of experiencing MACE. In addition, patients discharged to IRF (versus SNF) had significantly lower odds of experiencing important vascular outcomes, including ICH recurrence and vascular death within 1 year of follow-up. These outcomes represent clinically meaningful vascular events, which are potentially life-threatening and can lead to prolonged hospitalization, and are major drivers of healthcare costs and patient burden.^36,37 However, we also observed that the association between discharge disposition and MACE outcome varied significantly by age, with IRF discharge being linked to lower odds of MACE among those <65 years but not those aged ≥65 years.

Previous studies have shown that treatment in IRFs, compared with SNFs, is associated with a range of positive outcomes for patients recovering from stroke, including greater functional recovery, improved physical mobility and self-care function, lower readmission rates, and a higher likelihood of discharge to the community.^{11–14,38,39} Our study extends these findings by demonstrating that even when compared with patients discharged home (with or without home health services), patients treated in IRF had significantly lower odds of experiencing MACE. However, it is imperative to acknowledge that our study and the prior ones are observational in nature and, therefore, vulnerable to unmeasured confounding. Nonetheless, these findings provide valuable insights into the potential benefits of IRF care and highlight the need for further research, particularly randomized controlled trials, to comprehensively assess the impact of IRF interventions on secondary prevention of adverse cardiovascular events among ICH survivors.

Furthermore, we observed that the association between discharge disposition (home versus IRF) and the odds of experiencing MACE was significant only for those aged under 65 years. Multiple chronic comorbid conditions and cognitive impairments may complicate recovery and limit the effectiveness of rehabilitation programs in the elderly population.^40,41 Furthermore, frailty, functional decline, depression, social isolation, and reduced physical resilience may further hinder their motivation and engagement in rehabilitation programs, as well as their ability to participate in and benefit from intensive rehabilitation.⁴² These factors could potentially contribute to the lack of significant association between discharge disposition and MACE in older patients. Conversely, it is important to note that patients with higher levels of frailty, functional impairment, and multimorbidity may be excluded from inpatient rehabilitation, particularly so because the Medicare eligibility criteria for IRF coverage stipulate that patients must require and be reasonably expected to actively participate in and benefit from an intensive rehabilitation therapy program to be eligible for IRF coverage.⁴³ Consequently, patients who present with less physical disability, lower levels of frailty, and fewer comorbid conditions may be more likely to be admitted to inpatient rehabilitation and, therefore, may be less likely to experience vascular outcomes within 1 year. Further studies are, therefore, needed to delineate the key drivers of the age-related differences in outcomes observed in this study.

A panel of experts commissioned by the American Heart Association/American Stroke Association recommended, among other things, that the provision of patient education regarding secondary stroke prevention, individually tailored training to improve mobility and activities of daily living, and exercise and fitness programs specifically tailored to individual needs and abilities should be used as key performance metrics to evaluate the quality of inpatient stroke rehabilitation.⁴⁴ Theoretically, these interventions should instill a long-term commitment to healthy behaviors, including physical activity, blood pressure control, and healthy diet among IRF patients long after their discharge from the facilities, thereby reducing their likelihood of experiencing recurrent or new vascular events. Recent data suggest that structured rehabilitation approaches improve self-management behaviors among patients with stroke.^38,39,45,46 Such behaviors have led to increased levels of health education and the strengthening of self-confidence in disease management.⁴⁷ Furthermore, Chan et al³⁸ demonstrated that patients with stroke who received IRF care, compared with those who only received home health/outpatient care or SNF care, exhibited improvements in mobility, cognition, and self-care. These improvements address some of the most common barriers to effective secondary stroke prevention.⁴⁸ However, assessment of specific rehabilitation interventions or programs that may directly influence a lower likelihood of MACE among ICH survivors is beyond the scope of this study. In addition, there is a lack of studies on long-term compliance with secondary prevention activities among patients discharged to IRF and other settings. Therefore, future comparative effectiveness research is required to identify structured rehabilitation interventions potentially lowering the risk of recurrent vascular events among patients with stroke in general and patients with ICH in particular. Moreover, the extent to which IRF adheres to the American Heart Association/American Stroke Association panel of expert’s recommendations needs further investigation.

Our study has several strengths. To our knowledge, this is one of the largest and most racially, ethnically, and regionally diverse studies to evaluate the risk of MACE among ICH survivors, with a particular focus on assessing the contribution of discharge disposition on the odds of vascular events and all-cause mortality. Notably, although, our findings may not be generalizable to other states not included in this study. Furthermore, across several outcome measures, our findings consistently indicated that patients who received structured rehabilitative therapies in IRF settings had better outcomes, generating hypotheses that the benefits of inpatient rehabilitation may extend beyond improving functional outcomes, community integration, and activities of daily living. Moreover, beyond comparing the vascular outcomes across discharge destinations, this work also provides a modern population-based description of the discharge destinations among a large, diverse sample of patients with ICH. Nonetheless, our findings also need to be evaluated in light of certain limitations. First, the observational nature of our study design and reliance on administrative data sources impose limitations related to selection bias and the specificity and completeness of clinical information. Despite the utilization of proxy markers of disease severity and conducting sensitivity analyses based on ICH location, the possibility of residual confounding cannot be eliminated. In real-world settings, these factors may, in part, guide the decision-making process for discharging patients to IRFs and hence may influence the observed outcomes.^43,49,50 International Classification of Diseases codes for determining ICH location are available. However, to our knowledge, the validity of these codes has not been extensively evaluated in prior studies. Moreover, even with these codes, the location of over 40% of ICH cases is classified as unspecified (I61.9). Therefore, we conducted additional analyses that included the location of ICH among patients whose ICH location was specified, as well as treatment variables indicative of ICH severity (invasive ventilation, tracheostomy, and percutaneous endoscopic gastrostomy). These adjustments did not alter our conclusions regarding the differences in experiencing MACE between patients discharged to IRF and those discharged to SNF. Furthermore, we did not have data on specific patient-level clinical characteristics such as control of blood pressure and adherence to medications relevant to secondary stroke prevention, indicating that the contribution of unmeasured confounding in some of our estimates is uncertain and cannot be ruled out. Also, we lack data on out-of-hospital deaths and out-of-state emigration or in-state immigration, which could potentially lead to the overestimation of the population at risk and the underestimation of ICH recurrence risk. Nonetheless, these limitations are unlikely to affect the differences in event rates between patients discharged to IRF and those discharged home or SNF. Finally, our definition of vascular deaths for MACE, although similar to the definition used in prior literature,²² may not have exhaustively included all fatal vascular events. Furthermore, our data do not provide the specific cause of death. However, it is highly likely that in-hospital deaths in close temporal proximity to major vascular events are influenced by direct or indirect effects of these events. Despite these limitations, we think that the robust sample size and the consistency of our findings across various sensitivity analyses strengthen the validity of our conclusions.

Conclusions

In a large population-based ICH cohort, we demonstrate that postacute care provided at IRF is associated with lower odds of experiencing vascular events, including MACE, recurrent ICH, and vascular death, particularly among patients <65 years. Our findings highlight the potential importance of IRF treatment in the secondary prevention of recurrent vascular events. Further comparative effectiveness research to identify the specific rehabilitative interventions contributing to reduced risk of major vascular events among ICH survivors needs to be prioritized.

Article Information

Sources of Funding

The Institute for Rehabilitation and Research (TIRR) Memorial Hermann provided research infrastructure support for this work.

Disclosures

Dr Li reports grants from the National Institutes of Health. R. Abbott reports service as vice chair, member of the board of directors for American Medical Rehabilitation Providers Association, and compensation from Memorial Hermann Rehabilitation Hospital–Katy for other services. The other authors report no conflicts.

Supplemental Material

Tables S1–S15

Figure S1

STROBE Checklist