Geographic Disparities in Case Fatality and Discharge Disposition Among Patients With Primary Intracerebral Hemorrhage

Abstract

Background

We evaluate nationwide trends and urban–rural disparities in case fatality (in‐hospital mortality) and discharge dispositions among patients with primary intracerebral hemorrhage (ICH).

Methods and Results

In this repeated cross‐sectional study, we identified adult patients (≥18 years of age) with primary ICH from the National Inpatient Sample (2004–2018). Using a series of survey design Poisson regression models, with hospital location–time interaction, we report the adjusted risk ratio (aRR), 95% CI, and average marginal effect (AME) for factors associated with ICH case fatality and discharge dispositions. We performed a stratified analysis of each model among patients with extreme loss of function and minor to major loss of function. We identified 908 557 primary ICH hospitalizations (overall mean age [SD], 69.0 [15.0] years; 445 301 [49.0%] women; 49 884 [5.5%] rural ICH hospitalizations). The crude ICH case fatality rate was 25.3% (urban hospitals: 24.9%, rural hospitals:32.5%). Urban (versus rural) hospital patients had a lower likelihood of ICH case fatality (aRR, 0.86 [95% CI, 0.83–0.89]). ICH case fatality is declining over time; however, it is declining faster in urban hospitals (AME, −0.049 [95% CI, −0.051 to −0.047]) compared with rural hospitals (AME, −0.034 [95% CI, −0.040 to −0.027]). Conversely, home discharge is increasing significantly among urban hospitals (AME, 0.011 [95% CI, 0.008–0.014]) but not significantly changing in rural hospitals (AME, −0.001 [95% CI, −0.010 to 0.007]). Among patients with extreme loss of function, hospital location was not significantly associated with ICH case fatality or home discharge.

Conclusions

Improving access to neurocritical care resources, particularly in resource‐limited communities, may reduce the ICH outcomes disparity gap.

Nonstandard Abbreviations and Acronyms

AME

average marginal effect

aRR

adjusted risk ratio

E‐LoF

extreme loss of function

ICH

intracerebral hemorrhage

NIS

National Inpatient Sample

Clinical Perspective.

What Is New?

• Between 2004 and 2018, in‐hospital mortality (case fatality) among patients with intracerebral hemorrhage (ICH) declined, whereas the rate of home discharge increased.

• Patients with ICH admitted to urban hospitals had a lower likelihood of in‐hospital mortality and a higher likelihood of home discharge.

• ICH case fatality is declining significantly faster in urban hospitals; however, although the rate of home discharge is increasing in urban hospitals, it is not significantly changing over time in rural hospitals.

What Are the Clinical Implications?

• The increasingly widening urban–rural disparity in ICH case fatality and home discharge rates highlights the need for improving the availability of and access to neurocritical care resources and poststroke care facilities, particularly in rural areas.

• Ensuring that ICH critical care pathways in rural hospitals meet current practice guidelines may help close urban–rural disparity gaps in postacute transition of care among patients with ICH.

Wider implementation of quality‐of‐care measures for patients with primary intracerebral hemorrhage (ICH) and improved access to neurocritical care is envisioned to reduce ICH case fatality (in‐hospital mortality); however, contemporary population‐wide estimates of the temporal trends in ICH case fatality are either lacking or, at best, conflicting. ¹ , ² Furthermore, nationwide trends and urban–rural disparities in postacute transition of care have not been evaluated for patients with ICH. Therefore, we sought to comprehensively evaluate the nationwide trends and urban–rural disparities in case fatality and hospital discharge dispositions among patients with ICH during a contemporary 15‐year period (2004–2018).

METHODS

Ethics Statement

Per Houston Methodist's policy, this study is exempt from review by the institutional review board, because it used publicly available and deidentified data. We followed the Reporting of Studies Conducted Using Observational Routinely Collected Data guidelines. ³

Data Availability

The National Inpatient Sample (NIS) data are available for online purchase through the Health Care Utilization Project's central distributor (https://www.distributor.hcup‐us.ahrq.gov/) to qualified researchers upon completion of data use agreement training.

Study Design, Data Source, and Case Identification

We conducted repeated cross‐sectional analyses of adults (≥18 years of age) with principal ICH diagnosis using the NIS (2004–2018). ⁴ The NIS is a representative sample of >90% of US hospitalizations. ICH diagnosis was ascertained using the validated ⁵ International Classification of Diseases, Ninth Revision (ICD‐9) code 431 and Tenth Revision (ICD‐10) codes I61 (I61.0–I61.6 and I61.8–I61.9). We excluded encounters with coexisting diagnoses of head trauma, arteriovenous malformation, or intracranial aneurysms, as well as encounters with missing discharge disposition data.

Outcomes

Case fatality and discharge disposition are consistently defined in the 2004 to 2018 NIS. We categorized discharge disposition as (1) home discharge; (2) transfer to short‐term hospital; (3) transfer to long‐term or intermediate care facility, including skilled nursing facility, inpatient rehabilitation facilities; and (4) transfer to home health.

Independent Variables and Covariates

The NIS designates the hospitals' urban or rural location based on the hospitals' Core Based Statistical Area type. ⁴ Hospitals are considered to be urban if they are located in counties with a metropolitan Core Based Statistical Area type, and hospitals are classified as rural if they are located in counties with a Core Based Statistical Area type of micropolitan or noncore. We categorized the analysis timeline into 5 consecutive 3‐year periods (2004–2006, 2007–2009, 2010–2012, 2013–2015, and 2016–2018) and treated the transformed time variable as a continuous predictor. ⁶ Length of stay was divided in quartiles (0–1, 2–4, 5–8, and ≥9 days). We grouped patients with ICH with extreme loss of function (E‐LoF) and minor to major loss of function using the All Patient Refined Diagnosis Related Group severity of illness score.

Statistical Analysis

Patient characteristics, by urban–rural status, are reported using means and proportions. We used logistic regression to report the crude odds ratio (OR) and 95% CI for factors associated with admission to urban (versus rural) hospitals. We fit a series of survey design multivariable generalized linear models (Poisson family with log link function) to report the adjusted risk ratio (aRR) and 95% CI for factors independently associated with case fatality and each discharge disposition category, evaluating time–hospital location interaction for each model. ⁷ , ⁸ Our variance estimation method, which is based on the first‐order Taylor series linear approximation, is robust to overdispersion. ⁹ We also conducted stratified analyses for patients with minor to major loss of function and E‐LoF. All multivariable models were adjusted for demographic, clinical, and hospital characteristics (covariates are listed in Table S1). Variables included in the adjusted models were chosen based on evidence from prior literature indicating their association with discharge outcomes among patients with ICH. Length of stay quartiles were included in the ICH case fatality model. α was set at 0.05, and analyses were conducted using Stata version 16, incorporating survey design analysis to yield nationally representative estimates. Trend weights were used to account for NIS design change in 2012.

Sensitivity Analysis

To assess if poor outcomes in rural hospitals may have been driven by selective transfer of patients considered to have worse prognosis to rural centers, we performed sensitivity analyses (for each model) in which patients who were admitted through interhospital transfer from acute or primary care settings were excluded from the analyses. Also, we additionally fit models in which time was treated as a categorical predictor. Accordingly, outcomes in period 1 (2004–2006) were compared with period 5 (2016–2018) to assess if violation of our assumption of a linear relationship between time and outcomes will change the direction or significance of our study findings.

RESULTS

Descriptive Characteristics

We identified 908 557 eligible ICH hospitalizations with an overall mean age (SD) of 69.0 (15.0) years, 445 301 (49.0%) of whom were women. Rural hospitals accounted for 49 884 (5.5%) ICH hospitalizations, whereas urban hospitals accounted for 854 428 (94.0%) hospitalizations, and hospital location information was not available for 4245 (0.5%) hospitalizations (Tables S1 and S2). The overall crude ICH case fatality was 25.3% (urban hospitals: 24.9%, rural hospitals: 32.5%).

Urban–Rural Differences in Patient Characteristics

Patients admitted to urban (versus rural) hospitals were younger (mean age: 68.6 versus 75.8 years), predominantly men (51.4% versus 44.0%), more frequently of minority race/ethnicity category (NHB, AAPI, Hispanic, or others) (36.6% versus 14.6%), and not insured via Medicare (40.0% versus 20.7%; Table S1).

Patients admitted to urban hospitals were more likely to have hypertension (OR, 1.65 [95% CI, 1.57–1.74]), complicated diabetes (OR, 1.93 [95% CI, 1.71–2.17]), and hyperlipidemia (OR, 1.29 [95% CI, 1.22–1.36]). Conversely, patients admitted to urban hospitals were less likely to have congestive heart failure (OR, 0.88 [95% CI, 0.82–0.94]), atrial fibrillation (OR, 0.86 [95% CI, 0.82–0.90]), and long‐term (current) anticoagulant use (OR, 0.92 [95% CI, 0.85–0.99]). Patients admitted to urban (versus rural) hospitals also had higher acuity, demonstrated by a higher likelihood of E‐LoF (OR, 3.19 [95% CI, 2.92–3.47]). This coincided with a higher use of intensive treatment, including extra ventricular drain placement (OR, 6.12 [95% CI, 4.87–7.70]), craniectomy/craniotomy (OR, 4.54 [95% CI, 3.33–6.20]), invasive ventilator support (OR, 3.41 [95% CI, 2.86–4.06]), noninvasive ventilator support (OR, 3.27 [95% CI, 2.26–4.74]), tracheostomy (OR, 4.53 [95% CI, 3.43–5.98]), and gastric tube placement (OR, 2.69 [95% CI, 2.36–3.06]; Table S1).

Urban–Rural Differences in ICH Case Fatality and Discharge Disposition

Overall, patients admitted to urban (versus rural) hospitals had lower case fatality (aRR, 0.86 [95% CI, 0.83–0.89]; Table and Table S3). ICH case fatality in the overall population declined over time (aRR, 0.83 [95% CI, 0.82–0.83]); however, case fatality declined at a significantly faster rate in urban hospitals (average marginal effect [AME], −0.049 [95% CI, −0.051 to −0.047]) relative to rural hospitals (AME, −0.034 [95% CI, −0.040 to −0.027]; Table S4 and Figure). Furthermore, urban hospital patients had a higher likelihood of home discharge (aRR, 1.07 [95% CI, 1.02–1.13]; Table and Table S3). However, although the likelihood of home discharge is increasing in urban hospitals (AME, 0.011 [95% CI, 0.008–0.014]), it is not significantly changing over time in rural hospitals (AME, −0.001 [95% CI, −0.010 to 0.007]; Table S4 and Figure S1). Overall, the likelihood of transfer to a short‐term hospital is significantly lower for urban hospitals (aRR, 0.32 [95% CI, 0.28–0.37]), and this likelihood has been declining significantly over time (aRR, 0.95 [95% CI, 0.91–0.99]; Table and Figure S2). In contrast, the likelihood of transfer to a long‐term or intermediate care facility was higher for urban hospitals (aRR, 1.10 [95% CI, 1.07–1.13]), although this likelihood has declined significantly in urban hospitals (AME, −0.020 [95% CI, −0.024 to −0.016]) but not in rural hospitals (AME, −0.006 [95% CI, −0.016 to 0.004]; Table S4 and Figure S3). Finally, the likelihood of transfer to home health is increasing over time (aRR, 1.09 [95% CI, 1.07–1.11]; Table and Figure S4). Details of other sociodemographic and clinical factors independently associated with discharge outcomes are available in Table S4.

Table .

Adjusted Risk Ratio (95% CI) for Select Predictors of Case Fatality and Discharge Dispositions Among Patients With Primary Intracerebral Hemorrhage

Variables	Case fatality	Home discharge	Transfer to short‐term facilities	Transfer to long‐term and intermediate facilities	Transfer to home health
Overall
Hospital location, urban vs rural§	0.86 (0.83–0.89)‡	1.07 (1.02–1.13)*	0.32 (0.28–0.37)‡	1.10 (1.07–1.13)‡	1.05 (0.97–1.14)
3‐year period, trend§	0.83 (0.82–0.83)‡	1.04 (1.03–1.05)‡	0.95 (0.91–0.99)†	0.97 (0.96–0.97)‡	1.09 (1.07–1.11)‡
Hospital location, 3‐year period interaction||	0.92 (0.90–0.94)‡	1.05 (1.01–1.09)*	0.95 (0.88–1.03)	0.98 (0.96–1.00)*	1.06 (0.99–1.12)
Minor to major loss of function
Hospital location, urban vs rural§	0.89 (0.85–0.92)‡	1.08 (1.02–1.14)†	0.32 (0.28–0.37)‡	1.10 (1.07–1.14)‡	1.09 (1.00–1.19)*
3‐year period§	0.81 (0.80–0.82)‡	1.05 (1.04–1.06)‡	0.96 (0.92–1.00)*	0.96 (0.95–0.96)‡	1.09 (1.07–1.11)‡
Hospital location, 3‐year period interaction||	0.86 (0.84–0.89)‡	1.04 (1.00–1.09)*	0.95 (0.88–1.03)	0.97 (0.96–0.99)*	1.06 (1.00–1.13)
Extreme loss of function
Hospital location, urban vs rural§	1.00 (0.95–1.07)	1.15 (0.81–1.65)	0.41 (0.29–0.57)‡	1.09 (1.03–1.16)†	0.76 (0.57–1.02)
3‐year period§	0.89 (0.88–0.89)‡	0.99 (0.95–1.04)	0.91 (0.86–0.97)†	1.00 (0.99–1.01)	1.11 (1.06–1.16)‡
Hospital location, 3‐year period interaction||	0.99(0.95–1.03)	1.25 (0.93–1.69)	0.94 (0.74–1.19)	0.98 (0.94–1.03)	1.03 (0.86–1.24)

^* P<0.05.

^† P<0.01.

^‡ P<0.001.

^§ Model includes all covariates listed in Table S3, no interaction term.

^|| Model includes all covariates listed in Table S3 plus interaction between time and hospital location.

Figure . Marginal probability of intracerebral hemorrhage case fatality, overall and stratified by severity category.

Urban–Rural Differences in Outcomes by All Patient Refined Diagnosis Related Group Categories

Among patients with ICH with minor to major loss of function, admission to urban hospitals was associated with a lower likelihood of ICH case fatality (aRR, 0.89 [95% CI, 0.85–0.92]) and a higher likelihood of home discharge (aRR, 1.08 [95% CI, 1.02–1.14]); however, among those with E‐LoF, hospital location was not significantly associated with case fatality (aRR, 1.00 [95% CI, 0.95–1.07]) or home discharge (aRR, 1.15 [95% CI, 0.81–1.65]; Table). Also, among patients with minor to major loss of function, the likelihood of home discharge increased (aRR, 1.05 [95% CI, 1.04–1.06]), whereas the likelihood of transfer to long‐term or intermediate care facility (aRR, 0.96 [95% CI, 0.95–0.96]) decreased; however, among patients with E‐LoF, the likelihood of home discharge (aRR, 0.99 [95% CI, 0.95–1.04]) and transfer to long‐term or intermediate care facility (aRR, 1.00 [95% CI, 0.99–1.01]) did not significantly change over time (Table).

Sensitivity Analyses

Excluding patients admitted through interhospital transfer and treating time as a categorical predictor did not significantly change our main conclusions with respect to urban–rural differences in case fatality and home discharge (Table S5 and S6).

DISCUSSION

We report that nationwide ICH case fatality has declined, on average, by 17% per 3‐year period during the contemporary 15‐year period. However, these gains in mortality reduction are disparate and seem to be associated with greater proportion of ICH care provided in urban (versus rural) hospitals. As previously reported, our analyses also indicate a higher likelihood of ICH case fatality among patients admitted to rural hospitals. ¹⁰ We additionally report a widening trend of urban–rural disparity in ICH case fatality and home discharge. However, in stratified analyses, urban–rural differences in ICH case fatality and home discharge were only observed among patients with minor to major loss of function. Given that urban hospitals have better access to advanced neurocritical and postacute stroke care resources, ¹¹ , ¹² , ¹³ this increasingly widening urban–rural disparity in ICH case fatality and home discharge rates potentially reflects advances in neurocritical care for ICH management, as well as highlights the need for improving the availability of and access to poststroke care facilities, particularly in rural areas. We may also surmise that ICH severity likely moderates the beneficial effect of access to advanced neurocritical care resources, particularly given that urban–rural differences in ICH case fatality and home discharge were not observed among patients with E‐LoF. However, further evaluations using ICH‐specific severity scores may be needed to ascertain this hypothesis.

Our limitations include use of ICD codes for ICH identification, which may lead to misclassification and potential exclusion of nonhospitalized patients with ICH. However, the reported positive predictive value of ICD codes in ICH identification is as high as 100%, ⁵ and the reported proportion of nonhospitalized patients with ICH is <5%. ¹⁴ Finally, All Patient Refined Diagnosis Related Group score may not truly reflect the degree of ICH severity because it is administratively derived.

CONCLUSIONS

Although nationwide ICH case fatality has declined, urban–rural disparities in ICH case fatality and home discharge rates continue to widen. Improving access to neurocritical care resources and ensuring that ICH critical care pathways in rural hospitals meet current practice guidelines may help close the disparity gaps.

Sources of Funding

None.

Disclosures

None.

Supporting information

Tables S1–S6

Figures S1–S4