Discharge Disposition After Stroke in Patients With Liver Disease
Background and Purpose—Liver disease is associated with both hemorrhagic and thrombotic processes, including an elevated risk of intracranial hemorrhage. We sought to assess the relationship between liver disease and outcomes after stroke, as measured by discharge disposition.
Methods—Using administrative claims data, we identified a cohort of patients hospitalized with stroke in California, Florida, and New York from 2005 to 2013. The predictor variable was liver disease. All diagnoses were defined using validated diagnosis codes. Ordinal logistic regression was used to analyze the association between liver disease and worsening discharge disposition: home, nursing/rehabilitation facility, or death. Secondarily, multiple logistic regression was used to analyze the association between liver disease and in-hospital mortality. Models were adjusted for demographics, vascular risk factors, and comorbidities.
Results—We identified 121 428 patients with intracerebral hemorrhage and 703 918 with ischemic stroke. Liver disease was documented in 13 584 patients (1.7%). Liver disease was associated with worse discharge disposition after both intracerebral hemorrhage (global odds ratio, 1.28; 95% confidence interval, 1.19–1.38) and ischemic stroke (odds ratio, 1.23; 95% confidence interval, 1.17–1.29). Similarly, liver disease was associated with in-hospital death after both intracerebral hemorrhage (odds ratio, 1.33; 95% confidence interval, 1.23–1.44) and ischemic stroke (odds ratio, 1.60; 95% confidence interval, 1.51–1.71).
Conclusions—Liver disease was associated with worse hospital discharge disposition and in-hospital mortality after stroke, suggesting worse functional outcomes.
Recently, we found that liver disease is associated with an increased risk of intracranial hemorrhage.1 A similar relationship with ischemic stroke has not been observed.2 However, liver disease, particularly cirrhotic liver disease, is complicated by an imbalanced coagulopathy with implications for both hemorrhagic and thrombotic processes.3 Clinical outcomes after stroke in patients with liver disease have not been well studied. We therefore sought to assess the influence of liver disease, including cirrhosis, on outcome after stroke as measured by hospital discharge disposition. Our pre-specified hypothesis was that liver disease adversely impacts discharge disposition after stroke.
We performed a retrospective cohort study of patients with liver disease and stroke using publically available patient data, from the Agency of Healthcare Research and Quality, extracted from administrative claims from California, Florida, and New York. The Weill Cornell Medicine Institutional Review Board approved our analyses.
We included all adult patients with a first-recorded discharge diagnosis of stroke between 2005 and 2011 in California, 2005 and 2013 in Florida, and 2006 and 2013 in New York. We defined stroke based on validated International Classification of Diseases, 9th Revision, Clinical Modification code algorithms (Methods in the online-only Data Supplement).4 These algorithms were found to be 82% sensitive and 93% specific for the diagnosis of nontraumatic intracerebral hemorrhage (ICH) and 86% sensitive and 95% specific for ischemic stroke.4
The primary exposure was liver disease documented at the time of stroke. Liver disease was defined by an International Classification of Diseases, 9th Revision, Clinical Modification coding algorithm designed to identify patients with liver disease comprehensively: cirrhosis, cirrhosis decompensation events, acute liver failure, chronic viral hepatitis, hepatic infarction, and various forms of chronic liver disease (Methods in the online-only Data Supplement). This algorithm has 99.5% specificity and 80.2% positive predictive value for identifying liver disease.5 We also separately examined liver cirrhosis—both alcohol related and nonalcohol related (Methods in the online-only Data Supplement).
The primary outcome was discharge disposition assessed on an ordinal scale consisting of discharge to home, discharge to any institution (nursing or rehabilitation facility), or in-hospital death. This measure is a good surrogate for functional outcomes as measured by the modified Rankin Scale at 3 or 12 months after stroke.6 The secondary outcome was in-hospital mortality.
Baseline variables between groups were compared with the t test or χ2 test as appropriate. The association between liver disease and discharge disposition was analyzed with ordinal logistic regression. In a secondary analysis, we used multiple logistic regression to assess the association between liver disease and in-hospital mortality versus discharge to any location. Both analyses were adjusted for potential confounders: demographic variables, vascular diseases, and their risk factors (Methods in the online-only Data Supplement), alcohol use, and the Elixhauser Index. The Elixhauser Index can be used to reliably adjust estimates of in-hospital mortality risk after stroke for comorbidity burden.7 In addition, interaction terms were used to analyze whether the association between liver disease and discharge disposition varied by stroke type (ischemic versus hemorrhagic). All covariates were included independent of statistical significance to maximally isolate the contribution of exposure. The threshold of statistical significance allowed for an α-error of 0.05. All analyses were performed by HK using Stata/MP (version 13; College Station, TX).
We identified 121 428 patients with ICH and 703 918 patients with ischemic stroke. Patients with ICH were younger than those with ischemic stroke, had lower rates of traditional vascular risk factors, and had higher rates of liver disease and overt coagulopathy (Table I in the online-only Data Supplement). Liver disease was documented in 13 584 (1.7%) patients and cirrhosis in 6568 (0.8%) patients. Of patients with ICH, 3443 (2.8%) patients had liver disease and 2066 (1.7%) had cirrhosis. For ischemic stroke, 10 141 (1.4%) had liver disease and 4502 (0.6%) had cirrhosis. Patients with liver disease were younger than those without liver disease; had higher rates of alcohol abuse and overt coagulopathy, and lower rates of hypertension (Table II in the online-only Data Supplement).
After ICH, 29.7% of patients were discharged home, 43.2% were discharged to a facility, and 27.1% died before discharge. Of patients with ischemic stroke, 48.4% were discharged home, 43.6% were discharged to a facility, and 8.0% died. After adjustment for demographics, vascular risk factors, and Elixhauser comorbidities in ordinal logistic regression models, liver disease was associated with worse discharge disposition after both ICH (global odds ratio [OR], 1.28; 95% CI, 1.19–1.38) and ischemic stroke (global OR, 1.23; 95% CI, 1.17–1.29). Cirrhosis, specifically alcohol-related cirrhosis, seemed more strongly associated with worse discharge disposition after ICH (OR, 1.51; 95% CI, 1.32–1.74) and ischemic stroke (OR, 1.63; 95% CI, 1.46–1.82; Table). For all comparisons, interaction terms revealed no variation by stroke subtype in the association between liver disease and discharge disposition; in other words, liver disease was associated with worse discharge disposition to a similar extent after both ischemic and hemorrhagic stroke.
The crude rates of in-hospital mortality after stroke were higher for patients with liver disease (Results in the online-only Data Supplement). After adjustment for covariates, liver disease was associated with in-hospital death after ICH (OR, 1.33; 95% CI, 1.23–1.44) and ischemic stroke (OR, 1.60; 95% CI, 1.51–1.71). Interaction terms revealed that liver disease was less strongly associated with in-hospital mortality after ICH than after ischemic stroke (P<0.001).
In a large, heterogeneous cohort of patients with stroke, we found an independent association between liver disease and worse hospital discharge disposition and in-hospital death. This association seemed more pronounced in patients with alcohol-related cirrhotic liver disease. The impact of liver disease on hospital discharge disposition after stroke by ordinal analysis did not vary by stroke type. However, in secondary analysis, liver disease was more strongly associated with in-hospital death after ischemic stroke than after ICH.
Our findings are consistent with a small, retrospective study that found increased in-hospital mortality after ICH in liver disease (OR, 7.4; 95% CI, 1.3–41.5).8 In contrast to our cohort, this study was a single-center cohort of patients.8 With respect to ischemic stroke outcomes, unlike a limited analysis of monoethnic patients with nonalcoholic fatty liver disease,9 our findings suggest that liver disease is also associated with a worse outcome after ischemic stroke.
The observed influence of liver disease on discharge disposition after stroke may be multifactorial. Poor outcomes after ICH in this population may, in part, be because of increased hematoma size or expansion in patients with liver disease.10,11 Alternatively, patients with liver disease may be otherwise prone to decompensation because of non-neurological conditions,8 such as alcohol withdrawal or frailty. Last, possible interactions between liver disease and stroke treatment may play a role directly or limit provision of therapies.
Our study has several limitations. First, data on hematologic parameters, stroke severity, hematoma size, and antithrombotic medications were unavailable. These variables may mediate the observed association; therefore, it is not clear that they should be adjusted for in models. Second, given the reliance on inpatient International Classification of Diseases, 9th Revision, Clinical Modification diagnosis codes, ascertainment of liver disease and covariates may be incomplete. Similarly, some patients may have been included at the time of a recurrent and not first stroke. Imprecise ascertainment is expected to occur to a similar extent in the control group (patients without liver disease), but there may be differential ascertainment of some variables, such as alcohol abuse, which may cause residual confounding. Third, it is possible that some patients developed liver disease as a result of stroke treatment; however, this is expected to be rare as most liver diseases included are chronic, progressive disorders. Last, albeit a validated measure,6 discharge disposition is a crude surrogate of functional outcome.
In summary, outcomes, including survival to discharge, after stroke are worse in patients with liver disease. Further research into the role of liver disease on stroke outcomes may lead to strategies to improve recovery in this vulnerable subset of patients.
Sources of Funding
Dr Kamel acknowledges funding support from the NIH (grant K23NS082367) and the Michael Goldberg Research Fund.
Dr Navi acknowledges funding support from the NIH (grant K23NS091395) and the Florence Gould Endowment for Discovery in Stroke.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.116.016016/-/DC1.
- Received August 24, 2016.
- Revision received August 24, 2016.
- Accepted November 11, 2016.
- © 2016 American Heart Association, Inc.
- Parikh NS,
- Navi BB,
- Kumar S,
- Kamel H
- Tirschwell DL,
- Longstreth WT Jr
- Tziomalos K,
- Giampatzis V,
- Bouziana SD,
- Spanou M,
- Papadopoulou M,
- Pavlidis A,
- et al
- Niizuma H,
- Suzuki J,
- Yonemitsu T,
- Otsuki T