American Heart Association


Spinal Cord AVMs Are Getting Attention They Deserve

Daniel Korya, MD

Lee YJ, Terbrugge KG, Saliou G, and Krings T. Clinical Features and Outcomes of Spinal Cord Arteriovenous Malformations: Comparison Between Nidus and Fistulous Types. Stroke. 2014

In the 1800’s, three of the most recognized names in medicine: Virchow, Emmanuel and Luschka, described their findings on intracranial arteriovenous malformations (AVMs), for the first time. Since then, AVMs have been discovered in the spinal cord as well; however, spinal cord AVMs are more rare than intracranial AVMs, and due to their infrequent occurrence in the general population, clear guidelines about an accepted approach to diagnosis and management are lacking. 

In the 1950’s, Actor Ricardo Montalbán was diagnosed with a spinal cord AVM that became aggravated when he was thrown from his horse during the filming of the film “Across the Wide Missouri”. This caused him pain for many years, which led to an elective 9 ½ hour surgery in 1993 that left him paralyzed and wheelchair bound.
For several decades now, experts have debated the nuances of spinal cord AVMs.  As more cases were reported, two subtypes of spinal cord AVMs emerged: the nidus and fistulous type.  Accordingly, the authors of this paper first sought to describe and differentiate between the two types of spinal cord AVMs.  They observed the occurrence rates and treatment modalities for these AVMs and reported on the obliteration rates, based on their experience.
Essentially, the authors explained, fistulous AVMs are located superficially and only rarely possess intramedullary components, while the nidus AVMs are located in the spinal cord parenchyma. 
The researchers were able to gather forty-four consecutive patients with spinal cord AVMs.  There were 26 patients with a nidus-type and 18 patients with a fistulous AVM. Treatments were performed with embolization (n=23), surgery (n=13), combined embolization-surgery (n=3), or conservative management (n=5).
Overall, patients with nidus-type AVMs were younger at presentation and had a higher rate of spinal cord hemorrhage than the fistulous-type. The researchers also confirmed that complete obliteration could be achieved more often in the fistulous-type (72%) than the nidus-type (27%).
Hemodynamics and depth of the lesions seamed to be important factors affecting clinical outcome and obliteration rates. The nidus-type AVMs were less likely to have complete obliteration and had more complicated hemodynamics resulting in higher rates of morbidity.
In short, the researchers had a respectable number of cases for a rare disease, and they reported their findings in a concise and clear manner. Although their results cannot be considered universal, they are certainly a step in the right direction and provide a foundation for clinicians to build on. 

Increase risk of stroke in middle aged Japanese women who are overqualified for their jobs.

Duy Le, MD

Honjo K, Iso H, Inoue M, Sawada N, Tsugane S.Socioeconomic Status Inconsistency and Risk of Stroke Among Japanese Middle-Aged Women. Stroke. 2014

Inconsistency of socioeconomic status has been implicated as a risk factor for poor cardiovascular health. Honjo et al undertook a prospective cohort study evaluating inconsistency of socioeconomic status as a risk for increased incidence of all types of “stroke.”

61,595 Japanese female residents, ages 40-59 located in 15 different districts were self-administered questionnaires in 1990. They were asked to report on socio-demographic information and past medical history. 82% of the participants responded to the questionaire.  14,742 participants were deemed to be eligible and constituted the cohort group. Patients who had a malignancy, baseline cardiovascular disease were excluded from the cohort. A large number of patients were otherwise excluded were not explicitly accounted for.

The patients were measured on a scale of 1-4 in terms of highest level of education; 1) junior high; 2) high school; 3) junior college; 4) college graduate. Occupations were treated in a similar fashion where a score of 1 was assigned for manual labor; 2) sales and service; 3) office work; 4) professional and management.  Status inconsistency indicated a difference of 2 points between the level of education and the occupation. Women who were qualified for the job had a difference of less than 2 between their job and education. Patients were followed out to 20 years and the endpoint was stroke (Intra parenchymal hemorrhage, ischemic stroke, SAH) was considered an endpoint.

Baseline comparisons showed that Japanese women with higher levels of education were likely to be married, have higher self-reported psychological stress, be more physical active, less likely to be overweight and have less incidence of HTN and DM.

The results showed that there was an overall trend of Japanese women with lower education to work a manual job. Those with higher education were likely to have a professional and management job. However, there were still plenty of Japanese women with high level of education working manual jobs. The mismatch between education and career in this study is likely a specific reflection of women in the Japanese culture taking a career break in their  20’s and 30’s for family reasons and rejoining the work force in labor during their 40’s. Cox proportional hazard regression analysis was employed on the data and showed that women who were overqualified for their job had a higher stroke risk compared to women who were qualified for their job (2.06 odds ratio with [1.13,3.78 95% CI]). Women who were qualified for their job as defined above had an odds ratio of having a stroke set to a standard of 1.0. Underqualified women had an OR of 1.01 (0.69, 1.48) of having a stroke.

While the study makes an interesting finding, the generalizability of the results may be limited due to the cohort being constrained to Japanese women. The idea nonetheless is that dissatisfaction and stress at a job can potentially increase vascular events such as stroke.

Desmopressin: keeping beds dry and keeping hemorrhages small.

Vikas Pandey, MD

The treatment of intracerebral hemorrhage currently involves a lot of “watchful waiting” and blood pressure management, with the exception of those with devastating bleeds requiring surgery or those with intraventricular hemorrhages requiring intraventricular thrombolytics. The ultimate goal in these patients is to minimize hematoma growth which is thought to improve overall patient outcome. What if, instead of clinical dismay at not being able to treat with tPA when seeing a hemorrhage on CT scan during an acute stroke evaluation, there was an option to improve platelet activity and prevent further hematoma expansion. The authors of this article had a similar dream.  Desmopressin, a vasopressin analogue, has been used for conditions such as nocturnal bed-wetting and central diabetes insipidus, however has also been used in coagulation disorders such as von Willebrand disease and Hemophilia A due to the interesting effect of stimulating release of vWF from endothelial cells and also increasing the survival of Factor VIII due to increased vWF complexing.

The authors prospectively enrolled 14 patients with either reduced platelet activity on point-of-care testing, or aspirin use, or both and they received desmopressin 0.4 mcg/kg intravenously over 30 minutes. The infusion was started 12.25 (5.7-23.1) hours after the symptom onset. They measured the change in platelet function assays and vWF antigen activity and found shortened Mean PFA-EPI results from 192 +/- 18 seconds to 124 +/- 15 seconds (improved platelet activity) one hour after infusion as well as increase in vWF antigen from 242 +/- 96 to 289 +/ 103 percent activity, both of which were significant results. Peculiarly, one patient had paradoxically increased PFA-EPI.  Hematoma volume was found to be decreased with a range of -1.4 to 8.4 mL (two had hematoma growth). Median change in hematoma volume was -0.5 mL.  Modified Rankin Scale at 3 months showed, out of 12 patients for whom follow-up was obtained, four had no disability, one had a mRS score of 3 and three had mRS of 4. Sodium decreases with the infusion were only in six patients and were in the range of 1-3 mEq/L. Of the seven patients who received the infusion within 12 hours of symptom onset, 2 had hematoma growth.

The article demonstrated safety and efficacy in improving measures of platelet activity and vWF antigen and, though with small numbers, demonstrated some effect of halting hematoma growth, especially if given within 12 hours of symptom onset. The article lacked measures of baseline hematoma volumes, GCS levels, hematoma location and intraventricular involvement. These “ICH score” markers should not be overlooked especially given their importance toward determining patient outcome. The obvious side effects of hypertension and hyponatremia due to water retention were not seen at high rates in this small trial but given the impact these may have toward hematoma expansion and herniation, larger trials are needed to determine if this will be a factor. The authors also allude to the PATCH trial testing platelet transfusions in intracerebral hemorrhage and if positive, this may be a possible co-therapy with desmopressin. The study provides the skeleton for a larger randomized prospective trial and the authors should be applauded for having the proper approach to an area where stroke neurologists feel somewhat helpless.


Does Cerebral Amyloid Angiopathy increase ICH risk after rtPA?

Michelle Christina Johansen, MD

Reuter B, Grudzenski S, Chatzikonstantinou E, Meairs S, Ebert S, Heiler P, et al. Thrombolysis in Experimental Cerebral Amyloid Angiopathy and the Risk of Secondary Intracerebral Hemorrhage. Stroke. 2014

With an aging population facing the stroke neurologist and the increasing evidence that patients 80 years or older have a statistical benefit from IV tPA therapy, the understanding of the pathophysiology of cerebral amyloid and the impact it has on treatment of acute ischemic stroke grows more critical. Symptomatic intracranial hemorrhage (ICH) is a severe side effect that must be discussed with all patients prior to administration of tPA, but are patients with cerebral amyloid angiopathy at increased risk?

It is recognized that the deposition of amyloid in the cerebrovascular system increases the incidence of spontaneous lobar hemorrhage and advancing MRI techniques have made us increasingly aware of the presence of cortical microbleeds in these patients. Given the predisposition of the vessels towards dysfunction, Reuter and colleagues set out to investigate the risk of secondary intracerebral hemorrhage in an animal model after treatment with an adjusted dose of rtPA. In this study, APP23-transgenic mice and wildtype littermates underwent induced ischemic stroke via MCA occlusion, were treated with rtPA and functionally were assessed 24hrs after occlusion.  The brains of both cohorts underwent histological processing to evaluate infarct size and degree of acute intracerebral bleeding. 

In their study, the transgenic mice (9/13) displayed a higher risk (p=0.05) of developing ICH after stroke and thrombolytic therapy compared to the wildtype (3/11). A higher severity of bleeding (Grade 2 or 3) corresponded to an increased infarct burden in the amyloid cohort.  The authors appropriately raise a point of caution that the models were not evaluated in the absence of rtPA thereby potentially leading to an overestimation of the hemorrhagic burden caused by tPA.

Notably neurologic deficit and mortality were not statistically different between the amyloid and wildtype animals. The authors observed no intracranial hemorrhage after rtPA-treatment outside of the infarct area and offer this as an explanation for the lack of increased functional deficit in the amyloid cohort.

Although the n is small and the data is derived from an animal model, the study raises several points of discussion.  Would the results be influenced by a greater delay prior to administration of rtPA?  What if the study had been conducted using a model looking at posterior circulation stroke? Would the outcome scores have differed more than 24hrs after occlusion?

While the incidence of hemorrhage in patients with amyloid may be increased, it would appear from this data that there is no increased risk of mortality after administration of rtPA thereby freeing the clinician to use the last known normal and standard contraindications to guide care of the elderly. Certainly only further investigation can help us move towards a better understanding of using thrombolytics in patients with cerebral amyloid deposition but this study provides a launching point.  

Intracranial hemorrhage mortality in atrial fibrillation patients treated with dabigatran or warfarin

Rajbeer S. Sangha, MD

Alonso A, Bengtson LGS, MacLehose RF, Lutsey PL, Chen LY, and Lakshminarayan K. Intracranial Hemorrhage Mortality in Atrial Fibrillation Patients Treated With Dabigatran or Warfarin. Stroke. 2014

Warfarin has long been the agent of choice for the reduction of ischemic stroke secondary to AF.  Recently, new anticoagulant agents including dabigatrin, rivoxaraban and apixaban have been approved by the FDA for the reduction of stroke secondary to AF.  While these agents may be more effective in the reduction of stroke, the lack of commercially-available antidotes has been a limitation and a noted major disadvantage.  The authors of this study did a retrospective analysis of healthcare utilization date, in-hospital mortality in atrial fibrillation (AF) patients using oral anticoagulants who presented with intracranial bleeding (ICB).

Alonso et al. analyzed 2391 patients with AF who were admitted with ICB (2290 on warfarin and 101 on dabigatran), looking for in hospital mortality.  The results showed that in hospital mortality was similar in patients who were originally on warfarin (22%) vs dabigatran (20%).  Further statistical analysis showed that the propensity score-adjusted RR of morality in dabigatran users was 0.93.  The associations were similar for all varying subtypes of ICB.  Likely, due to a lack of data for all the patients the authors were not able to conduct analysis of patient outcomes following discharge and had to limit the study to in-hospital mortality. 

The authors of this study address an important issue which is at the center of debate for the prevention of stroke secondary to atrial fibrillation.  While the number of patients being analyzed in this analysis for dabigatran are low (101) – which the authors also recognize – the analysis is consistent with the RE-LY trial (Dabigatran versus Warfarin in patients with atrial fibrillation).  It would have been interesting however to see the analysis of patients regarding their outcomes of mortality in a one year period as well as three month outcomes of modified rankin scale.   Clinicians should continue to weigh this data when choosing an anti-coagulant while we await a commercially-available antidote.  

Toward a Simple MRI-Based Predictor of Successful Stroke Treatment

Mark N Rubin, MD
Yan S, Hu H, Shi Z, Zhang X, Zhang S, Liebeskind DS, and Min Lou M. Morphology of Susceptibility Vessel Sign Predicts Middle Cerebral Artery Recanalization After Intravenous Thrombolysis. Stroke. 2014

Although initial evaluation and management of stroke is typically straightforward, the dizzying array of “next steps” is matched in intensity only by the mountain of evidence that, in total, only equivocally supports any particular intervention beyond intravenous thrombolysis (IV tPA). We all just want our patients to get better, and it can be difficult if not impossible to make completely evidence-based management decisions in the timely fashion necessary for acute stroke.

All that said, stroke providers stand to benefit from the identification of biomarkers – ideally acquired during a standard clinical evaluation so as not to waste time – that inform prognosis and/or decision-making. Biomarkers of any sort would do, but there has been a focus on high-resolution neuroimaging with MRI in the hyperacute stroke setting in search of some such sign(s). This will surely continue in the era of 6-minute stroke MRI (not to be confused with the 7-minute workout), and MRI was the modality of choice for the investigators from China and the USA who contributed this biomarker study. They sought to clarify the controversy over the prognosticating value of the susceptibility vessel sign (SVS) in acute stroke.

The SVS in acute stroke, which seems to be the radiographic sign of an erythrocyte-rich or “red clot,” has been a tough nut to crack to this point. There have been conflicting studies through the years, with some showing the SVS is a predictor of recanalization and others suggesting the contrary. The SVS has all the makings of a simple, easily acquired and interpreted acute stroke biomarker, previous work suggests “red clots” are particularly amenable to tPA (at least in coronary arteries), and previous studies fail to definitively answer the question of the potential clinical use of the SVS, thus the interest in further research in this field. The investigators in this study performed a retrospective review of prospectively-collected MRI and clinical data from patients with MCA occlusions treated with tPA within 6 hours of symptom onset in order to determine potential prognostic value of the SVS.

The investigators were able to include 72 consecutive patients with acute stroke treated with IV tPA. The SVS was noted in 50/72 patients (~69%) and recanalization was only seen in 33 (~45%) patients at 24 hours. The mean clot length was ~14mm and shape irregularity was noted in 25/50 clots (50%). In brief, both SVS length and irregularity were independently associated with absence of recanalization at 24 hours. The cut-off length for prediction of recanalization was 14mm but no patients with a clot >20mm experienced recanalization.

This study suffers from a modest sample size and retrospective design, but represents the most “pure” sample of patients with SVS receiving tPA to date and an important step forward in our understanding of the SVS and its role as an acute stroke biomarker. Overall, the results suggest a long and/or irregular SVS predict absence of recanalization, something on the spectrum of “tPA failure.” There are many directions one can go with this information – and one should be cautious if considering any clinical decision-making based on these particular data – but the natural thought-leap, which the authors bring up in their discussion, is to best therapy for patients with an acute stroke and the SVS. Should a long and/or irregular clot prompt triage to the angiography suite post haste, with or without tPA, for endovascular reperfusion? Is an adjunctive antiplatelet infusion indicated in this setting? Might these be a subset of patients who benefit the most from sonothrombolysis? To parrot an oft-used phrase, further study is required. Let the hypothesis generation (and study design) begin!

Predicting Outcomes after Endovascular Therapy for Anterior Circulation Large Vessel Occlusions

Duy Le, MD

Rangaraju S, Liggins JTP, Aghaebrahim A, Streib C, Sun CH, Gupta R, et al. Pittsburgh Outcomes After Stroke Thrombectomy Score Predicts Outcomes After Endovascular Therapy for Anterior Circulation Large Vessel Occlusions. Stroke. 2014

While there have been multiple prior models which predict outcomes prior to thrombectomies as an effort to evaluate which patients are good thrombectomy candidates; Rangaraju et al have developed a prognosis model that carves out a unique niche. Rangaraju et al developed the The Pittsburg Outcomes after Stroke Thrombectomy (POST) Scale, which evaluates how patients will do after receiving a thrombectomy in hopes of providing information to families regarding prognosis to help guide patient management.

In this retrospective validation study, data was collected from a database at Grady Memorial Hospital (GMH) in Atlanta, Georgia. 247 patients were evaluated between 2009 and 2013. These patients met the criteria as follows: they were greater than 18 years of age and underwent endovascular therapy in the anterior circulation with large vessel occlusion (type of endovascular therapy was not specified) within 8 hours of last well known time. Baseline characteristics were measured, and after multivariant regression analysis, only age, final infarct volume and presence of hemorrhage (defined as presence of PH-1 or PH-2) were deemed to be independent predictors of good outcome. They then derived the following; POST Score = Age + 0.5xFIV + 15xH. A good outcome was defined as a modified rankin score of 0-2 at 3 months. 79% of GMH patients had successful recanalization (mTICI 2b/3). The POST score was deemed an excellent predictor of good outcome when evaluated on the GMH group (area under curve = 0.85). Scores that were <60 had a 91% chance of this good outcome, whereas a score of 60-89 carried a 59% chance of having a good outcome. A score of 90-119 had a 25% chance of a good outcome, while there was only a 4% chance of a good outcome with a score of greater than or equal to 120.  The POST score was then validated against two other registries; an institutional endovascular database (UPMC) and the DEFUSE-2 data set. Again, the score accurately predicted good outcomes in these registries as well. Additionally, the POST score was evaluated as a prediction model in those over 80 years of age and those who received the procedure later than 8 hours of last well known time. The general trend held true for these patients as well; a higher POST score was associated with a poorer outcome.

Some weaknesses of the study include the fact that the validating and derivative populations had significant baseline differencs. NIHSS, ASPECT score, IV-tPA, PH-1 and PH-2 rates as well as FIV were significantly different in the GMH, UMPC and DEFUSE-2 groups. UPMC patients had lower rates of IV-tPA (43.5% vs. 53% in the other two groups) and UPMC patients also had higher infarct volumes compared to the other two groups.  Additionally, 20% of patients in the derivation group and 8% in the validation group were excluded due to missing data points.

The goal of this score is to help guide the management of post thrombectomy patients, as many of these patients undergo gastrostomies or tracheostomies. Surrogate decision makers often base their decision to continue vs. withdraw care depending on the likelihood of a patient achieving functional independence.  While it is tempting to use the POST score as a surrogate marker in thrombectomy study patients to help dictate management; we must be careful in withdrawing care on these patients based on retrospective validation, as it may skew the results of the thrombectomy studies. To truly have a model that predicts outcome after thrombectomy, we will have to wait for a prospective validation study. And even then, we will have to await the results of this last wave of thrombectomy trials to see how applicable the POST score will ultimately be.

Hemodynamic quantification in brain arteriovenous malformations with time-resolved spin-labeled MRA

Daniel Korya, MD

Raoult H, Bannier E, Maurel P, Neyton C, Ferré JC, Schmitt P, et al. Hemodynamic Quantification in Brain Arteriovenous Malformations With Time-Resolved Spin-Labeled Magnetic Resonance Angiography. Stroke. 2014

Ruptured brain arteriovenous malformations (AVMs) represent about 2% of hemorrhagic strokes, but they can be quite devastating and often affect young people.  What if there was a way for us to predict which AVMs are more likely to rupture by quantifying some parameter?  That would be something.  Well, that is what Raoult and colleagues set out to do in this recent publication.

They used a 3T unenhanced time-resolved spin-labeled MRA (4D-SL-MRA) sequence of brain vessels to determine hemodynamic quantitative parameters in order to correlate them with progression of disease and rupture risk. 
In total, there were 16 patients in the study who were divided into groups based on AVM characteristics and prior rupture history.  One of the groups was considered to be the “High-Rupture-Risk” (HRR), while another was determined to be “Low-Rupture-Risk” (LRR).  Patients in the HRR were those with a history of ruptures (including asymptomatic bleeding signs on MRI), or exclusively deep venous drainage on DSA, or deep location.  There was another grouping that separated patients into hemorrhagic and non-hemorrhagic sub-groups. 
Essentially, what was observed was that patient’s in the HRR group (a majority of which had previous hemorrhages), were significantly more likely to have lower venous-to-arterial time-to-peak (TTP) ratio values.  The venous-to-arterial TTP ratio was below 2.0 in all cases that were more likely to rupture, and was thought to reflect higher velocities and exposure to higher vascular pressures within the AVM nidus.
Furthermore, by using the 4D-SL-MRA, the investigators argued that they were able to achieve temporal resolution in the range of 50-100 milliseconds, as compared with the usual 1-second times achieved by the current methods of contrast-enhanced MRA.  This provided for significantly better visualization of the vessels and limited vessel superposition. 
This modality allowed for accurate determination of the AVM’s main arterial feeders and draining veins as close to the nidus as possible, the nidus center, and the contralateral arteries corresponding to the main arterial feeders.  The level of detail obtained in a non-invasive way is certainly commendable and may be useful in not only predicting rupture risk, but also planning for treatment. 

Family History as a Risk Factor for Carotid Artery Stenosis

Rizwan Kalani, MD

Mahyar Khaleghi, Iyad N. Isseh, Hayan Jouni, Sunghwan Sohn, Kent R. Bailey, and Iftikhar J. Kullo. Family History as a Risk Factor for Carotid Artery Stenosis. Stroke. 2014

The impact of family history on ischemic stroke and its risk factors has been demonstrated. The risk conferred beyond that of traditional vascular risk factors, however, has been challenging to elucidate.

In this study, Khaleghi et al evaluated whether family history of stroke or coronary heart disease (CHD) was associated with carotid artery stenosis (CAS). They compared 864 patients with CAS (defined as 70% stenosis) and 1698 controls (who did not have CAS or atherosclerotic vascular disease history) that completed carotid artery doppler ultrasound over a six year period at a single institution. History of stroke/CHD in first-degree relatives before age 65 was obtained by questionnaire at the time of study recruitment. The principal finding was the prevalence of family history of stroke and CHD were significantly higher in patients with CAS compared with controls (stroke: OR 2.02; CHD: OR 2.01); sibling history of stroke/CHD was a stronger risk factor than parental history. The association was still significant after adjusting for age, sex and vascular risk factors (stroke: OR 1.41, CHD: OR 1.69). Having a greater number of affected relatives was associated with a higher odds of having CAS, independent of family size. Patients with the combination of 2 relatives with CHD and 2 relatives with stroke had an OR >7 compared to those who did not have this history.

This study demonstrates the role of family history of premature stroke/CHD in CAS risk. It suggests that there are likely genetic and environmental factors contributing to the development of CAS and that these are likely to be shared with CHD. Important limitations include the possibility of recall bias and predominant patient population being non-Hispanic Caucasian.
The authors note that future studies should address the utility of screening asymptomatic individuals for CAS who have a significant family history of vascular disease. I think that it also provides further evidence that the combination of next-generation sequencing methodologies and international collaborative efforts may identify novel genetic determinants of CAS. The potential to identify distinct pathophysiological mechanisms involved in CAS and allow for improved study of gene-environment interaction could pave the way for new stroke preventative and therapeutic options in the future.

Does the state you live in affect post-stroke discharge to rehab?

Ali Saad, MD

Skolarus LE, Burke JF, Morgenstern LB, Meurer WJ, Adelman EE, Kerber, et al. Impact of State Medicaid Coverage on Utilization of Inpatient  
Rehabilitation Facilities Among Patients With Stroke. Stroke. 2014

It may if you’re on Medicaid. This study suggests that residence plays an unfortunate role in a patient’s post stroke care. The authors examined 20,392 patients from the 2010 NIS (Nationwide Inpatient Sample) and followed whether they were discharged to IRFs (inpatient rehab facilities) versus other discharge destinations. All patients had Medicaid as their primary insurance and were of working age (18-64).

Of the 42 states analyzed, Medicaid did not cover IRFs in 4 states: TN, TX, SC, and WV. Having a stroke in one of these non-Medicaid IRF states conferred a 41% chance of not being discharged to an IRF compared to Medicaid-IRF states. No state-specific difference was found in non-Medicaid patients. It is especially concerning that this phenomenon is taking place in the “stroke belt”, which has a higher prevalence of stroke compared to the rest of the country.

The authors note that the accuracy of whether patients qualified for IRF is based on information from state-specific Medicaid websites. Nevertheless, a significant disparity remains. An important implication of this study is worsening health care disparity among working age minorities, although race/ethnicity was not available in the NIS database. Other factors including proximity of IRFs and patient preferences were not available in the NIS database.

Possible follow up studies include analyzing data from random hospitals in Medicaid-IRF versus non-Medicaid IRF states as the hospitals may have more patient demographic information than the NIS.

We have known for years that minority groups are at greater risk for the complications of stroke. Now stroke in the young is on the rise. Medicaid-IRF coverage requires further investigation to determine whether there is another target for improving health care disparities.