Mechanical Thrombectomy for Acute Ischemic Stroke in Patients With Dementia

Authors: Hamidreza Saber, Jason Hinman, Katherine Mun, Naoki Kaneko, Viktor Szeder, Satoshi Tateshima, May Nour, Radoslav Raychev, Yinn Cher Ooi, Reza Jahan, Gary P. Duckwiler, Jeffrey L. Saver, David S. Liebeskind

Abstract

Background

To compare stroke severity, as well as usage and outcomes of mechanical thrombectomy (MT) for acute ischemic stroke (AIS) in patients with and without dementia.

Methods

Study of the US hospitalizations with AIS and documented National Institutes of Health Stroke Scale scores from October 1, 2016, to December 31, 2017. AIS hospitalizations with documented dementia were compared with AIS hospitalizations without dementia for stroke severity at onset, use of MT, and outcomes at discharge. Outcomes included favorable discharge disposition (home without assistance), in‐hospital mortality, and radiographic intracranial hemorrhage.

Results

Among 179 665 AIS admissions with recorded National Institutes of Health Stroke Scale score during the 15‐month study period, 18 255 (10.2%) had preexisting dementia (62.5% women). The median age at stroke was 84 (interquartile range, 78–89) versus 69 (interquartile range, 59–79) years; and median presenting National Institutes of Health Stroke Scale score was 7 (interquartile range, 3–15) versus 4 (interquartile range, 2–10) in hospitalizations with versus without dementia. There was a stepwise association between increasing severity of the index stroke and the prevalence of dementia. In multivariable analysis, dementia was associated with significantly lower (odds ratio [OR], 0.43 [95% CI, 0.35–0.53]; P<0.001) odds of receiving MT, but not with the use of intravenous thrombolysis (OR, 0.95 [95% CI, 0.86–1.06]; P=0.4). In‐hospital mortality occurred in 21.3% versus 11.2% following MT in hospitalizations with versus without dementia (P<0.001). However, after adjustment for stroke severity, no associations were found between dementia and in‐hospital mortality (OR, 1.48 [95% CI, 0.90–2.43]; P=0.15) or favorable discharge outcome (OR, 0.61 [95% CI, 0.33–1.32]; P=0.12) following MT. Dementia was associated with increased likelihood of intracranial hemorrhage after MT (OR, 1.57 [95% CI, 1.03–2.40]; P=0.04).

Conclusion

One in 10 among all, and 1 in 4 among those aged ≥80 years had preexisting dementia at stroke onset. While patients with dementia were more likely to present with more severe deficits, MT was less frequently used in this population. Dementia was not independently associated with in‐hospital mortality beyond stroke severity at onset, but increased the risk of intracranial hemorrhage following MT.

Stroke is a major cause of mortality and disability and a major contributing factor to long‐term cognitive dysfunction. Recent data have demonstrated a higher risk of dementia following stroke. ^1^ It has also been suggested that a preexisting dementia is associated with higher risk of stroke and worse stroke outcomes. ^2^ Recent advancements in reperfusion therapies have dramatically improved outcomes in patients with acute ischemic strokes (AISs). ^3^ , ^4^ Dementia is not a contraindication for reperfusion therapies in AIS, but most prior trials excluded or underrepresented patients with dementia, ^5^ contributing to the inconsistencies in guideline recommendations in AIS. ^6^

Given the growing population of patients with cognitive disorders, it is critical to better clarify usage and clinical factors associated with outcomes of reperfusion therapies in this population. Prior studies evaluating reperfusion therapies in AIS are limited by lack of modern endovascular treatments and inconsistent finding. While some studies reported increased mortality and poorer functional outcomes following AIS in patients with dementia regardless of the use of reperfusion therapies, ^7^ , ^8^ other studies did not find similar associations. ^9^ , ^10^ Because of this inconsistency and fear of cerebral hemorrhage, it is likely that patients with dementia are less likely treated with reperfusion therapies including mechanical thrombectomy (MT).

More recently, National Institutes of Health Stroke Scale (NIHSS) score measures were included in the International Classification of Diseases, Tenth Revision (ICD‐10) and became available in national claims data, allowing for determination of stroke severity at onset. ^11^ A recent study demonstrated reliable representation of the range and patterns of NIHSS scores reported in the National Inpatient Sample (NIS) and their convergent validity with patterns seen in a prior large data set. Given these gaps in knowledge and the recent availability of stroke severity indices in US claims data, we aimed to examine the (1) the association of preexisting dementia and stroke severity at onset, (2) rates of usage of reperfusion therapies in patients with versus without cognitive disorder, and (3) outcomes of reperfusion therapies in patients with versus without cognitive disorders using a real‐word representative database.

Methods

Study Design and Population

All relevant data are available upon reasonable request from the corresponding author. We performed a retrospective cross‐sectional study using data from the largest US all‐payer inpatient claims‐based database, the NIS, from October 1, 2016, to December 31, 2017. As of October 1, 2016, administrative reporting of the presenting NIHSS score for patients with AIS was made available with the collection of ICD‐10, Clinical Modification (ICD‐10‐CM) codes. The NIHSS codes (R29.7xx) can be used in conjunction with acute stroke codes (I63) to characterize the patient's neurological status and the severity of the stroke. The code assignment should reflect the initial NIHSS at presentation and may be acquired on the basis of the documentation from clinicians who were part of the treating team. To identify AIS hospitalizations, we used ICD‐10 codes I61, I63, and I64. ICD‐10 codes for identifying dementia are provided in Table S1. Intravenous thrombolysis (IVT) administration was identified using the procedural code 3E03317 and MT using the procedural codes 03CG3ZZ. Radiographic intracranial hemorrhage (ICH) was identified using the ICD‐10 codes for nontraumatic intracerebral hemorrhage and nontraumatic subarachnoid hemorrhage. This study met US Department of Health and Human Services criteria for exemption from institutional review board approval and from individual patient informed consent, as all NIS data are publicly available and deidentified.

Outcomes

Main outcomes were in‐hospital mortality and favorable outcome at discharge. Favorable discharge outcome was defined as discharge disposition to home, an administrative data field that correlates strongly with excellent disability outcomes at 3 months. ^12^ , ^13^ , ^14^ Poor discharge outcome was defined as discharge to a facility or death. For the analysis of favorable and poor discharge outcomes, we excluded patients with admission source from a chronic care facility (4.8% of MT group); however, these patients were included for the analysis of other outcomes (ie, in‐hospital mortality and ICH). Discharge destinations were derived from the meta‐label DISPUNIFORM in NIS. Transfers from chronic care facilities and transfers to a second acute care facility were not included. In‐hospital mortality was identified directly within NIS.

Statistical Analysis

Categorical variables were expressed as sum of weights (weighted proportion) and continuous variables as weighted mean (±SE). Patients with and without NIHSS value recorded were compared using standardized differences. A standardized difference greater than 25% was considered meaningful and a standardized difference between >10% and ≤25% potentially meaningful. ^15^ Multivariate logistic regression analyses were used to assess the relationship between dementia status and receiving reperfusion therapy, as well as the association between dementia and reperfusion therapy outcomes. Initial regression models were adjusted for age and sex. Next, we added NIHSS score at onset as a measure of stroke severity at onset, as well as other possible hospital‐ and patient‐level confounding factors and comorbidities. Covariates were chosen to reflect a burden of comorbidity, which might affect the decision to use reperfusion therapy. The 12 patient‐level variables were age, sex, race or ethnicity (White, Black, Hispanic, and other [including Asian, Pacific Islander, and Native American]); NIHSS score; atrial fibrillation, coronary artery disease, chronic kidney disease, hypertension, diabetes, congestive heart failure, arrival mode (emergency department, in‐hospital transfer); and weekday versus weekend admission. The 3 hospital‐level variables evaluated were hospital size (small, medium, and large); hospital type (rural, urban community, and urban teaching); and hospital geographic region (Northeast, Midwest, South, and West). Hospital size categories are predefined in NIS and based on hospital beds and are specific to the hospital's location and teaching status, based on short‐term acute care beds set up and staffed in a hospital. ^16^ Statistical analysis was performed using SAS version 9.4 (SAS Institute Inc., Cary, NC).

Results

Characteristics of the Study Population

Based on the representative Nationwide Inpatient Sample, a total of 809 468 ischemic stroke admissions occurred in the United States during the 15‐month period. Overall, NIHSS scores were reported for 22.1% of these AIS hospitalizations. Patients with and without administratively recorded NIHSS scores were similar, with none of the 14 non‐NIHSS patient‐level and hospital‐level baseline variables showing a definitely meaningful difference and only the 3 hospital level variables showing potentially meaningful differences (higher NIHSS recording rates at large hospitals, urban academic hospitals, and hospitals in the Western United States) (Table S1). Among the patients with AIS with documented NIHSS, 49.7% were women, median age was 71 (interquartile range [IQR], 60–81) years, and median NIHSS score was 4 (IQR, 2–10).

Among all patients with AIS with NIHSS‐documented hospitalizations, 18 255 (10.16%) presented with dementia (62.5% women). Table 1 shows characteristics of hospitalizations with AIS for those with versus without dementia. The median age was 84 (IQR, 78–89) versus 69 (IQR, 59–79) years (P<0.001), and the median NIHSS was 7 (IQR, 3–15) versus 4 (IQR, 2–10; P<0.001) in hospitalization with versus without dementia. Among White people with stroke, the prevalence of dementia was 11.0%, whereas 8.5% of Black people and 8.6% of other races had a documented dementia at stroke presentation. Among all AIS hospitalization aged ≥80, 24.7% (12 760/516 999) had preexisting dementia, and the median NIHSS was 8 (IQR, 3–16) versus 5 (IQR, 2–13; P<0.001) in hospitalization with versus without dementia among this elderly population.

Dementia and Stroke Severity

There was a clear stepwise association between increasing severity of the index stroke and the prevalence of dementia, ranging from a prevalence of 7.45% among patients with an NIHSS score <6, 12.62% among hospitalizations with an NIHSS score of 6 to 15, 15.2% among those with NIHSS score of 16 to 20, and 17.2% among those with an NIHSS score >20 (P<0.001). A similar trend was observed after restricting the analysis only to hospitalizations among patients aged ≥80 years, with prevalence of dementia ranging from 20.3% among patients with an NIHSS score <6, 29.9% among hospitalizations with an NIHSS score of 6 to 15, and 30.9% among those with an NIHSS >15 (P<0.001). Among stroke hospitalizations with the use of reperfusion therapies, the median NIHSS score was significantly greater in those with dementia (Figure 1).

Use of Reperfusion Therapies

Among AIS hospitalizations with dementia, the frequency of treatment with IVT was 14.35% (2620/18 255), and the frequency of treatment with MT was 3.37% (615/18 255).

Overall, 16.8% (3075/18 255) of patients with dementia and 19.1% (30 895/161 410) of patients without dementia received reperfusion therapy. Specifically, 14.4% (2620/18 255) of patients with dementia and 14.7% (23 765/161 410) of patients without dementia received IVT (with or without MT), whereas 3.4% (615/18 255) of those with dementia and 6% (9600/161 410) of those without dementia received MT (with or without IVT). When analysis was confined to those with documented dementia, the median age was 85 (IQR, 79–89) versus 84 (IQR, 77–89), and the median NIHSS was 13 (IQR, 6–20) versus 6 (IQR, 3–13) in hospitalization with versus without reperfusion therapy. Figure 2 shows the proportion of MT and IVT use in acute stroke hospitalizations based on the presenting stroke severity.

There were no differences between hospitalization with or without dementia by geographic area, rural or urban/academic setting, hospital size, or primary payor (results not presented). In the fully adjusted multivariable model, dementia was associated with significantly lower odds of receiving MT in AIS hospitalization (odds ratio [OR], 0.43 [95% CI, 0.35–0.53]; P < 0.001) (Table 2). However, dementia was not associated with the use of IVT (OR, 0.95 [95% CI, 0.86–1.06]; P=0.4) in AIS hospitalizations. Similar associations were observed when analyses were stratified by age, where dementia (OR, 0.49 [95% CI, 0.38–0.63]; P<0.001) but not IVT (OR, 0.98 [95% CI, 0.86–1.12]; P=0.8) was associated with lower odds of MT use among AIS hospitalizations of patients aged ≥80 years.

A separate analysis was performed to examine reperfusion therapy usage among all AIS hospitalizations regardless of NIHSS documentation. Similarly, dementia was associated with lower odds of MT use (OR, 0.43 [95% CI, 0.35–0.53]; P<0.001) but not associated with IVT utilization (OR, 0.95 [95% CI, 0.86–1.05]; P=0.35).

Dementia and Stroke Outcomes

In‐hospital mortality occurred in 6.3% versus 4.3% in hospitalizations with versus without dementia (P<0.001). In multivariable logistic regression analysis adjusting for age; sex; race; presenting NIHSS score; hospital size and hospital status; and clinical history of hypertension, diabetes, coronary artery disease, congestive heart failure, atrial fibrillation, and chronic kidney disease, there was no association between in‐hospital mortality and dementia among AIS hospitalizations (OR, 0.87 [95% CI, 0.74–1.03]; P=0.11). Similarly, in‐hospital mortality was 8.9% versus 4.4% following IVT, and 21.3% versus 11.2% following MT in hospitalizations with versus without dementia (P<0.001 for both). However, after adjustment for stroke severity and other patient‐ and hospital‐level factors, no associations were found between dementia and in‐hospital mortality among cohorts receiving MT (OR, 1.48 [95% CI, 0.90–2.43]; P=0.15) or IVT (OR, 0.97 [95% CI, 0.67–1.41]; P=0.90) (Table 3). In the fully adjusted model, the odds of discharge to home were significantly lower in AIS hospitalizations with dementia (OR, 0.68 [95% CI, 0.62–0.74]; P<0.001). However, no associations were observed between dementia and favorable discharge outcome among AIS hospitalizations treated with MT (OR, 0.61 [95% CI, 0.33–1.32]; P=0.12). Overall, among patients with dementia and acute stroke who received only IVT, 729 (29.8%) achieved favorable outcome, whereas 265 (43.4%) among this population with presenting NIHSS score <6 achieved a favorable outcome.

Radiographic ICH occurred in 6.0% of AIS hospitalizations with versus 5.7% without preexisting dementia (P=0.11). Radiographic ICH occurred in 30.1% versus 21.2% among MT treated cohort with versus without dementia (P<0.001). In multivariable model, history of dementia was independently associated with increased likelihood of ICH among AIS hospitalizations treated with MT (OR, 1.57 [95% CI, 1.03–2.40]; P=0.04). Similar independent association was observed between dementia and radiographic ICH among MT‐treated patients aged ≥80 (OR, 1.96 [95% CI, 1.14–3.36]; P=0.04). In patients who received MT, the incidence of sepsis (4.8% versus 2.4%; P<0.001) and acute kidney injury (19.5% versus 11.7%; P<0.001) was higher in those with versus without dementia; however, associations disappeared after adjustment for age and stroke severity.

Discussion

In this large nationwide analysis of hospital admissions for AIS, we observed a significant association between preexisting dementia and stroke severity, with a clear stepwise increase in severity of the index stroke deficits at onset and the prevalence of dementia regardless of age. There was a lower use of MT in AIS hospitalizations with preexisting dementia. Among AIS hospitalizations treated with MT, dementia was not an independent predictor of in‐hospital mortality beyond stroke severity; however, MT‐treated patients with dementia were significantly more likely to develop radiographic ICH as compared with MT‐treated patients without dementia.

We observed a higher rate of severe stroke in AIS hospitalization with dementia irrespective of patients’ age, demographic, clinical, and hospital‐level factors. This may be related to the underlying pathophysiology of dementia and cerebrovascular system compromise including functional and structural changes across the entire cerebrovascular tree, from large artery atherosclerosis, and cerebral amyloid angiopathy to small‐vessel disease and the impairment of the blood–brain barrier. ^17^ These pathophysiologic changes, along with the adverse effects of Aβ vasoactivity, diminished cerebral perfusion, and impaired collateral reserve of the brain circulation, may play an important role in rapid deterioration and severity of stroke in these patients. ^17^ , ^18^ , ^19^ Alternatively, stroke severity as judged by the NIHSS might partly be attributable to confusion or poorer understanding of instructions when administering the NIHSS in patients with both dementia and AIS. Our findings in the US population are supported by the results from the prospective Oxford Vascular Study in a population of 92 728 individuals that demonstrated a stepwise association between increasing severity of the index stroke and the prevalence of preevent dementia, ^1^ ranging from a prevalence of 6.8% (95% CI, 5.4–9.0) among patients with an NIHSS score <3% to 20.6% (15.8–26.5) among those with an NIHSS score >10 (P<0.001) irrespective of age.

Current American guidelines for reperfusion therapies in dementia are subject to individual judgment. Prior reports had suggested ^20^ that patients with dementia and AIS are less likely to receive intravenous thrombolysis. Our analysis of data from 2017 suggested no association between intravenous thrombolysis use and dementia, likely representing an increased willingness among US physicians to administer thrombolytic therapies in AIS with dementia. However, patients with AIS and dementia were significantly less likely to be selected for MT. This finding may reflect a more conservative approach by physicians and interventionalists for endovascular therapies in patients with dementia ^21^ likely attributable to lack of randomized data to support efficacy of MT in this patient population.

Prior studies on the effect of dementia on the association between reperfusion therapies and stroke outcomes yielded conflicting results. Whereas some studies have reported that mortality after stroke in dementia is increased regardless of the use of IVT, ^7^ , ^8^ other longitudinal cohorts have not confirmed such associations. ^22^ Busl et al ^7^ reported that the likelihood of mortality and unfavorable discharge outcome increased in patients with more severe strokes (higher admission NIHSS score) and in the presence of prestroke dementia. Authors reported a strong independent association between prestroke dementia with in‐hospital mortality among elderly patients with acute stroke treated with reperfusion therapies. However, associations with unfavorable discharge outcome lost significance in multivariate analysis in patients who received intraarterial therapy. In our analysis of NIS data, we did not observe a significantly increased likelihood of mortality following MT in patients with dementia, although there was a marginally significant association between dementia and mortality in a subset of elderly patients aged ≥80. Similar to the study by Busl et al, our multivariable analysis did not show an association between dementia and discharge outcomes among patients who received endovascular therapies. It should be noted that the study by Busl et al included only the elderly population aged ≥80. In addition, this study did not compare treatment and nontreatment of elderly stroke patients as age‐matched patients with similar stroke severity who were not treated with thrombolytic agents were unavailable for comparison. In a recent longitudinal cohort study of the Swedish dementia and stroke registries, usage and outcomes of IVT were compared between patients with AIS with preexisting dementia (n=1356) versus matched patients without dementia (n=6755). ^22^ Authors reported no differences in mortality in IVT‐treated patients with dementia during hospitalization or at the 3‐ month follow‐up, but functional outcomes were worse. Our findings supported results from this study showing no increase in the rate of in‐hospital mortality but worse functional outcomes among patients with dementia treated with IVT.

Among AIS hospitalizations treated with MT, dementia was associated with a higher risk of radiographic ICH. The increased risk of ICH following MT may be related to the pathological changes in cerebrovascular tree in dementia such as amyloid deposition into the adventitia and media of the cerebral arteries. ^23^ Cerebral amyloid angiopathy has been reported in as high as 85% to 95% of patients with Alzheimer disease and is an important contributor to cerebrovascular pathology in dementia. ^23^ This process weakens the vessel wall and compromises the integrity of the blood–brain barrier, ^17^ which can lead to an increased risk of hemorrhages following endovascular therapies in patients with dementia. This association was clearer in the subset of patients aged ≥80 and may partly explain the nonsignificantly increased risk of mortality in this subset of elderly patients following MT. The rate of sepsis was significantly greater in patients with AIS with dementia. This finding is consistent with prior studies in hospitalized patients where a history of dementia was shown to be independently associated with sepsis. ^24^ , ^25^ , ^26^ Aspiration, self‐neglect, and urinary tract infections are quite common consequences of dementia and related causes of sepsis. In a large population‐based cohort study in Taiwan, dementia was associated with a 32% higher risk of acute organ dysfunction and a 50% higher risk of severe sepsis after controlling age, sex, surgical condition, comorbidity, principal diagnosis, infection status, hospital level, and length of hospital stay. ^25^ Our finding of a higher rate of acute kidney injury in patients with dementia may also be of interest for early preventive strategies. The outcomes of patients with stroke and dementia can be improved by health care professionals’ awareness of the higher risk of end‐organ failure and sepsis among this group of hospitalized patients with stroke.

The increasing number of the elderly population and anticipated increase in the prevalence of cognitive disorders warrant further research to better characterize factors associated with treatment usage and outcomes following strokes in this population. Overall, poor outcome is more likely in patients with dementia irrespective of MT. However, worse functional prognosis after stroke in older patients and in those with dementia could lead to therapeutic nihilism and withholding of treatment. On the other hand, excessive enthusiasm for IVT could expose individuals to serious complications. However, because stroke severity is generally greater in people with dementia and that stroke severity independently affects stroke outcomes, patients with cognitive disorders should not be generally denied the chance to benefit from reperfusion therapy. Further research is needed to identify clinical, imaging, and serologic biomarkers of response to therapy in this population to individualize the risk‐benefit analysis for patient selection in reperfusion therapy, in line with the principles of precision medicine in stroke. ^27^

Our study has limitations. We were not able to examine the association of stroke reperfusion outcomes in subtypes of neurocognitive disorders. The observational retrospective design of the study limits assessment of long‐term outcomes. We did not have data on all clinical factors such as blood pressure variations that might drive rates of complications and postreperfusion outcome. Though derived from an NIS, the relatively small sample size limits an extensive subgroup analysis by various underlying cognitive disorders. We followed the previously described approaches for identifying dementia hospitalizations and NIHSS scores, ^28^ but accuracy of case ascertainment is dependent upon accuracy of ICD‐10‐CM coding as indexed by hospital coders. Although NIHSS reporting performance is more likely in hospitalizations associated with endovascular thrombosis or thrombolytic use, ^29^ the NIHSS reporting performance for patients with dementia versus without dementia is likely nondifferential. However, our findings may not be fully generalizable to the hospitalizations without available NIHSS reporting. The observational retrospective design of the current analysis limits the ability to assess long‐term outcomes and associate stroke reperfusion outcomes with specific neurocognitive disorder subtypes. For instance, in patients with vascular dementia versus degenerative type of dementia, the risk for ICH postreperfusion therapies can potentially vary on the basis of the underlying neurocognitive pathology. Further analysis was limited by a lack of individual level data in the NIS including dementia severity, first‐time stroke versus recurrent stroke event, thrombectomy eligibility, time from onset to presentation, and baseline modified Rankin Scale and functional status. For example, it is possible that patients with higher baseline mRS scores (that are more likely to be in the dementia group) were more likely to be excluded from thrombectomy procedures.

Conclusion

Among patients with AIS, preexisting dementia is associated with stroke severity at onset. However, MT is used less frequently in AIS hospitalizations with documented dementia. Dementia is not an independent predictor of in‐hospital mortality beyond stroke severity at onset, although it increases the likelihood of radiographic ICH after MT. Sepsis is more prevalent following stroke reperfusion therapy in hospitalizations with dementia. Further studies are needed to clarify imaging and clinical markers associated with outcome in dementia patients to identify subgroups of patients with increased risks, and those who are more likely to benefit from reperfusion therapies.

Sources of Funding

None.

Disclosures

Dr Liebeskind reports other from Cerenovus, Genentech, Medtronic, Stryker, and Vesalio outside the submitted work. Dr Tateshima reports other from Medtronic, Stryker, and Cerenovus. Dr Saver reports personal fees from Medtronic, Stryker, Cerenovus, Boehringer Ingelhiem (prevention only), and Rapid Medical outside the submitted work; and the University of California has intellectual property rights in retriever technology for stroke.

Supporting information