Teleneurology and Artificial Intelligence in Clinical Practice.

ABSTRACT: As teleheath becomes integrated into the practice of medicine, it is important to understand the benefits, limitations, and variety of applications. Telestroke was an early example of teleneurology that arose from a need for urgent access to neurologists for time-sensitive treatments for stroke. It made a scarce resource widely available via video conferencing technologies. Additionally, applications such as outpatient video visits, electronic consultation (e-consult), and wearable devices developed in neurology, as well. Telehealth dramatically increased during the COVID-19 pandemic when offices were closed and hospitals were overwhelmed; a multitude of both outpatient and inpatient programs developed and matured during this time. It is helpful to explore what has been learned regarding the quality of telehealth, disparities in care, and how artificial intelligence can interact with medical practices in the teleneurology context.

Implications of Large Language Models for Quality and Efficiency of Neurologic Care: Emerging Issues in Neurology.

Large language models (LLMs) are advanced artificial intelligence (AI) systems that excel in recognizing and generating human-like language, possibly serving as valuable tools for neurology-related information tasks. Although LLMs have shown remarkable potential in various areas, their performance in the dynamic environment of daily clinical practice remains uncertain. This article outlines multiple limitations and challenges of using LLMs in clinical settings that need to be addressed, including limited clinical reasoning, variable reliability and accuracy, reproducibility bias, self-serving bias, sponsorship bias, and potential for exacerbating health care disparities. These challenges are further compounded by practical business considerations and infrastructure requirements, including associated costs. To overcome these hurdles and harness the potential of LLMs effectively, this article includes considerations for health care organizations, researchers, and neurologists contemplating the use of LLMs in clinical practice. It is essential for health care organizations to cultivate a culture that welcomes AI solutions and aligns them seamlessly with health care operations. Clear objectives and business plans should guide the selection of AI solutions, ensuring they meet organizational needs and budget considerations. Engaging both clinical and nonclinical stakeholders can help secure necessary resources, foster trust, and ensure the long-term sustainability of AI implementations. Testing, validation, training, and ongoing monitoring are pivotal for successful integration. For neurologists, safeguarding patient data privacy is paramount. Seeking guidance from institutional information technology resources for informed, compliant decisions, and remaining vigilant against biases in LLM outputs are essential practices in responsible and unbiased utilization of AI tools. In research, obtaining institutional review board approval is crucial when dealing with patient data, even if deidentified, to ensure ethical use. Compliance with established guidelines like SPIRIT-AI, MI-CLAIM, and CONSORT-AI is necessary to maintain consistency and mitigate biases in AI research. In summary, the integration of LLMs into clinical neurology offers immense promise while presenting formidable challenges. Awareness of these considerations is vital for harnessing the potential of AI in neurologic care effectively and enhancing patient care quality and safety. The article serves as a guide for health care organizations, researchers, and neurologists navigating this transformative landscape.

Descriptive Analysis of Patients Receiving Outpatient eConsults for Neurological Disorders in the United States.

Introduction: Interprofessional consultations ("eConsults"), which facilitate asynchronous specialist consultations, remain understudied in neurological disorders. We aimed to describe the patient population receiving eConsult services for neurological disorders nationwide and to conduct a comparative analysis between rural and urban patients within this eConsult cohort. Methods: We analyzed a dataset of U.S. outpatient claims from employer-sponsored commercial and Medicare plans. Using standardized mean differences, we compared clinical and sociodemographic patient characteristics between urban and rural patients within the eConsult group. Results: We identified 1,374 patients who had an eConsult order for a neurological disorder. Overall eConsult volume increased by 548.5% between 2019 and 2021. A majority of the cohort were aged 65 years or older (23.7%), had an eConsult order in 2021 (52.4%), and live in an urban area (90.4%). The primary diagnosis for our cohort was likely to be a sleep-wake disorder (21.9%), cerebrovascular disease (14.3%), neurological sign or symptom (14.2%), or headache (13.7%). In the secondary analysis, rural eConsult patients exhibited higher rates of primary diagnoses for traumatic brain injury, neuroophthalmic disorders, or neuropathy than their urban counterparts. Discussion: In this national sample of commercially insured patients, the utilization of eConsults for neurological conditions increased nationwide since 2019, especially for patients living in rural areas.

Healthcare utilization impacts of an eConsult program for headache at an academic medical center.

INTRODUCTION: Interprofessional consultations ("eConsults") can reduce healthcare utilization. However, the impact of eConsults on healthcare utilization remains poorly characterized among patients with headache. METHODS: We performed a retrospective, 1:1 matched cohort study comparing patients evaluated for headache via eConsult request or in-person referral at the Mount Sinai Health System in New York. Groups were matched on clinical and demographic characteristics. Our primary outcome was one or more outpatient headache-related encounters in 6 months following referral date. Secondary outcomes included one or more all-cause outpatient neurology and headache-related emergency department (ED) encounters during the same period. We used univariable and multivariable logistic regression to model associations between independent variables and outcomes. RESULTS: We identified 74 patients with headache eConsults who were matched to 74 patients with in-person referrals. Patients in the eConsult group were less likely to achieve the primary outcome (29.7% vs 62.2%, P < 0.0001) or have an all-cause outpatient neurology encounter (33.8% vs 79.7%, P < 0.0001) than patients in the comparison group. Both groups did not significantly differ by headache-related ED encounters. In multivariable analyses, patients in the eConsult group had significantly lower odds of having one or more headache-related or all-cause neurology encounters than patients in the comparison group (odds ratio (OR) 0.3, 95% confidence interval (CI) 0.1-0.6; OR 0.1, 95% CI 0.1-0.3, respectively). DISCUSSION: In comparison to in-person referrals, eConsult requests for headache were associated with reduced likelihood of outpatient neurology encounters in the short-term but not with differential use of headache-related ED encounters. Larger-scale, prospective studies should validate our findings and assess patient outcomes.

Latent class analysis of eHealth behaviors among adults with epilepsy.

OBJECTIVE: The objective of this study was to determine the proportions of uptake and factors associated with electronic health (eHealth) behaviors among adults with epilepsy. METHODS: The 2013, 2015, and 2017 National Health Interview Surveys were analyzed. We assessed the proportions of use of five domains of eHealth in those with epilepsy: looked up health information on the internet, filled a prescription on the internet, scheduled a medical appointment on the internet, communicated with a health care provider via email, and used chat groups to learn about health topics. Multivariate logistic regressions were conducted to identify factors associated with any eHealth behaviors among those with active epilepsy. Latent class analysis was performed to identify underlying patterns of eHealth activity. Survey participants were classified into three discrete classes: (1) frequent, (2) infrequent, and (3) nonusers of eHealth. Multinomial logistic regression was performed to identify factors associated with frequency of eHealth use. RESULTS: There were 1770 adults with epilepsy, of whom 65.87% had at least one eHealth behavior in the prior year. By domain, 62.61% looked up health information on the internet, 15.81% filled a prescription on the internet, 14.95% scheduled a medical appointment on the internet, 17.20% communicated with a health care provider via email, and 8.27% used chat groups to learn about health topics. Among those with active epilepsy, female sex, more frequent computer usage, and internet usage were associated with any eHealth behavior. Female sex and frequent computer use were associated with frequent eHealth use as compared to nonusers. SIGNIFICANCE: A majority of persons with epilepsy were found to use at least one form of eHealth. Various technological and demographic factors were associated with eHealth behaviors. Individuals with lower eHealth behaviors should be provided with targeted interventions that address barriers to the adoption of these technologies.

Disparities in telehealth utilization in patients with pain during COVID-19.

Introduction: The shift from in-person visits to telehealth visits during the COVID-19 pandemic presented unique challenges for patients with pain. Disparities in health care access already existed, and the impact of telehealth on these inequities has not been studied. Objectives: To identify sociodemographic characteristics of patients with pain obtaining care through video, telephone, and in-person visits as social distancing restrictions evolved during the COVID-19 pandemic. Methods: Using our institutional clinical data warehouse, we identified 3314 patients with pain receiving care at a large academic institution in New York City during a baseline period (September 23, 2019-March 22, 2020) and counted telephone, video, and in-person visits during the following conditions: a shutdown period (March 23, 2020-May 23, 2020), when nonessential in-person visits were strictly limited, and a reopening period (May 23, 2020-September 23, 2020), when restrictions were relaxed and in-person visits were available. Patients were categorized into 4 groups based on the technology used to complete a visit: (1) video, (2) telephone, (3) in-person, and (4) no visit. Results: Patients who were older, publicly insured, and identified as Black or Hispanic were overrepresented in the telephone visit group during shutdown and the in-person group during reopening. A video visit during shutdown increased the likelihood of continued video visit use during reopening despite the return of in-person visits. Conclusions: Results show differences in how patients with pain accessed clinical care in a socially distanced world and that flexibility in method of health care delivery may reduce barriers to access. Future research will identify factors (eg, Internet access, digital literacy, provider-patient relationships) driving heterogeneity in telehealth use in patients with pain.

Trends in the Utilization of Teleneurology and Other Healthcare Resources Prior to and During the COVID-19 Pandemic in an Urban, Tertiary Health System.

BACKGROUND: Patient groups traditionally affected by health disparities were less likely to use video teleneurology (TN) care during the initial COVID-19 pandemic surge in the United States. Whether this asymmetry persisted later in the pandemic or was accompanied with a loss of access to care remains unknown. METHODS: We conducted a retrospective cohort study using patient data from a multicenter healthcare system in New York City. We identified all established pediatric or adult neurology patients with at least two prior outpatient visits between June 16th, 2019 and March 15th, 2020 using our electronic medical record. For this established pre-COVID cohort, we identified telephone, in-person, video TN or emergency department visits and hospital admissions for any cause between March 16th and December 15th, 2020 ("COVID period"). We determined clinical, sociodemographic, income, and visit characteristics. Our primary outcome was video TN utilization, and our main secondary outcome was loss to follow-up during the COVID period. We used multivariable logistic regression to model the relationship between patient-level characteristics and both outcomes. RESULTS: We identified 23,714 unique visits during the COVID period, which corresponded to 14,170 established patients from our institutional Neurology clinics during the pre-COVID period. In our cohort, 4,944 (34.9%) utilized TN and 4,997 (35.3%) were entirely lost to follow-up during the COVID period. In the adjusted regression analysis, Black or African-American race [adjusted odds ratio (aOR) 0.60, 97.5%CI 0.52-0.70], non-English preferred language (aOR 0.49, 97.5%CI 0.39-0.61), Medicaid insurance (aOR 0.50, 97.5%CI 0.44-0.57), and Medicare insurance (aOR 0.73, 97.5%CI 0.65-0.83) had decreased odds of TN utilization. Older age (aOR 0.98, 97.5%CI 0.98-0.99), female sex (aOR 0.90 97.5%CI 0.83-0.99), and Medicaid insurance (aOR 0.78, 0.68-0.90) were associated with decreased odds of loss to follow-up. CONCLUSION: In the first 9 months of the COVID-19 pandemic, we found sociodemographic patterns in TN utilization that were similar to those found very early in the pandemic. However, these sociodemographic characteristics were not associated with loss to follow-up, suggesting that lack of TN utilization may not have coincided with loss of access to care.

Treating High-Risk TIA and Minor Stroke Patients With Dual Antiplatelet Therapy: A National Survey of Emergency Medicine Physicians.

BACKGROUND: Treatment with aspirin plus clopidogrel, dual antiplatelet therapy (DAPT), within 24 hours of high-risk transient ischemic attack (TIA) or minor stroke symptoms to eligible patients is recommended by national guidelines. Whether or not this treatment has been adopted by emergency medicine (EM) physicians is uncertain. METHODS: We conducted an online survey of EM physicians in the United States. The survey consisted of 13 multiple choice questions regarding physician characteristics, practice settings, and usual approach to TIA and minor stroke treatment. We report participant characteristics and use chi-squared tests to compare between groups. RESULTS: We included 162 participants in the final study analysis. 103 participants (64%) were in practice for >5 years and 96 (59%) were at nonacademic centers; all were EM board-certified or board-eligible. Only 9 (6%) participants reported that they would start DAPT for minor stroke and 8 (5%) reported that they would start DAPT after high-risk TIA. Aspirin alone was the selected treatment by 81 (50%) participants for minor stroke patients who presented within 24 hours of symptom onset and were not candidates for thrombolysis. For minor stroke, 69 (43%) participants indicated that they would defer medical management to consultants or another team. Similarly, 75 (46%) of participants chose aspirin alone to treat high-risk TIA; 74 (46%) reported they would defer medical management after TIA to consultants or another team. CONCLUSION: In a survey of EM physicians, we found that the reported rate of DAPT treatment for eligible patients with high-risk TIA and minor stroke was low.

A retrospective cohort study of clinical factors, visit patterns, and demographic factors associated with use of remote communications in patients with headache.

OBJECTIVE: To identify the characteristics associated with high utilization of remote communications (RCs) in patients with headache. BACKGROUND: Patients with headache frequently communicate with their providers using secure portal messaging and telephone calls. However, clinical and demographic factors as well as visit patterns associated with RC utilization remain poorly characterized. METHODS: We retrospectively analyzed data from patients with headache who were evaluated in the ambulatory neurology faculty practice at the Icahn School of Medicine at Mount Sinai in New York between January 1 and June 30, 2019. We extracted clinical and demographic characteristics, total office visits, secure MyChart portal messages, and telephone encounters from our institutional data warehouse. We defined high RC and MyChart utilization as the top tertile of RC and MyChart message volume, respectively, and assessed the relationship between patient characteristics and high RC (primary outcome), as well as high MyChart utilization (secondary outcome). We characterized the relationship between clinicodemographic characteristics and the ratio of MyChart messages to total RCs (secondary outcome). RESULTS: We identified 1390 patients, of whom 477 (34.3%) were high RC utilizers and 321 (23.1%) were high MyChart utilizers. High RC utilizers generated 3306/3921 (84.3%) RCs. The presence of chronic headache (aOR 2.31, 95% CI 1.75-3.03, p < 0.0001), cluster headache (aOR 18.3, 95% CI 5.0-71.7, p = 0.001), and migraine (aOR 3.82, 95% CI 1.93-9.3, p = 0.011) was associated with high RC utilization. Patients ≥65 years of age were less likely to engage in MyChart messaging as a proportion of RC (191/680, 28.1%) compared with patients 18-30 years of age (243/620, 39.2%, p = 0.049) and 30-64 years of age (1172/2721, 43.1%, p < 0.0001). CONCLUSIONS: A minority of patients with headache (477/1390; 34.3%) generated the majority (3306/3921; 84.3%) of RCs. Our findings should be validated in external patient cohorts with the objective of developing strategies to optimize RC utilization.

Usage Patterns of Web-Based Stroke Calculators in Clinical Decision Support: Retrospective Analysis.

BACKGROUND: Clinical scores are frequently used in the diagnosis and management of stroke. While medical calculators are increasingly important support tools for clinical decisions, the uptake and use of common medical calculators for stroke remain poorly characterized. OBJECTIVE: We aimed to describe use patterns in frequently used stroke-related medical calculators for clinical decisions from a web-based support system. METHODS: We conducted a retrospective study of calculators from MDCalc, a web-based and mobile app-based medical calculator platform based in the United States. We analyzed metadata tags from MDCalc's calculator use data to identify all calculators related to stroke. Using relative page views as a measure of calculator use, we determined the 5 most frequently used stroke-related calculators between January 2016 and December 2018. For all 5 calculators, we determined cumulative and quarterly use, mode of access (eg, app or web browser), and both US and international distributions of use. We compared cumulative use in the 2016-2018 period with use from January 2011 to December 2015. RESULTS: Over the study period, we identified 454 MDCalc calculators, of which 48 (10.6%) were related to stroke. Of these, the 5 most frequently used calculators were the CHA2DS2-VASc score for atrial fibrillation stroke risk calculator (5.5% of total and 32% of stroke-related page views), the Mean Arterial Pressure calculator (2.4% of total and 14.0% of stroke-related page views), the HAS-BLED score for major bleeding risk (1.9% of total and 11.4% of stroke-related page views), the National Institutes of Health Stroke Scale (NIHSS) score calculator (1.7% of total and 10.1% of stroke-related page views), and the CHADS2 score for atrial fibrillation stroke risk calculator (1.4% of total and 8.1% of stroke-related page views). Web browser was the most common mode of access, accounting for 82.7%-91.2% of individual stroke calculator page views. Access originated most frequently from the most populated regions within the United States. Internationally, use originated mostly from English-language countries. The NIHSS score calculator demonstrated the greatest increase in page views (238.1% increase) between the first and last quarters of the study period. CONCLUSIONS: The most frequently used stroke calculators were the CHA2DS2-VASc, Mean Arterial Pressure, HAS-BLED, NIHSS, and CHADS2. These were mainly accessed by web browser, from English-speaking countries, and from highly populated areas. Further studies should investigate barriers to stroke calculator adoption and the effect of calculator use on the application of best practices in cerebrovascular disease.

Teleneurology Expansion in Response to the COVID-19 Outbreak at a Tertiary Health System in New York City.

OBJECTIVE: To assess the implementation of teleneurology (TN), including patient and clinician experiences, during the coronavirus respiratory disease 2019 (COVID-19) pandemic. METHODS: We studied synchronous (video visit) and asynchronous (store-and-forward, patient-portal evaluation, remote monitoring) TN utilization in the Mount Sinai Health System Neurology Department in New York, 2 months before and after the start of our department's response to the pandemic in mid-March 2020. Weekly division meetings enabled ongoing assessments and analysis of barriers and facilitators according to the Consolidated Framework for Implementation Research and the Expert Recommendations for Implementing Change models. We used postvisit surveys of clinicians (from April 13 to May 15, 2020) and patients (from May 11 to 15, 2020) to determine technology platforms used, and TN experience and acceptability, using Likert scales (1 = very poor/unlikely to 5 = very good/likely). RESULTS: Over the 4-month period, 117 TN clinicians (n = 14 subspecialties) conducted 4,225 TN visits with 3,717 patients (52 pre- vs 4,173 post-COVID-19). No asynchronous TN services were delivered. Post-COVID-19, the number of TN clinicians, subspecialties performing TN, and visits increased by 963%, 133%, and 7,925%, respectively. Mean acceptability among patients and clinicians was 4.7 (SD 0.6) and 3.4 (SD 1.6), respectively. Most video visits were completed using Epic MyChart (78.5%) and Zoom (8.1%). TN implementation facilitators included Medicare geographic restriction waivers, development of clinician educational materials, and MyChart outreach programs for patients experiencing technical difficulties. CONCLUSIONS: A significant expansion of TN utilization accompanied the COVID-19 response. Patients found TN more acceptable than did clinicians. Proactive application of an implementation framework facilitated rapid and effective TN expansion.

Predicting adult neuroscience intensive care unit admission from emergency department triage using a retrospective, tabular-free text machine learning approach.

Early admission to the neurosciences intensive care unit (NSICU) is associated with improved patient outcomes. Natural language processing offers new possibilities for mining free text in electronic health record data. We sought to develop a machine learning model using both tabular and free text data to identify patients requiring NSICU admission shortly after arrival to the emergency department (ED). We conducted a single-center, retrospective cohort study of adult patients at the Mount Sinai Hospital, an academic medical center in New York City. All patients presenting to our institutional ED between January 2014 and December 2018 were included. Structured (tabular) demographic, clinical, bed movement record data, and free text data from triage notes were extracted from our institutional data warehouse. A machine learning model was trained to predict likelihood of NSICU admission at 30 min from arrival to the ED. We identified 412,858 patients presenting to the ED over the study period, of whom 1900 (0.5%) were admitted to the NSICU. The daily median number of ED presentations was 231 (IQR 200-256) and the median time from ED presentation to the decision for NSICU admission was 169 min (IQR 80-324). A model trained only with text data had an area under the receiver-operating curve (AUC) of 0.90 (95% confidence interval (CI) 0.87-0.91). A structured data-only model had an AUC of 0.92 (95% CI 0.91-0.94). A combined model trained on structured and text data had an AUC of 0.93 (95% CI 0.92-0.95). At a false positive rate of 1:100 (99% specificity), the combined model was 58% sensitive for identifying NSICU admission. A machine learning model using structured and free text data can predict NSICU admission soon after ED arrival. This may potentially improve ED and NSICU resource allocation. Further studies should validate our findings.

Comparative analysis, applications, and interpretation of electronic health record-based stroke phenotyping methods.

BACKGROUND: Accurate identification of acute ischemic stroke (AIS) patient cohorts is essential for a wide range of clinical investigations. Automated phenotyping methods that leverage electronic health records (EHRs) represent a fundamentally new approach cohort identification without current laborious and ungeneralizable generation of phenotyping algorithms. We systematically compared and evaluated the ability of machine learning algorithms and case-control combinations to phenotype acute ischemic stroke patients using data from an EHR. MATERIALS AND METHODS: Using structured patient data from the EHR at a tertiary-care hospital system, we built and evaluated machine learning models to identify patients with AIS based on 75 different case-control and classifier combinations. We then estimated the prevalence of AIS patients across the EHR. Finally, we externally validated the ability of the models to detect AIS patients without AIS diagnosis codes using the UK Biobank. RESULTS: Across all models, we found that the mean AUROC for detecting AIS was 0.963 ± 0.0520 and average precision score 0.790 ± 0.196 with minimal feature processing. Classifiers trained with cases with AIS diagnosis codes and controls with no cerebrovascular disease codes had the best average F1 score (0.832 ± 0.0383). In the external validation, we found that the top probabilities from a model-predicted AIS cohort were significantly enriched for AIS patients without AIS diagnosis codes (60-150 fold over expected). CONCLUSIONS: Our findings support machine learning algorithms as a generalizable way to accurately identify AIS patients without using process-intensive manual feature curation. When a set of AIS patients is unavailable, diagnosis codes may be used to train classifier models.

Calculated Decisions: Intracerebral hemorrhage (ICH) score.

The ICH score grades intracerebral hemorrhage severity and subsequent 30-day mortality based on age and CT findings.

Calculated Decisions: Secondary intracerebral hemorrhage (sICH) score.

The sICH score quantifies the likelihood of underlying vascular etiology in patients with intracerebral hemorrhage.

Emergent Large Vessel Occlusion Stroke During New York City's COVID-19 Outbreak: Clinical Characteristics and Paraclinical Findings.

BACKGROUND AND PURPOSE: The 2019 novel coronavirus outbreak and its associated disease (coronavirus disease 2019 [COVID-19]) have created a worldwide pandemic. Early data suggest higher rate of ischemic stroke in severe COVID-19 infection. We evaluated whether a relationship exists between emergent large vessel occlusion (ELVO) and the ongoing COVID-19 outbreak. METHODS: This is a retrospective, observational case series. Data were collected from all patients who presented with ELVO to the Mount Sinai Health System Hospitals across New York City during the peak 3 weeks of hospitalization and death from COVID-19. Patients' demographic, comorbid conditions, cardiovascular risk factors, COVID-19 disease status, and clinical presentation were extracted from the electronic medical record. Comparison was made between COVID-19 positive and negative cohorts. The incidence of ELVO stroke was compared with the pre-COVID period. RESULTS: Forty-five consecutive ELVO patients presented during the observation period. Fifty-three percent of patients tested positive for COVID-19. Total patients' mean (±SD) age was 66 (±17). Patients with COVID-19 were significantly younger than patients without COVID-19, 59±13 versus 74±17 (odds ratio [95% CI], 0.94 [0.81-0.98]; P=0.004). Seventy-five percent of patients with COVID-19 were male compared with 43% of patients without COVID-19 (odds ratio [95% CI], 3.99 [1.12-14.17]; P=0.032). Patients with COVID-19 were less likely to be White (8% versus 38% [odds ratio (95% CI), 0.15 (0.04-0.81); P=0.027]). In comparison to a similar time duration before the COVID-19 outbreak, a 2-fold increase in the total number of ELVO was observed (estimate: 0.78 [95% CI, 0.47-1.08], P≤0.0001). CONCLUSIONS: More than half of the ELVO stroke patients during the peak time of the New York City's COVID-19 outbreak were COVID-19 positive, and those patients with COVID-19 were younger, more likely to be male, and less likely to be White. Our findings also suggest an increase in the incidence of ELVO stroke during the peak of the COVID-19 outbreak.

History of Stroke Is Independently Associated With In-Hospital Death in Patients With COVID-19.

BACKGROUND AND PURPOSE: In December 2019, an outbreak of severe acute respiratory syndrome coronavirus causing coronavirus disease 2019 (COVID-19) occurred in China, and evolved into a worldwide pandemic. It remains unclear whether the history of cerebrovascular disease is associated with in-hospital death in patients with COVID-19. METHODS: We conducted a retrospective, multicenter cohort study at Mount Sinai Health System in New York City. Using our institutional data warehouse, we identified all adult patients who were admitted to the hospital between March 1, 2020 and May 1, 2020 and had a positive nasopharyngeal swab polymerase chain reaction test for severe acute respiratory syndrome coronavirus in the emergency department. Using our institutional electronic health record, we extracted clinical characteristics of the cohort, including age, sex, and comorbidities. Using multivariable logistic regression to control for medical comorbidities, we modeled the relationship between history of stroke and all-cause, in-hospital death. RESULTS: We identified 3248 patients, of whom 387 (11.9%) had a history of stroke. Compared with patients without history of stroke, patients with a history of stroke were significantly older, and were significantly more likely to have a history of all medical comorbidities except for obesity, which was more prevalent in patients without a history of stroke. Compared with patients without history of stroke, patients with a history of stroke had higher in-hospital death rates during the study period (48.6% versus 31.7%, P<0.001). In the multivariable analysis, history of stroke (adjusted odds ratio, 1.28 [95% CI, 1.01-1.63]) was significantly associated with in-hospital death. CONCLUSIONS: We found that history of stroke was associated with in-hospital death among hospitalized patients with COVID-19. Further studies should confirm these results.

A Multilevel Analysis of Surgical Category and Individual Patient-Level Risk Factors for Postoperative Stroke.

BACKGROUND AND PURPOSE: Many studies supporting the association between specific surgical procedure categories and postoperative stroke (POS) do not account for differences in patient-level characteristics between and within surgical categories. The risk of POS after high-risk procedure categories remains unknown after adjusting for such differences in patient-level characteristics. METHODS: Using inpatients in the American College of Surgeons National Surgical Quality Initiative Program database, we conducted a retrospective cohort study between January 1, 2000, and December 31, 2010. Our primary outcome was POS within 30 days of surgery. We characterized the relationship between surgical- and individual patient-level factors and POS by using multivariable, multilevel logistic regression that accounted for clustering of patient-level factors with surgical categories. RESULTS: We identified 729 886 patients, 2703 (0.3%) of whom developed POS. Dependent functional status (odds ratio [OR]: 4.11, 95% confidence interval [95% CI]: 3.60-4.69), history of stroke (OR: 2.35, 95%CI: 2.06-2.69) or transient ischemic attack (OR: 2.49 95%CI: 2.19-2.83), active smoking (OR: 1.20, 95%CI: 1.08-1.32), hypertension (OR: 2.11, 95%CI: 2.19-2.82), chronic obstructive pulmonary disease (OR: 1.39 95%CI: 1.21-1.59), and acute renal failure (OR: 2.35, 95%CI: 1.85-2.99) were significantly associated with POS. After adjusting for clustering, patients who underwent cardiac (OR: 11.25, 95%CI: 8.52-14.87), vascular (OR: 4.75, 95%CI: 3.88-5.82), neurological (OR: 4.60, 95%CI: 3.48-6.08), and general surgery (OR: 1.40, 95%CI: 1.15-1.70) had significantly greater odds of POS compared to patients undergoing other types of surgical procedures. CONCLUSIONS: Vascular, cardiac, and neurological surgery remained strongly associated with POS in an analysis accounting for the association between patient-level factors and surgical categories.