Introduction: Patients presenting to emergency departments (ED) with TIA and minor strokes (TIAMS) are often admitted for expedited evaluation, though outpatient care models have been proposed. We piloted a rapid outpatient evaluation protocol for patients presenting with TIAMS within 24 hours of ED discharge. We hypothesized that this approach would reduce hospital costs and length of stay (LOS). Methods: This analysis looked at patients presenting to our institution’s ED with TIAMS (NIHSS < 5) in calendar year 2017. We compared hospitalization LOS, costs and expected revenues between admitted patients and those referred for rapid outpatient evaluation. Patients eligible for outpatient evaluation were without disabling deficits, recurrent symptoms, new-onset atrial fibrillation, prior carotid imaging with >50% stenosis, and not receiving thrombolysis. Disabling deficits were defined as new gait impairments, significant motor weakness, hemianopia, dysphagia or severe aphasia. Cost data was obtained from our finance department and expected revenue was estimated using Medicare reimbursement data, assuming Medicare-Fee for Service as the primary payer for all patients. Results: We identified 92 patients referred to our rapid outpatient clinic and 90 admitted patients (mean NIHSS 0.8 vs 1.8 respectively). In comparison to patients who were admitted, patients referred to outpatient evaluation had shorter hospital stays, lower total hospitalization costs, and decreased net-losses after accounting for expected revenue (Table). Only one patient in the outpatient cohort was readmitted for further management. Overall, the one-year pilot cohort averted approximately 138 bed-days and $950,000 in hospitalization costs. Conclusions: For patients who presented to our ED with TIAMS without disabling deficits, rapid outpatient evaluation reduced hospital LOS and total costs. Further research is needed to incorporate costs to payers and patients.

May 7, 2019April 9, 2019Free AccessA hospital’s perspective: economic evaluation of hospitalization vs rapid outpatient evaluation for TIA and minor strokes. (P3.3-010)Steven Shapiro, Jorge Luna, Rachel Mehendale, Babak Navi, Benjamin Kummer, Sara Rostanski, Claudia Rosen, David Vawdrey, Bernard Chang, Eliza Miller, Mitchell Elkind, and Joshua WilleyAuthors Info & AffiliationsApril 9, 2019 issue92 (15_supplement)https://doi.org/10.1212/WNL.92.15_supplement.P3.3-010 Letters to the Editor

OBJECTIVE:We hypothesized that ILI is associated with risk of incident stroke, and that the risk would be highest closest in time to the event.METHODS:This case-crossover analysis utilized data obtained from the California State Inpatient Database of the Healthcare Cost and Utilization Project (HCUP). The outcome of interest was ischemic stroke. Exposure was defined as a visit to the emergency department or hospitalization for influenza-like illness (ILI) 365, 180, 90, 30, or 15 days before stroke (risk period) or similar time intervals exactly 1 or 2 years before stroke (control period). Conditional logistic regression was used to calculate the odds ratio and 95% confidence interval (OR, 95% CI).RESULTS:In 2009, 36,975 hospitalized ischemic strokes met inclusion criteria, and of these strokes, 554 (1.5%) had at least 1 episode of ILI in the 365-day risk period prior to their stroke. Using non-overlapping time intervals from ILI to stroke, the odds of ischemic stroke was greatest in the first 15 days post ILI (OR: 2.88, 95% CI: 1.86-4.47). The strength of the relationship decreased as the time from ILI increased, and was no longer significant after 60 days. There was a significant interaction (P = 0.017) with age and ILI; the odds of stroke associated increased 7% with each 10-year decrease in age (OR per 10-year age decrease 1.07, 95% CI: 1.03-1.35).INTERPRETATION:We found that ILI increases short-term risk of stroke, particularly in people under the age of 45, and therefore may be considered to act as a trigger for stroke.

Development and maintenance of order sets is a knowledge-intensive task for off-the-shelf machine-learning algorithms alone. We hypothesize that integrating clinical knowledge with machine learning can facilitate effective development and maintenance of order sets while promoting best practices in ordering. To this end, we simulated the revision of an "AM Lab Order Set" under 6 revision approaches. Revisions included changes in the order set content or default settings through 1) population statistics, 2) individualized prediction using machine learning, and 3) clinical knowledge. Revision criteria were determined using electronic health record (EHR) data from 2014 to 2015. Each revision's clinical appropriateness, workload from using the order set, and generalizability across time were evaluated using EHR data from 2016 and 2017. Our results suggest a potential order set revision approach that jointly leverages clinical knowledge and machine learning to improve usability while updating contents based on latest clinical knowledge and best practices.

•Comprehend the impact of palliative care consultation on cost of hospitalization.•Comprehend the methods used to estimate the impact of palliative care consultation on cost of hospitalization. Earlier palliative care consultations have been shown to reduce length of stay (LOS) and overall costs, while improving the quality of medical care afforded, in most, but not all studies. To estimate the unbiased impact of early initiation of palliative care on LOS and total direct costs. This observational study retrospectively assessed patients who utilized adult palliative care consultation services at an academic medical center between the dates of January 2015 and September 2015. Palliative consults, DNR status and ICU utilization were extracted from the electronic health record. Direct costs were extracted from the Allscripts EPSi accounting module. Comorbidity information was calculated using claims records processed with CMS-HCC (v2016). The effectiveness of early onset (0-3 days from admission), as compared to mid onset (4-10 days) and late onset (>10 days) palliative consults, on LOS and total direct costs were estimated using targeted maximum likelihood (propensity-score based) additive effect estimates. Palliative consultation was administered to 1425 inpatient visits, stratified by the timing of consultation yields: early (n=467), mid (n=523), late (n=435). Shifting onset of palliative consultation within the first 3 days of admission is associated with a cost reduction of $1,575 per visit (p=0.019) among the mid-onset group; and reduction of $2,073 per visit (p=0.042) among the late-onset group. Estimated impacts of LOS are 1.4 days reduction (p=0.004) in the mid-onset group; and 2.2 days reduction (p=0.011) in the late-onset group. Subgroup analysis among critical care units and oncology wards demonstrated larger cost savings. Early initiation of palliative care is associated with reduction of length of stay and total direct costs.

Background and Purpose— Infections have been found to increase the risk of stroke over the short term. We hypothesized that stroke risk would be highest shortly after a sepsis hospitalization, but that the risk would decrease, yet remain up to 1 year after sepsis. Methods— This case-crossover analysis utilized data obtained from the California State Inpatient Database of the Healthcare Cost and Utilization Project. All stroke admissions were included. Exposure was defined as hospitalization for sepsis or septicemia 180, 90, 30, or 15 days before stroke (risk period) or similar time intervals exactly 1 or 2 years before stroke (control period). Conditional logistic regression was used to calculate the odds ratio (OR) and 95% confidence interval (95% CI) for the association between sepsis/septicemia and ischemic or hemorrhagic stroke. Results— Ischemic (n=37 377) and hemorrhagic (n=12 817) strokes that occurred in 2009 were extracted where 3188 (8.5%) ischemic and 1101 (8.6%) hemorrhagic stroke patients had sepsis. Sepsis within 15 days before the stroke placed patients at the highest risk of ischemic (OR, 28.36; 95% CI, 20.02–40.10) and hemorrhagic stroke (OR, 12.10; 95% CI, 7.54–19.42); however, although the risk decreased, it remained elevated 181 to 365 days after sepsis for ischemic (OR, 2.59; 95% CI, 2.20–3.06) and hemorrhagic (OR, 3.92; 95% CI 3.29–4.69) strokes. There was an interaction with age ( P =0.0006); risk of developing an ischemic stroke within 180 days of hospitalization for sepsis increased 18% with each 10-year decrease in age. Conclusions— Risk of stroke is high after sepsis, and this risk persists for up to a year. Younger sepsis patients have a particularly increased risk of stroke after sepsis.

Introduction: Real-time identification of patients with acute ischemic stroke (AIS) in the electronic health record (EHR) can enhance care delivery systems, clinical decision support, and research subject recruitment. EHR data that is accessible during a patient’s admission may be used to identify patients with AIS, but established methods for characterizing which data to use have not yet been determined. Hypotheses: 1. An EHR “phenotype” of AIS can be identified using clinical EHR data. 2. Machine learning can identify the AIS phenotype using similar inputs with greater accuracy than clinician-specified identification algorithms. Methods: Two stroke neurologists selected generalizable AIS-related clinical data points from the Columbia University Medical Center EHR (clinical laboratory results and medication, imaging, and stroke service list orders) to identify the AIS phenotype, and determined pre-specified priority logic based on institutional practice patterns. Separately, a regularized logistic regression (RLR) model was applied to all available neurology-related order sets and clinical laboratory inputs. The classification accuracy of the two algorithms was compared using a “gold standard” data set, consisting of our institution’s ischemic stroke registry from January 1 st , 2015 to March 31 st , 2016. Negative controls were selected from all patients admitted to the neurology service at our institution during the same time period. Results: Our data contained 482 patients with AIS and 3,628 negative controls. The clinician-specified identification algorithm identified the AIS phenotype with sensitivity of 90.6%, specificity of 50.4%, and positive predictive value (PPV) of 93.5%. In comparison, the RLR-based algorithm had a sensitivity of 96.3%, specificity of 52.2%, and PPV of 93.8%. Conclusions: We determined an AIS phenotype that could be identified using clinical, non-claims data that is available during a patient’s admission, and used machine learning to optimize the classifying ability. While specificity is low, the high sensitivity may allow use for screening and clinical decision support. Further studies are needed to externally validate these findings and optimize algorithm specificity.

Background and Purpose— Case–control studies suggest that acute infection transiently increases the risk of childhood arterial ischemic stroke. We hypothesized that an unbiased pathogen discovery approach utilizing MassTag–polymerase chain reaction would identify pathogens in the blood of childhood arterial ischemic stroke cases. Methods— The multicenter international VIPS study (Vascular Effects of Infection in Pediatric Stroke) enrolled arterial ischemic stroke cases, and stroke-free controls, aged 29 days through 18 years. Parental interview included questions on recent infections. In this pilot study, we used MassTag–polymerase chain reaction to test the plasma of the first 161 cases and 34 controls enrolled for a panel of 28 common bacterial and viral pathogens. Results— Pathogen DNA was detected in no controls and 14 cases (8.7%): parvovirus B19 (n=10), herpesvirus 6 (n=2), adenovirus (n=1), and rhinovirus 6C (n=1). Parvovirus B19 infection was confirmed by serologies in all 10; infection was subclinical in 8. Four cases with parvovirus B19 had underlying congenital heart disease, whereas another 5 had a distinct arteriopathy involving a long-segment stenosis of the distal internal carotid and proximal middle cerebral arteries. Conclusions— Using MassTag–polymerase chain reaction, we detected parvovirus B19—a virus known to infect erythrocytes and endothelial cells—in some cases of childhood arterial ischemic stroke. This approach can generate new, testable hypotheses about childhood stroke pathogenesis.

Introduction: Retrospective identification of patients hospitalized with new diagnosis of acute ischemic stroke is important for administrative quality assurance, post-discharge clinical management, and stroke research. The benefit of using administrative claims data is its widespread availability, but the disadvantage is in the inability to accurately and consistently identify the clinical diagnosis of interest. Hypothesis: We hypothesized that decision tree and logistic regression models could be applied to administrative claims data coded using International Classification of Diseases, version 10 (ICD-10) to create algorithms that could accurately identify patients with acute ischemic stroke. Methods: We used hospital records from our institution to develop a gold standard list of 243 patients, continuously hospitalized with a new diagnosis of stroke from 10/1/2015 to 3/31/2016. We used 1,393 neurological patients without a diagnosis of stroke as negative controls. This list was used to train and test two machine learning methods of diagnosis and procedure codes analysis, for the purpose of ischemic stroke identification: one using classification and regression tree (CART) and another using regularized logistic regression. We trained the models using 75% of the data and performed the evaluation using the remaining 25%. Results: The CART model had a κ=0.78, sensitivity of 96%, specificity of 90%, and a positive predictive value of 99%. The regularized logistic regression model had a κ=0.73, sensitivity of 97%, specificity of 81%, and a positive predictive value of 98%. Conclusion: Both the decision tree and logistic regression machine based learning models showed very high accuracy in identifying patients with a new diagnosis of ischemic stroke, using ICD-10 code claims data, when compared to our gold standard. Applying these machine learning models to identify patients with ischemic stroke has widespread applications, especially in this period where national billing data has transitioned from ICD-9 to ICD-10 codes.

Introduction: Stroke research using widely available institutional, state-wide and national retrospective data is dependent on accurate identification of stroke subtypes using claims data. Despite the abundance of such data and the advances in clinical informatics, there is limited published data on the application of machine learning models to improve previously reported administrative stroke identification algorithms. Hypothesis: We hypothesized that machine learning models can be applied to claims data coded using the International Classification of Disease, version 9 (ICD-9), to accuracy identify patients with ischemic stroke (IS), intracerebral hemorrhage (ICH), and subarachnoid hemorrhage (SAH), and these models would outperform previously published algorithms in our patient cohort. Methods: We developed a gold standard list of 427 stroke patients continuously admitted to our institution from 1/1/2015 to 9/30/2015 using an internal stroke database and applied 75% of it to train and 25% to test two machine learning models: one using classification and regression tree (CART) and another using regularized logistic regression. There were 2,241 negative controls. We further applied a previously reported stroke detection algorithm, by Tirschwell and Longstreth, to our cohort for comparison. Results: The CART model had a κ of 0.72, 0.82, 0.59; sensitivity of 95%, 99%, 99%; and a specificity of 88%, 78%, 75%; for IS, ICH and SAH respectively. The regularized logistic regression model had a κ of 0.73, 0.80, 0.59; sensitivity of 95%, 99%, 99%, and a specificity of 89%, 78%, 75%; for IS, ICH and SAH respectively. The previously reported algorithm by Tirschwell et al, had a κ of 0.71,0.56, 0.64; sensitivity of 98%, 99%, 99%; and a specificity of 64%, 52%, 50%; for IS, ICH and SAH. Conclusion: Compared with the previously reported ICD 9 based detection algorithm, the machine learning models had a higher κ for diagnosis of IS and ICH, similar sensitivity for all subtypes, and higher specificity for all stroke subtypes in our cohort. Applying machine learning models to identify stroke subtypes from administrative data sets, can lead to highly accurate models of stroke subtype identification for health services researchers.

Background: C-reactive protein predicts prognosis after stroke, but relationships of other inflammatory biomarkers to prognosis is uncertain. We hypothesized that concentrations of interleukin 6 (IL6), serum amyloid A, tumor necrosis factor-α receptor 1 (TNFR1), CD40 ligand, and monocyte chemoattractant protein 1 predict recurrent major vascular events (MVE) after lacunar stroke. Methods: Levels of Inflammatory Markers in the Treatment of Stroke (LIMITS) was an international, multicenter, ancillary biomarker study nested within the Secondary Prevention of Small Subcortical Strokes (SPS3) Phase 3 trial in patients with recent lacunar stroke. Patients were randomized to aspirin versus aspirin/clopidogrel. Blood samples were collected at enrollment, and markers measured centrally using ELISA. Cox proportional hazards models were used to calculate hazard ratios and 95% confidence intervals (HR, 95% CI) for risk of MVE (stroke, myocardial infarction, vascular death) after adjusting for demographics, comorbidities, and statin use. Results: Among 1244 lacunar stroke patients (mean age 63.3 ± 10.8 years), there were 115 MVE. Risk increased with concentrations of both TNFR1 (adj HR per standard deviation [SD] 1.21, 95% CI 1.05-1.41) and IL6 (adj HR per SD 1.10, 95% CI 1.02-1.19). Compared with the bottom quartile of TNFR1, those in the top quartile had twice the risk after adjusting for demographics (HR 1.98, 95% CI 1.11-3.52), though this attenuated after adjusting for other risk factors (adjusted HR 1.68, 95% CI 0.93-3.04). There was an interaction between antiplatelet assignment and TNFR1 (p=0.008; figure) and IL6 quartiles (p=0.035); as biomarker concentrations increased, dual antiplatelets became less effective than aspirin alone. Other markers were not associated with prognosis. Conclusions: Among recent lacunar stroke patients, IL6 and TNF receptor concentrations predict risk of recurrent vascular events and efficacy of antiplatelet therapies.

Background and Purpose— We hypothesized that concentrations of interleukin 6 (IL-6), serum amyloid A, tumor necrosis factor-α receptor 1, CD40 ligand, and monocyte chemoattractant protein 1 would predict recurrent ischemic stroke and major vascular events after recent lacunar stroke. Methods— Levels of Inflammatory Markers in the Treatment of Stroke (LIMITS) was an international, multicenter, prospective ancillary biomarker study nested within the Secondary Prevention of Small Subcortical Strokes (SPS3) study, a Phase III trial in patients with recent lacunar stroke. Crude and Adjusted Cox proportional hazards models were used to calculate hazard ratios (HRs) and 95% confidence intervals (95% CI) for recurrence risks. Results— Among 1244 patients with lacunar stroke (mean age, 63.3±10.8 years), there were 115 major vascular events (stroke, myocardial infarction, and vascular death). The risk of major vascular events increased with elevated concentrations of both tumor necrosis factor-α receptor 1 (adjusted HR per SD, 1.21; 95% CI, 1.05–1.41; P =0.01) and IL-6 (adjusted HR per SD, 1.10; 95% CI, 1.02–1.19; P =0.008). Compared with the bottom quartile (tumor necrosis factor-α receptor 1 <2.24 ng/L), those in the top quartile of tumor necrosis factor-α receptor 1 (>3.63 ng/L) were at twice the risk of major vascular events after adjusting for demographics (partially adjusted HR, 1.98; 95% CI, 1.11–3.52), though the effect attenuated after adjusting for other risk factors and statin use (adjusted HR, 1.68; 95% CI, 0.93–3.04). Serum amyloid A, CD40 ligand, and monocyte chemoattractant protein 1 were not associated with prognosis. Conclusions— Among recent lacunar stroke patients, IL-6 and TNF receptor concentrations predict risk of recurrent vascular events, and they are associated with the effect of antiplatelet therapies. Clinical Trial Registration— URL: http://www.clinicaltrials.gov . Unique identifier: NCT00059306.

Background: The role of inflammation in cerebral small vessel disease remains uncertain. Tumor necrosis factor-alpha receptor 1 (TNFR1) has been associated with atherosclerosis and risk of stroke. We hypothesized that TNFR1 concentrations would be associated with cerebral white matter disease (WMD) and subclinical infarcts in patients with recent lacunar stroke. Methods: Levels of Inflammatory Markers in the Treatment of Stroke (LIMITS) was an international, multicenter, ancillary biomarker study nested within the Secondary Prevention of Small Subcortical Strokes trial (SPS3; www.clinicaltrials.gov unique identifier: NCT00059306), a Phase III trial in patients with recent lacunar stroke. Patients had blood samples collected at enrollment, and concentrations of inflammatory biomarkers, including TNFR1, were measured using ELISA at a central laboratory. Enrollment MRI scans were read centrally and interpreted for silent infarcts and burden of WMD using semi-quantitative scales. We compared proportions of patients with prior infarcts and WMD scores across quartiles of TNFR1, and used logistic regression to estimate odds ratios and 95% confidence intervals (OR, 95%CI) to examine the relationship between TNFR1 and subclinical disease after adjusting for demographics and comorbidities. Results: Among 1004 lacunar stroke patients with TNFR1 data (mean age 63.3 ± 10.8 years), 407 (40%) had infarcts besides the qualifying infarct; half the cohort had 0-4 white matter lesions, 27% had 5-8 lesions, and 23% had >=9 lesions. TNFR1 levels were associated with WMD score >=9 (OR per standard deviation (SD) TNFR1=1.2, 95%CI 1.0-1.4). Larger proportions of those in the top quartile of TNFR1, compared to those in the lowest, had ≥9 white matter lesions (29% versus 19%, p=0.026) and additional infarcts (45% versus 37%, p=0.28). After adjusting for demographics and comorbidities, the effect of TNFR1 on WMD score >=9 (adjusted OR per SD=1.1, 95%CI 1.0-1.3) and additional infarcts (adjusted OR 1.2, 95%CI 1.0-1.3) attenuated. Conclusions: Among recent lacunar stroke patients, TNFR1 concentrations were associated with prior or subclinical cerebrovascular disease. Future studies of TNF and TNF inhibitors in cerebral ischemic disease may be warranted.

Clinical teams in acute inpatient settings can greatly benefit from automated charting technologies that continuously monitor patient vital status. NewYork-Presbyterian has designed and developed a real-time patient monitoring system that integrates vital signs sensors, networking, and electronic health records, to allow for automatic charting of patient status. We evaluate the representativeness (a combination of agreement, safety and timing) of a core vital sign across nursing intensity care protocols for preliminary feasibility assessment. Our findings suggest an automated way of summarizing heart rate provides representation of true heart rate status and can facilitate alternatives approaches to burdensome manual nurse charting of physiological parameters.

BACKGROUND:Although designated stroke centers (DSCs) improve the quality of care and clinical outcomes for ischemic stroke patients, less is known about the benefits of DSCs for patients with intracerebral hemorrhage (ICH) and subarachnoid hemorrhage (SAH).HYPOTHESIS:Compared to non-DSCs, hospitals with the DSC status have lower in-hospital mortality rates for hemorrhagic stroke patients. We believed these effects would sustain over a period of time after adjusting for hospital-level characteristics, including hospital size, urban location, and teaching status.METHODS AND RESULTS:We evaluated ICH (International Classification of Diseases, Ninth Revision; ICD-9: 431) and SAH (ICD-9: 430) hospitalizations documented in the 2008-2012 New York State Department of Health Statewide Planning and Research Cooperative System inpatient sample database. Generalized estimating equation logistic regression was used to evaluate the association between DSC status and in-hospital mortality. We calculated ORs and 95% CIs adjusted for clustering of patients within facilities, other hospital characteristics, and individual level characteristics. Planned secondary analyses explored other hospital characteristics associated with in-hospital mortality. In 6,352 ICH and 3,369 SAH patients in the study sample, in-hospital mortality was higher among those with ICH compared to SAH (23.7 vs. 18.5%). Unadjusted analyses revealed that DSC status was related with reduced mortality for both ICH (OR 0.7, 95% CI 0.5-0.8) and SAH patients (OR 0.4, 95% CI 0.3-0.7). DSC remained a significant predictor of lower in-hospital mortality for SAH patients (OR 0.6, 95% CI 0.3-0.9) but not for ICH patients (OR 0.8, 95% CI 0.6-1.0) after adjusting for patient demographic characteristics, comorbidities, hospital size, teaching status and location.CONCLUSIONS:Admission to a DSC was independently associated with reduced in-hospital mortality for SAH patients but not for those with ICH. Other patient and hospital characteristics may explain the benefits of DSC status on outcomes after ICH. For conditions with clear treatments such as ischemic stroke and SAH, being treated in a DSC improves outcomes, but this trend was not observed in those with strokes, in those who did not have clear treatment guidelines. Identifying hospital-level factors associated with ICH and SAH represents a means to identify and improve gaps in stroke systems of care.

Background: Acute triggers of ischemic stroke (IS) remain poorly characterized. Emerging evidence suggests infections may promote short-term pro-inflammatory, pro-coagulant states that precipitate stroke. Objective: We hypothesized that exposure to influenza-like illness (ILI) is associated with increased risk of IS, as time intervals between events decrease. Methods: HCUP/AHRQ administrative claims from all nonfederal hospitals in California (2009) were queried for IS, using a published ICD-9-CM surveillance algorithm, identifying 41,148 unique individuals. A patient identifier was used to link IS patients across hospitalizations with a validated algorithm for ILI using inpatient and emergency department records. For each IS case, ILI events within 15, 30, and 90-day risk periods preceding IS in 2009, were compared to the same time periods one and two calendar years prior (i.e., 2007 and 2008). A 4-day buffer period was used to ensure ILI preceded IS. Conditional logistic regression models, matched on individuals, were used to calculate odds ratios and 95% confidence intervals (OR, 95%CI) for given risk period, after adjusting for monthly prevalence and mean age of ILI hospitalizations. Effects stratified by age are also explored. Results: Median (IQR) age of cases was 74 (62-83) years; 52.4% were female. There were more ILI during the 90-days preceding strokes in 2009 (n=439) than in the same stroke-free calendar period one year prior (n=303) and two years prior (n=81). ILI was associated with IS with decreasing magnitude of effect using windows of 15-days (adjusted OR 6.5, 95%CI, 2.2-19.7), 30-days (adjusted OR=3.7, 95%CI 1.7-8.3), and 90-days (adjusted OR=3.3, 95%CI, 1.9-5.8). The risk associated with ILI was higher among younger patients. Using the 30-day window, the association between ILI and IS was strongest among those 45-to-65-years (n=27,545, adjusted OR=2.53, 95%CI, 0.95-6.8). Conclusion: Influenza-like illness may contribute to a heightened risk of ischemic stroke during short term periods post-infection, especially among younger patients. Presentation with influenza represents an opportunity for targeted prevention of stroke.

Objective:Structural interventions can reduce HIV vulnerability. However, HIV-specific budgeting, based on HIV-specific outcomes alone, could lead to the undervaluation of investments in such interventions and suboptimal resource allocation. We investigate this hypothesis by examining the consequences of alternative financing approaches. Methods:We compare three approaches for deciding whether to finance a structural intervention to keep adolescent girls in school in Malawi. In the first, HIV and non-HIV budget holders participate in a cross-sectoral cost–benefit analysis and fund the intervention if the benefits outweigh the costs. In the second silo approach, each budget holder considers the cost-effectiveness of the intervention in terms of their own objectives and funds the intervention on the basis of their sector-specific thresholds of what is cost-effective or not. In the third cofinancing approach, budget holders use cost-effectiveness analysis to determine how much they would be willing to contribute towards the intervention, provided that other sectors are willing to pay for the remaining costs. In addition, we explore approaches for determining the HIV share in the cofinancing scenario. Results:We find that efficient structural interventions may be less likely to be prioritized, financed and taken to scale where sectors evaluate their options in isolation. A cofinancing approach minimizes welfare loss and could be incorporated in a sector budgeting perspective. Conclusion:Structural interventions may be underimplemented and their cross-sectoral benefits foregone. Cofinancing provides an opportunity for multiple HIV, health and development objectives to be achieved simultaneously, but will require effective cross-sectoral coordination mechanisms for planning, implementation and financing.

Objective: To determine whether high sensitivity C-reactive protein (hsCRP) predicts recurrent stroke and other vascular events among recent lacunar stroke patients. Background: Inflammatory markers have been associated with risk of first stroke. Their role in predicting recurrence is unclear. Methods: The Levels of Inflammatory Markers in the Treatment of Stroke study is an international prospective study of inflammatory markers among recent lacunar stroke patients enrolled in the NIH-funded randomized Secondary Prevention of Small Subcortical Strokes trial. Patients had blood samples drawn, saved at -80 degrees C, and run at a central lab for hsCRP using nephelometry. Cox proportional hazard models were used to estimate hazard ratios and 95% confidence intervals (HR, 95% CI) for associations of hsCRP with recurrence risk before and after adjusting for demographics, comorbidities, and statin use. Results: Among 1244 lacunar stroke patients (mean 63.3 ± 10.8 years), median hsCRP was 2.16 mg/L (interquartile range 0.93-4.86), and levels differed by age, sex, smoking, and LDL. Median time between stroke and hsCRP measurement was 60 days, and levels were inversely and weakly correlated with proximity to stroke date (r=-0.06, p=0.039). There were 83 recurrent ischemic strokes (45 lacunes), 16 hemorrhages, and 115 major vascular events (stroke, MI, vascular death). Compared to the bottom quartile, those in the top quartile of hsCRP (>4.86 mg/dl) were at increased risk of recurrent ischemic stroke (unadjusted HR 2.54, 95% CI 1.30-4.96), and the risk persisted after adjusting for age, sex, race, region, hypertension, smoking, prior history of stroke, diabetes, lipid levels, and statin use (adjusted HR 2.28, 95% CI 1.14-4.57). HsCRP was associated with an increased risk of major vascular events (top quartile adjusted HR 1.98, 95% CI 1.11-3.54). Results were similar using clinical thresholds of high risk hsCRP (> 3 mg/dl). There was no interaction of randomized antiplatelet treatment with hsCRP levels for stroke risk. Conclusions: Among recent lacunar stroke patients, elevated hsCRP levels predict increased risk of recurrent strokes and other vascular events. Levels of inflammatory markers did not predict a response to dual antiplatelet treatment.