Development of a decision analytic model to support decision making and risk communication about thrombolytic treatment

Advanced Machine Learning Models for Predicting Post-Thrombolysis Hemorrhagic Transformation in Acute Ischemic Stroke Patients: A Systematic Review and Meta-Analysis

Background: Thrombolytic therapy is essential for acute ischemic stroke (AIS) management but poses a risk of hemorrhagic transformation (HT), necessitating accurate prediction to optimize patient care. Methods: A comprehensive search was conducted across PubMed, Web of Science, Scopus, Embase, and Google Scholar, covering studies from inception until July 10, 2024. Studies were included if they used machine learning (ML) or deep learning algorithms to predict HT in AIS patients treated with thrombolysis. Exclusion criteria included studies involving endovascular treatments and those not evaluating model effectiveness. Data extraction and quality assessment were performed following PRISMA guidelines and using the Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis (TRIPOD) and Prediction Model Risk of Bias Assessment Tool (PROBAST) tools. Results: Out of 1943 identified records, 12 studies were included in the final analysis, encompassing 18 007 AIS patients who received thrombolytic therapy. The ML models demonstrated high predictive performance, with pooled area under the curve (AUC) values ranging from 0.79 to 0.95. Specifically, XGBoost models achieved AUCs of up to 0.953 and Artificial Neural Network (ANN) models reached up to 0.942. Sensitivity and specificity varied significantly, with the highest sensitivity at 0.90 and specificity at 0.99. Significant predictors of HT included age, glucose levels, NIH Stroke Scale (NIHSS) score, systolic and diastolic blood pressure, and radiomic features. Despite these promising results, methodological disparities and limited external validation highlighted the need for standardized reporting and further rigorous testing. Conclusion: ML techniques, especially XGBoost and ANN, show great promise in predicting HT following thrombolysis in AIS patients, enhancing risk stratification and clinical decision-making. Future research should focus on prospective study designs, standardized reporting, and integrating ML assessments into clinical workflows to improve AIS management and patient outcomes.

Thrombolytic Refusal Over Telestroke

Background

Tissue plasminogen activator (tPA) refusal is 4%-6% for acute ischemic stroke (AIS) in the emergency department. Telestroke (TS) has increased the use of tPA for AIS but is accompanied by barriers in communication that can affect tPA consent. We characterized the incidence of tPA refusal in our TS network and its associated reasons.

Methods

Patients with AIS who were offered tPA within 4.5 hours from symptom onset according to American Heart Association guidelines were identified within our Lone Star Stroke Consortium Telestroke Registry from September 2015 to December 2018. We compared baseline characteristics and clinical outcomes between patients who refused tPA and patients who accepted tPA.

Results

Among the 1,242 patients who qualified for tPA and were offered treatment, 8% refused tPA. Female and non-Hispanic Black patients and patients with a prior history of stroke were more likely to decline tPA. Patients who refused tPA presented with a lower NIHSS and were associated with a final diagnosis of stroke mimic (odds ratio [OR] 0.23; 95% confidence interval [CI] 0.15-0.36). Good outcome (90-day modified Rankin Scale 0-2) was the same among patients who received tPA and those who refused (OR 0.80; 95% CI 0.42-1.54). The most common reasons for refusal were rapidly improving and mild/nondisabling symptoms and concern for potential side effects.

Conclusion

tPA refusal over TS is comparable to previously reported rates; there was no difference in outcomes among patients who received tPA compared with those who refused. Sex and racial differences associated with an increased tPA refusal warrant further investigation in efforts to achieve equity/parity in tPA decisions.

Decision-Making Visual Aids for Late, Imaging-Guided Endovascular Thrombectomy for Acute Ischemic Stroke

BACKGROUND AND PURPOSE

Speedy decision-making is important for optimal outcomes from endovascular thrombectomy (EVT) for acute ischemic stroke (AIS). Figural decision aids facilitate rapid review of treatment benefits and harms, but have not yet been developed for late-presenting patients selected for EVT based on multimodal computed tomography or magnetic resonance imaging.

METHODS

For combined pooled study-level randomized trial (DAWN and DEFUSE 3) data, as well as each trial singly, 100 person-icon arrays (Kuiper-Marshall personographs) were generated showing beneficial and adverse effects of EVT for patients with AIS and large vessel occlusion using automated (algorithmic) and expert-guided joint outcome table specification.

RESULTS

Among imaging-selected patients 6 to 24 hours from last known well, for the full 7-category modified Rankin Scale (mRS), EVT had number needed to treat to benefit 1.9 (interquartile range [IQR], 1.9 to 2.1) and number needed to harm 40.0 (IQR, 29.2 to 58.3). Visual displays of treatment effects among 100 patients showed that, with EVT: 52 patients have better disability outcome, including 32 more achieving functional independence (mRS 0 to 2); three patients have worse disability outcome, including one more experiencing severe disability or death (mRS 5 to 6), mediated by symptomatic intracranial hemorrhage and infarct in new territory. Similar features were present in person-icon figures based on a 6-level mRS (levels 5 and 6 combined) rather than 7-level mRS, and based on the DAWN trial alone and DEFUSE 3 trial alone.

CONCLUSIONS

Personograph visual decision aids are now available to rapidly educate patients, family, and healthcare providers regarding benefits and risks of EVT for late-presenting, imaging-selected AIS patients.

Executive Summary of the Early-Onset Breast Cancer Evidence Review Conference

The Centers for Disease Control and Prevention launched the Bring Your Brave campaign to increase knowledge about early-onset breast cancer, defined as breast cancer in women aged 18-45 years. The American College of Obstetricians and Gynecologists convened a panel of experts in breast disease from the Society for Academic Specialists in General Obstetrics and Gynecology to review relevant literature, validated tools, best practices, and practice guidelines as a first step toward developing educational materials for women's health care providers about early-onset breast cancer. Panel members conducted structured literature reviews, which were then reviewed by other panel members and discussed at an in-person meeting of stakeholder professional and patient advocacy organizations in April 2019. This article summarizes the relevant literature, existing guidance, and validated tools to guide health care providers in the prevention, early detection, and special considerations of early-onset breast cancer. Substantive knowledge gaps were noted and summarized to provide guidance for future research.

Paramedic Acute Stroke Treatment Assessment (PASTA): study protocol for a randomised controlled trial

BACKGROUND

Despite evidence from clinical trials that intravenous (IV) thrombolysis is a cost-effective treatment for selected acute ischaemic stroke patients, there remain large variations in the rate of IV thrombolysis delivery between stroke services. This study is evaluating whether an enhanced care pathway delivered by paramedics (the Paramedic Acute Stroke Treatment Assessment (PASTA)) could increase the number of patients who receive IV thrombolysis treatment.

METHODS

Study design: Cluster randomised trial with economic analysis and parallel process evaluation.

SETTING

National Health Service ambulance services, emergency departments and hyper-acute stroke units within three geographical regions of England and Wales. Randomisation: Ambulance stations within each region are the units of randomisation. According to station allocation, paramedics based at a station deliver the PASTA pathway (intervention) or continue with standard stroke care (control). Study intervention: The PASTA pathway includes structured pre-hospital information collection, prompted pre-notification, structured handover of information in hospital and assistance with simple tasks during the initial hospital assessment. Study-trained intervention group paramedics deliver this pathway to adults within 4 h of suspected stroke onset. Study control: Standard stroke care according to national and local guidelines for the pre-hospital and hospital assessment of suspected stroke.

PARTICIPANTS

Participants enrolled in the study are adults with confirmed stroke who were assessed by a study paramedic within 4 h of symptom onset.

PRIMARY OUTCOME

Proportion of participants receiving IV thrombolysis.

SAMPLE SIZE

1297 participants provide 90% power to detect a 10% difference in the proportion of patients receiving IV thrombolysis.

DISCUSSION

The results from this trial will determine whether an enhanced care pathway delivered by paramedics can increase thrombolysis delivery rates.

TRIAL REGISTRATION

ISRCTN registry, ISRCTN12418919 . Registered on 5 November 2015.

National Practice Patterns of Obtaining Informed Consent for Stroke Thrombolysis

BACKGROUND AND PURPOSE

No standard approach to obtaining informed consent for stroke thrombolysis with tPA (tissue-type plasminogen activator) currently exists. We aimed to assess current nationwide practice patterns of obtaining informed consent for tPA.

METHODS

An online survey was developed and distributed by e-mail to clinicians involved in acute stroke care. Multivariable logistic regression analyses were performed to determine independent factors contributing to always obtaining informed consent for tPA.

RESULTS

Among 268 respondents, 36.7% reported always obtaining informed consent and 51.8% reported the informed consent process caused treatment delays. Being an emergency medicine physician (odds ratio, 5.8; 95% confidence interval, 2.9-11.5) and practicing at a nonacademic medical center (odds ratio, 2.1; 95% confidence interval, 1.0-4.3) were independently associated with always requiring informed consent. The most commonly cited cause of delay was waiting for a patient's family to reach consensus about treatment.

CONCLUSIONS

Most clinicians always or often require informed consent for stroke thrombolysis. Future research should focus on standardizing content and delivery of tPA information to reduce delays.

Racial disparities in refusal of stroke thrombolysis in Chicago

OBJECTIVE

To evaluate race differences in tissue plasminogen activator (tPA) refusal among eligible patients with acute ischemic stroke (AIS) in Chicago.

METHODS

Using the Get With The Guidelines-Stroke registry data from 15 primary stroke centers between January 2013 and June 2015, we performed a retrospective analysis of patients with AIS presenting to the emergency department within 4.5 hours from symptom onset. Patient or proxy refusal was captured as a reason for nonadministration of tPA to eligible patients in the registry. We assessed whether tPA refusal differed by race using logistic regression.

RESULTS

Among 704 tPA-eligible patients with AIS, tPA was administered to 86.2% (black race, 82.5% vs nonblack race, 89.5%; p < 0.001). Fifty-three (7.5%) tPA refusals were documented. Refusal was more common in black vs nonblack patients (10.6% vs 4.8%; p = 0.004). In multivariable analysis, the following were associated with tPA refusal: black race (adjusted odds ratio [OR] 2.5, 95% confidence interval [CI] 1.3-4.6), self-pay status (adjusted OR 3.23, 95% CI 1.2-8.71), prior stroke (adjusted OR 2.11, 95% CI 1.14-3.90), age (adjusted OR 1.04, 95% CI 1.02-1.07), and NIH Stroke Scale score (adjusted OR 0.94, 95% CI 0.90-0.99).

CONCLUSIONS

Among tPA-eligible patients with AIS in Chicago, over 7% refused tPA. Refusal was more common in black patients and accounted for the apparent lower rates of tPA use in black vs nonblack patients. Further research is needed to understand barriers to consent and overcome race-ethnic disparities in tPA treatment for AIS.

Shared decision-making in stroke: an evolving approach to improved patient care

Shared decision-making (SDM) occurs when patients, families and clinicians consider patients' values and preferences alongside the best medical evidence and partner to make the best decision for a given patient in a specific scenario. SDM is increasingly promoted within Western contexts and is also being explored outside such settings, including in China. SDM and tools to promote SDM can improve patients' knowledge/understanding, participation in the decision-making process, satisfaction and trust in the healthcare team. SDM has also proposed long-term benefits to patients, clinicians, organisations and healthcare systems. To successfully perform SDM, clinicians must know their patients' values and goals and the evidence underlying different diagnostic and treatment options. This is relevant for decisions throughout stroke care, from thrombolysis to goals of care, diagnostic assessments, rehabilitation strategies, and secondary stroke prevention. Various physician, patient, family, cultural and system barriers to SDM exist. Strategies to overcome these barriers and facilitate SDM include clinician motivation, patient participation, adequate time and tools to support the process, such as decision aids. Although research about SDM in stroke care is lacking, decision aids are available for select decisions, such as anticoagulation for stroke prevention in atrial fibrillation. Future research is needed regarding both cultural aspects of successful SDM and application of SDM to stroke-specific contexts.

CT and MRI-based door-needle-times for acute stroke patients a quasi-randomized clinical trial

OBJECTIVES

Door-Needle-times (DNT) of 20min are feasible when Computer Tomography (CT) is used for first-line brain-imaging to assess stroke-patients' eligibility for intravenous-tissue-Plasminogen-Activator (iv-tPA), but the more time-consuming Magnetic Resonance Imaging (MRI)-based-evaluation is superior in detecting acute ischaemia.

Development and Testing of Shared Decision Making Interventions for Use in Emergency Care: A Research Agenda

Decision aids are evidenced-based tools designed to increase patient understanding of medical options and possible outcomes, facilitate conversation between patients and clinicians, and improve patient engagement. Decision aids have been used for shared decision making (SDM) interventions outside of the ED setting for more than a decade. Their use in the ED has only recently begun to be studied. This article provides background on this topic and the conclusions of the 2016 Academic Emergency Medicine consensus conference SDM in practice work group regarding "Shared Decision Making in the Emergency Department: Development of a Policy-Relevant, Patient-Centered Research Agenda." The goal was to determine a prioritized research agenda for the development and testing of SDM interventions for use in emergency care that was most important to patients, clinicians, caregivers, and other key stakeholders. Using the nominal group technique, the consensus working group proposed prioritized research questions in six key domains: 1) content (i.e., clinical scenario or decision area), 2) level of evidence available, 3) tool design strategies, 4) risk communication, 5) stakeholders, and 6) outcomes.

Development of a computerised decision aid for thrombolysis in acute stroke care

BACKGROUND

Thrombolytic treatment for acute ischaemic stroke improves prognosis, although there is a risk of bleeding complications leading to early death/severe disability. Benefit from thrombolysis is time dependent and treatment must be administered within 4.5 hours from onset of symptoms, which presents unique challenges for development of tools to support decision making and patient understanding about treatment. Our aim was to develop a decision aid to support patient-specific clinical decision-making about thrombolysis for acute ischaemic stroke, and clinical communication of personalised information on benefits/risks of thrombolysis by clinicians to patients/relatives.

METHODS

Using mixed methods we developed a COMPuterised decision Aid for Stroke thrombolysiS (COMPASS) in an iterative staged process (review of available tools; a decision analytic model; interactive group workshops with clinicians and patients/relatives; and prototype usability testing). We then tested the tool in simulated situations with final testing in real life stroke thrombolysis decisions in hospitals. Clinicians used COMPASS pragmatically in managing acute stroke patients potentially eligible for thrombolysis; their experience was assessed using self-completion forms and interviews. Computer logged data assessed time in use, and utilisation of graphical risk presentations and additional features. Patients'/relatives' experiences of discussions supported by COMPASS were explored using interviews.

RESULTS

COMPASS expresses predicted outcomes (bleeding complications, death, and extent of disability) with and without thrombolysis, presented numerically (percentages and natural frequencies) and graphically (pictographs, bar graphs and flowcharts). COMPASS was used for 25 patients and no adverse effects of use were reported. Median time in use was 2.8 minutes. Graphical risk presentations were shared with 14 patients/relatives. Clinicians (n = 10) valued the patient-specific predictions of benefit from thrombolysis, and the support of better risk communication with patients/relatives. Patients (n = 2) and relatives (n = 6) reported that graphical risk presentations facilitated understanding of benefits/risks of thrombolysis. Additional features (e.g. dosage calculator) were suggested and subsequently embedded within COMPASS to enhance usability.

CONCLUSIONS

Our structured development process led to the development of a gamma prototype computerised decision aid. Initial evaluation has demonstrated reasonable acceptability of COMPASS amongst patients, relatives and clinicians. The impact of COMPASS on clinical outcomes requires wider prospective evaluation in clinical settings.