Diurnal variations in the quality of stroke care in Sweden

Developing machine learning models to predict multi-class functional outcomes and death three months after stroke in Sweden

Globally, stroke is the third-leading cause of mortality and disability combined, and one of the costliest diseases in society. More accurate predictions of stroke outcomes can guide healthcare organizations in allocating appropriate resources to improve care and reduce both the economic and social burden of the disease. We aim to develop and evaluate the performance and explainability of three supervised machine learning models and the traditional multinomial logistic regression (mLR) in predicting functional dependence and death three months after stroke, using routinely-collected data. This prognostic study included adult patients, registered in the Swedish Stroke Registry (Riksstroke) from 2015 to 2020. Riksstroke contains information on stroke care and outcomes among patients treated in hospitals in Sweden. Prognostic factors (features) included demographic characteristics, pre-stroke functional status, cardiovascular risk factors, medications, acute care, stroke type, and severity. The outcome was measured using the modified Rankin Scale at three months after stroke (a scale of 0-2 indicates independent, 3-5 dependent, and 6 dead). Outcome prediction models included support vector machines, artificial neural networks (ANN), eXtreme Gradient Boosting (XGBoost), and mLR. The models were trained and evaluated on 75% and 25% of the dataset, respectively. Model predictions were explained using SHAP values. The study included 102,135 patients (85.8% ischemic stroke, 53.3% male, mean age 75.8 years, and median NIHSS of 3). All models demonstrated similar overall accuracy (69%-70%). The ANN and XGBoost models performed significantly better than the mLR in classifying dependence with F1-scores of 0.603 (95% CI; 0.594-0.611) and 0.577 (95% CI; 0.568-0.586), versus 0.544 (95% CI; 0.545-0.563) for the mLR model. The factors that contributed most to the predictions were expectedly similar in the models, based on clinical knowledge. Our ANN and XGBoost models showed a modest improvement in prediction performance and explainability compared to mLR using routinely-collected data. Their improved ability to predict functional dependence may be of particular importance for the planning and organization of acute stroke care and rehabilitation.

Effect of admission time on provision of acute stroke treatment at stroke units and stroke centers—An analysis of the Swiss Stroke Registry

Introduction:

Rapid treatment of acute ischemic stroke (AIS) depends on sufficient staffing which differs between Stroke Centers and Stroke Units in Switzerland. We studied the effect of admission time on performance measures of AIS treatment and related temporal trends over time.

Patients and methods:

We compared treatment rates, door-to-image-time, door-to-needle-time, and door-to-groin-puncture-time in stroke patients admitted during office hours (Monday–Friday 8:00–17:59) and non-office hours at all certified Stroke Centers and Stroke Units in Switzerland, as well as secular trends thereof between 2014 and 2019, using data from the Swiss Stroke Registry. Secondary outcomes were modified Rankin Scale and mortality at 3 months.

Results:

Data were eligible for analysis in 31,788 (90.2%) of 35,261 patients. Treatment rates for IVT/EVT were higher during non-office hours compared with office hours in Stroke Centers (40.8 vs 36.5%) and Stroke Units (21.8 vs 18.5%). Door-to-image-time and door-to-needle-time increased significantly during non-office hours. Median (IQR) door-to-groin-puncture-time at Stroke Centers was longer during non-office hours compared to office hours (84 (59–116) vs 95 (66–130) minutes). Admission during non-office hours was independently associated with worse functional outcome (1.11 [95%CI: 1.04–1.18]) and increased mortality (1.13 [95%CI: 1.01–1.27]). From 2014 to 2019, median door-to-groin-puncture-time improved and the treatment rate for wake-up strokes increased.

Discussion and Conclusion:

Despite differences in staffing, patient admission during non-office hours delayed IVT to a similar, modest degree at Stroke Centers and Stroke Units. A larger delay of EVT was observed during non-office hours, but Stroke Centers sped up delivery of EVT over time. Patients admitted during non-office hours had worse functional outcomes, which was not explained by treatment delays.

Factors related to time of stroke onset versus time of hospital arrival: A SITS registry-based study in an Egyptian stroke center

BACKGROUND

High-quality data on time of stroke onset and time of hospital arrival is required for proper evaluation of points of delay that might hinder access to medical care after the onset of stroke symptoms.

PURPOSE

Based on (SITS Dataset) in Egyptian stroke patients, we aimed to explore factors related to time of onset versus time of hospital arrival for acute ischemic stroke (AIS).

MATERIAL AND METHODS

We included 1,450 AIS patients from two stroke centers of Ain Shams University, Cairo, Egypt. We divided the day to four quarters and evaluated relationship between different factors and time of stroke onset and time of hospital arrival. The factors included: age, sex, duration from stroke onset to hospital arrival, type of management, type of stroke (TOAST classification), National Institute of Health Stroke Scale (NIHSS) on admission and favorable outcome modified Rankin Scale (mRS ≤2).

RESULTS

Pre-hospital: highest stroke incidence was in the first and fourth quarters. There was no significant difference in the mean age, sex, type of stroke in relation to time of onset. NIHSS was significantly less in onset in third quarter of the day. Percentage of patients who received thrombolytic therapy was higher with onset in the first 2 quarters of the day (p = <0.001). In-hospital: there was no difference in percentage of patients who received thrombolytic therapy nor in outcome across 4 quarters of arrival to hospital.

CONCLUSION

Pre-hospital factors still need adjustment to improve percentage of thrombolysis, while in-hospital factors showed consistent performance.

Lifetime risk of stroke in the general male population

OBJECTIVES

Most previous studies of incidence rates of stroke are from register studies, while data from prospective cohort studies are limited. The aim of the present study was to describe hazard rates, prevalence and cumulative proportion free from stroke during a lifelong follow-up of a representative sample of middle-aged men sampled from the general population.

METHODS

A population-based sample of 855 men, all born in 1913, was investigated at 50 years of age and followed up with repeated medical examinations at age 54, 60, 67, 75 and 80. Data from hospital records and the Cause of Death Register were collected, and all stroke events during 48 years of follow-up were registered. Medical records were scrutinized in order to confirm and validate the stroke diagnoses.

RESULTS

One man was excluded because of stroke prior to baseline, while 176 of the remaining 854 men (20.7%) suffered a first-ever stroke during follow-up. The total 5-year stroke risk (hazard rate) increased with age, from 3.54 (95% CI: 0-7.55) per 1000 persons at risk at age 50 years, to 119.05 (95% CI: 45.39-192.70) at age 90 years. The stroke prevalence peaked at age 80 and older, with about 120 cases per 1000 years of observation. The survival rate (cumulative proportion free from stroke) at age 98 was 50.0%.

CONCLUSION

One out of five men in this population sample suffered a stroke of any type during follow-up from 50 to 98 years of age and the cumulative incidence was close to 50%.