For FUCKING STUPIDITY'S SAKE! Predicting cognitive decline DOES NOTHING FOR RECOVERY!
I'd have all of you fired for extreme stupidity! And this is from the AHA/ASA; one of our fucking failures of stroke associations!
Send me personal hate mail on this: oc1dean@gmail.com. I'll print your complete statement with your name and my response in my blog. Or are you afraid to engage with my stroke-addled mind? No excuses are allowed! You're medically trained; it should be simple to precisely state EXACTLY HOW predicting gets you recovered with NO EXCUSES! Your definition of competence in stroke is obviously much lower than stroke survivors' definition of your competence! Swearing at me is allowed, I'll return the favor. Don't even attempt to use the excuse that brain research is hard.
What’s in a Predictor? The Case of Brain Imaging Markers for Poststroke Cognitive Impairment
Poststroke cognitive impairment is common, but adequately validated prediction models are lacking.1 In this study, Sara Hassani and colleagues aimed to determine the predictive value of six brain magnetic resonance imaging (MRI) features for cognitive function after lacunar stroke.
The authors conducted a retrospective study of 134 participants enrolled in a small vessel disease cohort (the American Stroke Association Bugher Foundation Small Vessel Study) that recruited patients within 2 years of an acute lacunar stroke between 2007 and 2012 across four US hospitals. Brain MRI features were categorized and included number of lacunar lesions, lesion location, size of largest lacune, cerebral atrophy, and extent of white matter disease. Cognitive function was measured on average 2-3 months after stroke using tests for global cognition (short-form Montreal Cognitive Assessment) and executive function (time to complete TRAIL making test part B). About a third of patients initially part of the cohort were excluded due to missing data (no brain MRI or cognitive assessment), and people with dementia or cognitive impairment before stroke were not excluded. Participants had a mean age of 63 years (SD=11) at the time of stroke. In regression models adjusted for age, sex, years of education, stroke severity, and a few cardiovascular risk factors such as hypertension and diabetes, none of the brain MRI markers were associated with the measure of global cognition, but extent of white matter disease (graded semi-quantitatively) remained associated with the executive function test.
There is no doubt that a robust prediction model is needed to accurately identify people at highest risk of cognitive decline after stroke. Accessible and accurate statistical models for prediction of disease are the cornerstone of precision medicine. For example, such models can help researchers enrich clinical trials for patients at higher risk of poststroke cognitive decline (i.e., prognostic enrichment). For clinicians, simple scores can guide assessments and interventions based on predicted cognitive function after stroke.
In this study, the use of simple imaging markers readily accessible in several clinical settings is a strength and should facilitate external validation and replication in other data sets. These results also identify radiological biomarkers of cognitive decline which will guide future research. These are important first steps towards a more unified prediction framework for poststroke cognitive impairment.
The main limitation of this study, however, is that individual ‘independently associated’ predictive biomarkers are difficult to interpret clinically. Risk estimations in the clinic — including intuitions or ‘gestalt’ — are inherently multivariable. As such, focusing on a single class of features such as radiological markers evacuates predictors that may convey more information on poststroke cognition, such as premorbid cognitive decline, occupation, socioeconomic status, or novel fluid biomarkers that appear to strongly associate with vascular cognitive impairment such as plasma neurofilament light chain.2 Univariable associations also cannot fully recapitulate the complex and often nonlinear relationships between baseline variables and future disease. In addition, large odds ratios are usually needed to detect meaningful increments in predictive accuracy, such that significant associations in regression models are usually not good indicators of discrimination,3 while effect sizes cannot be interpreted as absolute risks, a more intuitive and actionable measure for disease prevention. These are some of the reasons why predictive features of disease often struggle to make their way to the clinic or influence clinical decisions. Future prediction studies may help accelerate clinical translation by implementing actions linked to test results (e.g., what threshold should be used to implement a cognitive rehabilitation intervention?) and by examining their impact on clinical outcomes in real-world settings.4
In summary, this study provides(It provides nothing towards recovery!) important information on potential radiological markers of poststroke cognitive decline, but these should be tested in more comprehensive clinical prediction models to help generate more robust and potentially clinically useful risk estimates.
No comments:
Post a Comment