A nomogram incorporating red blood cell indices to predict post-stroke cognitive impairment in the intracerebral hemorrhage population

Abstract

BackgroundPost-stroke cognitive impairment (PSCI) plagues 20–80% of stroke survivors worldwide. There is a lack of an easy and effective scoring tool to predict the risk of PSCI in intracerebral hemorrhage (ICH) patients. We aimed to develop a risk prediction model incorporating red blood cell (RBC) indices to identify ICH populations at risk of PSCI.MethodsPatients diagnosed with ICH at the stroke center were consecutively enrolled in the study as part of the development cohort from July 2017 to December 2018, and of the validation cohort from July 2019 to February 2020. Univariable and multivariable analyses were applied in the development cohort to screen the patients for PSCI risk factors. Then, a nomogram based on RBC indices and other risk factors was developed and validated to evaluate its performance in predicting PSCI occurrence.ResultsA total of 123 patients were enrolled in the development cohort, of which 69 (56.1%) were identified as PSCI, while 38 (63.3%) of 60 patients in the validation cohort were identified as PSCI. According to the multivariate analysis, seven independent risk factors, including three RBC indices (hemoglobin, mean corpuscular volume, RBC distribution width), as well as age, education level, hematoma volume, and dominant-hemisphere hemorrhage were incorporated into the model. The nomogram incorporating RBC indices displayed good discrimination and calibration. The area under the receiver operating characteristic curve was 0.940 for the development cohort and 0.914 for the validation cohort. Decision curve analysis and clinical impact curve showed that the nomogram was clinically useful.ConclusionRBC indices are independent and important predictors of PSCI. A nomogram incorporating RBC indices can be used as a reasonable and reliable graphic tool to help clinicians identify high cognition impairment-risk patients and adjust individualized therapy.

Machine learning in the prediction of post-stroke cognitive impairment: a systematic review and meta-analysis

“…The subject cohort size ranged from 72 to 22,950. The diagnostic criteria used in the included studies were Mini-Mental State Examination (MMSE) ( 33 ), Montreal Cognitive Assessment (MoCA) ( 34 ) ( n = 5) ( 20 , 21 , 28 , 29 , 31 ), MMSE ( n = 6) ( 12–14 , 18 , 24 , 32 ), MoCA ( n = 6) ( 17 , 19 , 22 , 25–27 ), Center of Cancelation (CoC) ( 35 ) ( n = 1) ( 15 ), Global Deterioration Scale (GDS) ( 36 ) ( n = 1) ( 16 ), Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE) ( 37 ) ( n = 1) ( 23 ), and vascular dementia criteria of the AHA/ASA scientific statement ( 38 ) ( n = 1) ( 30 ). The duration of follow-up was predominantly 3 to 12 months, and was 36 years in only one study ( 23 ).…”

“…There were 31 models in the included studies, which were constructed based on logistic regression (LR) nomogram ( n = 18) ( 12 , 14–16 , 18 , 19 , 22 , 25 , 26 , 29–32 ), random forest ( n = 1) ( 20 ), ridge regression ( n = 1) ( 23 ), LASSO regression ( n = 1) ( 21 ), mixed effects model ( n = 3) ( 24 ), support vector machine (SVM) classifier ( n = 3) ( 27 ), and decision trees ( n = 3) ( 28 ) ( Table 1 ). Modeling variables were selected using a multivariate approach.…”

Low HALP (Hemoglobin, Albumin, Lymphocyte, and Platelet) Score Increases the Risk of Post-Stroke Cognitive Impairment: A Multicenter Cohort Study

“…Stroke patients with low hemoglobin levels have been found to harbor increased risks of PSCI. 15 Studies reported that decreased hemoglobin levels might lead to brain hypoxia, mitochondrial dysfunction, and neuronal damage. Patients who have ischemia stroke with hypoxic conditions are much more susceptible to Alzheimer’s disease.…”

“… 17 Hemoglobin level is a significant independent predictor of PSCI. 15 Serum albumin is widely considered a surrogate of nutrition where deficiencies in serum albumin have been associated with poorer stroke outcomes, 12 explained in part by an impaired antioxidant capacity to protect neurons from ischemic injury. 18 The geriatric nutritional risk index based on serum albumin has been similarly implicated in PSCI.…”

Predicting post-stroke cognitive impairment using electronic health record data

“…8 Existing models to predict PSCI have demonstrated moderate accuracy but are not amenable to large-scale prospective implementation. Published models have largely focused on short-term follow-up (< 6 months) following stroke, [9][10][11] and often include predictors that are not available or easily accessible in electronic health records (EHRs). [10][11][12][13][14] Models that predict PSCI over longer periods may be more useful clinically to allow time for secondary prevention efforts.…”