A Study On Reliability And Validity Of Four Stroke Scales

Objective To study the reliability and validity of the four scalesin evaluating stroke patients, which include the National Institutes ofHealth stroke scale (NIHSS),the Barthel index (BI),the clinicalneurologic deficit scale (CNDS) and the Chinese scale of the activitiesof daily living (ADL).Sudjects and Methods Totally 222 consecutive inpatients with acutestroke onset were included in the study. They were assessed when admitted,at the 14th and 90th day of hospitalization by different rators using theNIHSS, BI, CNDS and the Chinese ADL. Intrarater,interrater andtest-retest reliability were determined by using Kappa correlation. Thesplit-half reliability and internal consistency were evaluated usingCronbach's a coefficient. Concurrent validity of the CNDS and theChinese ADL, and the predictive validity of the NIHSS and the CNDS weredetermined by spearman rank correlation coefficients. Construct validitywas assessed by the factor analysis and the construct validity of theNIHSS and the CNDS was measured according to the classifications of theOxfordshire Community Stroke Project (OCSP) ischemic stroke subtypes inthe patients with cerebral infarction.Results In the Chinese version of NIHSS, the items of "facialmuscle function" and "ataxia" have modest reliability and the internalconsistency is good as demonstrated by the score ofCronbach'sα(0.861).Eight items do not show good reliability in various assessments. The itemof "facial muscle function" has low validity. There dos not exist a specific item to evaluate the function of cerebellum and brain stem exceptthe item of "ataxia". The scores of NIHSS have a high correlation withstroke outcome (both P=0.000). In the Chinese version of BI, the itemsof "transfer" and "bathing" have low reliability and the internalconsistency is perfect as demonstrated by the high Cronbach' s score(0.941). Telephone assessment in the BI exhibits a high correlation withdirect evaluation. Factor analysis in the BI reveal two factors, whichreflect functions of motor, bowel and bladder in stroke patients. Inthe SCND, the item of "walking" has low reliability and the internalconsistency is good as reflected by the score of Cronbach'sα(0.886).Four items do not show good reliability in various assessments, whichis highly correlated with the NIHSS (both P= 0.000) in both total andsubtypes of stroke patients according to OCSP classification byconcurrent validity analysis. There is ahigh correlation between thescores of the SCND and stroke outcome (both P=0.000). In the ChineseADL, the items of "household duties" and "paying cash, cash counting"have low reliability and the internal consistency is extraordinarily highreflected by high Cronbach'sαscore (0.966). There is a highcorrelation between telephone assessment and direct evaluation.Concurrent validity analysis of the scale demonstrates a high correlationwith the BI in both acute and convalescence stage after stroke, factoranalysis displays two factors in the scale, reflecting the motor andcognition functions of stroke patients.Conclusions1. The Chinese version of NIHSS has a good internal consistency, butis not suitable for telephone assessment. It is consistently predictive of stroke outcome, focusing more on the evaluation of disorders of arteriacarotis interna than those of vertebrobasilar artery. Addition of theitems and/or weight to assess the function of cerebellum and brain stemis necessary.2. The internal consistency of the Chinese version of BI is extremelyhigh, indicating redundancy of some items or unnecessary items. It issuitable for telephone assessment. Two factors are demonstrated by factoranalysis, reflecting the motor function and bowel and bladder functionof stroke patients.3. The SCND has good internal consistency, but it is not suitablefor telephone interview. There is concurrent validity between the scaleand the NIHSS and could predict stroke outcome. Factor analysis of theSCND displays the best construct validity in the patients with partialanterior circulation infarction (PACI), and could be used to evaluatethe focus of vertebrobasilar artery despite its insensitivity.4. The Chinese ADL has an extremely high internal consistency,suggesting that some might be redundant or unnecessary., It is suitablefor telephone evaluation. Due to ambiguity, confusion andnon-specificity of some items, further revision is suggested. Concurrentvalidity exists between the scale and the BI. Factor analysis of theChinese ADL reveals two factors, which reflect the motor function andcognition function of stroke patients. It can thoroughly evaluate theactivities of daily living, therefore, the Chinese scale of theactivities of daily living is recommended for use after revision.