Confiabilidade de escalas de comprometimento neurológico em pacientes com acidente vascular cerebral / Reliability of neurological assessment scales in patients with stroke / Reliability of neurological assessment scales in patients with stroke

RESUMO

Escalas de comprometimento neurológico são utilizadas na avaliação de novos tratamentos, no acompanhamento da evolução clínica e em decisões terapêuticas. Estudamos a confiabilidade de três escalas a Escala de Rankin (ER), o índice de Barthel (IB) e a National Institute of Health Stroke Scale (NIHSS) versadas em português, em 51 indivíduos com acidente vascular cerebral, estimando a concordância entre resultados de diferentes examinadores. Utilizamos os coeficientes de concordância Kappa e Coeficiente de Correlação Intraclasse. A ER teve coeficientes moderados, substanciais ou excelentes. No IB estes foram substanciais ou excelentes nos totais e nos itens, assim como quando foi estratificado em agrupamentos prognósticos e funcionais. A NIHSS apresentou coeficientes excelentes nos totais, substanciais nos itens e moderados e excelentes nas estratificações em grupos de pontos e de itens afins. Estes resultados indicam que as versões em português da ER, o IB e a NIHSS apresentam adequada confiabilidade.

ABSTRACT

Scales for the assessment of neurological impairment are employed in clinical trials, outcome evaluation and in therapeutic decisions. We evaluated the reliability of the Portuguese version of the Rankin Scale (RS), Barthel Index (BI) and the NIHSS (National Institute of Health Stroke Scale) in 51 stroke patients, estimating the agreement of the results for examiners. The interrater reliability was measured using the Kappa Coefficient and Intraclass Correlation Coefficient. The RS showed moderate, substantial and excellent coefficients of agreement. The BI showed from substantial to excellent coefficients of agreement as a whole and also for its constituents. This scale was stratified in prognostic and functional groups, both presenting coefficients of agreement from substantial to excellent. The NIHSS presented excellent agreement on its total and substantial agreements on its individual items. The NIHSS was also stratified in groups of points, which presented moderate or excellent coefficient, and in groups of congruous items, wich showed a moderate to excellent agreement. These results suggest that the Portuguese version of the RS, BI and the NIHSS present adequate reliability.