Improved reliability of the NIH stroke scale using video training

Background and Purpose Despite the frequent use of clinical rating scales in multicenter therapeutic stroke trials, no generally acceptable method exists to train and certify investigators to use the instrument consistently. We desired to train investigators to use the National Institutes of Health Stroke Scale in a study of acute stroke therapy so that all examiners rated patients comparably. Methods We devised a two-camera videotape method that optimizes the visual presentation of examination findings. We then measured the effectiveness of the training by asking each investigator to evaluate a set of 11 patients, also on videotape. We tabulated the evaluations, devised a scoring system, and calculated measures of interobserver agreement among the participants in this study. Results We trained and certified 162 investigators. We found moderate to excellent agreement on most Stroke Scale items (unweighted k>0.60). TWO items, facial paresis and ataxia, exhibited poor agreement (unweighted k<0.40) and should be revised in future editions of the scale. Performance improved with video training compared with previous studies. Inclusion of the motor rating of the unaffected limbs in the total score did not affect reliability. Conclusions Video training and certification is a practical and effective method to standardize the use of examination scales. Two cameras must be used during the taping of patients to accurately present the clinical findings. This method is easily adapted to any study in which a large number of investigators will be enrolling patients at multiple clinical centers.