Accurate ALSFRS-R scores can be generated from retrospective review of clinic notes

Retrospective studies are important in ALS but require markers of disease severity to enable risk adjustment and to allow fair comparisons between patient groups. The ALSFRS-R could be used as such a measure. This study aimed to determine if accurate ALSFRS-R scores could be generated by reviewing clinic notes. Five investigators reviewed 100 de-identified clinic notes to generate estimated ALSFRS-R scores. These scores were compared to ALSFRS-R scores completed by patients within three months of the clinic note. The retrospective ALSFRS-R scores did not differ significantly from the actual scores (mean retrospective score 38.7+/-5 vs. actual score 38.4+/-6, p =0.5). The intra-class correlation coefficient between actual and retrospective scores confirmed reasonable agreement (rho = 0.53, p <0.001). Bland Altman analysis also confirmed good agreement between the actual and retrospective scores. This study indicates that ALSFRS-R scores can be accurately reproduced from information in clinic notes and should be considered as a marker of disease severity for use in retrospective studies.