Reliability associated with the abstraction of data from medical records for inclusion in an information system for persons with a traumatic brain injury

Abstract

Objective : This article presents the intra- and inter-rater reliability associated with the extraction, from medical rehabilitation charts, of data to be included in a head injury information system currently under development. Methods : A data collection form was developed to facilitate and standardize the data extraction. Two clinicians extracted information pertaining to 231 variables of the system from 15 charts of persons receiving rehabilitation services following a head injury. Results : Average percentage agreement was high and did not vary from one category of variables to the other (84-88%). Substantial intra-rater agreement ( s = 0.66) and moderate inter-rater agreement ( s = 0.56) were found to be associated with the extraction of the variables studied. Conclusions : The results suggest that clinicians using standardized procedures can reliably extract important data pertaining to personal history, impairments, and disabilities relating to sensorimotor function. Some potential sources of error are identified and recommendations are presented.