Evaluation of completeness of selected poison control center data fields

Poison control center data are used in research and surveillance. Due to the large volume of information, these efforts are dependent on data being recorded in machine readable format. However, poison center records include non-machine readable text fields and machine readable coded fields, some of which are duplicative. Duplicating this data increases the chance of inaccurate/incomplete coding. For surveillance efforts to be effective, coding should be complete and accurate. Investigators identified a convenience sample of 964 records and reviewed the substance code determining if it matched its text field. They also reviewed the coded clinical effects and treatments determining if they matched the notes text field. The substance code matched its text field for 91.4% of the substances. The clinical effects and treatments codes matched their text field for 72.6% and 82.4% of occurrences respectively. This under-reporting of clinical effects and treatments has surveillance and public health implications.