Implementation of a Macro/Autotext Format in FISH Reporting Significantly Reduces Errors
NR Shumaker, KK Reichard. University of New Mexico, Albuquerque, NM
Background: Error detection in anatomic pathology is well-described, and reporting errors may impact patient outcome (reference). Identifying sources of error and devising targeted approaches to reduce/eliminate error is critical. One approach is to use a standardized report (a.k.a. macros, autotext). In contrast to typical report preparation, our autotext program works by selecting items from an in-depth template rather than entering raw data. We evaluated the impact of using autotext on error reduction in fluorescence in situ hybridization (FISH) reporting.
Design: 250 consecutive FISH reports from before and after autotext implementation were identified (500 total). All reports were reviewed independently for errors (KR,NS). Raw data sheets from which these reports were generated were not reviewed. Errors were classified according to Zarbo et al. (ref). Categories included defective patient identification, specimen, report and interpretation. Error outcome severity included no impact on care; minimal harm; minor harm; moderate harm; major harm; and unknown.
Results: From pre- and post-autotext FISH reports, 27% and 15% had errors, respectively (Table 1). There were no errors of patient identification or specimen in either group. Pre-autotext errors included 51 defective reports (14 typographical, 3 punctuation, 12 date errors, 4 mathematical, 18 nomenclature) and 17 interpretation errors (4 false positive, 6 false negative, 7 not further classifiable). 14 had a minimal or minor impact. Post-autotext errors were typographical or grammatical, and none had clinical impact. 33 were due to integral errors in templates, while 5 occurred from manual entry.
|Type of FISH report||# cases||# (%) cases with errors||Types of errors (# cases)||Error outcome severity (# cases)(%)|
|Pre-autotext||250||68 (27%)||Defective report (51) Defective interpretation (17)||No impact (48)(70%); minimal (11)(16%); minor (3)(5%); unknown (6)(9%)|
|Post-autotext||250||38 (15%)||Defective report (38)||No impact (38)(100%)|
Conclusions: Implementation of autotext resulted in an almost 50% reduction in error rate in FISH reporting. This was due principally to decreased opportunity for error during manual entry. In addition, elimination of unnecessary and redundant data reduced the rate of interpretive error. Our study indicates that the overall clinical impact of error may be reduced to 0 primarily due to reduced interpretive error, although original data sheets must be reviewed to verify this finding. Zarbo RJ et al. Arch Pathol Lab Med. 2005:129.
Category: Quality Assurance
Monday, March 9, 2009 9:30 AM
Poster Session I Stowell-Orbison/Autopsy Award # 237, Monday Morning