Project to Improve the Transcription of Clinical Order Information into a Radiology Information System.

Context: Inaccurate and incomplete imaging order information presented to interpreting radiologists is a persistent problem in many radiology settings. Computerized Physician Order Entry processes in clinic-based settings are often inconsistent, and radiology transcription clerks continue to play a critical role in transmitting accurate content and information from referring physician orders to the radiology information system. (RIS) The purpose of this quality improvement project was to a) identify common transcription areas of deficient RIS imaging order information and b) test outcomes from an intervention to improve the content and concordance of transcribed patient information entered into the RIS.

Methods: A random convenience sample of 500 outpatient radiographic orders were categorized according to degree and quality of concordance between the transcribed patient information documented in the RIS and the corresponding original imaging order information. During Phase I, the authors used a root-cause analysis to determine the possible etiologies for discordance between the information in original imaging orders and the information transcribed into the RIS. The intervention that was delivered included a short education session with radiology transcription clerks with placement reminder posters at transcription workstations. During Phase 2, a second random sample was obtained following the intervention, with data collection and analyses replicating the process from Phase I. A set of inferential comparisons were conducted using chi-square tests to examine for statistical significance.

Results: There was an overall 44% decrease in transcription discordance (p < 0.001), and the number of cases with perfectly concordant RIS order indication documentations increased by 21% (p < 0.001). A total of 34% of transcriptions from Phase I were partially discordant due to an inadequate imaging study indication, compared to 15% during Phase II (p < 0.001). There was also a 22% increase in the number of completely concordant transcriptions free of grammatical errors (p < 0.001).

Conclusions: A short education session with radiology transcription clerks along with placement of reminder posters may significantly improve both the concordance and quality of transcribed imaging order information presented to interpreting radiologists using the RIS.