Purpose: This study explores the impacts of a known adverse event on the utilization of electronic prescribing (e-Rx) technology at the point of care. We also propose an approach for a control system that will allow for early detection of system failures and rapid process improvement.

Methods: Interviews of practice physicians confirmed anecdotal evidence, based on patient complaints, of a failure in the e-Rx process flow, primarily to a high volume retail pharmacy chain. We have explored the impact of this failure by analyzing data from 86 weeks of prescribing activity resulting in 36,131 prescriptions generated by six physicians from June 2005 to January 2007. 15,118 prescriptions met inclusion criteria for analysis. Three key process elements: the proportion of the number of electronic submissions to a pharmacy to the number of all the prescriptions that are written on a handheld (%e-Rx), the market share of e-Rx transmissions to each pharmacy (%Ph), and the number of prescriptions that are written on a handheld (PDA) are defined and computed using the data. We use the Chow test and What's Strange About Recent Events (WSARE - a machine learning approach), to verify and evaluate changes in physician prescribing behavior over time. Finally, we use quality control charts to develop an early detection system by simultaneously monitoring and prospectively evaluating established patterns of the three process elements.

Results: The Chow test and WSARE reveal an overwhelming shift in physician prescribing behavior toward printing vs. electronically transmitting prescriptions (significant decrease in %eRx) at week 76, which temporally matched the anecdotal evidence of transmission failures. However, these failures to a single retail pharmacy chain result in a consistent decline in e-Rx to all regional chains. Furthermore, the Chow test shows that the problem pharmacy was not adversely affected. The Cumulative Sum chart identified the duration and location of the out-of-control process for monitoring and managing the failure.

Conclusions: A critical e-Rx technology breakdown, such as a communication failure to a dominant pharmacy, can have a significant impact on the extent to which physicians utilize the technology. Statistical process control mechanisms, such as control charts and control procedure designs, have the potential to help identify system breakdowns quickly, provide the necessary feedback to establish meaningful action plans, and facilitate high technology utilization.