Does Computerized Physician Order Entry (CPOE) Reduce Patient Safety?

Writing this title, even in the form of a question, on a technology-oriented Healthcare blog feels like sacrilege; like an employee making a politically incorrect statement (choose one of your favorites here) at the annual policy meeting of the HR leadership. CPOE, including ePrescribing, is the holy grail of IT! Leapfrog endorses it and we hold it up as the prime example of what computing can do to improve patient safety. I know that many of us are implementing CPOE to do just the opposite, to increase patient safety, as we speak.

Recently a Microsoft colleague emailed me a link to an article that appears in the 6 December, 2005 edition of Pediatrics, entitled “Unexpected Increased Mortality After Implementation of a Commercially Sold Computerized Physician Order Entry System”. I saw this article referenced again in yesterday's Healthcare IT News. It reveals that in the Department of Critical Care at Children's Hospital of Pittsburgh the mortality rate increased from 2.8% before CPOE implementation to 6.6% after CPOE implementation. This is a highly significant difference in the wrong direction! It is not the first time I have heard the suggestion that CPOE or ePrescribing can increase medical errors and that got me thinking about the topic for my first blog entry.

Microsoft has remarkable technologies that truly enable CPOE. For example, Windows Mobile on the PocketPC or Smartphone is an ideal platform for electronic order entry and offers the developer an innovative and robust environment to quickly build CPOE applications. Dr Crounse wrote about the Windows Mobile platform for Healthcare in his most recent blog entry on December 13th. Microsoft has the best end-to-end platform and tools for implementing CPOE applications, but I strongly caution against implementing CPOE in isolation without considering the implications for the entire process and re-engineering it as necessary. I believe that computerizing bad processes for the sake of automation is very dangerous, however easy we make it with our tools and platforms. In fact automation alone makes the same bad processes more efficient, producing the same errors faster and more frequently. This could be one of the more important reasons for the increased mortality rates in the article sited above (and increased morbidity being reported by others who implement CPOE). There are clearly additional reasons for increased morbidity and mortality with CPOE including implementation problems and less vigilance after introducing a new and supposedly improved IT system.

I attended the most recent Microsoft Healthcare Executive Forum held in Bellevue, Washington, along with more than 120 healthcare professionals from the Seattle region. At the reception after the event I had the fortune to meet a very insightful pharmacist. She suggested that Microsoft might be able to help her hospital's CPOE process, even before they implement a commercial solution which was still two years away. We spoke about Microsoft Office features like spell-check, Smart Tags and even the Information Bridge which can go out and check entries against many different internal and external data sources. Microsoft has extensive experience with many tried and tested ways of indicating potentially incorrect or mismatched data. This type of checking and testing is what she felt is needed to improve patient safety with CPOE and ePrescribing. We also spoke about Microsoft's experience with checking and warning for spam, virus and spyware and we talked about bar-coding and RFID. Lets also not forget about how MS Money provides the ability to write checks with numbers and words in order to reduce errors when the paper check gets to a financial institution. Learning from our experience with these and other features and functions, we recognize that "good" CPOE for Healthcare will undoubtedly require reengineering the entire process, probably adding new steps in the process to set up double and triple cross-checks on the drugs, patients, dosages, etc. This will need to be implemented in a flexible way so that it can be adjusted from provider to provider as their own processes are customized (key to so many Microsoft products).

So my answer to the question that my colleague posed above, “Does CPOE Reduce Patient Safety?”, is "It depends!". Correctly implemented CPOE and ePrescribing will do more for patient safety than we can even imagine. I would be interested to get your feedback and hearing what you think of these findings. How many of you know of CPOE implementations that are simply layered over the top of an existing process? Is it integrated into the entire process or just a standalone? And what kinds of double or triple checking are you seeing to improve patient safety?

Clifford Goldsmith, M.D., US Director, Provider Industry, Microsoft Healthcare and Life Sciences