More on Clinical Decision Support
William A. Hyman
Professor Emeritus, Biomedical Engineering
Texas A&M University, w-hyman@tamu.edu
Read other articles by this author
It should not be a surprise that the way that information is presented can impact how readily that information is digested and then converted into action. Some of us remember when this was much discussed in regard to NASA’s use of overheads and the Challenger accident. This was later reflected in the 2003 NY Times article: “How PowerPoint Makes You Dumb”. I for one have seen many a PowerPoint presentation that if didn’t make me dumb, certainly didn’t enlighten me. (As have others, I wrote a short guide to PowerPoint presentations that I would be happy to provide on direct request to me.)
The relevance of this to EHRs generally should be obvious. EHRs may have a great deal of information, but this does not always translate into that information being presented to the user in a manner that facilitates its effective use. Herein also is a trap. When the lack of use of available information leads to an adverse event, it can be shown that the information was indeed right there and that the user should have seen and used it. The counter arguments that yes it was there but it was buried, on a different page, one of a long list, etc., are not likely to seem justified.
In this context an interesting study¹ has been reported on the effectiveness of EHR based Clinical Decision Support (CDS) alerts in impacting clinical behavior. In this case the issue was reminders about reassessing the ongoing need for urinary catheters in order to encourage removal when appropriate, as suggested by existing guidelines. The reminder interval was based on the original reason for ordering the catheter.
The original alert, which included an option for a removal order within the alert, caused 2 percent of the catheters to be removed. As a result catheter-associated urinary tract infections dropped from 0.84 to 0.70 per 1,000 patient days. With a new alert, said to be quicker and easier to use because it provided a direct yes/no option, the proportion of catheter removals increased more than seven-fold to an average among 3 hospitals of 15 percent (with a range of 10 to 22%) and catheter-associated infections dropped to 0.50 per 1,000 patient days. The authors note in conclusion that minor changes in the format of a reminder can have a dramatic impact on provider actions, and that given the state of user interface for reminders, the implementation of a CDS should be monitored to assess its actual clinical impact.
Of interest for the effectiveness of all hospital quality interventions, the initial impact of the better design on guideline compliance was followed by a subsequent drift back to earlier (poorer) levels. Perhaps a bolder approach would be for the alert to say that a removal order would be automatically entered unless the physician said not to.
While the new CDS alert developed in this study had at least an early positive effect, it is perhaps disturbing that a straightforward guideline such as to use a catheter only when necessary, and if used leave in place only as long as needed, would need the benefit of a CDS to obtain compliance. Worse perhaps is that behavior deteriorated after even the better CDS had been in place for awhile. On the other hand, getting healthcare providers to wash their hands also remains a challenge.
1. Usability and Impact of a Computerized Clinical Decision Support Intervention Designed to Reduce Urinary Catheter Utilization and Catheter-Associated Urinary Tract Infections. CA Baillie, et al, Infection Control and Hospital Epidemiology, Vol. 35, 1147-1155, 2014.