Skip to content

No More NUPOR Mooing and Musings #4


The ongoing series on No More NUPOR can be seen by clicking the link to the left.  It is part of a new and improved site.

The following statement was put forth at the Am Soc Health System Pharmacist’s (ASHP) House of Delegates in Seattle in June of 2008.

Use of Clinical Decision Support to Limit Need for Near-Universal Pharmacist Order Review — ASHP should advocate study of the potential use of clinical decision support in the place of the current practice of near-universal pharmacist order review.

It was turf’ed to the ASHP’s Council of Pharmacy Management for guidance. Their report has recently become available (page 8).

My ‘glass is half full’ interpretation of the Council’s report is that they fully and unequivocally support this recommendation. This is fantastic and hope that research interest and dollars follow.

Now the rest of the story….. Here is the write up with my comments inserted, followed by comments of the current and future Chairperson of the ASHP Section on Pharmacy Informatics and Technology.

Use of Clinical Decision Support to Limit Near-Universal Pharmacist Order Review.

In response to a Recommendation from the ASHP House of Delegates, the Council discussed the desirability of using clinical decision support systems (CDSS) to limit universal pharmacist order review. The Council discussed the current status of CDSS. According to a 2007 ASHP survey only 12% of US Hospitals have implemented CPOE with a robust CDSS. Up to 90% of hospitals are looking at this technology in the next three years. While every CPOE computer system includes commercially developed CDSS, extensive local customization is required to achieve optimal performance and patient outcomes.

When implemented and properly customized with dedicated pharmacist resources there is substantial evidence that CDSS can have positive patient outcomes.

[Unlike some other popular technologies being advocated i.e. BCMA]

However, the extensive customization required by these systems has limited the widespread use of CDSS,

[not sure this is the case, but ok]

especially for the purpose of limiting pharmacists’ review of medication orders.

[At least that is getting published. Many are doing it, none the less.]

The Council supported further research and pilot projects to demonstrate the value of CDSS. Research validating CDSS algorithms, as well as human factors research

[kudo’s to note human factor research – not enough of this either]

in the application of CDSS, would be valuable.

The impact of CDSS on the pharmacist review of orders should be aggressively evaluated.

The Council did not support the use of CDSS to replace pharmacist review of medication orders at this time.

[That is why we do research]

The Council did not believe that the technology has evolved to a point where it could replace the pharmacist’s role in medication review.

[I completely disagree with this, but did I mention that is why we do research]

The Council also did not think it would be wise to create policy that conflicted with Joint Commission requirements for pharmacist medication order review.

[Unless you have a death wish]

The Council did support the use of CDSS to improve medication use, believing that there may be more value in focusing efforts on the use of CDSS in improving the use of high-risk medications.

Responses to Section of Pharmacy Informatics and Technology Chairs

I like the rationale. We simply do not have enough adoption of EHRs with robust CDSS to make widescale recommendations on limiting the RPh role. However, this is not to say the “elite and progressive” health systems shouldn’t be pioneering and researching the possibility.


The attributes of a CDSS system that would serve as a filter for pharmacist review are decidedly different than the attributes of a CDSS that interacts with us on every order we review. It is clear that the council does not appreciate the difference.

Most, if not all, of our complaints about CDSS right now are that it is “unintelligent”, that it fails to recognize when it shouldn’t alert us, that it causes “alert fatigue”. I know of very few cases where the CDSS failed to detect something it should have known.

That very property makes it an ideal anti-NUPOR filter. You WANT to be certain that nothing escaped detection, no matter how absurd. If you have that confidence, you can proceed to auto-approve any orders for which there are no alerts. There will still be orders we look at and ask “why did it make me look at that?”, but there will be a significant number that just sail through.

But, to echo John’s earlier sentiment, it is not clear at all that current CDSS are insufficient to the tasks of screening orders. It is only clear that we find them annoying when they tweak us during the process of order entry.

My personal opinion is that the standard FDB/Medispan based CDSS is what most of the country uses, and what we are referring to when we talk about mainstream adoption. This is no where near being useful enough clinically to limit NUPOR. The advanced CDSS goes much further, but adoption is very spotty. An alert for a patient on warfarin that has not had an INR in the past 14 days is a decent example. These have to be custom built into most systems these days.

Perfect example. I guarantee you that warfarin orders on patients that have not had an INR in 14 days are not being questioned and are being verified without question at any number of Medical Centers. With CDS it would be caught, questioned and corrected. Hence better care.

Your thoughts?

Posted via web from RxDoc.Org

4 Comments leave one →
  1. Dennis Tribble permalink
    03/31/2009 5:29 pm

    The discussion of the Warfarin with no INR in 14 days is a really interesting example for a couple of reasons:

    1) It raises the question about whether or not NUPOR would catch it (that is, would NUPOR provide better care than a system in which the CDSS failed to catch the fact that appropriate laboratory follow-up was not occurring);

    To your point, NUPOR as it exists in many hospitals across the country, including some very well-known and well-regarded ones, would not capture this issue (“I guarantee you that warfarin orders on patients that have not had an INR in 14 days are not being questioned and are being verified without question at any number of Medical Centers”) so there is no evidence in this example that relying on a CDSS that would not capture this problem is any worse than relying on humans who don’t check.

    This could well be a case of our letting “perfect” stand in the way of “pretty darn good”.

    2) I raises the issue regarding whether or not CDSS is only an order-based phenomenon.

    Pretty clearly, the issue of a patient currently taking Warfarin without an INR in a long period of time is no less urgent than the processing of a new order. Why should we wait for order entry to become aware that the patient is not being properly followed? Why would not an advanced CDSS simply alert us that intervention might be required?

    Unfortunately, as was pointed out, not all CDSS out there can capture something like this (perhaps most cannot). But if the performance in the presence of NUPOR is no better, then what benefit do we think it provides?

  2. Dennis Tribble permalink
    03/31/2009 5:32 pm

    PS… credit where credit is due… The notion that pharmacists often let perfect get in the way of pretty darn good is a Siskaism (one of several pithy quotes from Mark Siska).


  1. Nearly Universal Prospective Order Review « Pharmacy Informatics
  2. Nearly Universal Prospective Order Review « Rx Informatics

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: