Rx Error (was Re: Learning to Love the War)

[From Rick Marken (2003.04.07.2020)]

Kenny Kitzke (2003.04.07.2030EDT)]
<Rick Marken (2003.04.07.1710)>

<The RTP analogy works (for me,
anyway) only after the war has started because before the war the use
of coercive force did not have “institutional legitimacy”. RTP is implemented…>

Sorry, Rick. Perhaps someone
wants to debate with you the war or whether or not it comports with RTP
as you perceive them, but I don’t and won’t.

I wish I had used my Delete key.
I guess I should take your advice even more than I usually do.

You wouldn’t even have had to use the delete key if you just hadn’t asked
me the question to which what you quote above is part of my reply.

Relating to your
Rand work, I visited my old client last week who runs a mail-service pharmacy.
They have their external (patient experienced) prescription error rate
down to about 50 per million. What is the lowest error rate you have
verified in your research? Do you have any advice for them unique
to PCT in cutting it in half again?
The error rates I’m dealing with are errors that occur when physicians
write a prescription. Call them Rx errors. The Rx error rates in the literature
are on the order of 1/1000 prescriptions, which is quite a bit higher than
50/1,000,000. But it sounds like you are talking about Rx errors that get
through to the patient. Only a fraction of Rx errors get though to
the patient because most are intercepted by the pharmacy (including mail-order
pharmacies, I imagine. I think pharmacies catch on the order of 90%
of Rx errors before they get to the patient. So this is getting pretty
close to the 50/1,000,000 error rate. I guess I’d want to know exactly
what your client’s error rate actually is and, more important, how they
measure it. I’d also like to see a time series look at the error
rate. Has the error rate become substantially lower over time or is 50
per mission the ambient level of Rx errors that get though to patients
via the mail order pharmacy. In order to get an accurate estimate of an
error rate of 50 per million requires observing many millions of prescriptions
accurately. If many errors are made in detecting errors the estimate
could be off by quite a bit. So it’s really important to know how the error
data was collected and, of course, what the measured error rate is over
time.
Still, it sounds like your client has an impressively low error rate.
The advice I would give is to not worry about cutting the error rate in
half. There is no guaranteed way to reduce such a low rate to an even lower
level except, perhaps, by taking more time checking the prescriptions and,
given the number of Rxs they are processing, I can’t imagine that that
would be cost effective. I would tell the client to keep doing whatever
he’s doing because whatever he does to try to reduce error further is as
likely to increase as decrease the error rate somewhat.

But I’d really like to see the data.

Best regards

Rick

···

Richard S. Marken

MindReadings.com

marken@mindreadings.com

310 474-0313

[From Kenny Kitzke (2003.04.07.0920EDT)]

<Rick Marken (2003.04.07.2020)>

<The error rates I’m dealing with are errors that occur when physicians write a prescription. Call them Rx errors. The Rx error rates in the literature are on the order of 1/1000 prescriptions, which is quite a bit higher than 50/1,000,000. But it sounds like you are talking about Rx errors that get through to the patient. Only a fraction of Rx errors get though to the patient because most are intercepted by the pharmacy (including mail-order pharmacies, I imagine. I think pharmacies catch on the order of 90% of Rx errors before they get to the patient. So this is getting pretty close to the 50/1,000,000 error rate.>

If I understand correctly, you are only dealing with errors in the prescriptions written by doctors? Then we are talking about opposite ends of the prescription drug fulfillment process.

My work on error reduction deals with the filling of the prescription written by the doctor by a pharmacy. Further, it deals with a mail order pharmacy filling millions of prescriptions each month. It does not relate to a local retail drug store, or the food store pharmacy popular in some regions, with a few or even just one licensed pharmacist doing the prescription filling.

<I guess I’d want to know exactly what your client’s error rate actually is and, more important, how they measure it. I’d also like to see a time series look at the error rate. Has the error rate become substantially lower over time or is 50 per mission the ambient level of Rx errors that get though to patients via the mail order pharmacy. In order to get an accurate estimate of an error rate of 50 per million requires observing many millions of prescriptions accurately. If many errors are made in detecting errors the estimate could be off by quite a bit. So it’s really important to know how the error data was collected and, of course, what the measured error rate is over time.>

First, this is naturally rather sensitive information to the client. I am bound by confidentiality covenants in my Retainer Agreement not to disclose such information to third parties without prior written approval. So, I can only answer in generalities.

The actual external RX error data is collected continually from patients. Most is by toll free telephone number calls. For example, a patient calls the mail order pharmacy number and says they received blue pills and the ones they usually take are white. What’s up? Such calls are logged and investigated as appropriate. It could be the generic pill is blue and is the same as the brand named white pill previously prescribed and filled by that pharmacy. Then, it would not be recorded as a prescription error, only as a customer service/satisfaction issue.

Now, verified errors are coded as to severity and cause when determined. For example, a wrong pill is probably the most serious error (could be life threatening). Others like wrong strength or wrong dosage in times per day taken are also recorded as serious errors. If the patient’s name is spelled wrong on the label, Markin instead of Marken, or the bottle contains 49 pills instead of 50, those would be errors needing correction, but are not as serious in terms of harm to the patient.

Now, the cause is also important as it helps focus on what corrective action is needed. Again, there are categories. It could be a mailing label error; a filling error; an error in reading the prescription and even a doctor error in writing the Rx (the ones you are dealing with).

The pharmacy manager claimed that the majority of the external errors they are now recording are as a result of physician errors or in recording the prescription into the largely and ever-more computer and automated process of filling and mailing. They make many calls to physicians to question the prescriptions and do find some of the errors. I do not believe that is their valid job, and if they do it, they should charge a fee to every doctor called. But, they provide this “service” for free and believe it is their ethical responsibility.

They were not interested in trying to cut this rate in half. I offered to help them do it using advanced statistical methods and suggested we might get some help from you as to the human causes and solutions. As you suggested, because the error rate is so low, and the actual harm very rare, and because of insurance for those cases, they have no economic incentive to act. I believe ethically even one serious error is a reason to act. And, I would do it just to show it is possible to run at or very near zero. But, that is me and they is they. We have different system references.

BTW, we all recognize that the actual external error rate is higher than the recorded rate. Not all patient errors are perceived or recorded (especially by the patients who died, smile). No one knows for sure how much higher but it does not seem important anyway.

My work concentrated more on the internal Rx error rate. This is the data recorded at the final pharmacist prescription check, just before the cap goes on the bottle. It is 100% inspection. It is not perfect however and external errors still get through. However, if the fraction of internal errors goes down, the external errors tend to also go down (and the data shows this). In the seven years I worked with this client, the measured internal error rate steadily dropped by a factor of 20 and in half again the past five years. They are now approaching the 6-sigma quality level first made famous at Motorola (about 5 errors per million opportunities).

<Still, it sounds like your client has an impressively low error rate.>

Yes, I believe it is the best in the world. I keep encouraging them to take competitive advantage of it. But, errors are not something the marketing people like to talk about.

One encouraging thing I learned is that the regulatory people seem willing to allow someone other than a relatively high paid, licensed pharmacist, do the final Rx quality check. I begged the Operations Executive to take the lead in this change and save far more money per year just from that than they ever spent with me. The check adds very little value. And, their competitors would be caught flat footed and stuck with the extra cost until they too could get their error rates down low enough (took us 7 years!).

He felt it was opening a can of worms and would not act on it. He has been retired for about five years and the new executive (who liked my services and training more than anyone else there) has been pursuing this change with good success.

<The advice I would give is to not worry about cutting the error rate in half. There is no guaranteed way to reduce such a low rate to an even lower level except, perhaps, by taking more time checking the prescriptions and, given the number of Rxs they are processing, I can’t imagine that that would be cost effective.>

You are absolutely wrong on that. There are guaranteed ways that are cost effective but they are not in using more checking. What they already do is not cost effective. That theory is wrong. It is like the ineffective process for solving psychological problems that relies on a faulty theory of cause and effect behaviorism. It is one reason health care and prescription drug costs are ever more expensive. They have faulty and rudimentary error elimination theories.

<I would tell the client to keep doing whatever he’s doing because whatever he does to try to reduce error further is as likely to increase as decrease the error rate somewhat.

But I’d really like to see the data.>

Well, the client has taken your advice, but I suspect for different reasons. If he took my advice, we might all (including you) be able to look at the real data, the real causes of errors (including human errors through PCT glasses). Your expertise on the doctor prescription errors are relevant to their external error performance. They don’t believe they can (or perhaps should) do anything about it. We would believe differently, hey? Also, what would help the doctors would also help their review and checking pharmacists. But, I could not persuade them.

The good news was that the President and Operations VP both expressed an interest in becoming better leaders and managers–ones able to maximize human performance well above what they currently achieve. So, I think they will attend my one-day workshop which is based upon PCT/HPTC and the Bible).

They were so excited about a recent “pay for performance” system (which we originally called the Quality Quotient in 1993 that combined quality and productivity performance into a give yourself a pay increase system). I listened in a “your preaching to the choir” attitude. But, when I asked them why this system I showed them so long ago worked, they did not know for sure. They each had different guesses (see the world through different behavior glasses).

When I asked if they would like to know the verifiable reasons (so they could change this pay for performance system before the results turned around on them) the President quickly quipped, “I’ll be retired before that happens.” I wish I thought it was funny. I think the Operations VP wanted to learn, but he has a boss who sets his references for keeping his job. :sunglasses: Perhaps he will get the top post someday, or perhaps the workshop (which addresses “incentive pay”) might trigger some advanced people behavior knowledge needs.

I would have sent this privately, but I know that Fred had also shown some interest in the topic of error reduction. It might be something other than the war to think about and discuss which we have some ability to control and increase value everyone finds worthwhile.