April 1 was the first of four “milestone” dates by which hospitals must comply with one of the Joint Commission’s 2008 patient safety requirements: Reduce the likelihood of patient harm associated with the use of anticoagulation therapy. This new requirement, part of the Joint Commission’s goal to improve the safety of using medications, has a one-year phase-in period. By April 1, hospitals were to have assigned oversight and coordinating responsibilities for the requirement; full implementation is expected by January 2009. Mark Wurster, MD, clinical associate professor, Department of Internal Medicine, Ohio State University College of Medicine, spoke at a Roche-sponsored Webcast last fall on what’s being done to reduce the likelihood of harm to patients in OSU’s anticoagulation clinics. An edited transcript of his remarks follows.
This is a fairly large set of patient safety requirements. From a guy who does coagulation for a living, this is probably the biggest development we’ve seen in the last 10 to 15 years in terms of jump-starting a process. What I think is going to be a fairly realistic and doable approach for managing 3E requirements is taking the expertise we’ve been able to develop on the outpatient side and translating some of that to an inpatient environment.
First, a few housekeeping details: I’m on a medical advisory panel that helps with the educational processes at Roche. I did the original design work on a Standing Stone application called CoagClinic. I am not here to sell anybody on any of these products. But a lot of my research has involved this combination of testing by point-of-care instruments and some sort of electronic medical record.
Now if you look at how you’re supposed to approach this type of encyclical requirement, there are so many different things going on here. In the traditional approach, a committee gets assigned, you divvy up the tasks, and each team goes off and does its own thing, comes back with solutions in a few months, and everybody talks about it. The problem with that approach is that this cuts across so many different departments and so many real-time patient care issues that trying to assign individual tasks is not going to work.
What I’m trying to tell people is that we have things that already work. It’s just a question of modifying some inpatient practices and adapting some outpatient modalities. By that I mean something called systematic anticoagulation management. The clotters like me, who’ve been working in this field for a number of years, have found that if we do some very simple things we can drastically cut down complication rates in the 80 to 90 percent range. If you look at the CMS national standards-of-care data from just a few years ago, there are 40,000 strokes a year that are totally preventable if we just manage warfarin and heparin correctly. It’s a huge problem. I can see why the Joint Commission went after it.
Now, systematic anticoagulation management, or SAM. It’s an approach that combines computerized decision support and point-of-service testing. The basic principles we’re talking about here work equally well in an inpatient environment as they do in an outpatient setting. And there are all sorts of nifty things that start happening, from a business-case standpoint as well as a cost-effectiveness standpoint, to help reinforce it. It’s very customizable. There is no single solution for any anticoagulation clinic out there. Same thing in an inpatient environment. That doesn’t mean this won’t work.
I’ll use outpatient encounters as an example. Our average encounter in one of the OSU clinics will be a 15-minute visit. (We do everything by scheduled visit. Of course, we tell people, “If you’ve got new bruising or bleeding, we’ll make an exception for you, because we don’t want to wait until you’re in the emergency room five days later.”) The patient comes in and gets a finger-stick done by an INR device. While that device is humming along, we ask four standardized questions regarding any new symptoms, any medications, any dietary changes since we last saw the patient, anything else they need to report. The patients get it very quickly. Within four to six weeks, most of these patients are able to anticipate the questions before we ask them and become part of the fixing process rather than part of the problem.
The first thing I found when we were developing the electronic version of the applications that would make this work is [the importance of] automatic reminder systems and decision support. First try to eliminate the error of commission or omission by the health care provider. Keep them on the straight and narrow. We use, in my clinics, the American College of Chest Physicians guidelines, and we’ve programmed them into the application. All of these guidelines are usable and doable in both an inpatient and outpatient setting.
The first thing we ever put into that application was automatic reminders. So if you fire up your application in the morning, getting ready to see your patients for the prothrombin times, it tells you that as of 5 PM the day before, Mr. Smith and Mrs. Jones didn’t show up for their INR visits. Attached to that ticker for each patient is their phone number and their contact information. We call them before we ever start the clinic. We don’t wait until bad things happen to them. Therein lies the active management part of that process.
Patient education is remarkable to see in operation. We give everybody a printout, sort of a layman’s version of the progress notes from that visit. As they leave the clinic, they have their new dose of warfarin now on the paper. It says here’s how you take it, or here’s how you take your heparin. Here’s when we need to see you back again; here’s what you watch out for. The combination of face-to-face contact and hard-copy information is extremely helpful in making those messages stick.
When we set up our new clinics, I’ll go baby-sit them once or twice, just to make sure they’re following through with the applications. [On one visit,] the last patient of the day was coming in. He was a 91-year-old gent, about five feet tall, who kind of hobbles in. I thought, “Uh-oh, this is going to be a problem.” He was sort of infamous in the clinic. He had been in the hospital repeatedly for bleeds and thrombotic episodes for the past five years while he was on warfarin. So we do his finger-stick, hand him the papers, start trying to educate him, and he says, “My wife makes all those decisions. I don’t do that.” I said, “Okay, bring your wife in.” She’s shorter than he is, and she’s moving slower than he is, and I thought, “Uh-oh. We’re really in trouble.” She got that paper, and we went through it. It took about three minutes. You could see her eyes go from field to field on that piece of paper. At the end of it, she looks up at us, and says: “That’s how we’re supposed to take the medicine? I had no idea.” He’d been on it for five years; they just hadn’t been getting the message. And ever since, his INR compliance is much more appropriate.
Now, what works in one spot had better work in other places—otherwise you haven’t accomplished much. And you can’t wreck an entire inpatient service and structure just for the sake of transplanting something. So rule No. 1: If you have a structure or organization or active protocol you’re using, keep it, don’t scrap it. But what you can do is make simple changes that will put you in full 3E compliance.
If you have an electronic medical record anywhere in your system, and I mean down to and including a PDA, that is so much more efficient and safe for the patient than any pencil- and paper-based system. There is no comparison. It’s real important that any system have something in it to remind health care providers when they’re supposed to start doing something and when they’re supposed to stop doing something, and to keep them on the straight and narrow in terms of not committing that first error.
Point-of-care testing is extremely important on the ambulatory side, and I think it’s going to be a very integral part of achieving a lot of these 3E goals as well. I don’t think you have to scratch your traditional teams or nursing services to accomplish most of this. A lot of it can be done with very cheap and cost-effective changes in order sets. However, I do think a consult service is going to be helpful in a lot of sites for several reasons, the first of which is that we know it’s a very efficient model on the outpatient side. You can bill for the services, so it at least covers its own costs.
A couple of years ago, we realized that at OSU we were not doing a terribly good job of recognizing deep vein thromboses and/or patients who needed deep vein thrombosis prophylaxis. As a matter of fact, we were doing a really bad job. We finally came up with a solution, which was to simply use the nurse-based assessment that’s done on every patient who’s admitted. It’s a detailed history analysis that’s done by the nursing staff. In that multi-page document, we put in five questions to help risk-stratify a person for developing a DVT. We would then trigger an automatic order that would put an order set on the patient’s chart for the doctor to follow up on. It drastically improved the number of patients who were receiving sequential compression devices [to prevent DVT], heparin, you name it, because the doctor had to sign off on it. If they did not want to do it, that was a signed order, not just an error of omission. We found that when the path of least resistance was helping them do the right thing, it was much easier to make them participate.
You can take that same principle and go a step further. You can have standardized order sets that are already there. I think there are 40 order sets you can find using Google alone for managing unfractionated heparin infusion. Take any of those protocols, and you can very simply adjust, for instance, one of the 3E requirements that’s going to be a big one—heparin-induced thrombocytopenia. It’s almost an allergic reaction to heparin. It’s clinically silent until it causes a major thrombosis or death, and therefore you need to screen platelet counts. In the 3E requirements, you need to be doing platelet counts on a regular basis on anybody who’s on any type of heparin or unfractionated heparin infusion. It’s one additional line on your standardized order set. On the inpatient side, it works like a charm. The problem we’re running into now is, we’re victims of our own success. The orthopedic surgeons have finally started to prophylax patients after their procedures, which is marvelous. We’ve been after them to do that for years. That raised an issue, though: How do we continue to monitor these people if they’re going to go on heparin for two weeks at home? The risk is not zero. You still need to check those platelet counts. We’re still trying to work out the logistics of how to get a platelet count done on an outpatient basis.
There are about a million patients who are receiving warfarin nationally. In our outpatient setting, we’ll see patients anywhere from every two weeks to every four weeks. However, it’s a different situation on the inpatient side. These people need to be tested daily, and tested promptly. If you wait two days to adjust somebody’s warfarin dose, particularly during a loading phase before you find out what your INR is, you are not treating that INR anymore.
We know there are some risk factors that are probably genetic, and there are age and racially based risk factors that may be reason to adjust the initial dose. That number of variables is pretty small. You can have standardized order sets that would simply start with recommended dose and adjust up if you have a certain risk factor and down if you lack a certain risk factor. So you could order your daily INR and pro time. You can order a consult if you have consult teams that do family and patient education.
Education is a problem. We find it’s extremely helpful on an ongoing basis on the outpatient side. It’s a little tougher to do on the inpatient side because you have someone who’s never seen the therapy or has not been on it for a while and needs a crash course. This is where a consult service is going to be extremely valuable. Education is a difficult thing for a nursing staff to do. You’d have to educate the entire nursing staff to the level of an expert. If you look at warfarin, it interacts with 650 medicines. It’s going to be a lot easier to have a person who’s used to giving this talk four or five times a day go in and see the patient, run a quick check on their medications, and find out about those medication interactions before they cause problems. It’s not only going to start you off on the right dose; it’s also going to trigger appropriate education efforts before the patient gets discharged.
Transition outpatient treatment is the nightmare of anticoagulation therapy. I do a fair amount of expert witness work—only for the defense, not the plaintiffs—and this is something that happens daily: Patients who get discharged from the hospital on warfarin or heparin don’t get a prompt followup for an INR. It could be happening for all sorts of reasons—they simply forgot to show up, somebody forgot to order it, it was ordered but didn’t get processed properly, or the result didn’t get back to the doctor. You’ve got to be proactive and aggressive about communicating how that patient’s going to be followed up. I don’t know of any easy way around this other than using an electronic medical record. If you’re going to stick with faxes, you had better make confirmation of receipt mandatory. If you don’t do that, bad things will come back to haunt you.
In my Kansas City days [St. Luke’s Hospital, 1987–1992], we had an Internet-based application on the inpatient service. We’d start the patient on warfarin or low-molecular-weight heparin. Once we put the demographics in, they were available to our outpatient clinic because the clinic was using the same application. We would use the scheduling function to tell the outpatient clinic, ‘This patient’s coming in at 9 AM, here’s their goal INR, here’s how long they’re supposed to be treated, here’s their starting dose of warfarin, and here’s the things you need to watch out for.’ It makes life a lot easier. The other thing we found is that we could discharge people a couple of days early and save hospitalization costs.
Systematic anticoagulation management is well established as the closest thing to a standard of care that you’ve seen on the outpatient side for the past several years. The active outpatient clinics can help the inpatient situation by providing bridging therapies, so they know the patient’s going to go in for a procedure, they hold the warfarin appropriately, they check the INR to make sure it’s getting therapeutic according to schedule, and they can complete that medication loading and monitoring after discharge.
On point-of-care testing, I admit I’m prejudiced drastically toward this. Again, I’m not pushing any particular device. All of the point-of-care CLIA-waived devices have been through rigorous protocols, and the internal controls on these machines are reliable and standardized. Where we see problems is almost always in the first two to three weeks of a clinic being set up, and we’ve learned to get back in and give refresher sessions. That’s the only preanalytic variable we’ve run into that could cause problems. You get rid of a whole bunch of other problems, though. No transportation, no delay, and no loss of samples because you’re doing everything reliably on a face-to-face visit in real time. The big problem with point-of-care devices that were out previously was that there was no way to get those data into your LIS. The newest generation of devices are HL7-compatible, and they have interfaces that are being written for them. You can get those data in just like you can from your glucose machines.
Just to finish up with something nice and scary: A very interesting study was done a few years ago by Alan Jacobson, MD, of the Loma Linda, Calif., VA. There are at least 300 different combinations of INR testing reagents and lab testing devices that you can use in this country in an in-hospital setting. He picked about 30 and did a simple study. Each vertical line represents a single blood sample from a patient on warfarin at various doses. (See box below.) For organizational purposes, he went left to right in terms of increasing INRs. What’s being plotted with those little colored dots that have kind of a scatter to them is the variation on that INR from a single sample from that patient when tested at the same time by these different testing methods. And the numbers are fairly astounding, especially as you get up toward the higher range of therapeutic levels. The average variation from high to low is on the order of 100 percent coefficient of variation. Now that compares to all of the point-of-care devices, on which you can measure coefficients of variation in the single digits.
The No. 1 question I’ve been asked over the years whenever we set up one of these clinics is, “How does the POC device compare to my in-hospital lab?” What this taught me was, we’re probably doing this backward. “How does my lab compare to the POC device?” is what we ought to be looking at. The variation and discrepancy that can be seen with different testing methods and different ISI reagents is huge and has a very definite effect on therapeutic decisions made based on those INR levels. Especially if you’re going from this to a POC test or another variation, you are not testing that patient with two compatible methodologies. Pick one and stick with it.