[This Transcript is Unedited]
National Committee on Vital and Health Statistics
Subcommittee on Standards and Security
November 4, 2004
Hubert H. Humphrey Building
Room 800
200 Independence Avenue, S.W.
Washington, D.C. 20201
Proceedings by:
CASET Associates, Ltd.
10201 Lee Highway, suite 160
Fairfax, Virginia 22030
(703) 352-0091
P R O C E E D I N G S [3:13 p.m.]
MR. HOUSTON: — I am still trying to locate a large provider who might want to testify that we’ve had some trouble with that, but right now that’s what we’re on track for, again just two panels in the morning, I think which is what we talked about Simon, so its taken a lot of effort, frankly more then I ever thought to get these two panel together but it seems to be finally shaping up somewhat.
MR. REYNOLDS: When you said HIPAA did you mean e-prescribing?
MR. HOUSTON: No, e-prescribing is the day before, we decided to merge the e-prescribing with security for one day just to talk about medical equipment.
MR. BLAIR: Specifically you mean the 12th?
MR. HOUSTON: Is that Friday? I believe that’s —
DR. COHN: We’re talking about the 18th and 19th I think —
MR. HOUSTON: Next week is NHII which is the 12th and then the 18th, that’d be the 19th.
MR. BLAIR: Thank you.
DR. COHN: Okay, and we’ll talk to you offline if we have some ideas about additional presenters —
MR. HOUSTON: Again, I really was hoping to get one large idea and somebody or a large hospital and I had a couple and they all fell through so it’s sort of been a little frustrating trying to get somebody lined up.
DR. COHN: Okay, well we’ll work with you on that and see if we can come up with some names. Harry, do you have any comments? No, okay.
Well why don’t we move on then to the, as you know we all had a conference call talking about work plan and all of this and decision support is some part of the work plan and I think this is going to help us I think further investigate and I think we can reflect on the information and see where it fits in after the presentation. So Jonathan thank you for coming.
DR. TEICH: Thanks very much, Simon. It’s good to be here and thanks very much for having me and now reach the stage where I can’t quite see all of you and my own screen at the same time without having my glasses just so so I will try and get this appropriately adjusted so I can talk with some, so I can present this vision with some vision.
I’m Jonathan Teich, I’ve been before this group before, I’m a physician at Brigham and Women’s Hospital, I’m chief medical officer for Health Vision which is a services and information systems company, and professor at Harvard University in the medical school. And I’m here this afternoon to talk about some preliminary findings, near complete, but preliminary findings in a white paper and a projects that’s been undertaken to understand and recommend the appropriate levels of clinical decision support necessary for electronic prescribing in particular.
This a sum of a work that’s actually been chartered by several different government and industry agencies all together and was really a group that was chartered to look at both what things needed to be in place for what kind of years, what kind of things needed to be justified possibly when certification and incentives come into play, and also what sorts of specific structures, standards and other kinds of major projects need to be done on the underside to make sure that clinical decision support can be appropriately applied and appropriately used for the benefit of the patients.
I’m sure people are familiar in this group what clinical decision support is, this is just a typical screen one might see in the middle of an electronic prescribing application, in this case someone has prescribed aspirin and there are several warnings which appear on the screen, this is probably the most commonly known kind of clinical decision support, specifically a reactive alert, and this is, there’s a variety of pieces of advice on the screen telling you that this aspirin you’re prescribing may have a few problems, number one you’re already prescribing something in the same family so there might be a therapeutic duplication, number two down at the bottom the patient is also on warfarin which has a pretty whomping serious drug interaction with aspirin. And the screen tells you the particular warning in one sentence, gives you enough information to try and make an intelligent decision, and in many cases leaves you the option of taking various action such as continuing the prescription or changing it in one way or another.
So that’s one form of clinical decision support, there are really many other different kinds of things that are included in this. It includes decision support for a variety of different reasons, it can be proactive support that’s given to help guide ordering, to help guide the routing of orders to different pharmacies for example or to different staff, it could be done for safety reasons, for improved quality reasons, it could be done for reimbursement reasons in the case of formulary warnings, it could be done for presentation of reference information.
So to give an example of the latter on this particular example in the middle of prescribing the drug Prozac the doctor had, the prescriber had a question about it and on the screen you’ll see a little box there marked knowledge links, are a series of typical questions that one often asks when one is prescribing and in this case the doctor checked off warnings and a pop up came up and showed that here’s the relevant information for warnings of Prozac. So it’s a nice way of bringing in information, things like this are a little bit friendlier then simply getting reference books whether in paper or online because they’re available right at the moment of prescribing and take a lot less time and studies show that people use these much more.
And clinical decision support in prescribing is applicable to both the desktop and handheld, sometimes to alerts on cell phones, alerts on pagers, references which may come into even a printed sheet that comes when the patient comes in in the morning in the case of certain kinds of guidelines. So there’s a variety of different ways to disseminate these, to publish them, to build them but all these really come under the same category of clinical decision support of giving you additional information or additional alerts to help you make the right decision at the right time.
Although there are simple forms such as drug allergies, drug interactions, therapeutic duplications, people are familiar with, there are more complex ones which it seems if you look at some of the early studies are in fact a significant part of what is actually necessary to make a big difference in the care of the patient. So drug disease interactions, patients with diabetes for example should be cautioned before using beta blocker drugs on them heavily. There’s many other similar things, use of drugs for preventing errors of omission, patients who have certain kinds of problems, certain kinds of chronic diseases, if you think about the six chronic diseases that the IOM has asked us to look at most of those require medication therapy at certain points and one role of clinical decision support is not just to prevent an error but to stimulate someone to use a certain drug if necessary for certain situations.
There’s other kinds of feedback channels, I can get information possibly from pharmacies that a patient has or has not refilled a drug in time indicating the patient is not using it possibly because the patient can’t afford the drug, but obviously a patient that’s not actually taking the drug is a patient that may as well not have been prescribed it so that’s another one which has been suggested. The ability to have complete medication history from a number of different sources, whether it’s available in your office or not, many other things down the line become more complex forms of clinical decision support.
So that introduction really serves to get everyone on the same page of what we’re talking about when we speak of the term. The fundamental single statement that our group can say is that a significant degree of clinical decision support is essential to the value of e-prescribing. E-prescribing that does not have any clinical decision support, e-prescribing which is simply a clerical aid and a communications aid can improve processes but is missing a substantial part of the value of safety, improved quality, and in fact improved cost that can be brought about. Therefore clinical decision support should be a part of all mandates, recommendations, support policies, or certifications that go into systems or electronic prescribing. And really the rest of what I have to say is kind of exposition of that single statement.
We know that clinical decision support can achieve disease management goals, we have research to indicate that it prevents certainly the most egregious of adverse drug events, can help make sure that patients get renewals, can help improve patient satisfaction with care, and can do a lot of things in terms of simply communicating the right information to the right person. In addition CDS can be used for improving practice efficiencies and reducing costs, in particular most studies that have looked at early CDS in electronic prescribing has indicated that it reduces call backs from the pharmacy by as much as 70 percent, which is a significant time and resource saving. It promotes adherence to formularies which can be of financial value to the patient, the physician, or the payer, and can reduce administrative costs for pharmacies and for PBMs or other kind of payer mechanisms.
We indicate that various forms of CDS are necessary and a lot of our recommendations will deal with the specifics of that. The other part of the coin is that there are various kinds of support structures that are necessary, not immediately or absolutely necessary but important in terms of improving quality and adoption. When I was here in March speaking about the eHealth initiatives work on electronic prescribing I talked a lot about specific standards that were necessary, some of these have been included in the preliminary findings of this organization.
There are three real primary areas in which supporting structure is necessary, one is continuation of promoting appropriate standards, which we’ll talk about, one is in establishing a basis for storing and exchanging appropriate kinds of knowledge so that the various vendors who produce these systems, the various prescribers who use them, the various pharmacies, all have access to the latest knowledge, all have access to a reasonably consistent set of rules and knowledge that can be turned into rules. And finally it’s important to have supporting dictionaries, it’s important that for example the drug dictionaries that are being used right now for electronic prescribing are highly varied, they’re very often not prescriber friendly which results in a lot of free text getting used in communications between prescribers and dispensers, so appropriate dictionaries are necessary as well, and we’ll talk about those towards the latter part of this talk.
So I’m going to give a brief history of what we’ve been working on and then we’ll talk about some of our recommendations. Certainly the recent history includes the substantial findings on the inpatient CPOE experience, work that’s been done at several different institutions, reported in a number of landmark papers over the last six to ten years, indicates as you all are aware that adverse drug events are numerous in the inpatient setting, that they are dangerous, and that they can in large degree be ameliorated or prevented by the use of inpatient order entry systems.
There has not been as much work to date in ambulatory electronic prescribing simply because its come up later, simply because its substantially less of a control environment obviously, but what we do know now is that there’s quite a bit of data that shows that ambulatory adverse drug events are probably as numerous and probably as dangerous as many of the things that happen in the inpatient environment. We know that a great many inpatient admissions come from these, I know personally as an emergency physician that several patients per day come to my own care because of some sort of adverse drug event which is responding poorly inside their system.
The Center for Information Technology Leadership Report on ambulatory CPOE established that CDS was of significant value, both in quality terms and in monetary terms. And there have been a number of pilot studies now on e-prescribing which certainly have immediately shown usability gains and efficiency gains, and which I now have seen are starting to show documented safety and quality gains but many of those are in the publication process and can’t be fully discussed just yet.
In early 2004 I was part of a group, I chaired a group from the eHealth initiative to establish a large multi-stakeholder consortium, about 75 folks participated from vendors, from pharma, from the pharmacy world, from the prescribing, from many of the organizations, to try and make some statements about appropriate adoption and appropriate value that can be coming from e-prescribing. And this established goals and recommendations in usability and standards, clinical decision support, and implementation and incentives. Also established the notion that there was a pyramid that people could come in on a relatively modest level but should have their eyes on the prize which was to quickly get themselves up to the more advanced forms of electronic prescribing, which certainly seem to have a sharply sloping increase in value. The same is true for clinical decision support, this is the electronic prescribing report which is available on the internet.
There was a study done by UCLA and RAND, there were actually two studies, both of them chaired by Dr. Douglas Bell, and these are primarily have been Delphi process models to review a number of the desired features of electronic prescribing, as part of that study there was considerable discussion given to various forms of clinical decision support, both the actual rules and their delivery, and those have been published in Health Affairs, there’s more findings that are in publications, and there’s certainly further work in progress going there.
There was the publication earlier this year of the Clinical Decision Support Implementers Workbook, this came out of HIMSS, the Healthcare Information Management System Society, and this group was essentially a how to practical workbook for clinical decision support implementation and it starts to break down the various kinds of rules, the various kinds of delivery of these and really in its quest to show people exactly what they should do to do this practically has managed to make a reasonable taxonomy which I think can be useful here.
The current work in progress began with a request from the national coordinator’s office to support MMA related activities including those of this body, also to support Dr. Brailer’s strategic plan and other goals, and particularly to look at both certification friendly information as well as supporting structures. It was assigned to the Clinical Decision Support Task Force that had produced the Clinical Decision Support Workbook, its had further significant support from AHRQ and from the American Medical Informatics Association, as well as the natural ties to the many people in that group that were also on the EHI report.
We had a group of five persons who developed an outline for a white paper and developed a slate of recommendations, then there was a consensus panel held on September 9th including members of all the things I’ve discussed in the last four or five slides, and that panel discussed a couple of tables which included recommendations of which things should be happening, primarily in the sense of which things should be happening now, which things should be later, which things are low level and which things are absolutely mandatory. And there was also a discussion on the support structures. Right now that set of recommendations and that white paper is undergoing final review and I’m bringing you preliminary recommendations from that panel here today and you can expect to see final results in the next three to four weeks.
The scope of work of this group included particularly the e-prescribing context, it was pretty clear immediately that many of the things we were talking about could be applied to clinical decision support in general, that it wasn’t just about electronic prescribing, there are many other areas of electronic health records that probably need the same treatment and there are a number of, when we did our work there were a number of structures that we found could be applied to other things. However, the report specifically is about electronic prescribing and particularly about CDS in that context so many other valuable parts of electronic prescribing, such as for example electronic communication between pharmacies and physicians, were mostly held to be out of scope unless they specifically effected CDS. But it did effect, it did include all sorts of errors, all sorts of quality improvement issues, all sorts of things that involve DUR or cost reduction issues. And it applied across the scope of all parts of the prescribing chain.
So here’s what we’ve come to at this point, the preliminary recommendations are really based on two tiers and two dates. We have established that there’s basic and advanced in 2006 and 2008, the reason for those years, 2006 is really when under the MMA many of the demonstration projects are slated to be enforced, 2008 is when many of the real supports kick in for Medicare patients. These also coincide to some extent with initial results that we expect to come out, initial timetables for the certification commission that’s recently been established, as well as simply giving people a chance to ramp up.
So we recommend things that should be found in all systems in 2006 and then there are things which are highly desirable, may not be mandatory, or may not be basic primarily because we’re not sure that everyone can quickly come up to speed on them but some parts of the industry already have, so advanced 2006 were generally things which we knew were available in some sectors, which could conceivably be obtained by everybody but were probably a little bit higher ahead for people with a heavy work schedule. Many of the things that were advanced 2006 were then distributed among basic and advanced for 2008 as well as some interesting additional things for those things such as the addition of genomic information. The presumption is that if there’s some sort of incentive being associated with either the production or the use of these systems that presumably an advanced system or a system which meets advanced criteria may be associated with somewhat greater incentives however those are defined.
The core recommendations are really divided among four different sectors, the knowledge base, the database, features, and organization, we’ll talk mostly about the knowledge base which is essentially the classes of rules, the classes of inventions, the classes of support that need to be in place, so alerts for certain kinds of indications and so on. We also did make recommendations on the database, for example I had what stage does one have to have things like laboratory results, problem lists, and other things which are not traditionally associated with prescribing systems in place.
We talked about features because there are certain things which make clinical decision support hard to use, such as what happens when there are conflicting alerts or what happens when persons other then a full fledged physician is actually the one doing the prescription or the prescription renewal. And we talked about organization at the local level and governance issues such as who has to review things, who has to vet these things on an initial and on an annual basis and how communication has to take place. So I hope we’ve set up enough different dimensions to either inform or confuse everyone, and let’s talk about some of the hard findings.
Basic 2006 essentially creates a knowledge base including those things that are often associated with common well done systems today, it includes drug allergy interactions, includes drug/drug interactions, it does include drug reference information callable directly from the prescribing screen, similar to that screen I showed you earlier on, that knowledge link or info button. And that’s not available on all systems but is well we believe within the reach of all systems with current technology. The ability to select dosage and frequency and also form and strength from lists, although the strength itself would not necessarily be required because that’s not common to certain prescriptions. And the ability to have at least general recommendations concerning adjustments for age and weight and test results, so it would say gee when you’re ordering this particular sedative one should be careful with it in the use of patients over 70 for example without specifically saying that this individual patient fits those criteria.
DR. STEINDEL: When you say knowledge base are you talking about a physical knowledge base that would be available to everybody?
DR. TEICH: We are not talking about a specific individual knowledge base in this setting, in the setting of this slide, we’re talking here about what elements and what categories need to be well included in your knowledge base.
DR. STEINDEL: Thanks for clarifying.
DR. TEICH: We do talk later on about whether there should be such a thing.
DR. STEINDEL: Okay, thank you. Sorry to jump.
DR. TEICH: No, that’s fine, please ask questions during the presentation as you wish.
Other basic 2006 elements, a number of things were necessary in the feature area, number one on everyone’s list was the techniques to reduce alert fatigues, particularly as well associated with drug/drug interactions, there are so many that are included for protective reasons in manufacturing literature that it’s very difficult if one were in fact to be presented with all possible alerts then one would probably ignore all alerts including the most important ones. So at least there needed to be techniques, we didn’t specify what those were, there are some good examples in some of the references I talked about earlier, but we specified that there needed to be conscious attention to this and had to be measured somehow.
A prescribing system that seems obvious but we did state that it needs to enforce the production of a complete prescription, needed to include quick choice lists so that typing was minimized, all these things are really usability and adoption features related to clinical decision support. Had to be able to display and show the entire medication list and had to be able to give the patient a copy of their entire medication list so that they too could participate in this.
And then on the organizational side had to include initial and periodic local review of all alerts so that there had to be some sort of body established which would say yes we do accept the following alert or we do accept the following interventions, and then on an annual basis or appropriate other time various kinds of stakeholder owners would be required to review to make sure the content was still relevant and had not expired.
Large category of knowledge based pieces falls in the advanced 2006 category and you’ll see I’ve divided this into those which then later will filter into basic in 2008 and those which will filter into advanced. And these are ones which we have seen in advanced system, which we’ve seen in some commercial systems and many laboratory systems, and which are within technological reach and which we also have found based on existing evidence to have a significant impact on safety and quality. Interactions of drugs with lab results, prescribing lasics(?) when a potassium level is very low, prescribing dajoxin(?) when it’s very low or very high. Interactions between drugs and problem lists, again beta blockers prescribed for insulin dependent diabetics perhaps. Errors of omission, and this is really the whole area of what you might call disease management preventive care essentially when patients have specific problem lists one needs to be reminded that statins are indicated for patients with certain levels of LDL for example, that have not passed other criteria.
Patient instructions, which is fortunately included in most systems but not all today, there was considerable push that that should actually be included in basic but for a number of reasons, particular language issues, it was put in the advanced category for this level. Advice regarding necessary follow-up tests so when certain drugs are ordered the liver function test must be checked, kidney test must be checked later on and so on down the line and that needed to be included.
And in this level we first put in payer specific formulary warnings, this is both a cost and a safety issue, cost for obvious reasons, safety because when patients go into a pharmacy a find out that the cost is excessive they very often do not fill these and as we know from recent publication they often do not tell their physicians that they do so. So formulary alerts are very important, these were thought to be desired for basic but again listed as advanced primarily because of difficulties right now in assembling well organized payer lists.
The more advanced of the advanced, highlighting best doses so that not only would all dose be available but best dose would be indicated based on patient’s age and weight and later on by renal function. Order by indication, which it’s interesting because there are many prescribers in the country that do not use this mode at all and many prescribers who value it highly, the ability to simply say I have a patient with an ear infection or I have a patient with hypercholesterolemia and please show me all the drugs available for this. Ordering by guideline which is really a multi-step way of doing the same drug problem list interactions. Drug references that are targeted to likely questions and notification when renewals are due. So all these things that I’ve listed here, within technical capability, certainly found in scattered systems as we go across, and which we think are worthy of recognition and whatever incentive may be decided upon if these are available in 2006.
In addition in the area of database and functional requirements, advanced systems at this point should be able to show medications from all prescribers, whether this comes from payer sources, PBM sources, whether it comes from the establishment of significant regional organizations that are able to pool data, medications from all relevant prescribers for a patient should be included. Since many of the knowledge based items call for a drug problem list interactions obviously the problem list has to be there so many elements of the electronic health record have to be here —
MR. BLAIR: Does this imply that the format for medication histories and medication lists that might be derived from dispensers and from payers and from prescribers would not be a consistent format or data structure in order to be standardized and pulled together?
DR. TEICH: Yes, it absolutely would.
MR. BLAIR: Thank you.
DR. TEICH: We’ve long said and we’ve had it in both the EHI report and on other reports from this body, it’s long been established that the more that electronic prescribing connects to the full electronic health record the better of everything is and so at A2006 level we call for some significant elements of the electronic health record to be tightly wedded and that’s something that’s going into this report which is the combined HIMSS/ONCHIT/AMIA/AHRQ report.
The ability to handle multiple simultaneous alerts without fatiguing people, doctors get frustrated when they hit an alert and decide, they say they can go past that and then they have another alert which probably would have obviated their choice on the first one. Usability for mid-level clinicians, and this is sort of a nod to reality in that many actual prescriptions are written, particularly renewals, by the hands of persons other then the official prescribers and there needs to be a way to either account for that or put it into appropriate perspective and reconcile it so that this can be done safely so that appropriate limitations and appropriate permissions are given.
The ability to display lab results immediately when a drug is being prescribed so that in fact one can see the lab results in advance and not have to worry about making a mistake, we would really like to start seeing advanced systems head off some of these alerts by showing you things up front. And a more direct involvement of patients, the ability of patients to see their medication list and perhaps to see them in an online form that allows them to make corrections, that allows them to suggest that well you have me on these five things but I’m also taking these two others that you don’t know about, the ability of the patients to really be as the most interested party to suggest corrections that could be accepted in the practice that say I’m on these two extra drugs or I’m not taking that one anymore, probably one of the best checks we have to make sure that someone is taking things correctly. There’s no doubt in my work as an emergency physician that I see many patients coming in with three of the same, bottles of the same drugs in different names and different bottles from different prescribers, a very dangerous situation to be sure.
Move into 2008 many of the things that were on the A list for 2006 move into 2008 but in addition in 2008 this is where we insisted that a prescriber friendly drug dictionary needed to be fully used, so Jeff this goes to your point, it goes to the point of communication from prescriber to pharmacy and back, it goes purely to the ability to use decision support effectively, when drugs are in such a sub-optimally organized dictionary it’s very hard to apply consistent rules to them and so we want to make sure that this is the point at which drug dictionaries, we suggest later on that perhaps the Rx-Norm/NDF-RT effort is the way to go with this but this needs to be in place and well used by all systems in all circumstances.
In 2008 is the time when full integration with the electronic health record should be there, there should be nothing really that is separate between electronic prescribing and electronic health by that time, obviously the sooner the better. Codified allergy and sensitivity reactions so that people can wrestle with the frequent problem of deciding between an allergy to a drug that causes stomach upset versus one that causes anaphylactic reactions. And then the ability to both receive reports on one’s aggregate performance and to be able to publish reports about local adverse drug events side effects, hopefully greatly improving the ability of the FDA and other organizations to collect these and to make some sense out of a much greater body of data.
The last section of our recommendation tables, for features in the advanced for 2008, really looks at some things which are really not available right now in any commercial systems but what you see on a trial basis in various laboratory systems and academic systems. Drug laboratory result interaction that’s triggered by the laboratory result, so that I’m happily taking a drug but then I get m periodic lab test done and the lab test comes back with something alarming, even though I’m not even doing a prescribing at that time I need to be alerted that perhaps the prescription needs to be changed.
Dosing guidance, active dosing guidance based on renal function, nopatic(?) function, age and so on. Here’s where we call for notification re non-filling or non-refilling, in other words non-compliance. This is certainly the place at which this can be used, one caveat that we’re still working out now is the privacy involvement of this and what this means in terms of the ability of patients to be able to approve or disapprove of these. And the ability to suggest consequent orders and tests, the ability to suggest that certain tests have to be ordered today, the ability to suggest that certain kinds of follow-up order is to be done and actually to rope those directly into the full electronic health record and the full ambulatory CPOE system.
Not surprisingly therefore in advanced 2008 we also call for there to be a full-fledged ambulatory CPOE system including things that are not just medications so that laboratory orders can be placed so that if a lab order is needed because of a drug then that can be not only suggested but can be fed directly into the CPOE system. Also at this time we include pertinent genomic data, obviously genomics is both fascinating, promising, and a question mark at this time but as it becomes available and clinically relevant this needs to be well supported by existing prescribing systems.
And then another relatively controversial but important issue, the selective suppression of alerts, not just the alert fatigue of drug interactions but we do know that in certain circumstances something which would be an adverse drug event or something which would be perhaps a prescribing error in some cases would not be in others, prescribing of drugs for the hematology oncology population who commonly have many different abnormalities which would otherwise trigger alerts for example, prescribing for certain post surgical patients and so on, so there has to be a mechanism for supporting selective suppression of the alerts in those cases without allowing too much reign.
So this is the set, as I mentioned this is preliminary and this has gone through two rounds of review and we’re sending this out to a wider body to try and get final comment on this before putting this out fully but we do hope to have that out fully before the final meeting of this body in December. And this will be published I believe simultaneously by ONCHIT office and by AMI and HIMSS.
The other area we mentioned was supporting structures or different kinds of structural elements that had to be in place. Knowledge exchange, so really the question here is what do I need to do for two purposes, number one to make sure that all developers, vendors, and users of prescribing systems have reasonable access to a good foundation. It doesn’t mean that they all have to accept every single rule in the foundation, in fact we’re not even dictating the specific allergies or the specific interactions. But there should be someplace where people can get ready access to a set of rules because right now many people are building and rebuilding these things over and over again at their own institutions.
This also supports research, some of the work of the CDS Workbook that I mentioned early has developed kind of an organizational taxonomy of clinical decision support and we hope that this will help stimulate research and help research become applied so that when one hospital or one practice is able to study the effects of a certain intervention they can actually put it in a framework so that other people using similar interventions will be able to draw on that research. So being able to categorize these things, being able obviously to advance the direct state of practice because as new findings come into the literature, as new medical knowledge becomes available, we want this to be transmitted as quickly as possible to practical prescribing systems if in fact it’s held to have clinical relevance.
So this could take a number of forms, it could take the form of one or more clearinghouses of knowledge and we thought about, it was discussed should there be a single central clearinghouse, that’s a possibility but it’s more likely that there should be a single structure for clearinghouses, a single format by which people can read and review clearinghouse information and then the ability of a number of certified groups to publish clearinghouses. This allows for different organizations, different clinical bodies, that may have different reasonable opinions to be able to publish things. Allows people to make educated choices about the different rules that are available while at the same time being able to quickly incorporate them into their own system.
The group that’s writing this white paper has an interest technically in being able to support some of these, there have been some prior efforts to do this but many of them have suffered from possibly not being as practically oriented as possible and not having sort of an ongoing momentum and we want, there are ways to possibly reverse those. So there’s a discussion going on among the group that’s working on this white paper as to how we can work with ONCHIT and with the rest of the organizations to possibly create at least the structure for these.
Standards and dictionaries —
DR. COHN: I’m sorry, can I ask a question on that one? So I guess what I’m hearing from you is is that we’ve got a plethora of knowledge exchange standards but none of them are useful and that’s not something where you’re not seeing any low hanging fruit there.
DR. TEICH: There are a number of standards that have been discussed, certain HL7 has been working on this for quite some time, Arden(?) syntax which is incorporated in this was part of it, there were various organizations that tried to establish the body. What we’re really talking about is the actual home, the actual place where specific items can go. So standards are one thing and standards are important, we think that it’s important to establish solid practical framework on top of those standards for what one of these websites, one of these databases would look like, that an electronic prescribing system could periodically read, have reviewed, and directly incorporate into their own system. So we’re looking not only for the blueprint but also for the mechanism for actually developing the car that people will use.
MR. BLAIR: Just when that question was asked you were just about to start into the standards topic.
DR. TEICH: Yes, standards, we had discussed four standards earlier with this group and they were validated by the HIMSS/AMIA/ONCHIT group, and AHRQ group, it’s hard to put all those letters together, but the HIMSS/AMIA/AHRQ and ONCHIT group also looked at these and really came up with the same four key areas that are necessary specifically to clinical decision support. Certainly the most important on everyone’s mind is the establishment of some sort of “doctor level” or low granularity drug dictionary. Drug dictionaries used for prescribing are primarily organized by pharmacy packages and it becomes very difficult to establish the right one that one wants to prescribe and certainly the right one that one wants to transmit to the pharmacy.
We looked with interest, great interest, at the Rx-Norm project, it was held by most people in our workgroup that this is probably one of the best ways to do this, Rx-Norm has already worked with some of the major commercial drug dictionary vendors and we would really like to have their work further supported and out into I would say commercial usability, not that this need to be a commercial product but that it should be available and should be readily usable by the commercial producers of the prescribing systems, because this we think is going to be one of the biggest steps in establishing a common language that people can use, not only to have better communication, not only to have better safety, but also to be able to do better research so we understand how research done in Arizona relates to a doctor in Maine.
Formulary standards —
MR. BLAIR: Before you go on to that one, does this mean that one of the high priorities that NCVHS maybe should look at during the next several months would be a standardized message between the drug knowledge base vendors and National Library of Medicine with respect to the clinical drugs for Rx-Norm?
DR. TEICH: I think that’s possible, what I would have said was that there should be almost more of a bulk process whereby the makers, the First Data Banks and Multums and MediSpans who produce most of the drug dictionaries that are used in prescribing systems should be able to have bulk access to an Rx-Norm level dictionary and should be able to in fact include those in their own products.
MR. BLAIR: Does that need to be a national standard or can that be done the way it’s being —
DR. TEICH: I think there probably needs to be a national standard which is either Rx-Norm or something like it, I mean all of those drug dictionaries are pretty good, they have as I say granularity issues and they’re also slightly different from each other so that tends to hamper things. So yes, I think and our group thinks if I can speak for them that there does need to be a single standard which can be updated as drug development goes on.
Standards for formularies, understandably these were later on in coming, but right now it’s very difficult to do the cardinal problem is a patient comes in and you want to know whether this drug is on that patient’s formulary, in order to do that I need to know who is paying for the patient’s medications and I need to know what that particular payer’s formulary looks like. Both of those are relatively non-standard, it’s difficult right now, there’s no organized list of payers that makes it easy for me to understand exactly which plan a patient is on, and then the various plans publish their formularies with different structures, with different terminology, which certainly gives producers and developers of these systems headaches as they have to go frequently from one to the other and change all the formats to normalize them.
And it makes it probably much more likely that this kind of clinical decision support is not in fact being applied when it could be because it’s very difficult to assign the patients to the plans and then the plans to the right drugs. So attention to having standards for both a list of payers and also standards for how formularies are specified would certainly help that.
The third of four standards —
DR. STEINDEL: Can we discuss what the relationship between the drug dictionary and the formulary standard, because when you’re talking about the doctor level drug dictionary and you mentioned Rx-Norm you’re talking about one level of drug and the way it’s described. And then when you go to the formulary standard you’re usually talking about a dispensed drug, which generally speaking —
DR. TEICH: It’s true that Rx-Norm is used at the doctor level, the formulary, certainly formularies that I both receive as a physician and formularies that I see as a producer of some of these, formularies tend to be also at a doctor level as you seen them, I mean they sometimes they will specify various forms but in general the formulary calls for don’t use nexium(?), whatever it may be, there may be something else that’s out there. So in general the formulary is organized by a classification, usually a therapeutic classification of drug and then the various drugs in there —
DR. STEINDEL: Then let’s go one step further, how do we relate the doctor level drug to the dispensed drug? Which is usually an NDC code?
DR. TEICH: I think that the doctor level drug, certainly in Rx-Norm a doctor level drug would probably include a large number of NDC codes obviously, for something like amoxicillin there could be 50 or 60 of them. In the formulary the primary difference is that, they will say for example use a generic form of this as opposed to a trade form, which would be a level down, but most formularies don’t for example specifically say that it has to be this package or this particular strength, so that part wouldn’t be necessary. What we’d like to see in the formulary is this drug, perhaps this brand if necessary, in this form, and that’s usually as far as those go and I think that would match up fairly well to the doctor list.
MR. BLAIR: I guess I thought that the National Library of Medicine is mapping Rx-Norm as a clinical drug to the NDC codes —
DR. TEICH: It is.
MR. BLAIR: Then there was another thrust to your question that I’m missing there.
DR. TEICH: Well, the problem with doing that is that it’s a one to N mapping, so if I have 50 amoxicillins let’s say and they map to, and what the doctor can write is amoxicillin capsules, and then what happens is perhaps if only two or three of those NDCs were off formulary there would be no way to distinguish them in what the doctor has said so the formulary does have to be at a level that matches that so that if a doctor prescribes a drug which may have some on formularies and some off that they would be able to find a match.
DR. STEINDEL: Actually there’s been a lot of things that you’ve mentioned that we’ve talked about at this subcommittee, also are in the letter, and you acknowledged that actually at the beginning of your presentation. I was just wanting to ascertain that you wanted to make sure that the link between the doctor level drug and the dispensed drug also existed.
DR. TEICH: Exactly right.
DR. STEINDEL: That was the purpose of my question.
DR. TEICH: No, thanks for bringing that up because that is an important point to include and we probably should state that more explicitly.
There is the sig standard, the sig is the specification of dosage and frequency and perhaps other instructions for giving a particular prescription and it’s kind of a paradox, it would be a relatively easy thing to specify if one person were in a room and it’s the kind of thing which perhaps we need to get sort of one person in a room to do and then let everybody else take potshots at it. But certainly it’s important because right now without a sig standard I don’t really know how much of a given drug is being prescribed and in drugs where there’s a concern about limitation of doses, everything from chemotherapeutic agents to acinometaphine, it’s very important to know how much has been prescribed, very important that someone can specify that, and that that can be kept track of and counted. Also important when transmitting information to pharmacies and back to physicians that there be able to have that kind of interaction so that people can see what’s been prescribed, people can follow doses as they go up or down.
MS. FREIDMAN: One of the things that has come out of the NCVHS process is the industry is moving very aggressively to establish a sig standard and so are just recommending that there should be one or that if somebody comes up with one then it should be adopted?
DR. TEICH: We’re recommending that one be in place and be in full force. There’s no doubt by the way that I am saying some things which already did appear in the preliminary letter and those are by means of acknowledge and enthusiastic support from our group.
MS. FRIEDMAN: I just got an email about the status of the progress of industry sig standard group and they’re really booking, moving right along.
DR. TEICH: That’s good, you’re several weeks ahead of me on that and I would like to actually know more about it.
MR. BLAIR: I think what you’re hearing Jonathan at least on my part, and I’m reading that into Maria’s and Steve’s comments, is we’re struggling to identify what we should focus on in December and January for standards evaluation and recommendations that might relate to clinical decision support, the ones we haven’t done yet, so we’re sort of looking for okay what did we miss, what did you find that would be the appropriate next areas for us to focus on for standards related to decision support.
DR. TEICH: I think philosophically we’re trying to help prioritize as well and really what we’re trying to do is take a lot of the things that have been discussed, things which are in fact in industry standard process and get them out the door, get them out the door where they can be well used and put into play.
The other thing is that there’s certainly the standards on this particular slide are sort of real standards standards, I think we had this problem with definitions at a previous talk, these are standards that are actually real, I can write these down and put them into a list standards. I think everything in this talk is a standard, I think the rest of the things I’ve discussed are standards for what a good prescribing system should look like and I think that what we care most about is that any kind of mechanism that either incents or in some ways requires the use of prescribing systems needs to have certain minimal criteria which we’re going to call standards and that’s really, those tables that I gave up above, those really were the standards we care about. We are concerns as you are that there needed to be more clinical decision support into the “standard set” of functionality for prescribing systems and that’s probably the most vital soapbox I’m trying to produce today. But these standards which are standards standards are ones which we’ve talked about before, I think most of these things are now in play and you’ve acknowledged and given me more information on that but we just definitely want to see these things out because these are the ones which are stopping progress in CDS.
DR. COHN: Let me ask because you obviously haven’t gotten to the fourth one and I was just sort of curious about what you mean by, it says local state, I presume it’s board of pharmacy requirements. What do you mean by standards in that respect?
DR. TEICH: The fourth one, local state board of pharmacy requirements was a request that the, there are about 56 domains, 50 states plus various other administrative bodies that are allowed to dictate what goes on a prescription in their state, and they’re all philosophically the same and they’re just about all different as they’re written by rule. It can be something as simple as one state requires the words dispense as written to be used, another state requires the words do not substitute and another state requires the words no substitution permitted. One state requires that the address of the patient need to go up on the top, another state requires it be the bottom.
While these are not major problems when you look at them what they do is they slow down the development of these systems because people who produce these systems for common consumption have to go through all 56 of these and have to make 56 different sets of things. So some of these things, really like the formulary standard, they don’t make things impossible but they’re a kinetic problem, they slow down the process because people have to piece through every one of these things and decide how to build it and software gets more complex. So it’s a practical logistical problem.
MS. TRUDEL: Hi, it’s Karen Trudel. Jonathan, it seems pretty clear to me that the quality of decision support that goes into e-prescribing software can be no better then the quality of the underlying clinical rules that are used to establish those functions. Could you talk a little bit about how those rules should be determined and how we could measure and monitor the quality of those underlying rules and is that something that the certification commission might want to get involved in in the course of its deliberations?
MR. BLAIR: Certification commission?
MS. TRUDEL: For software.
DR. TEICH: We had a number of things which were listed as, which we labeled key controversies and you hit one of them. There were actually six key controversies most of which have been resolved by this time and the one that you’re speaking of was really should we be in the business, so it’s a little bit different, should we as a body of the people doing this white paper be in the business of trying to say not only that there should be drug allergy interaction alerts but which ones there should be, not only there should be drug/drug interactions but which ones there should be. We held that for the purposes of this scope, first of all it’s a scope issue, that it wasn’t in our scope to be able to do something that deeply that it was clear that there did need to be reasonable approval and reasonable rationality to the establishment of those things but that it wasn’t something that we were going to take on in this particular level.
And the second question was should there be some kind of independent piece. And I think this is a question of sort of formality versus practicality, there’s really two or three different types of actual rules suppliers, there’s the table based knowledge suppliers, the people that again usually First Data Base, Multum, those folks, who provide these drug dictionaries for prescribing and who also provide tables of specific allergy groups, specific drug interaction groups and so on. They do them independently, they have a large number of pharmacists and other folks at their disposal to help make these things, so that’s one group which produces the things for the table based interventions.
Another group is the people who provide reference knowledge, whether it be for the patient level or whether it be at the physician level of people who actually provide the drug information sheets and so on.
And then the third is kind of the one offs, the everything else, the drug problem interactions for which there really is no real hard single table, specific rules about renal failure dosing for which there are a number of academic projects but no real tables, so there’s a lot of things that are done at the grassroots.
Practically speaking the allergy tables and the interaction tables that you get from the commercial sources are pretty good, they of necessity the interaction tables tend to be overkill because they just have to sort of supply those. But they do in fact tier them, they do in fact have high level and medium level and low level and they do that reasonably well. So you could argue that in fact there should be some vetting of these and there probably should, it’s not a screening problem right at this moment but certainly something which someone should keep their eye on.
Similarly in the areas of, the areas of patient knowledge and prescriber knowledge, reference information, generally quality control is very good at the prescriber level and quality control can be quite variable at the patient level. So here there’s probably a means, there probably is calling out for a need to in some way certify these. As you know there was an attempt not long ago to assign sort of preferential website domain names to certify knowledge and there’s YURAC(?) certification and other kinds of bodies which probably should be included in this.
And then the third category of the kind of grassroots ones, I think those are the ones which one or more of these clearinghouses, central knowledge sources, probably should be in the business of taking in and in their tables say here’s a particular piece of knowledge, here is the source of it, here is who it has been vetted by so that perhaps I can filter out and say well I’ll look at this table but only those that have been vetted by the U.S. Preventive Services Task Force, or only those that have been vetted by AHRQ, whatever else I wish to do. So there probably is a need to have some kind of rationality that, I think they should all be shown but I think you should probably say who in fact is vetting some of these.
DR. HUFF: One of the things that I’ve spoken to Mark about is that one way to get at what’s really useful is to actually just start accumulating data from people who are actually using them. So in our particular case, I mean I think without too much trouble we could say which ones are triggering, which ones are being adhered to, which ones we installed and took out because they were causing a nuisance and you could, for a little bit of work you could come up with best practices list, so it would be more empiric then sort of proceeding from first principles to say what’s right and good, to say what’s useful and practical that people are actually doing right now and publish that broadly and approach it that way.
DR. COHN: These are certainly an interesting set of issues, I guess as I’ve listened to all this I guess I’m thinking about first principles, second principles, third principles, and I somehow think that we need to get data sort of moved around, we need to get ways to share knowledge, share best, share the information. I somehow think that the quality discussion is something that, I mean it’s an important discussion, I’m just not sure it’s the first one to have. As you commented I think there are a lot of pretty good data out there.
DR. TEICH: Certainly a personal goal of mine is to really try and work with other folks to establish this kind of framework so that again for research so that I can put down when I do a piece of research and say this intervention is no good I have a place to post it, I have a place to put it in context so that other people can look at it and say well maybe this intervention is no good also because it’s just like that. So I think that establishing a framework where I can put pieces of research like this and best practices like this is a big part of that. And establishing some way of disseminating that is important. I did mention that I didn’t think these things, the best thing was the absolute first thing that needed to come out but I do think that it’s going to be a major contributor to quality as things go forward.
DR. COHN: Jonathan, I need to do a time check with you, what is, I mean we’ve obviously gotten stuck on this slide —
DR. TEICH: I have only one more slide.
DR. COHN: Oh good, okay, I just wanted to make sure you had a chance to finish, that we didn’t wind up cutting you off.
DR. TEICH: I’m happy to continue this discussion any time, this is the last slide as always is further work. So I didn’t mention really sort of whose been involved with this, there were five folks, Jerry Osheroff(?) of MicroMedix and the chair of the HIMSS Clinical Decision Support group and also an AMIA member, has been co-chairing this particular workgroup with me and also Dean Siddick(?) and Robert Genders(?) and Eric Pifer(?) are also closely involved in the core of building some of these things. We’ve worked closely with Kelly Cronin and Karen, Mike Fitzmaurice, Karen Trudel for the record, Zebediah(?) Kimmel(?) who’s an intern in the ONCHIT office, and a number of those folks have been sort of closely working on this. Carolyn Clancy as well participated to some extent but Mike’s been doing most of the AHRQ heavy lifting.
We had at the group that assembled on September 9th was comprised of people from academia, people from vendor interests, people from pharmacy interests, Mark Levitt(?) and Jane Medsger(?) who are now both representing the Certification Commission, and who is obviously interested in this work. So those folks have been really sort of on the inside of this and we’ve been expanding, in the last review we’re expanding it to a wider body of folks that we think ought to have a last look at this before it goes by.
So we are reviewing and completing both the recommendations and the white paper over the next few weeks. We have been talking with Dr. Brailer’s office as to how one can start to establish further involvement in creating some of these structures over the longer term, and there have been some preliminary discussions among these various bodies, among HIMSS and AMIA, ONCHIT and AHRQ, EHI, in terms of how to collaborate further to really sort of bring as much, as many IQ points to bear and as many people from different sectors to bear, that can establish some sort of collaboration to establish some of these structures in the future.
That’s really where we’re going and obviously our short term is to contribute to the work that you folks are doing, to contribute to a reasonable set of recommendations for both the demonstration projects and the CMS supported real world e-prescribing projects.
DR. COHN: I think Harry and then Steve.
MR. REYNOLDS: Excellent job, Jonathan. In one of your early slides you had a device shown on the screen —
DR. TEICH: A hand held device?
MR. REYNOLDS: Yeah, one of the things that’s interesting to me is there’s that device, you’re sitting there in front of a lap top, there’s PDAs, there’s every other thing, there’s Blackberry’s, there’s everything else running around. Have you guys thought about whether or not you want these standards and the information flow and everything designed at a least common denominator? Or has that even entered your thought?
DR. TEICH: Its certainly entered our thoughts, it’s a great question. It’s really a proof of the pudding kind of answer, it doesn’t matter on first principles what device one is using, the device should be there to fit the need. People use, hand helds have gotten a lot of the buzz on e-prescribing over the past couple of years and hand helds are very useful in certain context and less useful in others, certainly if one is a traveling provider, goes from office to office very quickly, needs to do certain things, then one can use a hand held very well but there are some concerns with them tending to break and tending to walk away because of their small size. If one operates in a one or two receiving offices one might find it better to use a larger format like a desktop or laptop device, wireless laptops if one needs to walk around but still needs that kind of format. I think the short answer is that the device itself shouldn’t matter unless the inherent properties of the device make it impossible to do this properly.
MR. REYNOLDS: Let me challenge that a minute. But on the other hand if the design of the standard of the information that flows and the amount of information that flows could circumvent —
DR. TEICH: You’re saying that’s, are you saying that a certain type of devices may restrict the amount of information that you can make use of?
MR. REYNOLDS: Well, for example, you have no warnings up there, if you had a drug/drug interaction warning and it came out and it was an 8-1/2 by 11 sheet it would take you a while.
DR. TEICH: Right. There’s a lot to say about design in all, actually interestingly enough and it’s interesting that you didn’t see this is that there actually are warnings posted on this particularly screen, actually it says there are none it lists them explicitly and if there were some presumably I’d see it. But the fact that on a quick glance you didn’t spot it possibly indicates that maybe this should be even bigger or something else. It turns out that I think in your hand this thing actually is more noticeable from what I’ve seen from actual use. But even on a gigantic screen there are lots of ways to make these very difficult to use.
MR. REYNOLDS: That is correct.
DR. TEICH: So I would say yes, certain formats, clearly you can’t display as much information on a PDA as you can on a larger screen, clearly you can’t get as much text rapidly into it but it has other compensating values. I think that the important thing is that there needs to be baseline usability and we have, the EHI report actually talked a lot about how usability needs to be measured and scored, we didn’t think about it in the scope of this discussion.
DR. COHN: Steve?
DR. STEINDEL: Jonathan, your last slide you talked about a task force to create structures and yet I got a sense there was at least one item, which was decision support information exchange standards, where standards needed to be created. You mentioned that we now have decision support databases, they all have different structures, different formats, etc., and what we should be looking for is while we might have multiple knowledge base suppliers that the formats of the knowledge bases should be standard so people could use them. When I heard that word I expected to hear later that some standards needed to be created because they don’t exist today.
DR. TEICH: That would be correct and if I didn’t say it more explicitly I apologize but absolutely, there does need to be a common format, I’m one that hopes for these standards to somehow be able to be developed and agreed upon in a fairly rapid fashion but it is important to get the necessary stakeholders together to make sure that what you establish is something that’s going to work for everybody. So I’ll come back to you for I’m sure the necessary response but I think that it’s important to get these things in place, I think it’s important that there at least be a reasonable framework for how these things get exchanged and if there is to be some sort of clearinghouse format that one is going to have to be established. Now some of the these standards go quite quickly, the continuity of care record was sort of went through things quickly, there’s various pluses and minuses to how that was done but it’s been able to sort of move forward in a certain quick fashion. What I’d rather do is have things done quickly and well altogether.
MR. ROTHERMICH: I’m Phil Rothermich from Express Scripts, I have a comment and a question, the comment being sort of narrow, the question being broader. I wasn’t sure how your testimony fit with the testimony from the summer but I just wanted to point out you talk about standardizing formularies and one of the things that we advocated for this summer is Rx-Hub has created a standard mechanism for communicating formulary to vendors so that it doesn’t require standardizing formularies and formulary in and of itself isn’t enough, you’ve got to get the benefit and you need to leave the market with the creativity to do things differently, to be able to manage benefits different ways and to hold down costs so I just wanted to point that out.
The question really has more to do with the market generally, the types of things that you talk about with respect to decision support I think most of us would agree are beneficial in the process of prescribing. The question I have is did the group that looked at this think about the commercial implications given that each of the vendors doing this are still finding their own business model and some want to do the low cost basic alternative and some want to be the Cadillac and what are the implications for requiring each of these different systems to contain this minimum set of functionalities which may result in many of them looking very much alike, may result in many of these companies having to pay license fees for different functionalities they don’t currently have which will impact the cost of the applications, and does that in the end hamper in any way the commercial viability of electronic prescribing because it’s forcing the vendors to do things a certain way that doesn’t allow the competitive forces to drive the industry?
DR. TEICH: Sure, good questions. With regard to your initial statement I think this is another example perhaps of where there’s a move towards creating perhaps a defacto standardization of certain things and that can be, as that’s been moving forward and people have sort of fallen in behind it it may turn out that that’s, that what Rx-Hub has done will turn out to be one of the standard ways of doing this, I’m not going to comment on which will actually take over but that’s certainly another example of where these things are moving sort of more by momentum then by sort of a wide race standards effort.
Regarding your question about whether this produces too much homogeneity or too much of a struggle, in the use of the term basic or B level we really wanted to emphasize that, that basic meant basic, that it would be difficult to justify a system in 2006 that didn’t support the basic functionalities of 2006 or in 2008 that didn’t support those. So the group really was instructed to and really commented passionately on which things really needed to be considered to be basic and there were members of several vendors of prescribing systems at that table. So I think that interest at least a good segment of that field were well represented. So it was important to say that if something was listed as basic it was because the panel felt that if something, if that function wasn’t there, if that knowledge wasn’t there, then that system was unnecessarily missing something that was of significant importance.
It also was noted that when we looked at those things there really weren’t a lot of things there that blocked the differentiation of one vendor for the other, vendors compete on lots of different things, on additional functionality, on superior communication, on integration with health records, simply on usability and nicer product grounds, so there was, so nothing in this, while certainly this produces a little bit of adherence to a common framework nothing in there was deemed to be so onerous that it would force each vendor to look alike.
As to vendors having to struggle to get to a certain place the items that were listed as basic, again because they are basic were things which people felt deserved that struggle. And the items which were advanced deserve at least some struggle and deserve probably additional reward for having them because drug allergy and drug interactions do not make the whole picture, research shows that you need more then that to really help the patients out the best.
DR. COHN: I just wanted to do a time check with everybody at this point, I don’t know, are you being taken away by Caroline and Brent or are they here for other reasons? I just wasn’t sure whether you were going to escort him out.
Well, I just want to tell everybody that we have time for one more question then we need to sort of think about how this impacts our December hearings or what we want to do and then we need to transform into the NHII Workgroup in a very short period of time, so just, Jeff, do you have a quick question or a comment?
MR. BLAIR: Jonathan, this is a mental groping question which I think leads right into, leads into where Simon wants us to go because I am struggling trying to put some pieces together. If we are to spend our next few months here, NCVHS, trying to focus in on areas where we could make useful meaningful recommendations on the next set of standards that could help that might be applied in the pilot tests, one of the areas that obviously we’re interested in is things that could support decision support. And so I’m going to grope around here that maybe you could add some clarity to this and say this is what you should focus on. When we talk about decision support HL7 has decision support construct, messages, the Arden syntax, however, the way they are constructed is almost to facilitate local use as opposed to national use where they’re recording who authored it and when it was authored and things like that, and maybe that gives us part of the answer. And then you have, and that would be maybe for certain alerts and warnings but when you have clinical protocols I’m not sure that that’s the best construct for decision support for protocols and guidelines and that may be a different set. And a third set might be formulary information, so I’m not sure we’re talking about the same type of standard for clinical decision support in each of those three areas, and then you get to something where you’re comparing drugs to allergens and I think Stan had mentioned that maybe there’s content issues there, do we need to wind up having standardized nomenclature for allergens that we maybe should be focusing on that we could do in a short period. I have you four kind of examples there as I grope around, can you give us any guidance or priorities or clarification that could help us?
DR. TEICH: Jeff Grubbs and the rest of us have three weeks of homework.
Yes, I’m a little bit taken because it’s a very deep question, you bring up a lot of points. The answer is yes, you’ve mentioned a couple of specific standards that are a little bit or a lot different from each other, all of which contribute to the mix. There are ways one can prioritize them, I would say that the establishment of a doctor level drug dictionary is of higher immediacy then the establishment of coded allergen reactions and probably of higher immediacy then formulary standard only because one needs A for B but obviously I’d like to see that one quickly as well. On the one hand if people are developing these things in parallel with some good talking to each other then there doesn’t need to be a priority. If NCVHS needs to establish an absolute prioritization then there are ways that we can comment on that, I’m hoping that there are at least several different parallel channels to do some of this.
But yeah, there’s several different, in terms of strict standard codes for things you’re right, that guidelines are going to look a little bit different from one shot alerts. One shot alerts can be, I think people have a reasonable enough grasp of and can generally do them, there have been a number of ways to do it. Guidelines obviously involve steps if one organizes it like a flow chart then one has to think about action steps and decision steps and stop steps and so on, so there’s that thing. And there has been HL7 work on doing longer phase guidelines as well. So HL7’s decision support systems workgroup has certainly addressed many of these things, of necessity they addressed it in a rather general context and their goal is to create sort of the standard message.
You were getting down perhaps another level in which we’re specifying codification, we’re specifying vocabulary, we’re specifying some very specific terms that HL7 often doesn’t deal with, but which are highly necessary for practical application. So I think if I was to conclude I’d say yes there’s a way to prioritize these things, yes I hope that that isn’t absolutely necessary, and that really we’re talking about a level of practicality that needs to be layered on to some of the work that’s being done in other bodies.
Does that deflect some of that?
DR. COHN: I think we’re going to go on for a while longer myself —
MR. BLAIR: I don’t even know if I framed the question in the right way, I mean is that, should we be pursuing our selection of what we focus on in the manner that I articulated or am I, do I not have the right construct?
DR. TEICH: I think you’ve listed a lot of things and I think that to the extent that one has to prioritize you prioritize in the order, in many ways you take this same table, these B-2006’s, A-2006’s and you prioritize in that order, you say well which of the seven or eight things that you mentioned are going to be necessary to support a B2006 level platform. Well, in order to do that I’m going to need to have a reasonable drug dictionary, in order to support A2006 which has formularies I’m going to have to have a standard formulary. In order to support, I think that multi-step protocols don’t come in until I think the 2008 recommendations so that’s when I need those. So I guess I would use the recommendation of desired functionality as a monitor for the timetable for your standards.
DR. COHN: Well, thank you Jonathan very much and I think you’ve given us a fair amount to think about. On the other hand you’ve also, it’s good to know that we’ve actually made progress on a lot of what you consider to be key items, so I think that’s something that as I’m listening to you makes me feel a little better. I’d rather actually be groping and then prioritizing among, we’re obviously not just talking about decision support, we’re also talking about e-signature and authentication issues, we’re also talking about all of the other stuff that we’ve been sort of watching and so it’s a question in all of that trying to prioritize and indeed I think your presentation I think gives us really a thought for what we need to do in next steps around decision support, and we also need to obviously talk about how we prioritize them.
Now I guess I’m going to move the conversation to the subcommittee know that we’ve sort of run out of time. We obviously had a conference call a couple of weeks ago and I think from that Jeff I think tried to put things into sort of containers. I mean one of them is I think e-signature and authentication, which I think we all thought was going to be an important issue for us to further investigate and I believe we have sessions in December that we’re talking about focusing in on that because I think, that’s a state board issue, it’s also an issue if you’re really doing e-prescribing.
Now obviously the other piece which I think everybody agreed to was that we needed to in December and in January somehow deal with sort of following along all of the issues that we sort of got going, the sig issue, which I think we all agree is a pretty important issue, the Rx-Hub/NCPDP work that’s going on, I would mention some more but mostly my computer keeps logging off on me so I have the list of Jeff that I was going to be referring as we talked about all of this. But sort of those things sort of going forward.
I guess the question I would have is we’re still sort of left with what exactly it is we want to do and in what timeframe on decision support and are there other things that we should be trying to investigate either in the December or January timeframe knowing that we have a total of about four and a half days, four days all together to do work on and I’ll ask you from what you’ve heard from this conversation and Jeff trying to grope, but doing some pretty good groping mind you —
MR. BLAIR: I said intellectual groping now.
DR. COHN: Intellectual groping, okay. I guess I would ask the subcommittee is do you think that we’re sort of, is the construct that I’m describing to you signatures and authentication following up and then figuring out what else we need to do with clinical decision support, is that the right way of describing all of this first of all? Are we, everybody in agreement about that about spending our time? I see Jeff’s nodding his head and everybody else looks like they’ve been through a long day. Harry, do you have a comment?
MR. REYNOLDS: So basically you’re saying we already had some things in process that people are going to respond to, or we’re going to get updates from them, right? Then the e-signature, and then clinical support.
DR. COHN: Well, I’m asking sort of what else —
MR. REYNOLDS: I’m agreeing, if that’s what you’re saying I’m agreeing to that, I just wanted to make sure I didn’t, you kind of fuzzed the center part about the follow-up, I just wanted to make sure those are the three distinct areas.
DR. COHN: The follow-up was the issues of NCPDP/HL7 harmonization, the Rx-Hub piece, the sig piece, the FDA SPL into the NLM Daily Med piece, it’s basically sort of seeing, sort of continuing to sort of push those all along.
Now Jeff what is your sense on, and Maria, what is your sense on decision support at this point? I mean do you want to think about, do we need to come back to the subcommittee or just, actually I would ask the rest of the subcommittee what is your preference, do you want us to just do, make some decisions and go forward with putting some things together, do you want to conference call? Jeff?
MR. BLAIR: I think we’re going to need a little more time, I need a little more time unless somebody else really has some more clarification here in their mind as to how do we define what we explore for decision support for standards to facilitate decision support. The question that I asked Jonathan reflects the fact that we know the areas but I don’t know how to define them clearly enough that Maria can wind up pulling together testifiers where the testifiers know what questions we’re really asking. So I think we need a conference call to get to that level. Maria, yes?
MS. FRIEDMAN: Yes.
DR. COHN: And we can obviously begin to work on the other pieces that we know about —
MR. BLAIR: And Stan, maybe you could help us with this because maybe you are more familiar with what HL7 has done with decision support because I haven’t been in those sessions for a couple years.
DR. HUFF: I’m not the expert, I’m pretty familiar, I know the experts, I mean I could give you a pretty good summary of where they’re at but Bob Greenas(?) and some other folks are the true experts that are leading the activity there and they’re the ones who would be the authoritative voice. But I could give you a good summary because I do follow that very closely, that’s part of the joint activity, part of the agenda for the joint.
DR. COHN: Exactly. Steve, did you have a final comment?
DR. STEINDEL: Yeah, Simon, I agree we need further discussion on what to do about decision support, now the question that I have is from what I heard the full white paper is going to be coming out sometime before our December meeting, it sounded like from the timeframe.
DR. TEICH: That’s correct.
DR. STEINDEL: And I think we need to take a look at the full white paper to decide what we actually need to look at and probably it would be in the January meeting. So we could either discuss it at the December meeting or have a conference call but I think we ought to defer any discussion until the white paper comes out.
DR. TEICH: And we will do everything to make the December time, I know one of the reasons that I’m here today is because none of us could be at the December meeting so depending on when you’re having later meetings that may be a possibility if you’re doing something in January.
Also we want to be a resource to you, if there’s additional questions, Jeff if I didn’t really answer your question in context or if there’s other things come up when you reflect on this we certainly would be happy to take some emails and see what we can do with them or participate on phone calls as needed. We’re trying to supply whatever we can in this.
MR. REYNOLDS: I agree with Steve, I think whatever debate we have or whatever starting point we have on decision support I think that white paper should be the basis, it sounded like a nice cross sectional group of people put it together and rather from starting from zero I think that would be a good basis.
DR. COHN: Well and of course that’s why we had Jonathan here briefing us. You want to see its entirety first —
MR. REYNOLDS: Well, I’d just like to reaffirm that that’s what we’d use as a basis.
DR. COHN: Well said. Okay, well we will schedule a conference call, obviously you know about meetings coming up in December on the 8th, 9th, and 10th, and then we obviously have January dates, second week of January. So with that we will adjourn this meeting, give everybody a five minute stretch and then we come back as the NHII Workgroup.
[Whereupon at 4:42 p.m. the meeting was adjourned.]