Quality is not only in the domain of the EHR

One of the challenges with quality measures that I've mentioned previously on this blog is fitting it into the provider workflow.  The example used in the illustration previously has to do with aspirin and heart attack.  Other challenges have to do with quality measures around Venous Thromboembolism (VTE) and Stroke.

The problem that EHRs run into is that in order to measure quality, you need to know about certain condition specific data. Yet this data is often very hard to codify in an EHR.  I delve into some discussion on contraindications and exclusions in Quality Improvement Never Ends that I wrote last week.  And I also tweeted a statement someone made at the CDC Public Health conference.  Farzad Mostashari asked for some details in a response to that tweet, so here they are.

I'll first go back to some of the work that HITSP did on VTE and Stroke.  There are some key data elements used in exclusions for these measures:

• Comfort Measures Only
• Elective Carotid Intervention
• VTE Prophylaxis
• Reasons for No VTE Prophylaxis

Where would you find these data elements in an EHR?  We were challenged to find a code for comfort measures in SNOMED CT.  You might find a procedure code for Carotid Intervention, but would you find whether that was elective or medically required?  Whose definition of elective would you be using?  An insurers?  Someone's guideline?  If so, from where? Next up: VTE Prophylaxis:  There are three types, medication based, device based, or compression stockings (arguably a device as well).  OK, how is this routinely coded in the EHR?  Medications certainly are coded using RxNORM, but looking through ICD-9-CM (Procedure codes used in a hospital setting), I have trouble identifying codes for application/use of VTE prophylactic devices.

On reasons for no VTE prophylaxis, think about how the data would get there.  A provider very well versed in treatment of patients who are at risk for VTE but aware that prophylactic measures are contraindicated would have to take an extra step to indicate why the weren't indicated for a patient.  I'm not saying it shouldn't be done, but often workflow questions like these about "what didn't you do and why" don't get raised during the implementation of an EHR, and so the data doesn't get captured.  And it wouldn't necessarily be built in because in some settings, it might not even be needed.

Looking at the specific question of HAI, one of the components in the HAI report is the location of care.  This is documented as using the NHSNHealthcareServiceLocationCode code system.  This is a very detailed code system.  More detailed in fact, than is often used in patient registration systems.  Patient Registration systems, I also note, are not necessarily the same as the EHR, and while they may be connected, might contain information such as patient location type that isn't accessible through EHR capabilities.

Another example is in the "Infection Condition Observation".  In this case, the type of infection being reported also has detailed coding.  Some of the condition codes include superficial/deep primary/secondary surgical site infections.  Note that there are not even SNOMED codes at this level of detail, and it often wouldn't be captured in the EHR.  Instead, this information would be captured by the infection control officer in a separate Health IT system devoted to the capture of detailed information for HAI reporting.

So, there may be a system that is able to capture data at this level, but it is a system that is used by one or two specialists within the entire facility, and I would be hard pressed to call this an EHR system.  Thus, I quoted the speakers statement:  Data needed for HAI reporting are not typically captured in an EHR.

Personally, I might amend that statement if I weren't quoting someone else to clarify that "Some data" needed for HAI reporting ..., but for the most part, it is a true statement.  The EHR doesn't solve every healthcare problem.  The EHR is but one component of an ultra-large-scale system.  We need to be aware that there are other components in this system beyond the EHR, and that the EHR will not solve every single Healthcare IT problem by itself.  In fact, the HAI specifications produced by HL7 were specifically created for systems used by infection control officers.

Google-

I'm still not a big fan of Google+.  When I need to engage, I need to engage in as many places as possible.  G+ makes that extremely difficult for me.  I have to take extra steps to engage there.  Sure, I could hook up a set of tools and write all of my posts in G+, but then I'd need to switch away to access the other social networks where I have existing engagements to keep up with them.  I can't access them through G+.

Instead of making it easy for me to remain engaged with my already established social communities, they've decided that they want to own them.  But they, like me, aren't leaving the services that their own constituencies have already adopted.  Basically, it's like changing your phone number because you just changed phone companies.  Google Dudes/Dudettes ... how 20th century of you.

The other challenge I have with G+ is that because each plusser gets one and only one channel that I can see, I don't get to pick and choose which channel I want.  It's like being forced to watch Fox news because I also like to watch Bones.  That's a pretty steep price to pay.

On Twitter I have two channels.  One is very personal.  About 10 people see it.  The other is where I engage in my own advocacy for Health IT and change in the Healthcare system.  Some of my friends subscribe to both channels, and have chosen to.  But with Google Plus, I don't have the option of letting my readers pick which "Keith" channels they want to listen to.

I made a similar switch on Facebook not long ago.  Originally, I was on Facebook to connect with online friends and family from time to time.  Soon, my standards colleagues from around the world found me there.  I let one in, then another, then another ... and so on until I couldn't find the content that was the reason I was there in the first place.  So I "unfollowed" anyone not in my circle of friends, and created a new Facebook account for that other persona.  I then requested friendship with all the people I just "defriended" from the new account.  Having made that separation, I expanded my twitter tweets into that other Facebook space.

When I want to share something on G+, I feel like I'm back in high school.  It's not circles, but cliques.  I have reading cliques, and writing cliques.  What I read comes from one, what I write goes to another.  Something that is posted on G+ is stuck there.  Like Wes, I know the feedback I make is limited to the group that it is shared with, and sometimes I want to broaden the discussion.  I can reshare, but that's another click, and doesn't let those not on G+ get access to the discussion.

I'm glad that G+ released their API, and disappointed that it's still read-only.  I can only hope that Tweetdeck won't be too far behind with an update to support it even if it is just read-only right now.  At least then I can get back to one tool for monitoring all my social networks, instead of simply ignoring G+ most of the time.

My last complaint about G+ is being able to find good content.  It seems that people haven't yet figured out cliques well enough to enable meaningful sharing, and for some reason, G+ is unable to let me set good search Filters.  That just doesn't make any sense at all.  I like Twitter, I can see what I'm subscribing to before I subscribe.  For some folks, I'm willing to deal with a noisy channel because the occasional gold-nuggets are truly valuable.  For others, I'd rather not listen.  I want the ability to share widely and listen cautiously.  G+ just doesn't give me what I want.

Making Connections

My brother died because of a medical error: he was effectively anesthesized to death.  There error was in the construction of the new ED that he was admitted to, and the oxygen and anesthesia lines became mixed.  I like to believe that this sort of error could not possibly happen today.  I haven't thought of him in more than a decade.  He died more than three decades ago.  I thought about him today because I was on a round table discussion for the Kaiser Permanente Total Health Blog this morning.  When asked what topics the blog should speak to, one commenter talked about gag orders in settlements for medical errors, and a petition to stop the practice.

There was no gag order in my brother's case; As the news already had it, no gag order could have been imposed.  .  My parents found out because of a front page news article published in the local paper.  I found out because a friend told me about the story (my family had tried to hide it from us, I was 12 and my younger brother 9 when it happened). My brother probably would have died anyway without the mixup.  My parent's anger at the situation was more about how they found out, and the new grief imposed upon their children because of it.  If the hospital had contacted them, and told them about the mistake, they might very well have been forgiving.  My father would have no part of the money.  It was, as he called it "blood money". That settlement paid for my brother's education, and mine.  I like to think that the real justice in that was that my education  (such as it was) led me to the place that I am today, and that what I do in Health IT standards to prevent similar mix-ups in communication between providers that could equally as harmful.

One of the discussion topics that went on during the round table was the idea of connecting diverse communities into and around total health.  As @ReginaHolliday put it, you really cannot tell if a person driving a trash truck wouldn't have an interest in #HealthIT.  From my perspective, I never would have expected an artist to have an interest in the topic either.  But so she does, as I well know.

What I do know is the importance of these social connections that I make across diverse backgrounds.  I'm not a doctor, but every now and then I jump onto #MDchat just to see what Docs are saying.  I've become myself a member of the Society for Participatory Medicine because of my connections with other engaged patients.  Listening to what they are saying is vitally important to what I do.

The value of creating a connected, diverse community around the topic of total health is a terrific one.  After all, it is through community that we can make a better world.  The connection I made this morning is one that might not have happened had it not been for the round table discussion this morning.  It renews my spirit to rediscover part of my past that I had forgotten, and to see after thirty years a resolution that my father, also departed, might be proud of.  And damn it, yes, I am crying.

Rambling from the HL7WGM

Lots of new irons in the fire this week.  Like Grahame's Resources for Healthcare proposal, the HTML5 + Microdata proposal is gaining traction.  Not overwhelming support like Grahame's work, but significant forward momentum.  I'm working on updating a Project Scope statement for ITS to review in a subsequent meeting.  I now need to reach out to my former standards community (HTML and XML geeks), and start really coming up to speed on what is going on in that space.  I just finished writing a position paper I hope to present next month at a W3C Workshop in the Boston area comparing HTML5 + Microdata with CDA Release 2.0, noting the gaps and opportunities.

I will also be presenting an EHR Functional Profile proposal to the EHR TC next week to rapidly develop a functional profile supporting metadata expression requirements for exchange with HIEs.  This is an action item that the HL7 Policy Advisory Committee (PAC) will put into their response to the ONC Metadata ANPRM that I've written about several times over the last few weeks.  Many committees have provided input to the PAC's proposed response to that ANPRM, some were working on it even before the committee requested feedback.

One very quick and easy win this week from the PAC perspective was the recent HL7 announcement that it has taken ONC's Pledge for Non-Data Holders, and its encouragement for members to do the same. There will be an HL7 newsletter article going out on that soon as well.  This happened relatively quickly, the time between ONC announcement of the pledge, my presentation of the proposal to the board, their response and announcement through a press release was under 48 hours.

On other notes, I will never live down the time we spent talking about a real problem implementers have raised, and which I commented on in the CDA Consolidation ballot, about how to deal with people with only one name (now dubbed the "Rock Star" problem).  I don't mind, we now have sufficient guidance I can point to.  One time might be an oddity, but this was a question I have seen from three different sources in as many weeks.  The answer is pretty straightforward.  Put the single name where your organization requires it to go, and use a nullFlavor for the other components.  Doug Fridsma even had HL7 staff print up a badge with name and organizational details all saying "Not Applicable".  I can deal with it.  In fact, I have a "Rock Star" ribbon for him to place under that badge.  I don't need them any more as I will probably never wear one on my own badge again (at least at an HL7 meeting) ;-)

I taught the CCD class again this meeting.  I've used pretty much the same material for the last 4 years.  I'll need to update it for January, and again for May to address changes in CCD 1.1.  Yes, CCD 1.1 is part of the CDA Consolidation Guide activity.

There are yet more irons in the fire.  I have a mobile-health profile proposal to present in a couple weeks to a very hard audience, the IHE IT Infrastructure workgroup.  It addresses a use case for software stack constrained platforms often used for mHealth applications that want to access information from a Health Information exchange.  I expect lively discussion.

Having just finished teaching CCD to a group of students, I have to prepare for my next teaching event.  IHE will be offering a day of CDA training (limited seating) to members after the November Face-to-Face.  This will be a new class because my goal will be to teach members how to create CDA-based IHE profiles.  That should prove to be an interesting combination.  By offering members this opportunity to enhance their skills, I'm hoping to scale up.

BIN Counts in X12 and Canonicalization

Someone complained to me the other day about a problem transmitting claims attachments between systems.  While I don't usually get into X12 transactions, this is one that I'm pretty familiar with because it was a topic of great concern to the Claims Attachments SIG in 2007.

You see, attachments use EDI formats to exchange information.  Many of these systems still use "big iron", and translation between ASCII on one end and EBCDIC on the other, and visa versa are common.  This was a challenge because carraige return/newline pairs in one environment get translated to single characters in another, and so forth.

The X12 BIN Segment has two parts, the binary data, and a count of the length of that data.  The challenge is that the whitespace is being changed from 2 characters to 1 or visa versa.  And then when the system outputs the record in the new encoding, the length is off.  This problem is clearly stated in the 275 transaction that contains a BIN segment:

It has been noted that line constraints, transfer protocols ... may insert additional control characters ... If this occurs in BIN02, the senders stated count in BIN01 may no longer be equal to the received content of the data in BIN02.

This is a truly big challenge, because if BIN02 count doesn't match BIN01, business logic set up to detect errors could wind up rejecting the response to the request for an attachment.

There is an even more sinister problem in XML.  There are three different valid ways to represent the character A, each with a different length.  Once as the letter A, another using a decimal character entity, and the final as a hexadecimal character entity.  As far as an XML processor is concerned, all three are the same.  And then there are different character sets that an XML processor must support.

This would seem to provide a challenge for creating digital signatures, but XMLDSIG addresses that with the Canonicalization Algorithm.  The algorithm ensures that the digital signature is computed over the same content by ensuring a consistent octet stream is used for the computation.

In X12, the BIN01 serves a similar purpose as a digital signature.  It is a "check" that the data has not been modified (but it doesn't support non-repudiation or any other cool features of XMLDSIG.  Canonicalizing Base-64 is simple.  Ignore any whitespace.

Given that the specification notes the disparity, I would presume that receivers are expected to deal with it.  So, give an innaccurate count to start with.  Don't count the whitespace because it doesn't matter.  Better yet, use the octet length of the original CDA document.  The challenge here is that senders and receivers would have to change the counting algorithm for the BIN segment to make it ignore whitespace.  That might be hard, but the only other solution I can think of is to upgrade the operating systems and hardware.  I haven't run into an EBCDIC to ASCII issue in years.

The next Ad Hoc Award will never be repeated ...

Long time readers of this blog probably understand the rules.  For new readers, it's very simple.  I am the sole arbiter and judge.  There is no nominating committee, but nominations are certainly welcome.  I won't award more than one a year for the same kind of service (and since this nominee is unique, I'm certain that won't be an issue).  I try to keep the number of awardees to five per year (no reason, as I said, the rules are arbitrary).  In general I try to give awards to folks who would not otherwise be recognized for their contributions.  One tradition is that I have given out an award on the same day that the HL7 Ed Hammond awards are announced.  Not so much to compete, but more to build on a trend.

I don't see how the current award possibly could be duplicated; ever.  As I mentioned previously, I just finished reading Ten Faces of Innovation.  This next recipient is a collaborator extra-ordinaire, but even more importantly a story teller of unsurpassed skill in both word and deed.  I first learned about this individual when I watched the HHS announcement of the Meaningful Use regulation.  This awardee has done a great deal to raise the awareness of Health IT to patients across this country, and even Internationally.  Their work has seen HL7 appear in circles that nobody would ever expect.

Without further ado, I proudly welcome the next recipient:

This certifies that 

Regina Holliday, Patient Advocate and Artist

Has hereby been recognized for outstanding contributions to the forwarding of Healthcare Standardization through Art.

Regina, welcome to "The Riding Gallery".

For those of you who don't know Regina, see her blog for some truly amazing story telling in both art and print.  I'm a very proud member of The Walking Gallery which she created to raise awareness of patient needs.  Her work was recently featured on the cover of The Illustrator's Journal.  If you look closely at the upper right hand corner, you can see a reference to HL7.  If you happen to be at HL7 this week, and want to see an example of her art, I'll be wearing it.

Quality Improvement never Ends

I had an interesting discussion over dinner with an expert on quality measures earlier this week.  They spoke about how in quality measures, there appears to be a tendency to complicate the measure with exclusions.  There are many reasons to add exclusions to a measure.  I'll take an easy to understand example:

In treating patients with heart attacks, research has shown that amount of time between the event and being given aspirin is related to survival and future re-occurrences.  So, a quality measure could be developed to ensure that patients are given aspirin as early as possible in these cases when it is appropriate.

Now, there are a number of reasons why getting aspirin might not be a good idea for all patients in this case.  For some, they may have an adverse reaction;  for others, there may be other contraindications, such as concurrent use of other medications.  One could ask for exclusions for this reason or that reason because giving the patient aspirin might not be possible, or aspirin may not be in supply, et cetera.  So the measure can continue to be refined with more and more exclusions.

But what is the point?  The point of the measure is to improve the quality of care, not to enable organizations to get a perfect score.  Why do I say that?  Where do you go after you've reached 100%?  Surely those patients who are excluded from the measure are impacted by the issue.  Is it OK that we are able to provide quality care only to those patients for whom we have treatment for?  Surely not.  Quality improvement never ends, but the 100% score seems to indicate that there is the notion of "perfect quality".

As we get into the details, and start drilling into the last percentage points, do these additional exclusions really improve the measure?  I'm certain that there are other details about how care is provided in the institution that have a greater influence on the care that a patient receives than these small adjustments.  As you look at these measures and begin to compare them against those of other institutions, does a half a percentage point really matter, or are we into the statistical noise?

What about the complexity and cost?  Adding refinements to the measure requires capturing additional data that could make it even more costly to implement.  Which measure has more bang for the buck?  The one you can implement today, or the one you can implement next year?  That is another important consideration.

After all, perhaps the most important thing to think about in measuring quality is whether your quality is headed in the right direction.