Tuesday, June 24, 2008

Electronic Health Records for Non-owned clinicians - Sharing data among providers

At times, the business case for interoperability is not entirely clear. If data sharing reduces the volume of redundant lab tests, then the healthcare system as a whole wins, but someone loses revenue.

Over the past year, I've seen a remarkable change in attitude among clinicians in Massachusetts communities. They are demanding data sharing. Here's the history, the specifics of the clinician requests, and the plan for making it happen.

When we first conceived our hosted software as a service model to provide electronic health records for non-owned clinicians, we designed one way interoperability. BIDMC has an ambulatory record called webOMR which contains the problem lists, medication lists, allergy lists, notes/reports, labs, and imaging studies for 3 million patients. We worked with our community EHR vendor, eClinicalWorks, to create a seamless web service that links eClinicalWorksto webOMR such that community physicians can securely view BIDMC data from inside eClinicalWorkswithout having to login again or use a separate application. However, we did not design a link between eClinicalWorksand webOMR to enable a BIDMC hospitalist or ED physician to view individual patient identified private practice data.

We did design aggregate data sharing such that the medical director of the Physician's organization could query private practices to retrieve performance, quality and outcomes data in support of pay for performance contracts.

As we began to communicate the vision of a community EHR, our private practice clinicians starting asking three questions:
1. How does a Primary Care Provider send a clinical summary to a Specialist?
2. How does the Specialist close the loop with the Primary Care Provider by sending an electronic consult note?
3. How does a hospital-based physician such as an Emergency Department clinician, hospitalist, or anesthesiologist retrieve patient summary records from private practices?

My initial response was that private practice data sharing is such a novel idea, that it would have to wait until after our EHR rollout was complete to formulate a strategy.

Clinicians were not satisfied with that approach. Thus, we've decided to accelerate our work on private practice data sharing sharing by creating a clinical summary repository for all our eClinicalWorksusers using the eClinicalWorks EHX product.

Here's how it will work.
1. Whenever a patient visits one of our BIDPO community clinicians, the documentation of their visit will be done in our hosted software as a service eClinicalWorksapplication.
2. Patients will be consented by the clinician for community data sharing via opt in consent at the practice level. Consenting at one practice implies that data from that practice can be shared with other practices, but not visa versa.
3. When the encounter is complete, a summary record including problems, medications, allergies, notes, and labs will be forward to the eClinicalWorks EHX repository using the Continuity of Care Document format.
4. Other clinicians, who are credentialed members of BIDPO will be able to view summary records from this repository, assuming the patient has consented to sharing that data.
5. An audit trail of all such lookups will be available to enforce security

Such an approach solves the PCP to specialist clinical summary issue, the specialist to PCP communication issue, and the hospital-based viewing of private practice records issue. From a technology perspective, it's an elegant solution that reduces the number of interfaces. All practices send their summaries to a repository in a standard format, then all exchange is done from that repository.

A similar approach has been used in the Massachusetts eHealth Collaborative pilots in North Adams, Newburyport, and Brockton to enable secure, patient consented data sharing in those communities.

This approach needs one additional architectural element - how do you share data among EHX repositories, with non-eClinicalWorks EHRs or hospital information systems like Meditech.

MA-Share provides the grid infrastructure in Massachusetts to enable community to community data sharing. Today, MA-Share's Gateway can push Continuity of Care Document Summaries from one organization to another. Over the next two years, we'll work with eClinicalWorks to expand this capability to push clinical document summaries between instances of EHX. This means that BIDMC will be able to push a discharge summary or other clinically important information to a community repository, where with patient consent, the clinicians of a community caring for the patient will be able to view the data, ensuring continuity of care.

I expect all of this bidirectional data sharing to be a journey. We're purchasing the EHX product as part of our licensing of eClinicalWorks software and will use it initially for performance reporting. But we'll configure it so that sharing of data between clinicians and among communities will be possible. I expect all these features to be implemented by 2011.

I'm hopeful that our BIDPO clinicians will be satisfied by our strategy to embrace bidirectional data sharing in this incremental way - sharing data from BIDMC, sharing aggregate private practice data, sharing data among private practices using eClinicalWorks, then sharing data among communities and hospitals.

8 comments:

StephSup said...

I've been following this blog for a while now and i really like it alot!! As a college student interested in Health IT this really gives me a real world view of what's going on.. EHR talk is one of my favorites, im still not used to all the terms that professionals like you use on a daily basis, but i figure if i keep reading it'll all make sense.. thank you for starting this blog in the first place.

Unknown said...

Hi Dr H,

How did you decide which data elements would be included in the shared clinical summary?

citizencontact said...

(this is a comment I made to http://news.cnet.com/8301-13505_3-9976958-16.html )

This is issue has personal relevance since I have to apportion over 20-40 pills daily for a sick relative. There should have been a simple online system for helping manage this. The Google Health system API makes this more likely (there is already one application that does this, though seems very buggy. Open standards and open source would make this effort much better.

However, the United States makes this effort problematic by not having a comprehensive and universal health care system. Since health information is used to deny health care and employment (whether legal or not, since there is no guarantee which developed societies have as a given), the incentives run counter to normal networked systems and applications.

This is hard to prove in the abstract, but should become apparent if you understand the nature of networks that grow in power as they reach 100% of the potential audience and the analogous situation that a society is healthier person by person as closer to 100% is given the best care (e.g. vaccines, etc).

Nonetheless, we should support the effort of Dr. John Halamka, because by moving ahead on this open standards/source effort that may help explain the 100% is best rule for networks in the context of health care.

Graham Barden said...

I am the managing partner for a 10 doc, three-site pediatric practice in NC. I just found your blog after it was recommended by a doc in Australia...
I have been reluctant to consider EMR since our paper seems to work well for us. (Preprinted forms, state electronic vaccine registry, good PM system) We don't have patients on multiple medications since most kids are quite healthy. And in pediatrics, infant/child/adolescent problems can be very different for the different life stages so I am not convinced that I could practice better with an EMR. We did try electronic prescribing, but found it difficult to tailor the scripts for the varying dosages of drugs found in pediatrics.
(if the amount needed was 110 cc's, and I know the product came in only 100 cc bottles which would have been adequate, the program would "force" the pharmacist to fill 110 cc's - ie, charge twice as much and throw away 90 cc's...)

But my biggest reluctance stems from the lack of portability of the data once we spend the considerable effort to create it. If the vendor is acquired by another or decides to focus exclusively on another specialty, or dies, we would be stuck! I am very worried about entering into a business agreement without an exit. I feel that as per Business 101, if you cannot easily extract yourself from an unsatisfactory business relationship, you are essentially "owned" by the other entity. And there is no real incentive for an EMR vendor to design their product so you can leave easily.
I am not as worried about real-time exchange of data as I am with long term graceful extraction from a failed business decision.
Is headway being made on the problems associated with migration to a different EMR?
-Graham Barden

John Halamka said...

Thanks for all these great comments. For clinical summary exchange we elected to use the Continuity of Care Document which includes problems, medications, allergies, labs,results, and also next steps for followup. Here's an example of one of our production documents.

These summaries are exchanged from provider to provider via the MA-Share gateway and the eClinicalWorks EHX respository for eClinicalWorks users.

This standard format makes the EHR data completely transportable.

Anonymous said...

Hello, Doctor.

The doctor in NC seemed to be asking without knowing it for an online EMR system designed as software as a service. I am wondering if this where eClinicalWorks is heading. And do you think it would be the answer to getting the primary care doctors to buy into EMR so that the Personal Health Record could become more meaningful?

Danlj said...

As a primary care internist who has acquired electronic access to 2 sites to which I refer patients, I can say the following:
- Access to the most recent clinical note regarding any visit at which a diagnostic or therapeutic decision is made, is extremely valuable.
- Access to phone call notes, and other administrivia is chaff that delays care and decisions
- The medication record is potentially of value, but is NEVER CORRECT - must always be verified point by point (med name, dose strength and form, administration times, reliability) or sad errors will occur.
- Access to actual pathology reports is invaluable
- Access to FLOW-CHARTED lab results is extremely helpful, especially if subsets such as lipid profiles or hemograms or lytes can be viewed.
- Non-flow-charted lab results can be a nightmare to parse intellectually
- Access to imaging reports is extremely valuable; access to images is seldom crucial except by surgeons or radiation oncologists or neurologists
- Problem lists are seldom maintained carefully; to verify their accuracy, it's useful to have access to H & Ps:
- Prior H & Ps are valuable for their lists of chronic ongoing issues and surgical events. But past history cannot be assumed to be reliable unless specific dates (month/year, at least) and specific jargon is used ("wide excision of Clarks Level 2 melanoma of the left chest wall Feb 1998" not "melanoma a few years ago")
- Summaries of major diagnostic events such as sleep studies, pulmonary function testing, echocardiograms and graded exercise tests are seldom easily found remotely but are very helpful during episodes of acute illness.
- "allergies" need to be subdivided into "medication allergies", "medication non-allergic intolerances" and "environmental and food allergies"

For 10 years I have used a word processor for medication management, with success. Three-column format
1: med, dose, form
2: instructions ('sig')
3: purpose (symptom or diagnosis)
Columns 2 and 3 are typed in patient-readable form.
The sheet is faxed to the patient's pharmacy or handed to the patient; the Rx symbol and administrative details are easily stripped from the document and handed to the patient as an instruction. Changes are bold-faced.
Caveat: I type 50+ wpm, so this is faster than hand-written prescriptions; this won't work for the ham-handed provider.
It is more customizable than EHR boilerplate, as I can add narrative text as needed. (I store them on disk for future access. The stored medlist is automatically saved to a drive in the transcription area for inclusion - cut and paste - to the dictated note.

Tony said...

citizencontact, you say that "However, the United States makes this effort problematic by not having a comprehensive and universal health care system."
Please note that France, among other countries with universal health care, has been struggling for many years to develop shared health records - with no substantial success so far.