The Health Information Technology Platform Meeting

Today at Harvard Medical School, Zak Kohane and Ken Mandl assembled 100 folks to discuss "Substitutability" - the notion that EHRs of the future will be more modular and more iPhone app-like, possibly with common Application Programming Interfaces.

Some of the speakers at the conference were Mitch Kapor (Lotus founder), Sean Nolan (Microsoft), David McCallie (Cerner), Alfred Spector (Google), Rob Kolodner (formerly of ONC), David Liss (New York Presbyterian), Charles Friedman (currently at ONC), Mark Frisse (Vanderbilt), Clay Christensen (HBS), David Kibbe (Senior Advisor to AAFP), Aneesh Chopra (US CTO), Todd Park (HHS CTO), and Regina Herzlinger (HBS)

I served on the opening panel called Open or Closed Platforms? One or More Platforms.

Here are a few of the key points.

I suggested that standards can be required at different levels of specificity depending on the transaction. e-Prescribing transactions should be highly specified to ensure the right drug, right dose, right form, right patient are transmitted securely with data integrity from provider to pharmacy. Sharing an office note might be done with less specificity - a secure transport layer plus a human readable payload that might be XML such as a CCD document, PDF, or a scanned TIF image. Quality measures are very precise and require a high degree of specificity to ensure comparability among clinicians and institutions. Even though these examples have different levels of specificity required for interoperability, the architecture is left open to innovation - these transactions could be done in comprehensive EHRs, iPhone apps, or PHRs with point to point data exchange or via a healthcare information exchange.

Several of the panelists thought a common API in front of various EHR vendor products was unlikely. Exchange at the data layer was viewed by all as very possible and most thought the meaningful use plan for 2011,2013, and 2015 data exchanges made sense.

All agreed that common data transport with appropriate security is important.

The general theme of the panel was openness. Microsoft offers a standards-based way to enter and retrieve data from healthvault. Cerner is supporting XDR, XDM and XDS as well as Cerner proprietary APIs which enable custom plugin applications to work with their products. OpenMRS is an open source platform for electronic health record development, embraced widely in developing countries.

The move toward more openness - regardless of architecture, platform, or open source/proprietary product type - was refreshing. To me, being able to extract data out of an EHR (even via a proprietary API) combined with common data transport standards and translation into structured vocabularies gives us a path forward for health information exchange in the near term.

The industry is changing, motivated by meaningful use, a business case for data sharing, and patients expecting coordination of care. Add to that healthcare reform which requires quality measures and H1N1 surveillance which requires public health data exchange, and the pace will accelerate.

It's clear the we'll have many product choices in the future - some integrated, some interfaced, some modular, and some platform-based. My take home message from the conference is that innovation and data standardization can co-exist.

Preparing for ICD-10

In 2013, CMS will require the use of ICD-10 for coding of diagnoses in billing/administrative transactions. The code set allows more than 155,000 different codes and permits tracking of many new diagnoses and procedures, a significant expansion of the 17,000 codes available in ICD-9. How can we best prepare for this transition?

1. Improve electronic clinical documentation

The granularity of ICD-10 requires precise clinical documentation. Unstructured paper-based notes are unlikely to enable coding beyond the most general code for each diagnosis. ARRA incentives require ambulatory EHR implementation with structured problem lists, medication management, and clinical documentation as well as hospital CPOE use. These electronic systems will provide the foundation for the detail needed by billers/coders to accurately select the most specific diagnoses.

Here's an example - in ICD-10-CM, the code L89.133 is for a pressure ulcer of right lower back, stage 111. This single code in ICD 10 is specific to the right lower back and stage. Detailed electronic clinical documentation is needed to select the right code.

2. Train HIM professionals

The American Health Information Management Association (AHIMA) is the leading advocate for ICD-10 and training HIM professionals.

You'll find their training materials and courses online

Here's an example illustrating the training needed:

The ICD 10 code for Pathological fracture, right radius, initial encounter is M84.433A. This code is specific as to the location of the fracture, including laterality. The seventh character extension identifies the episode of care. Separate code categories are available for pathologic fractures specified due to a neoplasm or osteoporosis.

3. Embrace SNOMED-CT

Meaningful Use specifies ICD9 or SNOMED-CT for problem list management in 2011, ICD10 or SNOMED-CT for problem list management in 2013, and SNOMED-CT for problem list management in 2015. SNOMED-CT enables a clinician to document signs/symptoms, rule-out diagnoses, and problems, not just diagnoses. The structured clinical observations embodied in SNOMED-CT encoded clinical document will facilitate the accurate coding of ICD-10 diagnoses. The NLM's SNOMED-CT Core Set will include an ICD-10 crosswalk in the future.

4. Ensure we have the right vocabulary tools

Just as with the NLM SNOMED-CT Core set, it's important that the country has easy access to ICD9-ICD10-SNOMED-LOINC mappings that enable fluid translation of one codeset to another for various purposes. Many companies are working on such resources such as

Intelligent Medical Objects, which provides such services inside the EPIC EHR

AnvitaHealth (disclosure: I serve on the Board) which provides such services to Google Health

Apelon which has provided terminology services to Intersystems, Intel and the New Zealand Ministry of Health.

SNOMED Terminology Solutions not only maintains SNOMED CT on behalf of the IHTSDO but also provides mapping and consulting services to help provider organizations, vendors, various agencies of HHS, and other HIT-related companies with the successful implementation of terminology standards.

Over the next year, the HIT Standards Committee Clinical Operations Workgroup will discuss the vocabulary tools needed to support meaningful use.

5. Work with vendors to ensure EHRs and Hospital Information Systems are capable of supporting ICD-10 and X12 5010.

Existing EHRs and HISs will need to be upgraded to support ICD-10 coding and the transmission of transactions to payers via X12 5010 (replacing 4010). Vendors will provide one piece of the puzzle - software that is capable of supporting the new standards. Vendor efforts need to be supplemented with all the other strategies mentioned above to ensure successful ICD-10 implementation/meaningful use.

I realize that the dual transition of ICD-10 and SNOMED-CT over the next 5 years seems daunting. In my view, embracing SNOMED-CT for clinical observation encoding as part of electronic documentation provides the foundation for ICD-10 implementation by providing the clinical detail needed by billers/coders to accurately select the proper ICD-10 code. If we think of SNOMED-CT as the clinician facing vocabulary and ICD-10 as the administrative billing vocabulary for HIM professionals, adopting both codes is part of a single project plan to enhance the quality of healthcare data for all stakeholders.

Building the Nationwide Healthcare Information Network

I describe interoperability as a set of business partners with aligned incentives who exchange data to enhance efficiency, reduce costs, and improve coordination of care. Generally healthcare information exchange is local - hospitals, labs, pharmacies, clinician offices, and public health in a region exchange data for a specific purpose. Privacy and data use concerns are resolved locally. I do not believe that an architecture that requires a monolithic central database in the basement of the Whitehouse is going to be acceptable to stakeholders.

So what is the Nationwide Healthcare Information Network (NHIN) likely to be?

It will be a federated network of networks based on a common set of policies and data standards, enabling local, regional and domain specific (VA, DOD, Children's Hospitals) networks to connect with each other. Think of HIE's as similar to local phone exchanges and the NHIN as long distance service. What is required for a successful implementation of a "long distance carrier" for healthcare data?

1. Governance - A national framework for setting policy and technology for the NHIN. The HIT Policy and HIT Standards Committee could serve this purpose.

2. Education/Promotion - We need to ensure all state HIEs think of the NHIN as a connector between regional activities and understand how to use it. ONC could do this or partner with an organization such as the National eHealth Collaborative (NaeHC) or the e-Health Initiative (eHI).

3. Incentives - Meaningful use provides a powerful set of incentives to foster healthcare information exchange. Ideally, communications with Federal stakeholders such as CDC, SSA, FDA, and CMS would be done via the NHIN. This will incentivize all stakeholders to purchase EHRs and build HIEs which are compliant with NHIN policies and data standards.

4. Common transport, content and vocabulary standards - When EHR and HIE data exchanges are built, implementers have a choice of architectures and standards to implement. The work of HITSP and the HIT Standards Committee is architecture neutral, but provides enough constraints in the standards to reduce the number of choices, enhancing interoperability. Ideally, EHRs, HIEs, and the NHIN should should the same data transport (SOAP or REST over TLS), the same content (HL7 2.51, CCD, NCPDP Script 10.x, X12 4010 or 5010), and the same vocabularies (LOINC, SNOMED-CT, RxNorm, UNII) ensuring easy integration of regional and national efforts.

5. An agreed-upon set of security and privacy rules, including data use and reciprocal support agreements to which everyone who links to the NHIN must conform. Entities that link into the NHIN, and consumers who allow their information to be sent over the network, should be able to safely assume that some well defined, basic protection rules are enforced throughout, and that some well defined rules for representing, exchanging, and enforcing authorizations and consents are in place throughout the network.

Over the next year, the HIT Policy and Standards Committees are likely to work on NHIN related issues. I look forward to a secure nationwide network of networks with common policies and data standards that supports healthcare reform, public health, and the needs of patients, providers and payers. This is something we will create - we do not need to wait for our children to build it!

Cool Technology of the Week

As clinicians implement electronic tools to achieve meaningful use, it's likely that a diversity of approaches will be used in 2011 - some comprehensive EHRs, some hosted Software As a Service applications, and some modular applications. What are modular applications? Imagine that a clinician assembles a collection of iPhone apps and hosted interoperability services (Surescripts, Quest, Emdeon) to achieve e-prescribing, lab viewing, quality reporting, and administrative data exchange with payers. Such an approach would fall under CCHIT's notion of modular certification. Think of it as a "Project" rather than "Product" certification, ensuring that the collection of applications has the capabilities needed to achieve meaningful use.

This week, Quest introduced a six month trial of its web-based Care360 e-prescribing application, a Surescripts-certified solution which enables clinicians to access formulary information, route prescriptions, process refills and act upon FDA alerts.

Quest also announced that clinicians can now access these services from an Apple iPhone or iPod touch using Care360 Mobile. With Care360 Mobile, clinicians can create and send a new prescription from an iPhone in three simple steps and can also renew existing prescriptions. The application is available in the Apple App Store under Medical Applications and can be downloaded at no charge.

An e-prescribing application with formulary, routing, and refills that is part of a suite of web-based and iPhone products which assist with meaningful use. That's cool.

How I Eat

It's time for a Thursday blog where I turn introspective an examine my life experience.

I've written about what I eat and where it's from but not about how I eat.

What do I mean?

In the US, meals are often considered a meat-based main dish plus sides or trimmings i.e. we're having chicken for dinner.

Taking a lesson from Japanese cuisine, there is no main dish to any my meals. I typically have 4-5 or five small plates that include salad, soup, rice, and vegetables.

Here's tonight's dinner

1. A bowl of Coconut Barley Pilaf with Corn, Tofu and Cashews (I replaced the chicken in the recipe with tofu)

2. A bowl of Braised Yuba and Bok Choy from our CSA

3. A bowl of Curried Summer Squash with Yuba and Cilantro

4. A cup of Sencha Fukamushi green tea

I savor every dish equally and consider them part of the palette which makes up the meal. The family gathers to discuss the events of the day, the schedule for the next day, and the challenges/frustrations we face at school/office/studio. We often read the Dave Barry calendar page of the day at dinner and comment how the day's Dilbert parallels real life. We serve ourselves individually in the kitchen, picking from a large assortment of random Japanese bowls and plates, so that every meal is a completely different visual experience, sized to the appetite of the day, the seasons, and personal whim.

Grazing rather than eating, enjoying several small plates rather than a main disk, and making dinner the family time of the day works well to keep us together, keep us communicating, and keep us appreciating the wonderful myriad of foods available to vegans.

Meaningful Use for Specialists

I was recently asked how specialists, such as pediatric surgeons with few Medicare or Medicaid patients, can participate in ARRA and implement EHRs with meaningful use.

First, let's review how ARRA stimulus payments work:

Medicare
Medicare incentive payments are capped at 75% of allowable Medicare charges, up to $18,000 for the first payment year. Incentive payments are reduced in subsequent years: $15,000, $12,000, $8,000, $4,000, and $2000.

For eligible professionals in a rural health professional shortage area, the incentive payment amounts are increased by 10 percent.

Physicians who do not adopt/use a certified EHR will face reduction in their Medicare fee schedule of -1% in 2015, -2% in 2016, and -3% in 2017 and beyond. ARRA allows HHS to increase penalties beginning in 2019, but penalties cannot exceed -5%. Exceptions can be made on a case-by-case basis for significant hardships (i.e. rural areas without sufficient Internet access).

What does 75% of allowable Medicare charges really mean? A provider's office which has allowable Medicare charges totaling $24k or more can receive the full $18k in ARRA stimulus money. A provider with $13.3k in allowable Medicare charges is eligible for only $10k in ARRA stimulus money. Several providers have told me that an office needs about 7-8% of their patients to be Medicare beneficiaries to receive the full stimulus.

Medicaid
Office-based physicians whose patient mix includes at least 30 percent Medicaid beneficiaries are eligible for up to $63,750 over six years, as long as they are able to demonstrate "meaningful use" of healthcare IT. A lower 20 percent threshold applies to pediatricians.

Physicians who predominantly practice at Federally Qualified Health Centers and other settings can qualify if 30 percent of their patient base is characterized as "needy," including those covered by Medicaid, those who receive uncompensated care and patients who are charged income-related, sliding-scale fees.

Providers cannot participate in both the Medicare and Medicaid programs - they must choose one.

My interpretation of ARRA and everything I'm hearing from Washington suggests that the Interim Final Rule and Notice of Proposed Rulemaking planned for December are not likely to change this.

What is the alternative?

In my experience, the private sector and government need to collaborate to accomplish societal change. Private payers need to support clinicians who do not qualified for ARRA incentives. Why?

EHRs reduce cost and enhance quality via care coordination, reduction of redundant testing, and decision support that results in the right care at the right time. The largest portion of the financial benefits of EHRs accrue to payers. Payers should gainshare this savings with clinicians.

Malpractice insurers are another possible source of incentives. The Harvard affiliated clinicians are covered by a self insured risk management pool administered by CRICO/Risk Management Foundation. In discussions with CRICO, I learned that a large proportion of malpractice assertions arise from test results that are not reviewed/acted upon and by referral workflow that is never completed i.e. a PCP and specialist do not coordinate the patient's care. Meaningful use emphasizes the need to implement electronic lab workflow, decision support, and care coordination. If specialists, such as those with few Medicare and Medicaid patients, participate in EHR implementation and healthcare information exchange, it is highly likely that malpractice assertions will decrease.

This blog is a call to the private sector - private payers and malpractice insurers have much to gain from EHR and Healthcare Information Exchange adoption. It's time to gainshare and fill the ARRA donut hole, ensuring that all clinicians, including specialists with few Medicare and Medicaid patients, are meaningful users of healthcare information technology.

Guidelines and Protocols

I'm often asked about decision support capabilities in the BIDMC inpatient and outpatient EHR. I've previously written about our general principles of decision support and our priorities for implementing electronic guidelines and protocols.

I thought it would be interesting to show you the screens from our self-built Provider Entry System. I've included the protocol for Complex antibiotic ordering per protocol, our protocol for Heparin Dependent Antibodies, our Hydration Protocol to minimize risk of Iodinated Contrast Neprhropathy, Red Cell ordering, and TPN ordering.

I've also included an overview of our web-based ambulatory EHR, called webOMR. You'll see all the meaningful use features - vocabulary controlled problem lists, medication management (including medication reconciliation, e-prescribing, drug/drug interaction checking), clinical documentation, screening sheets which display all results and quality measures specific to treatment of a disease process, and ordering with complete decision support to ensure labs and radiology orders follow best practice guidelines.

The thematic definition of meaningful use is:

2011: To electronically capture in coded format, and to report health information, and to use that information to track key clinical conditions

2013: To guide and support care processes and care coordination

2015: To achieve and improve performance and support care processes and on key health system outcomes

Our inpatient and ambulatory systems are evolving constantly to meet these goals.