HIT Adoption – Alignment & Simplification


David Hartzband is a Lecturer in Engineering
Systems at MIT, teaching courses in large-scale software systems and Director of Technology Research at the RCHN Community
Health Foundation. In his role at the Foundation, Dr. Hartzband spearheads the
organization’s continued evaluation, assessment and findings
dissemination related to health information technology.

if we didn't know already, most of the leadership of Health and Human
Services has now weighed in on the importance of health information
technology (HIT) in realizing goals for health care improvement and
reform. HHS Secretary Kathleen Sebelius said in a House Ways and means
Committee hearing on May 6th that “health IT is critical
to health reform”. To her credit, she also said that “just shifting
our paperwork to computers won't work, unless we make sure they can
talk to each other.” We also know that substantial amounts of money
will be available through the ARRA and other sources for acquisition
of electronic heath care records systems (EHR) as well as incentives
to Medicare and Medicaid providers for meaningful use of such systems.
Those of us who have worked in HIT, for even short amounts of time,
realize that there is a step missing in this progression: acquisition,—–,
meaningful use. That missing step is the adoption of technology, and
adoption is considerably more difficult than either of these other steps.

Many studies have been done on what
impedes or facilitates adoption. The factors most often found are: 1)
technical – system complexity and lack of integration with existing
systems; 2) cost – initial investment, lack of funds for training, maintenance
etc., unclear ROI; 3) social or cultural – unprepared workforce, lack
of management commitment, privacy issues and finally; 4) alignment –
technology not well matched to work flows and work styles of users,
system not useful to users. AHRQ did a study several years ago on this
(2006) and found that the biggest impediments to adoption of HIT were:
cost-benefit misalignment, technological complexity, lack of data integration,
lack of workforce preparedness & lack of motivation on the part
of providers. Some things have changed since then, but not all that

OK – so we won't reach meaningful use
of EHR technology, let alone other necessary and productive health information
technologies, just by throwing money at the problem, even by paying
incentives to providers. How can we ensure that EHR and these other
technologies are adopted? After all, we'll realize no benefits from
HIT even if it is acquired and deployed. This is just a lost sunk cost
without adoption. As a technologist, I am most familiar with what can
be done on the technology side, so I'll make some suggestions there

Studies are fine, but most of these
studies on adoption were done based on general technologies, such as
the adoption of fiber-channel attached mass storage. Technologies are
not usually adopted – products are. Products have user interfaces and
workflows and feature sets that must be aligned with the workflows and
needs of the population using them. I had the great good fortune to
be Chief Technology Officer of a company called eRoom Technology (acquired
by Documentum in 2003 which was acquired by EMC Corporation in 2004)
that developed a product that provided collaboration and lightweight
process and document management among individuals and groups. This was
a product that wound up with 90,000 users at the Ford Motor Company
despite the fact that it was not initially supported by Ford's IT group.
This happened because eRoom was entirely aligned with its user population's
work style and workflows. People wanted to use it and even found new
ways to use it that we adopted back into the product. There are very
few products in the HIT space like this. Most practice management, EHR
and other HIT products make users work in a specific fashion. Users
must adapt to the product, not vice versa. This is the first thing that
we can do technically. Design HIT products that are actually aligned
to the needs of providers and associated health care workers so that
they want to use them, not that they have to.

How do we do that, I hear you say.
The only way I know of is to actually work with the user population.
There is a very broad range of how work gets done in healthcare organizations,
so there is a broad range of aligned workflows. Asking & observing
how providers use an EHR is an essential first step. Every organization
will tell you that their requirements, & therefore their workflows
are unique. There is some truth to this at a detail level. Primary care
workflow in a private hospital is different than in a Federally Qualified
Health Center than in a small private practice,… but, they are not
so different that you can't find common threads . The general outline
of what needs to be done is generally the same. Designing & developing
products with a high degree of configurability, ideally user controlled,
is essential to be able to create workflows that providers in each type
of organization will recognize & be comfortable with.

The second thing we can do technically
is to simplify how these products are integrated. Have you ever had
to talk to the IT manager at a health center about sharing data between
their practice management and EHR systems so they can do integrated
billing? That's hard enough without having to work NHIN into the conversation
as well. We have to make this type of information sharing easier, both
to understand and to accomplish.

The last time I did talk to the IT
manager at a health center about application integration & participation
in a RHIO (actually last week), the person was also the billing supervisor
& managed most of the back-office staff. She wanted to be able to
transfer the billing codes from their EHR back into their practice management
system so that they did not have to re-enter them to prepare their electronic
billing. Both products were HL7-compliant, but they had a software contractor
who estimated about 6 effort-weeks to get it set up. That's not simple…
They were also looking at joining a RHIO which would require them to
install integration engine software that would connect their patient
demographics & some clinical data to a shareable database, but would
not help their internal integration issues; both not simple & not

We have been doing this kind of process
(workflow) alignment & information sharing in supply chain management
& manufacturing for decades now. Pragmatic mechanisms were developed
for process modeling & sharing as well as information sharing that
would work well in healthcare. They worked well because the thinking
about information sharing & process integration changed to encompass
organizations outside of a single company connected by a network as
the unit of business. The current thinking about NHIN has to change
in order for large scale healthcare information sharing to take place.
We need to stop thinking about NHIN, or any medium to large scale information
sharing, as a massive enterprise integration project & look at it
as an actual network of information & function, that is a set of
healthcare organizations collaborating over a network. This would allow
the use of technologies currently being developed for 'cloud-based'
computing such as Google's Big Table (multiple petabyte distributed
data management) & MapReduce (massively parallel indexing) that
are designed to enhance the function of a network. This would truly
enable the kind of conceptual & implementation level simplification
we need to make healthcare information sharing, at any level, work.

Alignment & simplification work
on the policy side also, but that will be for another post.

Categories: Uncategorized

Tagged as:

6 replies »

  1. Fully agree with David’s points here.
    The challenge is persuading the IT team that user-friendly interfaces and work flows are not a “down the road” phase of the work that may or may not ever get done, but should be front and center in the initial phase of the work. If the technology turns off the users, it’ll never be readily accepted, much less advocated for.
    And if the data integration feeds and interfaces aren’t there and readily accessible, then where is the system adding value?

  2. As an IT professional in the healthcare field, I do not believe that HL7 is particularly “Difficult.”
    I also believe that managers should have practical experience in HEALTH CARE IT. Not supply and chain. Not finance.
    Barking about HL7 compliance is simply silly — they are saying “we have the capability to transmit using HL7,” not “We put the data in the exact field you want it in.” HL7 is a way of shipping data, not evaluating it for quality. Mapping exercises should NEVER be performed in a vacuum by a manager (especially not a manager, chances are they haven’t touched a real system in years).
    I can’t even START with all the things I find need more explanation in this post. Especially not since I have a mapping session this afternoon. For HL7.

  3. I don’t quite know where to start with this combination of platitudes and nonsense.
    What worked in supply chain management & manufacturing worked precisely because the value of the individual transactions and the margins in the industry are so small that internal workflows were changed to meet the demands of the information technology that made the transaction processing cost nearly zero. Management wanted it to happen, it HAD to happen. The job redesign (not computer system redesign!) that went with this pushed responsibility for recordkeeping out of the back room and onto the floor, and what the veteran stock picker or receiving clerk in the warehouse may have thought about it didn’t mean a damn. It was therefore successful. Why is it that in medicine, people must have warm fuzzy feelings about tools before they can be expected to meaningfully use them? If the job “nurse” or “physician assistant” or “medical transcriptionist” even “physician” should be re-designed to meet the recordkeeping needs of medicine and the available tech, so what? Yes, I know there is great disagreement about what those needs are but that’s beside the point.
    Bracketing the fact that there is no such thing as an “HL7-compliant” product because there is no such thing as “HL7-compliance” — this integration project you mention isn’t simple compared to what?!?! It is not particularly more difficult to store and transmit a Continuity of Care document than it is to store and transmit the EDI bills of lading or purchase orders that fly up and down supply chains all day long. Why should we expect integration in medical systems to be any simpler than the very successful EDI integrations used in supply chain management? Nobody expects a random IT manager at a distribution company to be able to map EDI transactions in and out of systems, this is a specialized sort of skill. So why should we expect it to be so very simple that a random health center IT manager ought to be able to map much more complex data in and out of medical information systems?
    And a RHIO that requires individual sites to install anything like a full-blown integration engine isn’t doing a good job thinking through the problem and negotiating licensing agreements with integration engine vendors. All that should be required at the sites is some kind of store-and-forward message queueing system and these are not that hard to write (I’ve done it. Twice.).
    How on earth will vaporware for vapor-based computing make the scales fall from our eyes and “conceptual simplifications” so apparent? Humidity?

  4. “Design HIT products that are actually aligned to the needs of providers and associated health care workers so that they want to use them, not that they have to.”
    The CPOE devices out there do not approach meeting this criteria.
    With the rapid proliferation of workarounds, doctors can still keep a patient alive despite being forced to use these disruptively defective gear.
    Even the US Army has learned about the intolerable state of interoperable HIT devices.

  5. Professor, I think the quality of product is an essential part of the reason the adoption of HIT has been low. I tell people that in my school days I wrote a program that worked perfectly for about 12 data points and any other resolution, it would go berserk. The working of the code for 12 points was no indication of the goodness.
    In the sameway, when people tell that the HIT softwares are good and working for many; I would challenge it. It is just working on these special cases…they need to work in typical cases.
    Kathleen is right….we need to focus on business and clinical process fundamentals and in the meantime IT can catch up in creating a robust product.