Thoughts on Adoption of Health Information Technology

12/11/2009

Computers in a Server Room
 My conversations with CIO’s about capabilities and adoption of health information technologies.

Since my comments on seeing computer systems at a leading hospital from a patient’s viewpoint, I have had a chance to carry on some email discussions with various CIO’s and others in Boston, and to reflect on the forces that lead to good interoperability or bad within institutions.

Dan Nigrin, CIO at Children’s Hospital Boston (CHB), wrote what I assume would be many CIOs’ heartfelt response to seeing a report of MGH’s struggles:
“It's particularly heart warming to me (tongue only partially in cheek) to
read that some of the problems that we have with our commercial system are one and the same with those that they have with their custom-developed system.

“Even at the places that have been ‘successful’ with this stuff, there's
still quite a long road to travel before it will be used efficiently and
without incremental burden.”
Indeed, our email discussion was triggered by an article in Business Week [no longer on-line] reporting a Harvard study suggesting that, thus far, moderately large investments in health information technology (HIT) have failed to save hospitals any money.  In early 2009, I was also part of an NRC study that suggested that HIT systems need to re-focus on the cognitive tasks of clinicians instead of just supporting the individual transactions of health care in order to reach their potential value.

The current state of affairs reminds me of the hand-wringing in business schools of the 1980’s, where scholars watched the investment of billions of dollars into new computer and information technology but were unable to detect any beneficial effects of this vast flow of money.  Only in the early to mid 1990’s did people like my MIT Sloan School colleague Erik Brynjolfsson detect the anticipated benefits, and it turned out that the key factor in resolving the “productivity paradox of information technology” was that the benefits came from changes in business practices enabled by the technology, not simply from applying new technology to old practices.  Yet much of what I see in today’s hospital spending on HIT falls just within this older paradigm, mainly automating existing practice. Only unlike in 1980’s businesses, there seems to be much less faith in the eventual payoff from these investments, so they are more half-hearted and will probably have to be driven significantly by Federal stimulus funds.

I also challenged John Halamka, CIO at Caregroup (and Harvard Medical School) to explain how I would have experienced the computer systems at Beth-Israel/Deaconess (BIDMC).  John says “At BIDMC, we've built most of our systems, so we have an integrated rather than interfaced clinical suite.” (Emphasis mine.) Here are my questions, based on my MGH experiences, followed by John’s answers:

“John, to put you on the spot a bit...
If we go down some of the examples I wrote about, how would these work at BIDMC?  I do know that PatientSite would give me far superior access to my own data, but what about the following:

  1. If I show up with my own x-ray or CAT scan in digital form, is there some way to get it into your hopper so that you could avoid doing another x-ray and instead use the one I brought you?  How is provenance handled?  How do you bill for reading a non-local x-ray?  Do docs actually use this kind of workflow?
  2. Are your ICU monitors integrated with other systems?  Do the flowsheets that MGH derives manually from these systems get collected automatically and do they become part of the EMR?  Are there isolated temperature sensors, blood pressure cuffs, pulse oxymeters whose reading are not automatically captured?  Is an EKG recorded through the system or through a separate portable EKG machine automatically captured to the EMR?
  3. How are nurses trained to use your integrated systems?  Do they indeed know how?  Are systems used differently in different parts of the hospital?  What sort of support does your system provide for "report"?  Are nursing notes written, typed, etc? Do they wind up in the EMR?  Are they used for anything other than legal defense or certification?
  4. Do nurses encourage or discourage patients from reading their own data?  Do they help interpret them? How about the doctors?  Are groups on rounds more accessible than the traditional "stealth" visits?
  5. Are the kinds of coordination problems I mentioned between eMAR and the pharmacy system avoided? Do you have more expressive mechanisms for contingent orders?
  6. Are doctors' notes captured in the EMR?  I assume there is a CPOE system in place.  Do you experience the problem I've seen in other institutions where it winds up being the resident who enters orders into the CPOE even though the decision on what to order has been made by a more senior doc?”
John’s response:
Pete - I met with my folks today and here's the "official" answers to your questions

  1. We upload exams from outside institutions  into PACS as "Reference Only" studies. These are often used as comparisons to studies performed here, and we currently upload over 1,000 exams per month.  The other scenario is one where the Radiologist agrees to provide a "Second Opinion" reading on an exam performed outside.  This primarily happens in the ED only and an order is generated for a 2nd opinion reading and coded as such so that only the professional fee is charged, and not the technical fee. We do less than 100 of these per month currently. On occasion, we do this with a non-ED patients study but only if the referring MD and the Radiologist confer about it and agree to proceed.
  2. We use the Metavision ICU system from IMDSoft.   All ICU devices are fully integrated via automated interfacing.   All charting is done electronically in the ICUs.   All ICU data is available to any clinician via the web.     EKGs done outside of the ICU are uploaded wirelessly and are available via the web to any clinician.
  3. Nurses are fully trained on all our systems.    Since nursing workflow is based on our integrated system throughout the hospital, there are not major departmental differences.  There are a few areas with unique workflow considerations (e.g., PACU, L&D) for whom we have made some design changes in our systems.
  4. We have 1 million square feet of free wireless.    Patients can view their results in real time via PatientSite . Pathology results have a delay before appearing on PatientSite to ensure that the patient has an opportunity to discuss first with a provider.
  5. All our major systems - POE, ambulatory EHR,  Pharmacy, OR are all one integrated system with a single database.   The medication administration record is currently done on paper using labels printed from POE/Pharmacy. When we implement electronic MAR, it will benefit from the same tight integration with our other systems.
  6. All ambulatory areas are fully electronic.   The only place we are not fully electronic is progress note writing on regular wards.   Those notes are scanned and made available on the web after discharge.   We're implementing an innovative wiki approach to progress notes that will go live in pilot in 2010.
Following up my discussions with Dan Nigrin, he provided the following answers to the same questions. Children’s has moved over the past several years from a combination of home-made and commercial systems to an integrated Cerner installation. According to Dan, this is now in place except that some Eclipsys systems remain in the ICU.

  1. Our Radiology Dept imports the images into our PACS system, and when necessary/asked, our Radiologists do a second opinion read.  Furthermore, we have provisions in place to allow clinicians to read the studies directly on local workstations from optical media in their clinical areas - a challenge when you also try to lock down those same  workstations to NOT allow reading of optical media, for security and virus protection.
  2. Our ICU monitors (including all the devices you mentioned) are all integrated directly to our EMR – no re-keying of values necessary.  The data (even the waveforms) are available via web interface throughout the organization.  EKG’s are linked wirelessly into the EMR.
  3. Nurses undergo rigorous training when they are new to the organization, or when a new system is introduced (or existing process changed).  There are regular newsletters and updates with tips, how-do-I’s, etc..  There are screensavers throughout the clinical enterprise that remind all clinicians about how to do certain things in the EMR system.  Most nursing documentation is electronic, including textual notes, MAR documentation, vital sign documentation, input/output documentation, etc...  They are referred to as necessary by any clinician – they are just as much a part of our integrated EMR as the notes from physicians, ancillary providers, etc...  We have some built in reports that both nurses and physicians use for “sign out”, that automatically pull labs, vitals, medications, etc.. in summary form.  We are in the process of enhancing these now, for both physicians and nurses, to include more features, like to-do lists, “things to check on”, and so on.
  4. In general our environment is one in which inclusion of patients and families in data sharing and decision making is considered paramount.  So I’d guess that more often than not, result sharing with patients/families is the norm.  Re: rounds – there are certainly some teams that need to round early in the morning, before many patients/families awaken.  But in general, engaging and communicating with the child/family is considered an essential part of managing the child, and so in those cases, I’d bet that someone circles back later in the day – I know I do!
  5. Our eMAR, Pharmacy and CPOE systems all use one and the same database.  All locations in the Hospital use the eMAR, and we are now in the midst of rolling out point of care barcode documentation using Bluetooth scanners (the units that do not yet use the bedside handheld scanning still enter the administration info data electronically, into a mobile  workstation).  You previously described an interesting problem that we too encountered initially, around orders being placed, the system assuming that the ordered medication was immediately available to the nurse to administer, and with the resulting problem that the nurse (who invariably would have to wait at least a little bit of time to get the medication from the Pharmacy), would have a “late” task on the eMAR.  We’ve since addressed this in our system.
    Furthermore, another unanticipated problem that we encountered in conversion of a paper-based MAR to an electronic one, is one of “control”; in the paper world, the MAR was managed by the patient’s nurse, especially with respect to medication administration times.  In the electronic world, the eMAR is now electronically controlled, whether it be by the ordering physician and the administration times/intervals she has prescribed in her CPOE order, by the Pharmacy, which may have adjusted the administration times based on standard schedules (e.g. “BID” = 8am/8pm), or by the nurse as before.  The practical realities though are that the nurse often has to adjust these administration times, to allow for instances when the patient is off the unit having a study done, or when there are multiple medications to be given within one time period, and so grouping them is desired so as not to disturb the patient multiple times.  Although the system allows for such dose adjustments, its not as straightforward to nurses as we would like.  Is the readjustment just for this dose?  Or for all doses?  Does rescheduling a single dose to later in time cause the subsequent dose to be administered too soon, potentially causing an adverse event?  This is difficult stuff that we are still improving upon – the unanticipated side effects of an electronic transition.
  6. All physician ambulatory notes, and virtually all physician inpatient notes are documented electronically, either by direct free text entry, templated direct entry, or by transcription.  All physicians are trained in order writing, and all have the ability to do so – however the decision as to who actually enters the order is made by each individual team.

John also asked Jim Noga, the CIO of MGH, to look over my earlier blog comments, and his response was:

Thanks for forwarding. Many of his comments are fair.
 
We rolled out the eMAR this year and the integration with Pharmacy worked as designed -- but as is always the case -- now that we are in production additional requirements have emerged and we are well aware of those pitfalls and working on plan to meet those requirements. In addition to the pharmacy, it is well integrated with Order Entry.  Some [of] the issues listed are more process issues (that are recognized as needing to be resolved) than technology issues (late meds). The good news there is a steering committee and while we have completed the roll out significant eMAR development and enhancements will occur over the next year.
 
He is right on inpatient documentation, it is still either manual or semi-manual (e.g. Word). I am sure you aware that MGH and BWH are undertaking the roll-out of Acute Care Documentation with iMDsoft as the underpinning and that will take several years inclusive of the necessary integration with systems above.
 
Regarding interoperability, I would say we have not achieved anywhere the interoperability that we would like to with biomedical equipment (iMDSoft will help) but our IT systems have a decent level of interoperability through the use of enterprise repositories such as the EMPI and CDR and to enterprise services which are in the early stages (allergies, meds, med reconciliation).
 
We are actively working on integration with web services of the GE Muse system providing links to the strips out of the CDR.
 
While MGH is filmless in Radiology and images are available through both our webresults and LMR apps we have not fully addressed outside images though this fall we are piloting a program in Neurosurgery through the guidance of Keith Dreyer.  In terms of sharing images on regional basis please see an exchange between Keith and on of our neurosurgeons below. [Elided.]
 
If you need more let me know but hopefully you get the picture. We are much better off than we were ten years ago (we are level 5) but despite best efforts interoperability continues to be a challenges especially as it relates to convergence of IT and Biomed.
 
While we progressively move towards homogeneity the challenge is meeting the needs of today while planning for tomorrow (meaningful use, 5010 ICD10, HIE etc).

So obviously different institutions set different short-term goals with different priorities.  From my patient’s vantage point, the most visible system at MGH was eMAR, which is only in planning at BIDMC, and implemented at CHB though with additional planned transitions in the barcode scanning technology.  Various institutions now seem attracted to IMDSoft’s MetaVision system to integrate ICU data, but BIDMC has pulled this trigger, whereas MGH (Partners) is still in the implementation stage.  Halamka’s focus on integrated design rather than a system based on many interfaces certainly sounds like it makes for more fully functional HIT. Similarly, Nigrin has bet on integration by choosing a single vendor to provide all systems, though at the (sometimes frustrating) cost of having to rely on an outside provider. Decades ago, such vendor systems were simply not functional enough, so leading institutions tended to build their own.  Today, the vendors have gotten better and more collaborative, and institutions are tempted by the ability to outsource development and maintenance to buy rather than build.

My sense is that most hospitals have not dedicated the resources to building or buying a complete integrated system.  Instead, they make do with a combination of independent, functional legacy systems and “best of breed” departmental systems from various vendors.  In that setting, they must innately deal with the problem of developing interfaces, and almost invariably suffer inconsistencies and bottlenecks in their internal information flows among these systems.  (Institutions taking this approach must essentially solve internally the same health information exchange problems that are now receiving national attention to facilitate inter-institutional communications.) 

I should end by pointing out that I’m not trying to judge the success of various Boston-area system deployments.  I have only directly experienced the systems at MGH, and only from a limited vantage point.  I have (fortunately) not also been a patient at BIDMC and CHB, so I cannot tell how those systems would appear to me as a patient.
Back to Blog

Accessibility