The pricing of voice-recognition software as one sign of distortions of the medical market, and other inside insights.
For background on the EMR saga, see this original article and previous installments one, two, three, four, five, six, and seven. Today, let's talk technical and business specifics of electronic-record software.
First, from someone in this business, a vivid and specific illustration of the overall distortion of the medical marketplace.
I'm an independent IT consultant, working mostly with solo practitioners and small (2-10 doctors) practices. My clients choose their practice management and EMR software (sometimes they ask me for advice, but usually the choice has already been made by the time I get involved) and I help them make it work.
Over the past few years, I've worked with about 15 different EMRs, and I've developed a theory: all EMRs suck; they just suck in different ways.
However, despite my frustrations, I'm convinced that this is a good and necessary thing to do, and will lead to advantages not only for wider patient care but for doctors themselves (though they'll kick and scream even while they benefit; it's just something they do.)
I thought I'd indulge myself (and bore you, no doubt) with a few observations:
- Software companies in "vertical" markets have never been magnets for top programming talent...
- Nowhere is the lack of star talent more glaringly obvious than in user-interface design. To be fair, there is an awful lot of information to be captured, and Medicare* frowns on too-great indulgence in boilerplating - but sometimes I am staggered by the sheer number of clicks required to get through even the simplest of screens, and there are far too many screens.
- Counterintuitively, some of the most physician-UNfriendly interfaces I've seen were designed by physicians. With very few exceptions, users are lousy at designing their own tools! One of the best I've seen is Practice Fusion, which is a relatively new company started by Silicon Valley/Web 2.0 types (breaking the old vertical-software paradigm.)
- The back-turned-to-the-patient issue is an easy one to solve: use a tablet, or a laptop on a rolling stand, and face the patient (or stand next to them.) The fact that such an easily-solved problem is so widely cited as a deal-breaker says more, I think, about the mindset of physicians than about the technology itself.
- Nuance Communications has a virtual lock on the voice-recognition market**, and they exploit it in ways that I frankly find appalling. Dragon Dictate Home Edition is about $50; Premium is around $100-150; Professional around $500... but Dragon Medical is $1500. The only real difference between Premium and Medical is a pre-trained vocabulary; I can see charging extra for that if the user wants it - but all non-Medical editions of Dragon check for EMR software and will not run if it's present. If you're a doctor, no edition of Dragon but Medical will run on your machine. Furthermore, updates for other editions are available on Nuance's website so that if you upgrade, e.g. from Windows XP to 7, you don't have to buy a new copy of Dragon - but Medical users are left twisting in the wind. [JF note: I agree. I like and use Dragon/Nuance software but have been astonished by the tiered pricing. For the record, I've bought and personally paid for the Professional version.]
- Data interchange between competing EMRs is laughable. There are national and international standards for this (HL7, CCR/CCD, etc.), but no EMR company takes this seriously - they generally do an OK job of exporting data but are completely clueless about importing it. (If CERN, ARPA, and the big universities had acted like these guys, there'd be no Internet.) The biggest player in "gluing" various systems/equipment/etc. together is an open-source software project called Mirth, and the company/foundation that looks after it (think Mozilla, basically.) Earlier this year, Mirth was purchased by NextGen, one of the largest EMR companies. I'm keeping my fingers crossed that NextGen will adopt Mirth's mission of connecting the medical world... but I fear that Mirth will simply wither and die.
* Medicare _and all the other insurers_, but Medicare's the one with real teeth so I use them as shorthand.
** There used to be several other players in the voice-recognition market - SpeechWorks, ViaVoice, Jott, Loquendo, Transcend, etc. - but Scansoft (now Nuance) bought them all and either killed them off or folded them into Dragon. Google's speech recognition engine is the only real competition left (Siri, of course, is powered by Dragon), and Google doesn't provide a product that works with EMRs.
Now, about the public-health advantages that may offset some of the individual-practitioner annoyances:
I've been following all the different pieces around the EMR/EHR work--and it feels like a lot of the folks who've been writing in are really missing the forest for the trees--everyone's really missing is how important these innovations are to population health. Even working in a medically underserved community, this has changed how I work with leaps and bounds over the past five years.
Want to know how many smokers there are in a specific zip code who are served by your clinic or hospital? Want to be alerted whenever one of your patients go into the ED? Want to see a panel of what percentage of patients have diet-related co-morbidities? Want to know who a patient's Primary Care Provider? Want to geographically hot-spot specific health problems? All these things are infinitely easier with the existences of both EMRs/EHRs.
In other words, we can know so much more in so much less time. Rather than sending some poor soul into stacks upon stacks of ill-organized and non-standard hard copy medical records to sort through items, you can simply find it through a relatively (though not completely) understandable electronic system.
An example from my work is telling...we work with a variety of hospitals and clinics on a large public health project, which requires them to pull data, quarterly, on how many diabetic patients they have and how many of those diabetic patients smoke. For those few facilities still using hard copy records, we can only ask them to pull a sample of their data--and it takes two full days for their entire team to pull that information. At our EMR/EHR facilities, one person can pull all of the necessary information in a fraction of that time.
Yes, its almost certainly more cumbersome for practitioners --but it makes a drastic improvement in the quality of care coordination and the quality of data collected.