« Yikes | Main | The Chicken and the Edwards »

July 20, 2007

Electronic Medical Records Will Clean Your House

In answer to last week's study suggesting that initial installations of electronic health records aren't proving as effective as hoped, Overheard in Providence offers up a corrective to the doubters, and looks into a future filled with health information technology and the many fruits of its abundance:

If EMRs [electronic medical records] become widespread, they open the door to a huge new area of medical research. Computer-aided diagnosis is going to get a lot better if millions of anonymized medical records become available. A few years ago I went to a talk by MIT professor Peter Szolovits. He was able to use a computer to diagnose certain heart conditions from audio recordings better than most doctors. EMRs would greatly facilitate the development of automated screening procedures. More importantly, when new procedures are developed, they could be applied retroactively to data collected years earlier, even when a patient stops seeing a doctor. There’s absolutely no way to do this now.

EMRs would also be a huge boon to public health research. Researchers could be given access to a huge data base of anonymous medical records, all in a standard format. It would be trivial to check if two conditions are correlated, or if one disease occurs more often in some segment of the population. The amount of data would be so large, a doctor could even search for records similar to their patent, and use those records as a guide for what health problems to watch for.

Finally EMRs make much better use of healthcare we already provide. If you go in for surgery, all sorts of equipment is used to monitor your well-being. This data should be recorded and reviewed by a doctor who isn’t busy cutting you open. If you go in for a 3D bodyscan, even more data is collected. The scan could easily be reviewed by experts in other parts of the country, provided they have access to your EMR.

I tend to plug EMRs in terms of their immediate and obvious benefits: Reduction in paper, reduction in cost, reduction in lost records, simple programs that prescription errors, etc. But this is all correct, too. The widespread implementation of EHRs could trigger enormous change in the way medicine is practiced, and could accelerate many, many types of research. The downside here is that there are obvious and clear privacy concerns, but as genetic risk profiling inches ever closer to reality, these are issues we're going to have to deal with one way or the other.

July 20, 2007 in Health and Medicine | Permalink



In your own mind, I'd flip the order of importance: quality of care is the greater impact of health IT (this is not just EHRs when you start talking about decision support), not just the cost reduction.

Unfortunately, as was said in a prior thread of health IT, doctors don't view these systems as the MIT professor does, but rather an oversimplification of their work that precludes what they believe is their greatest asset: judgment at the individual level. This is a clear fallacy, but one of the larger obstacles to implementation.

Of course, pay-for-perfomance initiatives are increasingly adding a health IT component, which could change this dynamic. This is also the other main benefit of health IT-- it could easily facilitate physician compensation on process and outcomes measures rather than # of visits.

Posted by: wisewon | Jul 20, 2007 11:58:11 AM

I am all about privacy (not that we've got much of it left these days), but I can see one obvious benefit to having everything stored electronically: time and paper savings. And by that I mean the three or four sheets of paper, per child, that I have to fill out EVERY TIME I take them to the doctor.

Me: But I've already filled these out a hundred times--our contact information is the same, our insurance is the same, we're all the same, and I still agree to letting other people in your office see these records...

Receptionist: "Mrs. Tornello, we still need you to fill them all out, completely, if you want to see the doctor. Now please sit down."

Posted by: litbrit | Jul 20, 2007 11:59:37 AM

Why are you silent on the topic of the day, the judge's dismissal of Plame's lawsuit? I demand you drop this dreary medical topic (which requires real knowledge of the business to have a reasonable opinion on) immediately, in favor of the failed lawsuit.

Posted by: yan d. kamecki | Jul 20, 2007 12:01:19 PM

Paper never goes away. In practice, healthcare IT creates new jobs (doesn't replace old paper-based ones) and new revenue streams, thus adding to bloated administrative overhead costs.

Posted by: mondo | Jul 20, 2007 12:09:03 PM

Having a comprehensive EMR system in place could be critical in the event of a large scale public health problem -- say, a flu pandemic.

If a pandemic were identified and contained via EMR's, you would see a hell of a cost savings right there. (the health benefits might not be too shabby either...)

Posted by: DukeJ | Jul 20, 2007 12:41:05 PM

> If a pandemic were identified and
> contained via EMR's

Talk to Megacorp and ask them how well their 10-year, $200 million implementation of SAP has worked out in terms of providing "transparent information across the enterprise". You won't be able to reach them because they are now in the third year of their Data Warehouse project (a paltry $50 million) designed to "unlock the wealth of information in our SAP system".

EMRs may or may not be good things. But I just have the sense here that people are mixing what software salesmen promise with some small-scale successes and assuming that (1) the software vendor's promises, themselves non-binding reflections of their customers' dreams, will be met (2) such systems will scale. Enterprise-class software implementation from 1992-2005 does not bear this out.

But hey, I had better start looking into these EMR things - the consulting work could carry me all the way into retirement.


Posted by: Cranky Observer | Jul 20, 2007 12:56:26 PM

I've been thinking seriously (in great detail too) about this EMR issue - based on a career's experience in the computer industry (Hewlett-Packard and others), in large corporate IT departments, in the IT portion of the USAF, and as a patient in a our state's academic medical center and clinics - Oregon Health and Sciences Univ is fully computerized for patient records.

This hoped-for idea, which appears to be good for patient health and medical cost control. But it will be a work of generations to achieve this.

Just consider the simple facts of a computerized system: each item of information must be coded in a particular format (a blood test result for lipids, for example). Each EMR system will encode this data in a different way, using its own input and display screens to access the data. A user of OHSU's system would have no way of knowing how to use another provider's system (even if they somehow had access, which in the general case they won't and shouldn't). Transfer of the item of information (and the entire record) from system A to system B would require compatibility of data formats or a translator that knew of both sending and receiving system's encoding.

If a patient was seeing a private MD for general practice, being referred to an academic center for digestive problems, and seeing a specialist on arthritis at another private provider, there would be no consolidated medical record, and no interchange.

About 4 decades ago I visited the premier medical facility in Sweden (as an IT guy from a computer manufacturer), and they displayed their national medical records for patients, which all providers had access to and which contained everything medical (including mental health info) about each patient, for a lifetime. They avoided the interchange problems I've discussed above by having only a single, national, medical records system that encompassed all providers.

There is NO way the US will accept or implement such a national system. However, without a national system, no individual (private) medical provider would be able to keep electronic records, particularly comprehensive records for all providers for all medical incidents.

People who point to the VA system (or to comprehensive systems like OHSU's) don't realize that these systems are incompatible in data storage formats, incompatible in means of updating, incompatible in means of retrieving data, and have no way to interconnect. The VA/OHSU systems work within their own universe but are by no means comprehensive - if a patient sees an MD outside their system (quite likely for most people) that medical information is lost to the comprehensive record in the VA/OHSU.

Given the record of the federal government and state governments in regard to establishing IT systems that work and are using current technology (a failing grade at best), the proponents of EMR smoking something funny. Only a national, comprehensive system could possibly work, and that is politically impossible, and technically probably not achievable.

For those who think that the insurance company systems might be a basis for EMR, they only contain procedure data for cost/payment management. Each is unique. They don't even interchange data with Medicare, which is also a payment management system only - not a comprehensive record.

This is a dream or a hopeless fantasy. We don't have the political and professional will to nationalize medical record keeping, and we probably don't have a way to build and maintain such as system using either private resources or public resources.

It won't happen.

Posted by: JimPortlandOR | Jul 20, 2007 12:59:55 PM

"Paper never goes away. In practice, healthcare IT creates new jobs (doesn't replace old paper-based ones) and new revenue streams, thus adding to bloated administrative overhead costs."

Spot-on. The only way EMRs will eliminate paperwork or reduce costs is if its developed by the federal government, standardized across all institutions, mandated by law, and handed down for zero cost by the federal government to all providers and institutions.

Thats NOT whats happening. Instead whats happening is about 10,000 private enterprise "solutions" to EMRs and Health IT which do none of the unifying goals I listed in the first paragraph, all trying to carve out their little fiefdom and take their pound of flesh... errr money from the healthcare system.

The private enterprise "solution" will cause healthcare costs to go UP, not down. After all you will have to siphon off hundreds of millions of dollars in extra costs to pay for bloated CEO salaries for all these competing companies.

P.S. I've heard for at least 20 years now that some enterprising IT/computer tech guy had invented a radiology software suite that could self-diagnose a bunch of things on CXR, CT and MRI better and cheaper than a radiologist. Every study that has compared human radiologists to these computers has shown the humans were better, hands down.

Posted by: joe blow | Jul 20, 2007 1:03:58 PM

Is hte MIT guy going to be held liable every time his computer mis-diagnoses somebody? Or does he have the gall to suggest that his computer is 100% accurate?

Posted by: joe blow | Jul 20, 2007 1:10:09 PM


Great comment... so true. How about that $40M CRM implementation that was to allow you to zig before your customer zags? Ohh by the way, you couldn't get your reps to us it!

What's often not discussed on this subject is risk, and the risk that an EMR implementation will not provide the value you think, but it'll cost twice as much as you forecasted.

EMR is far from my forte, but it seems to have even greater issues than CRM, which was more dependant on end user buy in than systems like Accounting and ERP. Like CRM, you have a strong end user that needs to see the system as adding value to their day-to-day job for them to adopt it, without which, the solution fails. If EMR cannot provide that, the Doctors aren't going to use it.

One question for those more familiar with this space: does someone offer this software as a service (ex: Salesforce.com or NetSuite)?

Posted by: DM | Jul 20, 2007 1:36:47 PM

"However, without a national system, no individual (private) medical provider would be able to keep electronic records, particularly comprehensive records for all providers for all medical incidents."

Jim, is there not an industry body that can develop a set of standards for the industry? For example, there’s a group within my space that’s helped the industry adopt a set of standardized XML documents to represent specific business data and transactions. Mine is very fragmented space with many small technology providers and this has gone a long we to help.

Posted by: DM | Jul 20, 2007 1:44:29 PM


You're really missing the fragmentation problem.

You need a standard data format FIRST. Right now, everyone could have an EMR, and you would not be able to get "info in a standard format" because it isn't in a standard format.

For an analogy, think of plain documents. In paper form, you can read all of them--but the underlying record could be in .doc, .rtf, .txt, .pdf, LaTex, .xls, .gif, .jpg, .html.... There's no easy way of getting standard data, just because they are all text documents and all electronic.

Posted by: SamChevre | Jul 20, 2007 1:56:14 PM

Is hte MIT guy going to be held liable every time his computer mis-diagnoses somebody?

hey, joe, learn to read. Svolovitz said that he was able to make a diagnosis using the computer (and that, interestingly, this diagnosis was more accurate). The point is that the diagnosis can be done because electronic records exist. It's much like a doctor saying that he can make a diagnosis using an MRI. Does anyone ask if the guy who invented the MRI is going to be held liable everytime a patient undergoing an MRI receives an incorrect diagnosis?

More intelligent Republicans, please.

Posted by: Tyro | Jul 20, 2007 2:09:37 PM

DM: is there not an industry body that can develop a set of standards for the industry?

Yes, this COULD be done. Can you imagine getting the fragmented medical establishment (with very big egos) agreeing between universities/medical schools, private hospitals, government hospitals, 1000's of private MDs, 1000s of medical specialists (radiology, etc.), and insurance companies? I can't imagine this.

In the 1970's when I was involved in US Air Force efforts to standardize information across all bases, they had a data dictionary to define each data element, specify its format, indicate which codes were acceptable for each element, provide for error checks, and indicate who were the 'owners' of that element, and what authorities could modify the info.

The data dictionary for just USAF personnel matters was literally binders full of pages, changing constantly, and using maybe 10 more binders for the particular codes to be entered in each field. It was probably better than illegible records, but the human cost to learn, use and maintain it was monumental as only that word could mean in the government.

I'm skeptical. XML only provides a way of handling part of the problem that the scope defines.

I've watched my MD try to navigate and use the OHSU system, and he's an Asst Prof of Med, eager to use technology, and smart as hell. He barely can manage it, and often has to give up because the system won't let him do what needs to be done.

We used to joke in the military about 'forms', which are required to do anything. A wise old E-9 (highest ranking enlisted guy) used to say: you can't die because there is no place of the form for that.

Posted by: JimPortlandOR | Jul 20, 2007 2:28:02 PM

Ok, so we've heard all about how private enterprise (sic) can't/won't get the job done. So what else is new?

Because, y'know, and I'm sure you will think this is straight out of Ripley's Believe It or Not!, this all happened before, when the railroads were built. All the gauges and couplers were different and private enterprise couldn't finance the roads so the public had to do it and then when they were a financial success business got control and built twice as much railroad as we needed trying to get monopolies.

It happened when Edison started selling electrical power and other companies started selling power with different frequencies and voltages. Heck, if you're over 40 you remember throwing away the last Beta tapes you owned, and the chances are good you've got floppies in your desk you no longer have a drive to read them with.

What's happening for humanity is the development of a new neural net, the same way that the brain of a child grows and organizes neural pathways in response to the accretion of experience and practice.

What most commenters are missing here is that we are talking about very large numbers and very exact tools to extract data. Sure, it's not as good as the salesmen say it will be, but 50 years ago you had to be an insane LSD-taking science fiction writer to even imagine the widespread use of the tools we have today.

Imagine you go to the doctor and and they hold a stethoscope to your chest and write down what they heard. Then the stethoscope is connected to a computer which compares what it hears with, 100 million other cases. And remember, the computer did not listen to really loud rock music while it was in college.

What if every adverse drug reaction were reported to a central data base, instead of the fraction of one per cent that gets reported today? What if you could do your own research and find out exactly what percentage of users had suffered each side effect listed, and what other characteristics they had or prescriptions they were taking that resembled your own?

Like I said, very large numbers. Now think back to your Statistics 105- what are the two things that improve the reliability of any study? That's right- the size of the sample, and the span in time covered.

Very large numbers.

Posted by: serial catowner | Jul 20, 2007 2:42:01 PM

Looking for common denominators, you could say that EMR, or the hoped-for efficacies of EMR, shines a light on what is - not solely, but essentially - a database problem. That is, gathering the data, vetting the data gathered, guaranteeing privacy and (the Holy Grail) sharing the data, to mention a few elements of the enterprise. Jim effectively sketches the history, hope and despair of working for and waiting on the Promised Land promises the technology tends to engender and sees correctly that the non-standardized nature of entrepreneurial health care management makes interoperability (pretty much crucial for solving database problems) for all intents and purposes impossible.

DM asks "is there not an industry body that can develop a set of standards for the industry"?, which is the best question to answer before arguing about prospective benefits, risks, hassles, etc. inherent to moving to EMR in broad way. Perhaps a consortium comprised of NIH, CDC, IEEE, AMA and other (not too many) organizations could be formed and given a deadline to hammer out mandatory protocols for QOS (quality of service), interoperability, wireless frequencies and so on. Perhaps fraught with contention and a certain amount of brain damage, such an effort nevertheless moves in a direction that fosters adoption in the long term and considerable innovation along the way. Could be that such a project is already in motion, but I haven't heard of it.

Posted by: BdubBellingham | Jul 20, 2007 2:46:26 PM

Another thing most people here are missing is that our records already are computerized, just not in a sharing way. Lab records are kept on computers. Billing is kept on computers. More and more of your x-ray images are digital.

In fact, the only thing that still shows up in the ur-format as a pile of paper is the actual patient record, to which anybody can add something that is not true, or drop the file and lose something that is.

Worst part of any appointment- "Nothing has changed since my last visit." ("Nurse" looks frantically through chart.) "Well, I can't find that here- you'll have to tell me again."

Right. Like that will work.

Posted by: serial catowner | Jul 20, 2007 3:09:45 PM

Maybe its wrong of me to leave this message here, but I figure so many health care policy wonks read this page I might get a real answer.

I'm soon to be among the many Americans who will have to buy my own health insurance. I'm looking at "Mega Life and Health Insurance Co." and can't tell if its a great policy or a scam. As they claim to be nationwide, any horror stories out there?

Posted by: Jared | Jul 20, 2007 4:24:20 PM

Another thing a lot of people are missing is the potential to allow patients to order their own refills of their prescriptions.

You require all the pharmacies to connect to a central database. They all use computers already, and the government already has more power here to issue orders and demand records than almost anywhere else in American life. This in itself would be a big improvement, as it would stop patients from going to different doctors and taking multiple prescriptions for the same problem to different pharmacies.

Then you set up a red flag system that triggers a "hold" order when the refill the patient asks for falls outside certain guidelines. This would obviously include stuff like too many refills, allergies, medications that interact adversely, too long since doctor visit, etc. Again, this would be a big improvement on the state of things as they are today.

The patient is the best, and sometimes final, authority on the medications they are taking. Patients are not qualified to diagnose, and poorly qualified to prescribe, for many serious conditions, but by age 50 the average American is taking one drug on a regular basis and the chances are good that they know more about how it works for them than the doctor or nurse they're likely to see during a typical appointment.

This would be such a time and money saver that I have absolutely loved this idea since the first time I realized it could be done- 15 years ago.

Posted by: serial catowner | Jul 20, 2007 4:41:57 PM

This is a case where Government could do a lot to help and so far has not. A key privacy issue is if your records could be sold to marketeers affiliated with your insurance company (or physician or hospital or pharmacy) so that they can "provide you with better product selection opportunities". It got fumbled under the Republicans to mush.

Clearly Government could provide the basis through standards definition using NIST and industry groups. (something that Hoover was famous for doing)
But Bush and the Republicans have done almost nothing.

Posted by: marc sobel | Jul 20, 2007 7:40:46 PM

Nothing I've seen in the Democratic cost containment proposals about a uniform standard either. Both Edwards and Clinton (probably Obama too, but I don't recall) specifically call for EMR, and interoperability, but neither says anything about how to set uniform standards and practices.

Posted by: Sanpete | Jul 20, 2007 7:59:46 PM

The only federal legislation that I know of for a national standard for EMR health records was proposed by former-Senator Santorum and a certain junior senator from New York. I'm not sure where that went, however that is one of the keys to doing a decent implementation of EMR. Otherwise, it's just more paper (or the electronically stored scanned paper that my current employer uses).

I was a systems and software engineer for 10 years before going to medical school. We built systems that helped other types of users transition from paper to computers. Believe me, those users were every bit as "arrogant" (or skeptical, as the case may be) as physicians. There are gigantic transition costs. In my first career I used to work with the military, but most private practices are very small businesses. It's not clear to me that they can afford the costs for marginal benefit.

Never the less, after 10 years as a physician, I just had my first interview for a job in medical IT. I'm a huge believer in EMR, but I completely agree with Cranky and Joe. It's not as simple or as transperant as you imagine it to be, Ezra.

Posted by: J Bean | Jul 21, 2007 12:05:23 AM

Sam and others,

While there is still a ways to go on health data standards, they are much further along than has been portrayed in this thread. HL7 (messaging/text), NCPDP (prescriptions), LOINC (labs), DICOM (imaging, e.g. MRI) do exist through public-private efforts over the last 5 years or so. These are beginning to be adopted by the IT industry, but some further clarity and completeness is needed to fully overcome this obstacle.

The real challenge with interoperability is not the IT standards, but actual agreement between health organizations to be interoperable. In every region, there are health centers that are leading edge and comprehensive. Like other businesses that bundle services, they typically have some offerings/health specialties that are truly best-in-region while others aren't-- but they rely on the convenience that patients get from the relative ease of sharing information across providers within their health center (even a purely paper-based center can share charts more easily than with another provider on the outside) to ensure a greater market share of patient services. This is true for both hospitals and physician practices. True interoperability is a clear threat to this market dynamic, so these stakeholders, which typically have the most influence in their health communities are more reluctant than others to participate in these initiatives.

A regulatory requirement mandating interoperability is needed to overcome this-- moreso for the competitve dynamics described above than the health IT complexities.

Posted by: wisewon | Jul 21, 2007 7:48:41 AM

Or, IOW, it used to make a HUGE difference to have all the services available under one roof. A doctor alone, or a small group, suffered a major handicap. Fully implemented EMR will to some extent level that playing field.

Posted by: serial catowner | Jul 21, 2007 8:32:50 AM

It is simple. There should be ONE format for the records which is tied to the ONE format for billing. Everything needs to be entered for billing, doesn't it?

ATM transactions cost 5¢ - why should medical transactions cost $40?? Doctors need to keep records and they need to bill - All it takes is the creation of a single format ( I'd assume it will be in some kind of XML ) for both. That way, all of the information will be there - including how much it cost and who got paid and when.

This is like the interstate highway system - It is a huge need that is too big to be created by anyone but the feds.

Posted by: fasteddie | Jul 21, 2007 5:22:41 PM

The comments to this entry are closed.