I really enjoyed this piece, by Barb Darrow, about the development of healthcare-related data management at the University Pittsburgh Medical Center (UPMC) for a number of reasons. First of all, as the article explains, UPMC is a leader in the area and is doing some really interesting things! Secondly, I always enjoy a good story about the economy of Pittsburgh, since it represents one of the best cases (the best case?) of transforming a US rust belt city into a 21st century City (Health care and robotics are among the key fields relevant to that success.). Finally, and most significantly, the issues raised in the article demonstrate that mastering new levels of data management do not lead to ease and simplicity but rather, to even greater opportunities and challenges.
In the case, of UPMC, getting a head start on developing Electronic Medical Records has led, first of all, to the challenge of coordinating independent systems across specialties. Being able to manage enormous amounts of granular data, and to understand how to deploy that data, is one of the frontiers beyond the systems coordination step:
Doctors now try to take a more holistic view of their patients, and that requires the ability to pull together data from different sources. Imaging data is separate from surgery notes, which is separate from pharmacy data.
“If we look at big data, the idea is how to interconnect multiple points of data across the broad, biological continuum,” Shrestha said. “If the patient is diabetic, you don’t just see an endocrinologist looking at the liver in terms of liver function tests or any scans but across the biological spectrum of organs and then down to a cellular level. We look at pathology slides, reports on molecular imaging and down to the genomic levels.”
Darrow explains that data can be broken down into three buckets: imaging data, which accounts for close to 50% of UPMC’s digital information; databases, which account for about 10%; and unstructured information, such as “postoperative notes, radiology reports, discharge summaries,” which accounts for the remaining 40%. The piece goes on to describe some of the specific technologies that are being used to address these various categories and concludes by pointing to another, even further frontier: the integrated management of pathology reports.
Big data, as the article in which I found the above reference would argue, is here to stay. The more we know, the more we that will become knowable. Personally, I find the challenge daunting and head-spinning. Do you?
I keep hoping we can make electronic records work, but to date, they suck. I always wonder who designs these things. They look like they were designed by computer geeks for computer geeks. Do they ever use real live medical people to help design them?
Steve
My own experiences with EMR were somewhat frustrating while, at the same time, very exciting. For example, when physicians used the prescription software, a patient’s prescription was waiting for them by the time they got to the pharmacy. In addition, when writing the prescription, you knew every other prescription the patient had received (in our system). Lab data were available where ever we were (home, clinic, or hospital). As image transmittal improved, images were available anywhere at any time. In addition, radiology images were no longer lost or unavailable (estimates were that 10 to 25 percent of radiology repeat images were made unnecessary. Because the computer can manipulate images, retaking the image because the technician made an error in exposure was eliminated (another 10 or more percent of images). Hospitals gained huge amounts of space since image storage in a computer takes a few square feet rather than a warehouse size room.
I found that doing quality assurance reviews was greatly improved and made easier.
The down sides mostly related to difficulty in moving written and transcribed information into the chart (progress notes, consultations, OR notes, etc). Few physicians are adept typists so either dictated or hand wrote their notes. In order to “blend” digital and analog data, everything had to be printed out. This meant that we tripled (at least) the amount of paper consumed and stored.
One small point about the article, she mentions how much data are generated by CT, MRI, ultrasound, and isotope scans. However, plain film xrays generate almost a magnitude larger data. CTs and the others are relatively low resolution where flat film (chest xray, mammagram, bone, etc) require high resolution. A chest xray generates about a megabyte of data. In the mid 80s when this all became feasible, data storage was pretty expensive and transmission was very expensive. Now data storage and transmission are very cheap. We would have needed a T1 line for digital radiology whereas now it would go through a 3G or 4G transmission.
On days when all of the stars are aligned, our system is often helpful fro some things. Other days, not so much. It is amazing how page after page of dietary info gets in the record, and things like cath reports and echo reports do not.
I have been pushing for what I call an active H&P to be included in every record. It should automatically include the latest lab tests, imaging studies and diagnostics. It should also include the most recent history and significant physical findings.
Steve
Thanks for the feedback. I admit that I have spent more time reading about EMR’s than using them. Also, when I have used them, the systems have appeared, as Steve notes, somewhat clunky and unimaginative. I think that there is a learning curve for software developers to get their products to approximate the actual thinking/needs of medical professionals. Unfortunately, that process can take years. Here’s my favorite past post on the subject: http://knowledgeapproach.com/2010/03/25/from-apple-to-apps-an-early-innovators-perspective-on-the-future-of-personal-health-records/#more-36
While the link above doesn’t answer all of your questions, Steve, it does point to the long arc of development ahead.
And, Edward, thanks for the point about x-rays.
My perception of the IT guys was that they came into HIS (health information systems) thinking it would be easy. Having no conception of just how complex health care is, they thought a few data bases would do the job. In actuality, there are numerous data bases, lots of narrative data, lots of historical data, literature (medical journals) data, and intuition all rolled up into minds somewhat like that idiot, Dr House except with some conception of courtesy and compassion that are essential to quality health care. The IT guys, after a couple months get very humble and begin to think that good doctors are rather remarkable.
Do not forget that real HIS will include rapid access to sites such as MDConsult where one can read texts and current journal literature real time. It isn’t much good having all the data but no help in interpreting it.
We were the beta site for DoD MIS and lab data were displayed immediately and permanently in several formats as desired by whoever was calling them up. I can’t believe that that is a problem at this point in other systems. As the chief of the medical staff, I could call up all the lab data a patient had had done for years and graph it and call the attending with any suggestions or questions (more questions than suggestions, generally). Lab results, radiology results, pharmacy, were incredible. H&P was not included at all nor were progress notes or other narrative info. Eventually, discharge summaries made it. I agree that those are essential to any real information system.
Thanks for sharing your personal experience as well as your point about MDConsult.
Also, I very much enjoyed your description of the IT-MD interaction. I’ve seen the same kind of initial confidence from IT fellows in other situations. Figuring out how to make that sort of interaction more seamless is what engages me professionally (technology marketing) as well as intellectually (considering how technology will enhance, or detract, from humanistic goals and endeavors).
RE: EMR–not sure what digitizing medical records will do for what I see as two of the most serious issues in that area:
(1) too many people have “legitimate” access to a patient’s medical information (by one doctor’s count ten years ago, 55 during a single hospital stay), and
(2) too few of those people actually READ the relevant parts of the medical info–Mr. Wired has a medical file the size of the Manhattan phone book, and I have yet to meet a doc who has even skimmed the whole thing, which is a shame, because his medical issues are very complex.
One huge advantage of EMR is being capable of sorting data and displaying on one screen those data one needs for a decision. For example, a chemotherapy doc can call up and display on one screen all the CBC (complete blood counts) done for the past several years. A radiologist can call up and compare every chest film in the file in seconds (used to take hours.. if all films were even available).
This sort of capability is critical for persons such as Mr Wired where the paper record is so huge and cumbersome as to make sorting impossible. I can’t tell you how many hours I spent in my residency searching for past lab reports on complex patients.
Access to data should be limited to those issued applicable passwords and need to know. One advantage of EMR is that the system keeps track of who accessed the record and when it was accessed. In the old paper records, 55 persons probably looked at records but no one knew that it had happened.