Cookie-cutter EHRs and Physician Productivity

In 1993 my institution implemented its first Hospital Information System (HIS). At the time the facility was large and busy. Over 400 were employed filing the medical records and there were about 50 people working on the computer systems. Everything required a paper form and each form require committee approval (a 6-12 month process). Every physician order was written in the notes, written on an order sheet, transcribed to the Kardex, and hand-carried (later faxed) to the performing area. The procedures that now make up HEDIS and other "quality" measures were important then, as now, and the staff made valiant efforts to remember to do everything that they were supposed to do - and - unless you maintained your own list of things to follow-up, by the time the results were ready, you were hard pressed to remember what to check.

The institution now occupies a new $1 billion facility. It is about half the size and sees about half as many patients as 1993. The number of doctors is up about 15 percent. The number of staff in medical records is about 50 percent lower but the number of IT staff is up by about 500 percent. Physician charting is now a hodge-podge of typed notes, dictation, point-and-click applications that get uploaded to the HIS in various ways. Some still chart on paper which is scanned. Forms are still ubiquitous and still require approval by the same committee. Many orders require a paper worksheet that the nurse uses when entering the order into the computer. The pharmacy still does not accept medication orders electronically. In addition to charting treatments in the traditional way, many must be entered in duplicate into other systems.

In 1993, if the patient's chart could be located you would have access to all of the notes and results. Now, portions of the chart are now stored in so many different places that the institution has published a matrix describing which of a half-dozen sources might need to be consulted depending on the patient type and visit type. There are no automated alerts or reminders; the staff must still remember to do everything that they are supposed to do. Since 1993, expenditures on IT have been in excess of $200 million.

In short, more people are working harder to see fewer patients; the institution is spending more on overhead and accomplishing less.

One could argue that the institution cited above is particularly dysfunctional and that the average situation is better. Kaiser Permanente is a leader in employing computer technology on a large scale to facilitate patient care. They have committed or spent in excess of $2 billion on their new EHR and have probably achieved some operational efficiencies and cost savings. The technology has not enabled their providers to see more patients than before; the productivity standard in some areas has been lowered from four visits per hour to three. A number of the practitioners spend several uncompensated hours per day completing the charting that they didn't have time to do during normal clinic hours.

Poor return on investment is not just a feature of gigantic organizations. It is not unusual to discover that a small to mid-size facility has spent in excess of $50 million implementing an EHR, however, one rarely if ever hears of any dramatic cost savings or productivity gains that resulted from the expenditure.

What has caused this gigantic disparity between expectation and reality? I blame high hopes coupled with inadequate experience and technical knowledge to temper them. The managers of healthcare organizations have witnessed the success that FedEx, UPS, Amazon.com, and many other organizations have had at creating efficiency and profit by the aggressive application of Information Technology and extrapolated, inappropriately, that if they adopted the technology available to healthcare, they would experience the same sort of business success.

Why "inappropriately?" Because, while the business processes and workflows that these other businesses were trying to improve are complex, the data needed to drive the processes is not. Furthermore, the programming languages, databases, and other software that are available to developers today are targeted at exactly this sort of simple data. Simply put, the development tools and methods are well matched to the problem.

In healthcare these happy circumstances to not apply. Healthcare processes themselves are vastly more complex than those found in the typical business environment. The complexity of the information itself is essentially infinite and totally unlike the simple data elements needed to describe a bank deposit or purchase at a Web-based store. There have been interesting projects that have explored specific facilities that programming languages and databases should provide to facilitate the development of medical application but none of these has been incorporated into a mainstream language or an industry standard database product.

It's not surprising, therefore, to find that the typical EHR looks and feels like a transaction processing application found in a bank. They are, after all, built out of the same raw material, by people who think that is how computer software should look and function. Unfortunately, when you put the end product in the hands of physicians, most of them find it awkward, illogical, constraining, and generally unpleasant to use. When you require them to use it you turn them into clerks. Is it any wonder their productivity goes down?

What must be done is to create an environment where "outside the box" thinking can flourish. There must be a mechanism to encourage those who can imagine what could be or what should be to get to work inventing or transforming the technology necessary to realize their vision. While every attempt will not be successful, many will, and some will be memorable. Consider Watt, Edison, Benz, Einstein, and Tim Berners-Lee.

To achieve this goal in a short time-frame will require something akin to DARPA or the Apollo Project - government support for radical innovation coupled with incentives to industry to implement the products of the research. The solution to the problem is not to mandate that everyone work under the same handicap.