Imagine a small community hospital in a tree-lined Detroit suburb, and you might think of bake sales and volunteer candy stripers – more "General Hospital" than "St. Elsewhere." What you might not expect is a computer room on the sixth floor filled with networked NeXT workstations. : Yet Mt. Clemens General Hospital is on the leading edge when it comes to administrative medical technology. The hospital has tabbed NeXT for its plan to cut costs and improve patient care by developing its own in-house medical-records system.
The first phase, with 100 NeXT machines used for administration and admitting, will go on-line in January of 1993. : "We think we are unique in our comprehensive strategy," says Vimal Chowdry, CFO of Mt. Clemens Hospital. "We're developing a system internally and it is very cost effective. The total budget, including NeXTstations, is less than buying a text-based system with dumb terminals."
As politicians and economists debate proposals for affordable health care in this election year, a growing cadre of far-sighted medical technologists have turned to NeXT as part of the solution. But medicine is not a market NeXT cultivated. "The way we found them is that they found us," said Steve Jobs at NeXTWORLD Expo in January. : Hospitals have a unique set of priorities. They need a competitive advantage in an increasingly tight market, but patient care must remain their primary consideration. The medical professionals who have decided to go with NeXT systems are convinced that the combination of its graphical user interface and powerful UNIX operating system will serve both requirements equally. : Third-party developers and major customers believe the NeXT will not only beat current mainframe systems in speed and connectivity, but that its object-oriented programming environment will permit painless adjustments to new and changing standards without systems being rewritten or administrative costs going up.
The biggest area for increased productivity is patient records, but the NeXT is also finding uses in imaging and diagnostics.
Glen Carbon Corporation, Champaign, Illinois: Emergency-room physician Rick Pionkowsky got so interested in the challenge of designing an interface that would appeal to doctors that he went back to school and got a master's in computer science. Once he saw a NeXT, he believed such an interface was possible. Glen Carbon's modular system will go into beta testing soon in the osteopathic hospital at Rush University/St. John's Medical Center in Chicago.
Logibec, Ile des Soeurs, Quebec, Canada: This database corporation has been in the business of medical-records systems for over ten years, using re-lational data-bases such as Progress and Sybase. It also developed DB Toolkit for the NeXT, an SQL-oriented precursor of NeXT's own DBKit. Logibec's NeXT-based system, called Clinibase, is beta testing at the Alberta Children's Hospital in Calgary. The software's bilingual interface and documentation is expected to give the company an edge in the Europe market.
Oceania Health Care Systems, Palo Alto, California: Two emergency-room doctors, a retired medical professor, and one savvy businessman have set out to create a system that organizes medical records like their paper counterparts. The beta site, San Diego Hospice, went on-line in August with 40 computers for administrative use. Oceania also has plans to involve portable computers for clinical use.
Tecor, Santa Cruz, California: Steve Hurwitz spent five years creating medical software on the Macintosh and Sun workstations as part of his medical-consulting firm's work. His last consulting client needed a custom program combining medical records and accounting features. The NeXT-based system he developed to meet their needs provided the foundation for a new company that is spawning beta sites in Chicago, Saudi Arabia, and Spokane, Washington.
NeXT itself has recognized that health care represents an important market opportunity. Early this year, the company held a conference for medical developers to discuss the issue of standards and compatibility between medical applications. Over 92 people attended from 36 medical-development firms.
"The first thing for them to realize was that they weren't alone," said Ken Rosen, NeXT's manager for emerging markets. And while most participants expressed an interest in compatibility between their products, no one has made the first move; each company intends to fulfill its own vision of medical-record systems before attempting integration.
But the world of medicine is alarmingly arcane. Hospitals were among the first to recognize the possible benefits of computerized record keeping, purchasing mainframes along with banks and the government as early as the 1940s. Lean times mean tight budgets, however, and improvements to facilities and staff provide more tangible advantages than a system upgrade or new hardware. Nowadays, few U.S. hospitals are without a collection of mainframes, minis, and micros. But they are hopelessly outdated and disorganized. They've been wired and rewired with a spaghetti of LANs, and few systems work well with each other, even within the same hospital.
The other main hurdle in the medical field is what Brian Rosen of VISUS, a Pittsburgh-based NeXTSTEP developer, calls the "fiddle-faddle factor": "If doctors can't figure it out in five to ten minutes, forget it." Most of today's systems require days, not minutes, of training. And as any business that maintains data knows, it doesn't matter how easy it is to get the information out if it doesn't get entered in the first place. Most developers working in this field agree that a choice of interface style is crucial if the systems are to adapt to physicians' personal idiosyncracies.
Under the current system, data-entry clerks type about 700 two-minute phone messages about patient status, medications, or tests into the hospital's records system every day. That's because most doctors don't type – they just phone in and leave all their messages with a message center.
It doesn't work very well: Few doctors call promptly after an examination, which means delays in updating a patient's record. A specialist called in for consultation may mistakenly order tests the patient had an hour before, or prescribe medication that reacts with what the patient is already taking. But so far, leaving voice messages on the hospital's computerized answering machine is a hospital's best bet, because doctors can't read each others' handwriting any better than anyone else can.
Typically, each hospital office, lab, and pharmacy has its own, highly specialized computer system. Each system must be separately logged in to and out of, one at a time, from a terminal at the nurse's station. There's no simple way to see the big picture: a complete record of each patient. "It isn't designed to support the decision-making process of the physician," says Dr. Doug Will, chief of staff at the Loma Linda Medical Center in California.
"Our goal is to create a united patient record," says developer Rick Pionkowski.
The new DBKit will make manipulating remote databases, like those in pharmacies and labs, much easier. Three of the five sites developing medical NeXT systems – Mt. Clemens, Oceania, and Logibec – plan to build their servers from hardware made by Sun Microsystems, accessing data stored in Sybase databases over Ethernet. It's a combination that makes perfect sense, says Pete Skinner of Oceania. By combining Sun with NeXT, he says, Oceania can take advantage of the best new technology available for both workstations and servers.
Not only can the NeXT talk to the database server, it can also communicate with all of the other networks already in place around the hospital, including Novell LANs, UNIX TCP/IP, and AppleTalk. Linking these systems might be the road to a paperless hospital. All information from medical procedures and doctor visits would be sent directly to the billing system, which would itemize direct and indirect costs. Prescriptions would be sent to the pharmacy system, and stat results from the lab would ap-pear on the patient record as soon as they were finished.
None of this addresses Rosen's "fiddle-faddle factor," however – the toughest problem to overcome and the key to success in medical-information systems. Improvements have been introduced before but have failed to capture doctors' attention. NeXT developers believe they can build systems that are responsive to each doctor's preferences and individual style of practice yet integrate into the hospital's overall information structure. "The primary customer of a hospital is the doctor," says Chowdry.
Attempts to beat fiddle-faddle vary enormously, but all focus on creating options for data entry. The system at Mt. Clemens will be used administratively at first, so keyboard entry is crucial. There, the most critical items for ease of use are pop-up lists generally associated with codes, doctor names, or medications. Each list narrows as you type in letters, finally entering the final choice and returning to the entry view. The pop-up lists are based on code written for the hospital's mainframe application; on a NeXT, the pop-up feature becomes an object that can be used over and over again.
Oceania's approach might feel more familiar for the practicing doctor. A window opens for each patient with a series of tabs at the bottom, replicating the tabs from a standard medical file. Clicking on a tab switches the user to that portion of the record. Tecor's Advantage, on the other hand, starts with a Digital Librarian–like patient record that spins off billing, insurance, history, and other information from the Shelf, with each item creating a new window.
Beyond pure administrative tasks, NeXT technology is also playing a role in medical diagnostics such as radiology and MRI (magnetic-reso-nance imaging). The NeXT's Display PostScript capabilities make digital imaging not only feasible but desirable. Several programs under development include the ability to enhance a scanned X-ray with color, sketches, or personal notes, while keeping the original image intact. Hardware companies are developing scanners for large-format film, microfilm, and micro-fiche, as well as digital imaging to incorporate currently used systems.
Chowdry of Mt. Clemens points out that the cost of film still beats the equivalent of disk space by $5 or more. But tape archiving costs less than 4 cents per image and will last considerably longer than the current two years for film. Administrative savings are also great. Under the old system, doctors requesting an image from the hospital X-ray library had to wait anywhere from a few hours to a day or longer. Jukebox-type devices holding multiple years' worth of archived images could provide fast access, while current images could be stored on disk for immediate retrieval.
Eventually, high-resolution imaging will make surgical planning and training tools much more effective. These tools already allow doctors to plan their use of microscopic instruments inside the body before making an incision, decreasing both the time spent in the operating room and the margin for error. They also give interns a chance to hone their skills. The high resolution would make for much greater accuracy.
"The key to success is acceptance of the product by end users," says Pete Skinner of Oceania. "Common wisdom says that docs will never use computers. The challange is to create a compelling user interface."
The history of computers in medicine has shown that hospitals aren't as interested in upkeep of proprietary systems as they are in effective information technology. Chowdry's decision to go with NeXT is a simple one: "If we are doing a system that the hospital must use for the next 12 to 15 years, do we want to do it text-based or with a GUI?"
Kristin Dyer is an assistant editor at NeXTWORLD and a healthy cyberspace kid. She can be reached at firstname.lastname@example.org.