If information systems are installed in sync with hospital operations, the return on investment can add up. In the case of Brigham and Women's Hospital, there's a lot of adding to do.
Ask for an accounting of computer benefits and you get a list of 40 separate initiatives in which a little information interjected at a critical moment has made a big difference.
By the way, it's important to get the latest version of that list because clinicians are thinking up new return-generating uses of the computer systems all the time.
The software system's development began in the late 1980s with a goal of harnessing computer processing power and making desktop computers vigilant assistants to physicians and other providers, says Jonathan Teich, M.D., director of Brigham's Center for Applied Medical Information Systems and Research.
As the home-grown software programs began to flex their muscles, "there clearly was a point where the power of the systems was really recognized, and the challenge now is keeping up with the demand (for new uses)," says Mary Finlay, corporate director of information systems.
The current list includes 21 computerized interventions that are saving a collective $7 million per year. Another dozen initiatives picked for their potential are still under study for quantifiable benefits. A few others have benefits that aren't quantifiable but add timely information judged valuable to improving or deftly changing the course of treatments.
The basis for much of the computer benefit is an order-entry system that handles about 12,000 orders per day at 7,000 personal computers in the hospital and at a growing number of clinics and physician offices, Teich says.
Designed to support the way clinicians do their work, the system has lured physicians to the computer workstations -- about 85% of all orders are entered by residents or attending physicians at the teaching hospital.
By getting physicians to direct their attention to computer screens on a regular basis, Brigham managers have put their key targets of information right where they want them. Computer programs present clinical feedback and gentle nudges on ordering preferences, triggered by what doctors order or ask.
"Surveillance programs" keep track of a patient's status throughout the institution by keying on orders entered, tests reported and observations made. Computers use databases of medical knowledge and patient history to analyze new data on patients and then prompt physicians to consider doing something in response.
The impact can be substantial. One program warns if a patient is allergic to the drug ordered -- an alert triggered about 1,500 times per year, according to Brigham's records. About 70% of orders are canceled after the warning, preventing an estimated 40 adverse events per year and $250,000 in costs of extended lengths of stay, additional tests and countermeasures.
Another program checks for hazardous interactions between the ordered drug and other drugs the patient is taking. Brigham estimates it saves $160,000 a year by avoiding the cost of treatment consequences.
Some programs intervene specifically to control costs. One shows charges for laboratory tests being ordered and suggests less expensive orders, on the hunch that physicians may choose a less expensive alternative once aware of the relative costs. The result: a 5% reduction in expensive tests. At an annual 700,000 orders, the savings amount to $1 million per year.
Another initiative is saving an estimated $500,000 to $1 million a year by reducing the use of expensive intravenous medications where equivalent oral medications can be taken.
Teich says intravenous drugs are ordered for incoming patients who are very sick and cannot swallow medicine. But at some point a patient gets better and can be taken off the IV bag. A computer program prowls nightly through reports to "find" patients still on IV medication who can take drugs orally. A dietary order, for example, would indicate that a patient is well enough to eat and therefore able to wash down a pill, Teich says.
About 20,000 such instances are displayed each year, of which 25% result in a change to oral medications.
A newer intervention customizes doses of drugs for individual patients to prevent adverse effects, reaping $640,000 in annual savings as a dividend. For classes of drugs that can put a strain on kidneys, a software calculation ensures the drugs "are properly adjusted for people whose kidneys aren't working very well," Teich says.
Normally such fine adjustments for a patient's age and renal function are too complex for caregivers to make along with so many other care duties, he says. But "the computer is very good at remembering very complex rules." Brigham estimated the intervention has the potential to prevent about 100 adverse events a year.
Brigham's proliferating interventions come mainly from ongoing care improvement teams that tackle specific diagnoses in pursuit of better processes of care, Finlay says. Clinicians work in concert with computer analysts to implement treatment improvements with a software assist, she says.
Brigham spends $3 million to $4 million annually on new information systems development, choosing proposed projects that have the most impact and are the most feasible. With the pump primed, decisions on computerization are less a concern about return on investment and "more of a question of how much can we handle in a given year," Finlay says.
And although much of the return can be measured in dollars, Brigham's approach from the start has been to target clinical improvement and let the savings fall into place. Conceived a decade ago, Teich says the idea was based on "a strong visceral belief that clinical systems could be beneficial -- long before we had these tables to prove it."
Besides, some benefits are not measured in anything but clinical terms. One, for example, displays lab results relevant to medications being ordered -- solely to prevent errors that may stem from overlooking those results when selecting a medication.
Another program taps into a computer-logic database, looks for abnormal readings on lab tests -- known as "panic lab results" -- and compares them with other patient information to decide whether it signals a crisis, Teich says. If so, it alerts caregivers so something can be done right away.
Dollar benefits still are under study. But the alarm has gone off 10,000 times a year to mobilize forces on behalf of a patient in distress. And the median response time has plummeted by two-thirds -- to 42 minutes from 126.