Making clinical decisions on facts, not 'feelings'
Meeting criteria easier with computerized records
What do utilization and quality managers dream about? They dream about standing at the physician's left hand with vast amounts of clinical information at their fingertips, ready to alert the physician to an alternative treatment if needed. They dream of improving quality at the point of service, not weeks or months later.
These dreams are coming true, in a sense, as computerized patient record storage and computer-assisted decision support systems start producing significant results, say officials at Salt Lake City-based LDS Hospital. LDS recently released the results of a seven-year study on how its computer-assisted decision support system helped improve antibiotic management in the hospital1. Here are some of the results of the January 1988 to December 1994 study:
* Antibiotic use decreased by 22.8%, even while the proportion of patients in the hospital receiving antibiotics increased from 31.8% to 53.1%.
* The percentage of surgical patients who received appropriately timed preoperative antibiotics increased from 40% to 99.1%. Antibiotics administered at two hours prior to surgery have proven to be the most effective.
* Antibiotic-associated adverse drug events decreased by 30%.
* Antimicrobial resistance patterns were stable, length of stay averages remained the same, and mortality rates decreased from 3.65% to 2.65%.
Financial outcomes were measured over the same time period, too:
* Hospital costs for antibiotics (adjusted for inflation) decreased from 24.8% ($987,547) of the pharmacy drug expenditure budget to 12.9% ($612,500).
* Antibiotic costs per treated patient (adjusted for inflation) decreased from $122.66 per patient to $51.90.
A decision-support system that provides information at the time physicians are making treatment decisions was the linchpin to success, says R. Scott Evans, MS, PhD, director of research for the Department of Clinical Epidemiology and a member of the Medical Informatics Department. The time sensitivity of such a system requires an integrated system that links all areas of the hospital and the physician offices, Evans says. "The more integrated information that you have, the more powerful it is," he says. (See successful systems require standardization, p. 112.)
Pharmacy, microbiology data integrated
For example, LDS has integrated pharmacy and microbiology information with regard to patients' medications. "If you have pharmacy data combined with microbiology data, then you can tell if people are having infections and are not being treated with appropriate antibiotics," Evans says. "If you have pharmacy data only or microbiology data only, then you are not able to do that."
Staff at all hospital levels collaborated on setting guidelines for the decision- support program and creating the integrated computer system. (See story on criteria development, p. 110.) Their efforts have resulted not only in clinical improvements but also in time savings, because chart reviews are handled by the computer.
At LDS, the computer checks to see who should receive antibiotics based on criteria such as lab results, coexisting diseases, severity of illness, vital statistics, and allergies. The computer assures that every prescription fits the profile and the prescription is appropriate before the order is made.
"If you have the integrated database -- with surgery, radiology, microbiology, urinalysis data and you have temperature and white [blood cell] counts -- you cover every reason why a patient should be on antibiotics," Evans says. "Then, the computer says, 'I don't see any reason why this patient should be on antibiotics.' If the physician chooses to ignore the alert, the system generates an alert with the pharmacist or quality assurance coordinator. On a typical day, the charts of five or six patients require review instead of hundreds."
Evans estimates it would take five full-time pharmacists doing chart reviews to handle the same volume of cases. And, unlike its human counterpart, the computer treats each case the same, he says.
Human chart reviewers are less effective in another way: Their alerts come after the fact. At LDS, on the other hand, every patient room and nursing station is equipped with a computer terminal. Orders are punched in on the floor or in the room. Physicians find out immediately if they have missed something or their prescription does not fit the guidelines and why.
Computerized records reduce blood use
New standards from the Joint Commission on Accreditation of Healthcare Organizations prompted other improvements that were achieved via computerized patient records. In 1989, the Joint Commission recommended that each hospital have a system of evaluating appropriateness of blood transfusions (JCAHO standard PI.3.2.3). To meet these standards, LDS uses a computer-assisted program to monitor the use of blood and blood products. It is called the Health Evaluation through Logical Processing system, developed by 3M Health Information Systems in Murray, UT.
The ability to track nearly every order for blood products has significantly changed blood ordering practices. A study of 13,082 transfusion orders for 5,847 consecutive patients showed over a one-year period:2
* a drop of 100% to 14% of orders requiring review by the quality assurance department;
* the average hematocrit reading of patients being transfused dropped from 28.6 to 25.5.
* accounting for blood use rose from about 25% to more than 98% of all orders.
Physicians and nurses must make blood orders through the system, specifying a reason for the order. The system compares the patient's clinical data with established criteria and, if there is a disceprancy, indicates the order is inappropriate. (See sample criteria, p. 112.)
The user may override the suggestion by specifying a reason, such as the patient is currently bleeding, but all overrides are reviewed by the quality assurance department, which also has access to laboratory and vital signs data at the time of the order. (See sample interface below.)
"We don't cut their fingers off," says Reed Gardner, PhD, professor of informatics at the University of Utah and co-director of medical computing at LDS Hospital. "But, we make them think about [their blood ordering] ... Now, there's a lot less blood used, and if there is less blood used, there is less risk to the patient."
A key to the success of the blood ordering system is the immediacy of the feedback to the physicians and nurses, Gardner says. Physicians and nurses are provided a record of blood orders within the past 24 hours -- and whether they have been dispensed or infused -- as well as the laboratory data for the past 48 hours.
With more traditional systems that rely on manual chart reviews, feedback on improper orders can take a month, says Gardner.
"That does not go too far toward changing behavior," Gardner says. "It's a nice historical record. You can go back and say, 'We didn't do very well last month,' and people can say, 'We sure didn't. We'll do better this month.' When you give the feedback instantaneously, it really makes a difference."
About one out of two times when an order is challenged by the computer because it does not meet criteria, the user decides not to order the blood product, Gardner reports.
Criteria grounded in reality
The medical staff at LDS set up the blood ordering criteria with typical cases in mind, not worst-case scenarios. One of the criteria for ordering blood is a hematocrit reading under 24. Many clinicians thought that was too low and posed a risk to the patients. Not disputing the opinions of the more conservative clinicians, Gardner looked back through three months of blood orders and found 30 patients with sub-21 hematocrit readings and found there was no complications resulting from the ordering guidelines. The one reported death in the group, which occurred in the emergency department, was a known drug user who died from an overdose before a transfusion could be made.
After Gardner's study, medical staff members, including some who voiced their reservations, more readily accepted the ordering criteria. "It's their set of rules, not from some computer guy or some blood-banking group. It's the operational, clinical staff of the hospital," he says.
LDS is working to reduce the number of units of blood ordered for total hip and open-heart surgeries using the data generated from the computer system. "Rather than order two or four units, we can go back and look at the statistics," Gardner says. "There might be only one time out of 200 [cases] when we used four units.
[Editor's note: For more information on the HELP system, contact 3M Health Information Systems, 575 W. Murray Blvd., Murray, UT 84157-0900; (801) 265-4400.]
References
1. Pestotnik SL, Classen DC, Evans RS, et al. Implementing antibiotic practice guidelines through computer-assisted decision support: clinical and financial outcomes. Ann Intern Med 1996; 124:884-890.
2. Lepage EF, Gardner RM, Laub RM, et al. Improving blood transfusion practice: role of a computerized hospital information system. Transfusion 1992; 32:253-259. *
You have reached your article limit for the month. Subscribe now to access this article plus other member-only content.
- Award-winning Medical Content
- Latest Advances & Development in Medicine
- Unbiased Content