All clinical decision support systems enable users to more effectively manage resources and productivity--but not all clinical decision support systems are created equal. The impetus to mine clinical data to business and clinical advantage has resulted in two separate decision support system types. Systems incorporating clinical data may be billed as clinical decision support systems whether they function by incorporating clinical data with financial information to derive costs and profitability information, or whether they are designed as active knowledge systems for use by caregivers nearer the point of care.
Cost- or care-based
The very name, clinical decision support system, can be fuzzy. Seemingly, if clinical data is incorporated into the system in any way, it qualifies as a clinical decision support system. Unfortunately, the name does not designate whether it is a system focused on business issues or on medical outcomes. Both types play significant but very different roles.
Business-focused systems tend to highlight cost accounting, cost allocation and fixed and variable costs, among other capabilities. Systems that focus on medical outcomes tend to use such terms as "knowledge-based" and "point-of-care." Although serving different purposes for different users, the two systems are not mutually exclusive. The closer the medically focused system is to the patient, the more opportunities for clean and direct data input to the repository, resulting in timely access to more accurate clinical data for use in business-focused applications.
Breaking traditional ties
The ties that bound traditional financial and administrative decision support systems to a single department--finance--and to a handful of users are broken. John Nunnelly, vice president and general manager of the Amherst Product Group, HBO & Company, Atlanta, reports that the move has been very rapid and has resulted in a radically different situation than just three to four years ago, as all members of top management--CFOs, CEOs and COOs--are now demanding the decision support system on their desktops. Once upon a time, such systems were used by the organization’s financial analyst who mined the data, ran the reports and delivered them to top management. Now it is not uncommon for organizations to have several hundred decision support system users.
Most institutions have already taken advantage of the operational gains through departmental staff management and resource management, says John Nunnelly. "The greatest opportunities now are in finding new, more efficient and more effective ways to treat patients and to direct them to appropriate settings of care. Unless the organization can take a broader, vertical view of the market, it can’t really see the opportunities to put patients in alternate, often less expensive settings."
Decision support systems have also become more user-friendly--many have a Windows look-and-feel common to other applications in daily use. As a side effect, this has resulted in the demise of the traditional executive information system, says Martin Grey, VP, JJO Enterprises, Miami, Fla., a decision support vendor. The whole concept of the executive information system is not cost efficient because, more often than not, the decision support system generated the information for it.
Financial and administrative decision support systems are well-established, allowing managers to attach value to the products and services and identify business opportunities through analyzing the data and past performance. Adding clinical data to the equation enables users to glean valuable risk-sharing and profiling information from organizational data, as well as to coordinate the often complicated work-flow requirements within the managed care environment. Market penetration for decision support systems among very large healthcare organizations is high, says Alec Karys, vice president of software development and technical services, InterQual, Inc., Marlborough, Mass., citing company statistics that indicate 80 percent to 90 percent of those organizations use the tools to manage resources and internal business operations.
Most vendors agree that the entrance of managed care into the local market, bringing a dramatic shift in revenue drivers, usually spearheads the implementation of a clinical decision support system that is capable of merging clinical data with financial data. But it is not the only one. If it is not managed care, it usually follows some other critical event that highlights the need for critical decision-making information, says Grey.
Although the business is much more complex due to the effects of managed care, the basic premises of healthcare decision support really haven’t changed dramatically, observes Nunnelly. An in-depth understanding of the most basic of business equations--revenue minus expenses equals margin--remains critical to running a successful business. Reflecting that rationale, his clients’ top reasons for purchasing systems tend to fall into four main categories, with cost control--accompanying revenue enhancement--and business process improvement topping the list. Prospective customers are also interested in broadening access to the decision support system across the organization and adding outcome management and measurement capabilities. However, even though these latter two factors are actively influencing purchasing decisions, Nunnelly reports that relatively few clients are implementing them.
Cost accounting functionality has regained its importance as a system function, says Grey, who attributes the change to the intense competition within the healthcare market. From a business perspective, the organization must know its costs and be able to assign them to the patient properly, he says.
Perhaps because clinical data is so intrinsic or perhaps because it has traditionally been collected for historical and liability reasons--not for strategic business planning--its impact on administration, quality of care and labor and expenses is difficult to quantify. Although Valerie Stevens, director of process, decision support services, Pittsburgh Mercy Health System, Pittsburgh, (see Case in Point, p. 34) suggests that it might be possible to attach dollars to some of the benchmark data from an operational and clinical perspective, she thinks it is very difficult, if not impossible, to measure all of the derived benefits from the clinical decision support system.
Care path-associated decision support systems, rather than traditional financial statement-type information systems, are in great demand at present, says Grey. And vendors are responding. Although support does not constitute an interactive system for point-of-care use, these systems can collect clinical data for ultimate use by care path developers within the organization. Political and cultural issues aside, many sites have been unable to build and implement them locally for lack of sound clinical data.
Demand for an enterprise master person identifier has mushroomed at clinical decision support vendor Transition Systems, Inc., Boston, where Randy Thomas, VP of corporate marketing, reports it as the most significant change in customer requests. Until very recently, organizations didn’t need to be able to track an individual across disparate systems. It doesn’t make sense--financially or culturally--to homogenize systems through replacement, she says, when it is possible to front-end disparate systems or provide wrap-around systems for extension across the enterprise.
Down the care path
While clinical decision support products that merge clinical data with financial and/or administrative data allow users to make decisions about clinical practice, they do not provide tools that get involved in the care process. Less than half of all vendors providing software solutions to the industry offer clinical content decision support systems, according to Karys.
The availability of rules and patient guidelines in electronic format unleashes the power of decision support tools for bedside--and near-bedside--applications to benefit the clinician and, ultimately, the patient at the time and point of need. Patient-focused systems’ users are nearer direct care and, ideally, so are clinical staff. The clinical decision support point-of-care model is designed to assess the needs of the individual patient and to assist the caregiver with a clinically based rationale for decision making at the point-of-care. Important for utilization and cost control, software tools can use patient-specific clinical criteria, provide prompts for key clinical management steps and verify therapeutic and diagnostic criteria to facilitate the referral process.
The standardization of clinical decision support criteria promises to bring objectivity and optimal standards of care into the caregiving environment. Once in an electronic, interactive form, these guidelines can be a key component of the electronic chart. The fact is, few of these knowledge-based systems are actually in use at the bedside. Julie Kees, VP of guideline development at the Institute of Healthcare Quality, Health Risk Management, Minneapolis, concedes that although some physicians are now using her company’s clinical guidelines at the point of care, they are not the majority of users. More commonly, the guidelines are used by the quality management department within the healthcare organization or case management and care management organizations. Karys says utilization review staff nurses are the primary users for InterQual’s tools. Once again, a step away from the patient.
The ideal user is, of course, the clinician at the point-of-care--as he or she evaluates the patient, orders lab tests, writes prescriptions and makes other treatment decisions. For optimal value in guidelines, emphasizes Kees, the tools need to be available at the time the clinician is contemplating care procedures. Credibility for patient guidelines is utmost, says Kees, who is seeing increased acceptance and credibility among providers. However, prescriptive pathways of care that outline patient treatment steps are very facility-specific and, consequently, are not well accepted by other facilities. For objectivity, Karys favors a nationally recognized standard that could provide a third-party external set of criteria for use as a benchmark.
An ongoing obstacle for guidelines proponents, as it is for others seeking to streamline the care and documentation process, is the reluctance by physicians to adopt any tool that might be characterized as interfering or might lower productivity. It is not a question of whether interactive decision support at the point-of-care will become an integral part of the medical practice, it is a question of when. Karys estimates that products meeting the technological challenge of not interfering with the task are still three to five years away.
Multiple pressure points are currently driving acceptance of clinical decision support tools, says Kees, one of which is certification organizations. Some have responded to physician concerns regarding decisions of coverage made by health plans, she notes, with clinical guideline mandates. The growth of programs from organizations such as the National Committee for Quality Assurance (NCQA) and Joint Commission on Accreditation of Healthcare Organizations (JCAHO) is bringing increased recognition of the need for evidence-based clinical decision support tools for the patient and as a basis for a quality improvement program.
Time expectations are pushing adoption from another direction. Reports built from last month’s data are too old; yesterday’s may be acceptable for strategic planning and review, but they’re not recent enough for many.
Richard Dick, PhD, chairman and CEO of ASCENTechnologies, a healthcare technology consulting firm, Alpine, Utah, is an active proponent of network-based, real-time clinical decision support. Data access across the enterprise and timeliness are major problems in implementing such an interactive system, he says. Some systems now emerging have embedded the medical logic in the application, making it unavailable to other clinical applications across the enterprise. Not only that, most of the applications available today are struggling to give a sub-second response time. The response time is certain to worsen, he predicts, with embedding another burdensome application on top of a computer-based patient record system or a data repository system.
Business at the bedside
Real value for the clinical decision support system comes with coupling strategic planning decision support with real-time clinical decision support, says Thomas. This concept blankets the organization with decision support capabilities from the finance department to the bedside. Organizations can implement protocols for the clinician at the bedside and deliver feedback in real time for variances. A step away from the bedside, users can review the variances in aggregate for trends. And, in the business office, senior level management have cost-based data analysis for historical review and business process re-engineering. This represents a classic continuous quality improvement cycle, she says, enhanced by real-time alerts.
Attaching dollars to real-time clinical decision support, Dick maintains that "Knowledge delivered in a rapid, sure-fire manner to the point-of-care can affect the bottom line very positively in addressing just three target areas--adverse drug events, prescribing appropriate medications and ordering appropriate laboratory tests." He further cites budget plans for one large medical center leader who intends to invest approximately $5 million per year in decision support technology because he expects yearly savings to exceed that amount.
Dick, and other advocates participating in the upcoming Third Annual Conference on Network-based Real-time Clinical Decision Support (NRCDS) Systems, propose to de-couple the clinical logic now embedded in the applications and make it available on a high-performance engine that sits on the network. The hope is to move clinical decision support more into the mainstream, says Dick. In addition to openly available clinical practice guidelines, he proposes open access to other medical knowledge bases to provide detailed and robust clinical support at the point of care. Right now, the availability of more robust standards is slowing the adoption of such real-time systems, he says. The absence of a standardized clinical vocabulary is a major issue remaining to be resolved. On a more positive note, a standard offering much promise in the area of real-time clinical alerts is the Arden Syntax (see Interop, p.17).
Mergers and consolidations have taken their toll on decision support, as they have on other systems in the healthcare organization. Many potential customers must realign their foundation systems before beginning the process of aligning more precise decision-making tools. Craig Castro, VP and CIO, St. Agnes Medical Center, a member of Holy Cross Health Systems, South Bend, Ind., is well-versed in that process (see Case in Point, p. 38).
As one of the few enterprise-wide systems in the organization, the clinical decision support system relies heavily on leadership and direction from top managers to achieve maximum effectiveness, says Thomas. "When the decision support application is most successful, senior-level management lead the way."
CASE IN POINT
Building a Physician Support Team
Well before it became commonplace,JoAnn Narduzzi, MD, executive VP of medical management, became interested in incorporating clinical data into business decisions. In 1989, when clinical decision support was still in its infancy at Pittsburgh Mercy Health System, Pittsburgh, she created a basic physician’s report card defining cost per case and length of stay. Few were interested then, but times have changed. Now she and Valerie Stevens, director of process, decision support services, are leading the Pittsburgh Mercy Health System--acute care hospitals, long-term care and skilled nursing facilities, home healthcare, even the behavioral health arm of the organization--into a new decision support era.
The turning point for the institution and for Narduzzi came in 1992 with a job change. As vice president of medical affairs, she suddenly had much more access to clinical data. As her awareness of the value of the data increased, so did her use of the data. Now she uses it almost daily to look at admissions, discharges and transfers and trending--by unit, by physician and by hospital. The advantages are significant, she says. Rather than discovering a problem after five or six months, changes--and challenges--emerge almost daily.
Like many other institutions, decision support at Pittsburgh Mercy began as a finance product, using very little clinical information. "The true value from decision support came with the merger of clinical and financial data," says Stevens. "We have now gotten it to the point where decision support addresses the critical success factors of the organization, not only from an operational standpoint, but from administrative, labor and expense and quality of care standpoints. These are not systems that can reside in one area; they have to meet the needs of the entire organization."
Longtime users of HBO & Company’s TRENDSTAR and an early adopter of the company’s recently released Pathways Decision Support product, Stevens sees great advantages in the "coming together" of decision support data and the opportunity to move it onto the desktops of the community. About 120 have used the new system, but Stevens anticipates that closer to 300 will ultimately be able to take advantage of system functions.
Power behind pathways
Clinical pathways are a major emphasis at Pittsburgh Mercy--even the institution’s quality programs that are based on clinical pathways. Beginning in 1993, a pathways development program targeted the 40 diagnosis related groups that comprised over half of the institution’s business and 65% of revenue. "The clinical decision support system has been the driving force behind our pathways," says Narduzzi, who uses the system as a major source of her information to derive evaluations on patient progress, service utilization, service units, cost per case, length of stay and variation by physician and specialty practice. Regarding physician compliance, she says, "External pressures have helped our internal processes."
External pressures may be one driving force, but Stevens adds, "I firmly believe that the success of these products has been because there has been a physician champion as well as a business champion." "With Dr. Narduzzi as a clinical champion, it was possible to begin the process of getting data into the physician community."
Clean data, clean slate
Gaining support from the physician community demanded a sound strategic offensive. Narduzzi cautions the importance of data presentation and interpretation. Perhaps the greatest battle a physician must fight on proposed implementation of such a system is fellow-physicians’ in-depth interrogations. "When you first present the data," she counsels, "the data must be absolutely clean and reproducible. You must be able to defend every piece of information presented because when you present it, you are going to be questioned and questioned. You must be able to defend every small piece of information that is there. Physicians will challenge you and if you cannot respond to the challenge, they won’t believe the data. If they don’t believe the data, they will never use it."
The challenge of presenting physician profiling data to physicians requires a great deal of diplomacy and diligent follow-ups. Physicians want to see the results, but they are afraid of them, notes Narduzzi. Expect them to challenge every issue. She advises addressing each and every challenge, then following up on each of them--getting back to every physician with every one of his or her questions and complaints. When Narduzzi does her second presentation, the physicians also challenge the data, but to a lesser degree. From that point, the veracity of her data is not in question.
As managed care has become more and more a part of Pittsburgh Mercy’s existence, Narduzzi reports that physicians’ quest for data has become astronomical. "We can hardly keep up with their requests for data now that they know they’re being judged. In the beginning, we wanted to prepare them for the kind of report cards they would be receiving from payors. We told them that outside parties were looking at their data, but they didn’t believe us. I think they believe us now."
Quantifying benefits has been very difficult with a system that is so ingrained in the organizational culture. Stevens says that the introduction of other types of benchmark data within the past couple of years has presented some opportunities to attach dollars from an operational and clinical perspective. "Although I don’t believe we can get our hands around what decision support data has done for this institution, I believe it is responsible for our business success. I think hospitals that do not have access and cannot evaluate their results in a timely manner are in trouble."
"We have taken decision support to the next level," says Stevens. The decision support system has enabled us to create a re-engineering area, and much of the process reworking and redesign is based on the decision support analysis. "It is important that people realize that decision support is all-encompassing--it is not just a financial piece, it is not just a clinical piece, it’s not just a productivity piece: it’s everything."
CASE IN POINT
Building The Role Model: Self-Development or Vendor Solution?
When Craig Castro, VP and CIO atSaint Agnes Medical Center, Fresno, Calif., struck out to find a decision support system that could meet the medical center’s needs in managed care country, he did not expect such a rough and rocky road to implementation. St. Agnes, the only member of the Holy Cross Health System Corp., in South Bend, Ind., is unique, requires more clinical, more benchmark and more patient satisfaction data than any of the Holy Cross Health System’s other nine member organizations located in other parts of the U.S.
St. Agnes has depended heavily on Health System’s decision support system--a home-grown financial/ administration system. The decade-old system has been dependent on technical staff located at corporate headquarters in South Bend for all development, maintenance and support. So long as the corporation used the common financially based system across its member organizations, compatibility between clinical systems wasn’t really important. However, an increased demand to input clinical data drastically changed the situation. Standardization of clinical data was very difficult. That, combined with the increased demands related to managed care, eventually brought the corporation to the realization that it was unable to keep up with commercial software development and support. "Our health system stock is not traded by our software development or system building capabilities," says Castro. "It’s how we use the systems to better our business."
Stop and go
Since St. Agnes was the dominant user in the health system, it was tapped to find, evaluate, negotiate, install and implement a system that might be appropriate for use throughout the greater organization. The need for a new decision support system was never in question, but the road to implementation was replete with dead-ends. Top management pulled the plug on the first system as it was ready to "go live," fearful that the technology was not advanced enough for use across the health system. Regrouping, Castro and his team reconsidered self development and submitted its criteria to the health system. Six months later, little progress had been made. "Self development may give you flexibility," he says, "but that is not our core competency."
After recognizing the shortcomings of an in-house development and support team, St. Agnes got another chance at a commercial product. The timing was good. The vendor selected on the first go-round, HBO & Company, now had a pilot for its new Pathways Decision Support System. "We knew there was a risk to being an early adopter," says Castro, "but we thought we could influence the product by being first out of the chute."
The next step, very important to the process, was to ensure that St. Agnes would not be an anomaly in the health system. Unlike the first false start, St. Agnes has a fully committed corporation behind the project. The corporate office has funded the project, has added three staff members in finance to bring cost accounting systems up to speed and is staffing database administration and all interfaces from the corporate office. In addition, if Pathways proves itself at St. Agnes, Holy Cross Health System plans to exit the decision support business and deploy the Pathways system throughout the organization.
A common well of clean data
One of the goals for the organization is ownership of a common repository with clean data feeds. It will be a retrospective repository (not a point-of-care interactive system) and, although it will extract clinical data from point-of-care clinical systems such as the emergency room and operating room, Castro does not plan to get closer than a 24-hour window, mainly because of extraction modes and feeder system operations. The master plan is to build clinical pathways that get next-day review--once sound data is available. "We have had difficulty building pathways because our current data is slanted for charge and financial applications and does not accurately reflect actual clinical interventions."
Castro has itemized benefits St. Agnes expects to gain from its new system. Three are fairly standard: manage clinical, cost and satisfaction across the continuum; integrate financial, clinical and market-based data; and improve and distribute data analysis and reporting capabilities. The fourth, to move to commercial software and support, however, tells of a path well trodden. "Development is time-consuming and capital intensive," says Castro, "but ongoing maintenance will kill you if you don’t commit staff to it. We couldn’t keep up."
The Pathways system went live in January, expanding its user base from a small group of trial users to a "super-user" group of service line managers, and users in medical records, strategic planning and financial departments. Roll-out to an anticipated user base of 20 to 30 is scheduled for second quarter 1998.
Charlene Marietti is senior technology writer at Healthcare Informatics.