College of American Pathologists
Printable Version

  Middleware sparks renewed ROI interest


cap today



February 2007
Feature Story

Anne Paxton

Return on investment is a classic finance ratio—money gained or lost compared with money invested—that evolved long before there was such a thing as information technology. When companies calculate ROI, they are usually talking about capital outlays for nuts-and-bolts products that—as the expression goes—when you drop them on your foot, it hurts.

It may not take a giant leap of faith for laboratories to spend money on a laboratory information system or other software. But the means to link savings or revenues to the purchase generally haven’t been there.

With the growing popularity of middleware—most frequently defined as software modules that sit between laboratory instruments and the LIS—it’s getting much easier to make that link, says Bruce A. Friedman, MD, a well-known IT consultant who coined the term “pathology informatics” in 1990.

“It’s very difficult to assign a dollar value to much of the functionality and benefits flowing from a classic LIS, which is a broad suite of applications including test ordering, test resulting, and quality control,” says Dr. Friedman, active professor emeritus, University of Michigan Medical School.

“You can propose a figure and then defend it by stating all the assumptions that entered into the calculations, but people are always going to resist the more subjective of them. Let’s say you can definitively show that automating laboratory processes with an LIS results in a reduction of laboratory errors by eight percent, so then what is the dollar value of avoiding one error?”

But now that more laboratories are sticking with their older LISs and just buying middleware—which has more focused and frequently more objective benefits—“there has been a resurgence of interest in return on investment calculations for laboratory software,” Dr. Friedman says.

One of the primary functions of middleware, autoverification, is a good example. “With autoverification, much of the checking of test results immediately prior to reporting on an automated line in hematology or chemistry that previously required a med tech disappears. You can just do a ‘before and after’ comparison and calculate the lesser manpower requirements couched in terms of FTEs [full-time equivalents], which translates very quickly into financial benefits.”

Personnel savings like that, or reduced cost of materials—by avoiding rejects on assembly lines, for example—are objective savings, he says. The value of a safety measure or an “error that is avoided” is a more subjective judgment.

Unfortunately, at most hospitals, the laboratory’s feet often get held to the ROI fire while other cost centers may appear to get a free pass, he notes. “When the CIOs and CEOs are writing checks for their personal projects, they don’t feel as cost-constrained. An EMR [electronic medical record] system, for example, may cost $60 to $80 million, and very rarely do they ever do an ROI calculation on that. They assume that the need is so obvious that it does not require financial justification.”

In its most pristine form, ROI represents costs avoided or revenues gained as a result of purchases of technology. “But because so many of the benefits of a broad multi-function LIS are subjective—quality and safety issues upon which it’s difficult to place a dollar value—your ROI calculation becomes suspect in a lot of people’s eyes.” Including his eyes, he says. In many cases Dr. Friedman has “gone through the motions” of calculating an ROI simply because the CIO required it.

Calculating ROI for the popular category of middleware known as laboratory portals—Web-enabled software used with outreach programs and reference laboratories to facilitate electronic order entry and results reporting from physicians’ offices—is relatively straightforward, he says.

“If a physician’s office were using paper requisitions for test ordering and you were faxing or couriering back to them hard copies of test results, when you deploy a lab portal software such as that offered by Atlas Medical or, your costs of those same transactions will almost certainly be lower. Although you have costs for the new technology, you’ll have fewer courier stops and cleaner orders.” The orders are cleaner, he adds, because the system will not accept an incomplete order and can check necessary information such as the insurance coverage of a patient.

The bottom line for the laboratory is ROI and payback—the time it takes to recover your costs for a software system, Dr. Friedman says. “For example, if you get a middleware module that costs $50,000 and your ROI calculations say that costs avoided or additional revenue is $25,000 a year, then your payback period is two years. To obtain an ROI of a year is considered very desirable. But when you are dealing with a vendor you can ask: Do you have other clients with similar configurations, and can you share with me their ROI calculations?” It is then up to the potential purchaser of the product to assess the validity of the ROI calculations of the other lab, Dr. Friedman says.

Jacques Baudin, executive vice president of middleware vendor Technidata America Medical Software, Tucson, Ariz., says he has not yet seen much demand for ROI calculations from potential customers. “The reason people believe it’s easier to calculate ROI with middleware is because it is usually more ‘intra-lab’ related, and therefore more measurable,” he says. “With an LIS, there are many parameters outside the control of the laboratory, relating to hospital management and logistics, collection of specimens, etc., that make it more difficult to calculate a direct ROI.”

“Middleware is really designed to address productivity issues inside the laboratory, and the laboratory is in control of most of those issues,” Baudin says. But a good ROI is not generated from a good software solution alone; good workflow and processes and optimized lab reengineering is essential to creating the maximum value, he adds.

He finds that when laboratories say they are considering a Technidata solution and appeal to him to supply cost justification, “it’s more a matter of working as a team in developing a sound and specific business case than providing ROI numbers from other organizations that may not be relevant.” More and more people are convinced of the benefits—operating, business, and intangible or strategic—of middleware solutions, he says.

The laboratory at Raritan Bay Medical Center, Old Bridge, NJ, which performs a million tests a year, was one of the first in the nation to use autoverification starting in 1997. But it was not a middleware module; it was included as an option in the laboratory’s LIS, initially a Siemens system and then a Misys system beginning in 2002.

The impact on technologist workload and turnaround time has been impressive, says laboratory director of operations Leonard K. Dunikoski, PhD, DABCC, CHE. But he estimates that only about 50 percent of U.S. hospitals have moved to autoverification.

“Most of the middleware that has been available so far has been used only for one vendor’s instruments. For example, you couldn’t use middleware to do autoverification on a Beckman and a Bayer instrument; it would only handle the Bayer instrument. I think that’s changing, and some new middleware is coming out with open systems so you could put any vendor’s instrument through it.”

Pressed by intense competition in New Jersey, Raritan Bay planned several years ago to make a transition to an automated track system, but in the end the laboratory tabled the idea because of ROI concerns.

For an investment in new hardware or software, Dr. Dunikoski says, “the financial people at Raritan Bay expect a payback in 30 months. I don’t know where that number came from, in terms of a decision to buy or not buy, but that’s the number they most frequently use. Ten years ago it might have been a five-year payback, but I think their feeling is now the technology is changing so rapidly you can’t put your eggs in a five-year basket, because in three years you might need a new product.”

In instrumentation, “the laboratory is one of the hospital departments that is particularly constrained by ROI issues,” Dr. Dunikoski notes. “It’s harder to make a case that a new laboratory instrument will result in more admissions or higher outpatient volume, but that’s not the case in diagnostic imaging and cardiology, where physicians see significant improvements in the quality of results in MRIs and CT scans.”

Kenneth E. Blick, PhD, sees middleware as the development that will finally help fulfill the promise of automation in the laboratory. Laboratories have spent millions and millions of dollars on laboratory information systems, but they still have a lot of Post-It notes on their computer terminal screens, says Dr. Blick, professor of pathology and director of clinical chemistry, the University of Oklahoma Health Sciences Center. “So the whole notion of getting more value out of software in the laboratory is an idea whose time has come.”

He understands that LIS vendors have been challenged by the demand to come up with a product suitable for all three sizes of laboratories. “But fortunately there have been some very enterprising vendors who developed software that runs between the instruments and the LIS. Almost overnight, a number of middleware vendors recognized the shortfall and moved into the vacuum, delivering this 20 percent to 30 percent of missing value that the LIS never delivered.”

Okalahoma University Medical Center’s Meditech track automation system for the core laboratory went live for chemistry in March 2004, and “after we got used to the track, about nine months later we started turning on all the autovalidation and other rules using the track middleware. We had so much luck with that in chemistry that we decided to do the same thing in hematology,” Dr. Blick says.

The laboratory was worried about going out and getting “another flavor of middleware,” and having to train medical technologists to deal with a different user interface in hematology. So “to get value out of our middleware we went with the same product [DataLink 2000]. If we’re going to share employees in chemistry and hematology, then the ‘touch and feel’ of all the software running in the core laboratory has to be the same.”

Now the laboratory is looking forward to putting the same middleware into infectious disease, urinalysis, and coagulation. “So instead of having somebody sitting and manually releasing results, we want rules firing in the middleware to tell if results pass muster. And if not, we want an alert in real time to appear on our ‘middleware dashboard’ that there is some kind of problem with a result that requires a technologist to get into the loop.”

A factor that doesn’t get enough attention in ROI calculations is the user interface, he says. “In the old days you might buy a hematology instrument because maybe you’d get a better deal on reagents, or you liked the sales rep. Now all of a sudden we have to add in middleware to this decision. And middleware may be more of a ‘showstopper’ than the instrument features. For example, in the past you might pick equipment based on a slightly better percentage on the coefficient of variation. But with middleware you need to look at where you’re trying to get in four or five years, and will your middleware be building rules the same way in hematology as in chemistry?”

At OU Medical Center, the hospital administration “tended to be highly suspicious when we asked for new things, and were generally not supportive. They often didn’t think they were getting good value for what they were spending.”

A high percentage of chief financial officers in hospitals see laboratories that way, he believes. “They feel they’re not getting a good ROI for what they’re spending on the laboratory. In fact, some CFOs see the lab as a sinkhole where we always want more money and more people.”

Since most hospital administrators also realize high-tech solutions like robotics are inevitable, they are now more amenable to automation, Dr. Blick says. Once the OUMC laboratory reengineered and automated, “our CFO got on board almost immediately because we were able to demonstrate significant improvement in our productivity figures.”

However, the hardware can’t be looked at separately from the software. “In our laboratory, you see all this automation and think, hey this is a hardware solution, but it’s not. We have seven computers talking to each other making the hardware work. Software is the elixir that makes this thing successful.” And, he believes, it was the missing factor five or six years ago when total laboratory automation’s skeptics held sway.

“The most important thing that’s happened to the clinical laboratory over the years has been automation—and middleware may be the biggest thing that’s ever happened to automation.”

Autoverification is one of the processes that have brought laboratories a significant return on investment, and for those with an LIS lacking a robust expert rules engine, middleware has often proved an effective solution, says Hal Weiner, a consultant based in Florence, Ore.

“I know of one multi-site facility that had 25 FTEs doing result review and verification for chemistry, hematology, and urinalysis. After they implemented a middleware solution that did autoverification, they actually were able to let go seven people, and they were able to take another four medical technologists and replace them with medical laboratory assistants,” Weiner says.

“They believe they reduced their personnel cost by 40 percent, and that’s a lot of money. And they also reduced their turnaround time by about 15 percent because delays from waiting for someone to review a test and verify it were eliminated. What was more interesting is after a year of doing this, their stat orders have reduced by 35 percent because they improved their TAT.”

“So there’s an area where you have real, proven ROI with a middleware solution.”

Weiner is not suggesting that middleware is essential. Some LISs can perform autoverification or implement rules for reflex testing—but the older legacy systems may not be fast enough.

“In reality, people are keeping their legacy systems longer and longer. I did a quick survey and there are more than 1,600 LIS systems that I personally know of that are over 10 years old. That doesn’t mean they’re out of date. Rather than purchase a brand-new system, some people are finding they can acquire this middleware add-on to extend the capability of their existing legacy system.”

Another middleware application that has brought his clients significant ROI is physician office link systems to handle order entry and result returns to physicians’ offices. “Some of my clients install physician office link products in their clients’ offices, and they do medical necessity checking, appropriate CPT coding, and correct patient demographics at the point of order entry,” Weiner says. One client as a result has seen a reduction of re-bills by 40 percent and a reduction in days receivables outstanding (called DSO) from 62 to 32 days.

All better than getting a handwritten piece of paper on which half of the information is missing, he says. “If you put a system in doctors’ offices that forces them before placing the order to capture the patient SSN, and get the right name and address at the front of the process, you have a much better opportunity of reducing errors on the back end when producing the bill.”

Somewhat outside the traditional definition of middleware—but also likely to bring ROI dividends—is software that helps laboratories improve profitability by managing their business proactively, Weiner adds. He calls it the “third leg of the stool of ROI.”

“If I can reduce the cost of producing the test, get paid appropriately for the test and maximize payment, that’s two-thirds of the stool. The third is improving the profitability of the business I am getting.”

“This can come from an add-on product with an effective set of rules you can generate that can personalize services to clients but also monitor what’s going on within your laboratory,” he explains. “For example, who is the doctor who ordered 100 Pap kits a month but never sends you a Pap test?”

He stresses that using IT to improve ROI doesn’t require the purchase of middleware. “If you are making the most effective use of what you have already, you can in fact improve your ROI without having to buy anything else.”

Health care institutions in general, laboratories included, tend to estimate ROI before they make purchases, but they rarely go back later and say, “Did I get the returns I expected?” says Dennis Winsten, an independent consultant specializing in laboratory information systems. “We get a lot of a priori analysis that says if we do this project we will save this amount of money, or increase quality or reduce errors, and so on, and we make a projection. But after the system is bought, usually nothing happens. We’ve got other issues to deal with and we just have to accept what we have. And my biggest concern is, yes, the money has been spent, but wouldn’t it be useful to determine where we haven’t achieved the benefits and take some remedial action?”

As a rule of thumb, he says, “If you look at laboratories’ use of their information systems, they are probably using about 70 percent of the capacity of that system, and there’s another 30 percent of benefits to be gained. That’s true to some extent with the middleware piece as well. There are a lot of good applications for middleware that have a direct impact on productivity, on throughput, on being able to process more samples through a system, which of course, particularly in the case of reference labs, is a measure of how much revenue you can produce. And that’s another potential ROI on this type of product.”

While vendors’ ROI projections may be looked at skeptically because they have a vested interest in making the ROI look good, in many middleware situations there are quantifiable throughput and labor savings, Winsten argues, that can be cited as hard ROI data.

“But I’m not sure I’ve ever seen a strong retrospective ROI analysis, and the reason is that most institutions are unwilling at the front end, before the automation or middleware is put in, to carefully measure a number of operational cost parameters, such as person-hours to do a specific function, or what is the error rate, so that we have a baseline.”

“Then after you’ve had the system installed for some months and given it the opportunity to shake down the rough edges, you can go back to the baseline parameters and compare,” Winsten says.

Among hospital executives reviewing capital requests from laboratory managers, there is a renewed interest in quantifying the ROI not only before but also after middleware products have been purchased and installed, Dr. Friedman says. He believes middleware solutions like automated specimen retrieval from storage, results autoverification, and positive patient identification lend themselves to relatively easy measurement, and could help clarify the ROI picture.

Measures such as mean time to retrieve a specimen for test add-ons, test turnaround time by shift, and numbers of specimen identification errors could be made available real-time in the same way that some new cars display current miles-per-gallon performance during operation, Dr. Friedman says.

“Middleware could similarly calculate laboratory management and quality measures on-the-fly, allowing for experimentation by lab managers to improve some specific performance measure. For example, one could change staffing on a particular lab shift and look for improvements in TAT.”

Some 15 or 20 years ago, Winsten notes, “I think the attitude of laboratories in hospitals was more that of the ‘show-me’ state. But in the last 10 years it’s been generally felt by lab administrators and hospital people that automation and middleware are good things and have a positive ROI.”

The middleware does work, he emphasizes. “The companies developing the logic—Dawning Technologies, Data Innovations, Technidata, and so on—know what they’re doing, and they do provide a lot of value added.”

“But what’s particularly interesting is what will happen with labs in the future with regard to what their role is and what the role of the LIS is, because now with EMR being developed so quickly, a lot of the work that LISs used to do by way of reporting, for example, may no longer be applicable because it would be done with an EMR system.”

Similarly, Winsten says, “more and more orders are not directly associated with the LIS but come with the hospital information system or EMR system or practice management system, so a lot of people are projecting that the LIS in its current form is going to contract and be much more oriented very specifically to the internal operations of the lab, and much less to external context outside the lab.”

“So in the future one could imagine that a lot of what lab systems will be doing will be more in the domain of what middleware systems do now, i.e. obtain samples, process them through instrumentation, perhaps control automation, do autovalidation, perform checking. The LIS would become much more of a ‘feeder’ to larger systems, and much less responsible for order processing at the front end and results reporting at the back end. Obviously it will take some time, but the LIS functions are going to change,” he says.

For now, Winsten urges laboratories purchasing IT to do a good retrospective assessment of ROI.

“I can understand their reluctance to do it, because supposing you do it and find you didn’t come anywhere close? But it’s useful because you may be able to take remedial actions to get that ROI and not just accept substandard results. I think the industry needs to do these assessments more rigorously—and it will also provide feedback to the vendor to help them make better products.”

Anne Paxton is a writer in Seattle..