College of American Pathologists
Printable Version

  Smarter, smoother decision management through middleware


CAP Today




February 2010
Feature Story

Anne Paxton

It’s happening in large hospitals and small: Each year, increasingly nimble and user-friendly versions of middleware are letting more clinical laboratories augment their laboratory information systems with sophisticated decision management and advanced rules-based intelligence.

At the 200-bed Memorial Hospital in Tampa, Fla., for example, the latest version of Dawning Technologies’ JResultNet Interface Engine Software has solved multiple problems the laboratory was facing, says Alvin Conchas, MT(ASCP), director of laboratory services.

Memorial Hospital has had McKesson Star as its LIS since 2000. “When we got McKesson, we were introduced to middleware because we needed it to interface all our instruments to the LIS, but at that time it was mostly a pass-through interface,” Conchas explains.

“Our LIS is very limited. We can’t create rules or change values, and we don’t know when a result has been sent or accepted.” That was the main reason, about three years ago, the laboratory started with Dawning’s JResultNet; it recently upgraded to the middleware’s 3.2.0 version as beta tester. JResultNet 3.2.0 includes an improved graphical user interface, an offline Rules Development Kit, and a new User Management system that tracks and allows for the control of individual user permissions, as well as new search functions.

As a Java application, JResultNet is portable and can be run across a spectrum of hardware platforms. It can handle high-volume configurations involving multiple hospitals or, more commonly, can be installed on desktop PCs in the lab to manage all the instrument connections within the LIS environment.

The key change in the latest version was replacement of the Excel-based rule entry tool with a new expanded Wizard with several dropdown templates available for generally simple, frequently used rules. Creating rules that are more complex requires the JResultNet Rules Calculator. Dawning provides templates for test cases that can be used to test any rule created or imported into the system.

“The new version is a lot easier as far as creating rules goes,” Conchas says, noting that it’s also simpler to manipulate and easier to understand the screens. “With the middleware’s graphic user interface, it’s more visualit’s actually animated, so when you look at the screen you can actually see results crossing over.”

As an example of rules: “On calculated LDL, if the triglyceride value is greater than 400 it will send a message ‘unable to calculate.’ Or one of the new rules was on troponin. If it’s high, it actually changes the result to ‘wait for re-run.’” Vancomycin dosing and monitoring in adults is another area the middleware handles. “Our LIS is limited as to whether this one value is a peak or another one is a trough. Some LIS systems are more capable, but we’re working with Dawning so that if a certain value is a peak we send one value, and if it is a trough we send another result.”

“There are a lot of mistakes that techs make from day to day, like hemolyzed patients. Now you can just put the tube on the analyzer, then you really don’t have to look that often. And I have a rule in there that if potassium is greater than 6 or hemolysis is greater than 2 plus, it removes the potassium result and makes sure you look at the tube and check that it’s not hemolyzed and that nothing’s wrong with the specimen.”

A more complicated rule that Conchas is creating for CBCs depends on the laboratory review criteria and is still being perfected. “It would automatically remove the automated differential, so you have to perform a manual differential or a smear review. The rule fires on certain conditions as established by our laboratory, such as WBC greater than 30,000, neutrophils greater than 90 percent. A smear review will be triggered by our hemoglobin and platelet review criteria.”

If the rule is not working, he says, “Dawning will look at it and give me ideas on how to fix it. They help so you’ll be able to do it on your own.” However, he would appreciate more rules and more robust rules. “It’s very hard, with everything going on in the lab, to go in there and make changes in the rules and test them over a number of days.”

Another feature he likes is that the system sends the actual instrument ID to the patient result. “It’s an internal result, but you can keep it. That’s one of the CAP requirements: If you have two instruments performing the same test, you should be able to tell which one did it. The JResultNet keeps track of both automatically for you.”

Conchas is pleased with what Dawning provides. “It’s actually not cheap—but it’s cost-effective. We’re not reducing FTEs, but it has made my turn-around time faster. Before I came here troponins were almost an hour and 15 minutes. Now they’re probably averaging 45 to 50 minutes, and I’d like to make that better.”

Memorial Hospital is part of Iasis Health Care, and the entire Iasis system—with 15 hospitals in Arizona, Florida, Nevada, Louisiana, Texas, and Utah—is using Dawning right now, he says. All the hospitals have been able to benefit from middleware’s most important benefit: preventing errors. “Middleware is a necessity because of our LIS, but the rules allow us to easily convert results and make it easier on our technologists so they make fewer mistakes.”

When OhioHealth implemented the latest release of Sysmex’s WAM, the middleware solution for hematology instruments, it acquired all the features she had ever wanted in middleware, says Kim Moser, MT(ASCP), core laboratory supervisor at OhioHealth’s largest hospital, Riverside Methodist. The integrated delivery network OhioHealth, based in Columbus, has five hospitals in Central Ohio, each with its own core laboratory. Riverside, with 800 beds, handles a large amount of outpatient work, while the system is in the process of getting a sixth hospital up and running with a satellite lab. Together the hospitals process nearly 2,500 hematology orders per day.

For about seven years, OhioHealth used another middleware product with its Coulter instruments for CBCs. “In 2007 we actually switched to Sysmex analyzers; that’s why we switched middleware,” Moser says. The system started with Sysmex’s WAM 3.0 version and then, in April last year, became a beta tester for the newest version, WAM 4.0.

OhioHealth is one of Sysmex America’s largest and most complex implementations and a proving ground for future customers. The Sysmex WAM system (formerly known as MOLIS WAM) interfaces four hospitals’ Sysmex hematology automation lines, plus analyzers from the fifth hospital, to OhioHealth’s two laboratory information systems: McKesson’s Horizon Lab and Soft Computer’s SoftLab. WAM Version 4.0 is supporting OhioHealth in achieving its long-term strategies of improved turnaround time and lower review rates in the laboratory.

From Moser’s standpoint, the major benefit of WAM is its management of all result and reflexing contingencies for instruments, manual differential, and smear results. “If you put a sample on the instrument, there are rules written in the system to tell it to go ahead and rerun, to do delta checks, or if you want the sample to run and quantitate NRBCs, that’s another reason it repeats automatically. It also has rules to tell the analyzer when to send the sample to a slide-maker/stainer to make a slide.”

Autovalidation is a big plus with the middleware, she says. “We’ve autovalidated about 85 percent of our results here. The system ensures standardization of how the techs do their work through ‘op alerts’ that pop up to tell them how to handle the sample. It’s a huge savings in time because with the ‘op alerts’ the techs are not having to review slides so intensely.”

Middleware alleviated staffing pressure, Moser says. “We didn’t cut down on FTEs when we got WAM because we were already at minimum staff, but we have other testing, such as more special coag testing, and our people were able to be redeployed to help in those other areas, not for a whole shift but for parts of the shift.”

The issues OhioHealth has run up against in implementing middleware have been fairly foreseeable ones, she says. “Our newest hospital is Dublin Methodist. It went live two years ago and is not as experienced doing differentials. With WAM, they send their tech-level review to us; we can see the results on our screen and change the differential however we need to. So we can help them on a specimen level, or if they have questions about why a sample is held, I can go in and see what rules fired and help them along, explaining why it’s doing what it’s doing.”

The biggest chunk of time is taken up with validating all the rules, Moser notes. “When we first went with WAM 3.0 we had to rewrite all the rules, which is a huge ordeal, and they have to be validated with a lot of testing, and that took a few months.” Sysmex’s emulator, which provides ‘virtual’ samples, helps in the process, she says. “But you still have to do a lot of wet testing to validate all the rules, all the criticals, and all the delta checks.”

It has been a bit of an adjustment working with Sysmex’s Unix operating system as opposed to a Windows-based system. “It’s a little harder than with our previous middleware to divide work up between the person running the instrument and the person doing the differentials. Sometimes we have two people looking at the same sample.” OhioHealth is working with Sysmex on this now.

The hospital may someday consider adding Sysmex WAM for coagulation testing. The middleware already exists for it as part of WAM 4.0. Ohio-Health may also consider Sysmex WAM for A1cs. But first it has to resolve space issues so it can add the A1c analyzers to the line. “Right now we don’t have room to add the two analyzers,” Moser explains.

Sysmex WAM’s redesigned user interface software features histograms, full-color scattergrams with “zoom-in” capability, and screens with a single view of current, previous, and rerun results. All five hospitals are happy with the 4.0 version, she says. “It’s much more user-friendly than 3.0. It has a lot more color coding, which makes the interface easier to understand.”

Sysmex recently announced availability of its WAM eLearning program for its approximately 120 middleware user facilities in the United States. It’s a standardized, on-demand electronic training tool for new WAM users and continuing education tool for existing users. With online instructor-guided training, it covers WAM system navigation, results validation, query ordering, work-flow diagrams, and other frequently asked questions.

Glen McIver, managing director of the laboratory at Centra Health of Lynchburg, Va., says the core laboratory’s adoption of Data Innovations’ middleware contributed to Centra Health being named recently by Hospitals & Health Networks Magazine one of the “most wired” hospital systems in the country. The company, which operates three hospitals in central Virginia and runs more than 4 million laboratory tests a year, uses middleware to bridge the gaps between its existing Sunquest LIS and multiple instruments. But the middleware’s benefits go far beyond that.

About five years ago, Centra Health decided to move most of its testing to a core lab at its largest hospital, Lynchburg General. “We had a corporate mandate that every department will be in the top quartile of the Solucient efficiency and productivity surveys by the end of the year,” McIver says. But when the consolidation to one campus occurred, “it resulted in more throughput than our techs could keep up with, and we needed more automated IT.” LIS manager Mary Rogers was the one who came forward and said she had to have tools to make the software more flexible. “She was the one who originally proposed that we give serious consideration to middleware. Our ideas on how to improve operations and efficiencies were coming in more quickly than she could keep up with, given the complex-ity and restrictions that trying to program the LIS was entailing,” McIver says.

The lab was upgrading instrumentation in readiness for conversion anyway, so when Centra Lab put up hematology, urinalysis, and coagulation analyzers, it acquired the LIS interfaces from Data Innovations along with DI’s other middleware capabilities, instead of acquiring the classic interfaces from Sunquest. “With DI, you’re essentially buying a connection and can continually re-use it for any different instrument you want to forever. You just download another driver if you want to use that connection point for another instrument. This ability to recycle that interface connection is unlike the standard LIS interfaces we used to get, which had to be repurchased each time you upgraded or replaced an instrument.”

Once the middleware interfaces were in place, the laboratory started building in rules. “We started with coagulation as our first autoverification site, because it has a limited menu and has very clean and simple results... either pass or fail. Then we brought up urinalysis, where the techs have to integrate machine-generated results with some manually obtained results and keyboards. The techs got an opportunity to be much more fluent with the middleware, and the LIS staff got an opportunity to build increasingly more sophistocated rule sets.”

All of this was a prelude to installing Centra Health’s Abbott APS automated track system, which has been operational for a year. “There’s no way a lab tech can keep up with a track system like this unless it’s fully autoverified,” McIver points out. “The DI middleware brings all this together; it gives us the tools so we can combine different instrumentation platforms and still have the same rules set, the same scrutiny with QC, and the same algorithms regardless of the manufacturer or the menu. It’s much cleaner, easier, and more manageable with the middleware solution.”

The middleware proved itself again, he says, when Centra integrated its third hospital. “When we brought up Southside Community Hospital’s lab, not a single instrument there (outside of a Clinitek urine dipstick machine) was the same as the instrumentation in the two Lynchburg-based hospital labs. There we are using Coulter instead of Sapphires for hematology, Ortho chemistry analyzers instead of the Abbott Architects, Siemens coagulation instead of Stago. There wasn’t anything the same, yet all the exact same rules that we initially wrote were now applicable to these instruments as well.”

The middleware from DI also allows the laboratory to police its quality control effectively. “You can more timely release your QC, your variances, your outliers, and you can pull out data more handily. And the really sweet thing is it’s now tied into EP Evaluator,” which DI acquired recently.

EP Evaluator is a lab standard for doing correlation testing, verifying reference ranges, and establishing normal values, and all of the CAP requirements for checking calculations requirements are covered under that software, McIver says. “It doesn’t make sense to check calculations every six months as per the minimum requirements. You should be in the background checking for errors constantly, and that’s what we’re evolving to—this dynamic, constant evaluation of population means and of moving averages in chemistry. We don’t want to overlook trends or have narrow focuses just based on a lab tech looking at a control value and saying ‘Is it in or is it out?’ With EP Evaluator you can do a lot more than that.” When the Abbott track system went live, more than 80 different analytes were brought in and all were run through EP Evaluator.

Another plus for this middleware is its connection to Bio-Rad, a dominant QC company in the lab world, McIver says. “They may not be the cheapest way to go, but their data reduction and peer group comparisons are second to none. With DI, we have a direct connection to them called ‘Unity Real Time’ QC. As the name implies, it does comparisons in real time as your controls are being run on your Internet connection, so we’re becoming capable of picking up on more things when they happen.” Centra Health is constantly in the process of revalidating all of its reference ranges. “And with this new combination of DI plus Bio-Rad plus EP Evaluator, you can now start to do that with real-time results data coming through the interface. You don’t have to set up blocks of patients and run them all in triplicate or quadruplicate and manually do all the math. It can be done as you are running your normal patients.” (JResultNet, too, has a QC module for Bio-Rad users.)

Calculations that Centra Health now offers include GFRs routinely reported with all chemistry panels and, most recently, a calculated serum osmolality. “We are now routinely reporting both of these calculations with all comprehensive and basic metabolic screens.” With the middleware, “It doesn’t matter which instrument it’s run on or which facility does it. We were unsuccessful in getting that to work very handily solely within the confines of the LIS, but outside of the LIS, the flexibility and intuitiveness of the DI middleware allowed us to bring this up in a matter of days. Having the ability to orchestrate communications and rule sets between the various instruments, the LIS, and the middleware, picking the best path based on either the individual or collective strengths of these three—that is where the advantages of having a very robust middleware tool become very clear.”

Not every lab will have an LIS expert on staff, but DI offers a range of help, he notes. “You can buy from them everything from a turnkey solution to just the software.” Rule writing, too, can range from basic to sophisticated. “For instance, one can go on the Internet and download a standard rule set for hematology that most labs agree on. It can be as simple as: ‘If white count not greater than x or less than y and you don’t get an error message, then take the result.’ Depending on how tight you want to make the algorithms, 50 percent to 60 percent of your CBCs could possibly go through just based on instrument flagging rules. But if you really do it right, you can have over 90-plus percent of your CBCs go through and you only deal with the important stuff.” In addition, he says, “the objective is for the lab techs to only have to sign in to one system and on one screen be able to see the cell counter’s scattergrams, and on that same screen see the cell counts and their manual keyboard differential tally. And be able to do that for any and all analyzers, even if in other buildings. It’s the sophisticated stuff that only with the middleware will we be able to do.”

With these sorts of middleware capabilities, Centra Health’s LIS in chemistry and hematology has become somewhat more of a repository. “It handles reports and we use it for outreach billing, but otherwise the laboratory doesn’t really need to interact with it,” McIver says. “That is not true for microbiology or blood bank or pathology, of course. But the LIS alone really does not help as much where there are large instrument platforms or combinations of different instrument sets. It has no way to fully manage all of that. It tends to look at individual instruments, whereas DI tends to look at everything more from a systems standpoint.”

Those are a few of the reasons that other large laboratories that are contemplating a middleware solution visit Centra Health, McIver says. Middle-ware’s benefits become obvious when observers see how the system has worked for Centra Health: “It gives you a whole tool set that is outside the limitations of your LIS and that lets you accomplish things much more easily.”

Anne Paxton is a writer in Seattle.