Vol. 15 Issue 11
Autoverification can transform hematology laboratories and improve patient care. The key is understanding which strategy is right for your lab.
It's a common scenario in the laboratory: The staff is working hard to release hundreds of results per shift, but the phone keeps ringing, stat samples continue to arrive and a stream of people stop by to ask questions or check the status of a test.
Zoom out to a macro level and you see that, on average, test volumes are climbingjust as a significant percentage of lab technologists are preparing to retire. As busy as clinical labs are today, they're about to get even busier. And in this environment, being able to deliver high-quality results will become even more challenging.
Looking for Answers
Faced with this reality, a growing number of labs are asking important questions: "What can we do to maintain our high quality in the midst of these challenges? Is autoverification technology the answer?"
With autoverification (sometimes called autovalidation), all results deemed "acceptable" according to the lab's criteria are automatically released. No manual review is required.
Labs that decide to autoverify will find they have three options from which to select. They can enable automated decision-making through:
the hematology analyzer,
the laboratory information system (LIS).
Each method has its advantages and the right choice will differ from lab to lab, depending on a variety of needs. But no matter how it's enabled, autoverification has been shown to transform hematology laboratories, allowing labs to do more with less while improving patient care.
Turning to Autoverification
If there's more work to be done, why not just add more people? Budget issues aside, there is another problem with this strategy: There simply aren't enough qualified technologists to fill this need. For years we have been hearing about the dearth of medical technologists. Fewer medical technology schools exist, so fewer individuals are graduating from each program per year than 20 years ago. At the same time, more than 50 percent of all clinical laboratorians will be eligible for retirement in the year 2010a statistic that foretells an alarming gap.
If hiring more people is not a viable solution, what is? A growing number of lab administrators are agreeing with Gerald Davis, BS, MT(ASCP), MPH, senior clinical technologist at University of Michigan Hospitals.
"The only way to do more with less is to automate," he says. "That's exactly what autoverification does."
Yet, when it comes to autoverification, concerns typically surface. Will it be difficult to implement? Will the system "get it right" every time? Won't technologists do a better job since they can make nuanced judgments? There is also a fear among medical technologists that jobs will be lost if result verification is automated in their laboratory.
But labs that have embraced autoverification are quick to dispel these myths. Indeed, they point to the compelling benefits they've realized, which include:
consistency of decision rules application,
decreased turnaround time and
better utilization of staff.
Consistency of Decision Rules Application
Consider the CBC with differential. With 20 parameters and multiple rules for each (dilution, rerun, manual diff, etc.), the average technologist would be applying scores of decision rules in their head for each sample. The likelihood of every technologist reviewing all of those rules for each sample in the same manner is not realistic. Interruptions, fatigue and stress may alter their ability to apply the same rules to each sample consistently.
"Autoverification is like taking the script that runs through your best tech's head and making it play around the clock for every result," says Vicki Parsons, assistant laboratory director at University of Kansas Hospital.
By automating the laboratory's own decision rules, the same high-quality decision criteria are applied across all shifts at all times. Every sample is treated the exact same way, no matter the time of day or day of the week or any other variable.
In some cases, the lab's review criteria may not be completely captured in writing and this makes autoverification all the more valuable. "Many labs don't have written review policies," says Davis. "Instead, every technologist in the lab thinks he knows how to apply the policies. The problem is that not everyone follows the same standards."
Autoverification takes out the potential decision-making variability and reduces the risk of error by standardizing the review process. Depending on the technology used, labs can even standardize very specific situations. For instance, rules can be created for certain physicians or patient locations.
The bottom line of this consistency is less risk of error and better patient care.
Decreased Turnaround Time
Instrument throughput is an important feature. But what is the true value of an analyzer that can process 100 samples per hour if the technologists can only review and release 50 samples per hour or if the results are waiting to be reviewed while the technologist is on the phone or attending to other duties? By eliminating the need to review every result manually, autovalidation increases throughput and reduces turnaround time. And rapid results deliver more value to the caregiver, who can act on those results faster. If your process delays results from being released, the test results diminish in value to the caregiver. They will tend to make decisions without the laboratory information, using it as "supporting information" rather than action-oriented information.
In some cases, delayed test results can seriously impact patient care. According to a Sentinel Event Alert issued by the Joint Commission of Accreditation of Healthcare Organizations (JCAHO), delayed test results are the second most common reason for delays in treatment.
There is no doubt that short turnaround time is essential to patient care, but Davis points out an additional benefit. "With better turnaround time, the phone stops ringing," he says. "Physicians have the results, so they don't need to follow up."
Better Utilization of Staff
For many technologists, the word "autoverification" stirs up fears of job loss. In reality, autoverification can enhance technologists' jobs by creating more time to focus on abnormal samples or complex testing proceduresactivities that utilize their specialized knowledge. With this increased focus on abnormal samples, technologists are able to make a more significant impact on the patient outcome.
Many autoverification systems are able to alert the technologist to perform specific "next steps," which helps standardize these actions. This gives the generalists, as well as the hematology specialists, the tools they need. The overall stress level in the laboratory typically decreases when autoverification is introduced. Instead of constantly working through a backlog of results, technologists can focus on only the samples that are held for manual review. Spending more time on these abnormal samples leads to a higher quality manual review and reduces risk of error.
"Reviewing all results can be a mindless and numbing process," notes Davis. "And this can lead to mistakes."
Parsons agrees, adding, "There's so much going on in the lab, and everyone is expected to multi-task. Autoverification is just a safer way to do things. Things don't get overlooked."
Three Avenues to Autoverification
The overall advantages of autoverification are compelling, but specific benefits will vary depending on which technology a laboratory selectsinstrument, middleware or LIS.
Route 1: The Analyzer
Some hematology analyzers are designed with built-in decision-rule capability, giving laboratories a fast, convenient option for enabling autoverification. This is the approach used at the Chinook Health Laboratory in Lethbridge, Alberta (Canada). Several years ago, the Chinook Health Region was established, uniting 11 acute- and continuing care facilities, 15 community health sites and all community outreach under one umbrella.
As a result, testing volume in the Chinook Health Center lab rose steeply, from 150 samples per day to 700 per day. To manage this increase without hiring new staff members, the laboratory enabled autoverification (on its COULTER® GENS™, Beckman Coulter®). Through the analyzer, the lab could program multi-level decision rules, with up to four conditions for each rule.
"We were able to define many situations," says LIS Coordinator Monica Kearns. "It was a simple, inexpensive process."
For delta checking, the laboratory turned to its LIS (Meditech). The key to success, says Kearns, was to link the analyzer and LIS, so the lab would take advantage of the best autoverification features on each system.
To do this, the lab created a non-numeric test in Meditech. If a sample in the GENS system generates a flag, the result is sent to the LIS with the message, "Sample not validated." In this case, a manual review is required. If there are no flags in the GENS, the result is sent with a message that says, "Autovalidated," which gives the LIS approval to auto-release after a delta check verification.
Prior to autoverification, the lab's manual verification process was time consuming and prone to error; now it automatically releases 70 to 80 percent of its samples.
Kearns notes that autoverification has become a technology the staff can't live without. The laboratory is investigating next-generation analyzers and plans to continue autoverifying results using a hematology analyzer. Newer analyzers will offer more features, such as the ability to print a list of "next steps" based on user-defined criteria.
There is, however, a limit to this strategy: The information used in the autoverification algorithm can only come from that instrument. For labs that want to autoverify using results from multiple instruments, a better choice might be middleware or the LIS.
Still, at Chinook Health, the laboratory is pleased with the results, as well as with the fast, easy process of programming decision rules. In essence, the analyzer offers many of the same features as middleware, but with the added convenience of being built directly into the instrument.
"This strategy definitely works for us," says Kearns. "We are able to do everything we want to do and we know that the rules are applied consistently every time."
Route 2: Middleware
Middleware is software that sits between the analyzer and LIS. The hematology decision rules are written in the middleware software. Information flows from the analyzer to the middleware, where decision rules are applied. The sample results that are within the laboratory's acceptable validation limits are released to the LIS, while results not within the validation limits are held in the middleware system.
One of the chief advantages of middleware is the flexibility and specificity it offers when it comes to writing rules.
The laboratory at Kansas University uses Aqueduct software (Orchard Software Corp., Carmel, IN), and Parsons notes that the lab is able to write rules in as complex a manner as is necessary. "We are able to add or edit rules whenever needed," she says.
Depending on the vendor, rules can be specified for a wide range of options, including:
critical values for IP and OP;
relationship of parameters (e.g., RBC, HGB, HCT);
dilution needs or manual processes;
laboratory review policies;
pathology review policies; and
specific units or physicians.
With middleware, labs can customize their decision criteria and this can yield a higher pass-through rate. The results that do not pass through are held in the middleware program, ready for technologist review. The technologists can view the patient's complete hematological record, including all histograms, scatterplots, flags (even instrument flags, definitive flags and suspect flags) and any decision rule comments on the "Results Review" screen.
Middleware also allows labs to connect multiple instruments, identical or different, and apply the same autoverification criteria for results across a variety of analyzers. All the checking and reviewing occurs in a single place.
At Mount Carmel Medical Center in Columbus, OH, it was this convenience that prompted the laboratory to switch to middleware (DL2000, Beckman Coulter, Brea, CA) after roughly five years of autoverifying results in its LIS.
"With the LIS, some instruments could utilize rules programmed in the LIS instrument manager, while others relied on the LIS rules package," says Denise Scaduto, MT(ASCP), manager of Laboratory Information Resources. "In addition, some error codes coming from the instruments were not recognized in the LIS."
With middleware, the lab brought all of its rules and test results together in one place, so autoverification would happen in a central location instead of having some results checked in the instrument and others in the LIS. The information is easier to manage and the laboratory still has control of all of the sample data. In addition, the middleware is able to keep track of sample repeats and dilutions, something the LIS couldn't do.
Middleware is typically managed directly in the laboratory, which creates both a benefit and a potential issue. The benefit is that the laboratory has complete control over the rules and how autoverification is applied. The potential issue is the time required to learn and manage the software.
Still, the amount of time saved can be significant. Research has shown that with an auto release rate of 80 percent, labs can save 2.81 hours per 1,000 samples. That equates to a 44 percent time savings.
"We autoverify about 70 percent of our samples," says Parsons, from the laboratory at Kansas University. "Our pass through rate would be even higher, but we still report bands. The staff supports autoverification because the extraneous work is out of the way, yet we are still reviewing everything we need to review."
Indeed, many labs find that middleware can autoverify roughly 80 percent or more of all results, creating more time for technologists to focus on abnormal samples.
"To continue improving the quality of laboratory testing, we knew we needed to incorporate automation," says Rick Brant, MBA, DLM(ASCP) CPHIMS, administrative director of Mount Carmel West Hospital. "Autoverification is an essential step in our long-term plan to improve efficiency and patient care, and middleware was the right solution for us. We have one autovalidation platform across hematology and chemistry, which is helpful for staff cross-trained in both areas."
Route 3: The LIS
LIS autoverification can be further sub-divided into two categories:
instrument-driven autoverification and
LIS rule packages.
All LIS vendors now provide instrument-driven autoverification, either with the standard LIS package or for an additional price. In this configuration, whenever data passes from the instrument into the LIS, it is checked against criteria that are hard-coded into the LIS software. Typically, there is a range of criteria options that laboratories can select, including:
Delta check types;
the ability to suppress results under a variety of conditions; and
the ability to use multiple logic levels for various conditional criteria that may apply to the single result, all results in a group or all results in that accession that come from that specimen draw.
"With the LIS approach, this is where the bulk of autoverification takes place," says Davis.
However, because the information is hard-coded, it cannot be changed. Laboratories need to work with the review criteria options built into the LIS software; they cannot add new criteria. That's where "rule packages" come into play.
With LIS "Rules Package," typically available for an additional fee, labs have more flexibility than instrument-driven autoverification. Labs can write their own rules specific to their user-defined criteria. They can also use the instrument-driven criteria and the rules package together.
At the laboratory for University of Michigan Hospitals, instrument-driven rules and LIS rules packages work in tandem to produce pass-through rate of between 81 and 85 percent.
"We autoverify CBCs at the instrument level and handle auto-diffs with a rules package," notes Davis. "We've been fine-tuning our strategy over time and it's easy to make changes as we go."
As with middleware, the rules are easy to write. "You don't need to be a programmer to write autoverification rules," Davis adds.
Typically, laboratories are able to write rules and program review criteria parameters on their own, without working through an LIS representative. However, if a laboratory has outsourced this maintenance work to the LIS vendor, it could be difficult to make changes in a timely manner.
Making the Right Decision
Laboratory decision makers should assess the needs of automatic result verification for their own laboratories (Table). Autoverification within an analyzer may be the right choice for a small laboratory with focused operations, particularly with the ease-of-use and convenience it offers. Meanwhile, larger laboratories may need the flexibility and breadth that middleware or LIS systems can provide.
For laboratories that want to create a wide range of rules, including very detailed rules, middleware often provides the most flexibility and customization. But some LIS systems can be powerful as well, with an added advantage of consolidating result review.
With all of the choices available, a growing number of labs are investigating autoverification and this number will increase in the coming years. From a regulatory standpoint, the College of American Pathologists (CAP) has developed standards for the autoverification of laboratory results. Laboratories must adhere to these standards to ensure a quality system. The standards can be found in the CAP General Checklist, Laboratory Computer Services section, under the heading "Autoverification" heading.
Far from being scary or new, autoverification is becoming a proven, dependable technology that delivers measurable improvements. More laboratories are investigating this option and, as Davis sees it, this trend is destined to continue.
"If you look at what's going on in the industry, it's clear that laboratories will continue to be asked to do more with less," he says. "The laboratories that want to decrease turnaround time and improve qualitywhile also adding more valueneed to consider autoverification."
Stefanie McFadden is a clinial lab and hematology consultant.
Table: Assessing the Need for Automatic Result Verification
Is autoverification a viable solution for your laboratory?
Where are your techs spending most of their time?
What is the test volume in relation to the staffing level?
Which review criteria would you standardize?
Which result review issues are difficult to standardize in a manual review mode?
If autoverification will be utilized, which method is suitable for your lab?
How many hematology analyzers are in use?
What functionality does your LIS provide?
How much flexibility do you want when it comes to creating and editing rules?
Which system can provide you with the highest autovalidation rate (i.e., has the most flexible rules) for your patient population?
Do you want to eliminate paper completely?
Which middleware products work with your analyzers and what are the options from independent vendors?
Will the LIS vendor support the middleware product and how much will this cost?
table/courtesy Stefanie McFadden
Efficiencies of Autoverification
The biggest gain in laboratory operational efficiency is the ability for middleware to correlate the instrument data with the manual differential data without all the paper, explains Anne Tate, product manager for Sysmex's hematology data management system, MOLIS WAM. Previously, the operator had to compare the instrument graphical and numeric data on the instrument paper printout with the manual differential results. With the ability to perform a manual differential in the MOLIS WAM application, for example, all data is now in one view. "The laboratory can gain at least an hour of productive time each day by consolidating data into one screen for review," Tate claims.
The rules engine is the core of the middleware product. A library of system rules drives efficiencies in laboratory operations. Rules can be simple or complex based on the laboratory's best practices and unique workflow. For instance, rules in Sysmex's MOLIS WAM can be written to include non-numeric as well as patient demographic data with the absence of a test order to drive the automatic addition of another test. In this example, a NRBC is ordered on a neonate based on these specific data inputs.
Labs can achieve dramatic results when combining many variables to build their rules. They can be very specific and may be adapted as the laboratory changes.
Additionally, the power of middleware software can solve difficult integration issues that laboratories may encounter with multiple LISs. Some software has the capability to interface not only multi-sites, but multi-LISs into the same database. The orders are directed to the correct instrument based on the specific site and results are transmitted back to the correct LIS. Patient delta checking can occur because all patient results are located in one database regardless where the sample originated or which LIS transmitted the order.
Data management systems should provide adequate scalability to meet the needs of dynamic laboratory changes and offer flexibility without the need to change software applications.
Deb Hewett-Smith is director, Health IT Systems, Sysmex.
Demystifying Autoverification Rules
How do you know what rules to apply with autoverification? And how do you go about creating these rules?
The International Consensus Group for Hematology Review conducted a two-year study to establish and validate review criteria for CBCs with automated differentials. The result of that work is 41 recommended decision rules, which are available at www.islh.org (under "Consensus Group") as well as in a published paper.
In addition, the Clinical and Laboratory Standards Institute (www.nccls.org) is issuing official autoverification guidelines this fall, called "Autoverification of Clinical Lab Test Results; Proposed Guidelines" (CLSI index number Auto10P).
With this recommendation, lab managers can outline the specific rules they want to apply, knowing that their decisions are based on proven research.
The method for writing rules will differ depending on where autoverification is taking place (analyzer, middleware or LIS). But in most cases, creating rules is easy and does not require any specialized information technology knowledge.