By Elizabeth S. Roop for For The Record
Two case studies shed light on the issues to consider before making a purchasing decision.
As the transition to ICD-10 edges closer, hospitals are being bombarded with information on how computer-assisted coding (CAC) can ease the process. Conference exhibits, sales calls, and direct mail campaigns can make it feel like CAC vendors are everywhere, ready to spring with unassailable proof of why their solutions are more thrilling than a summer blockbuster.
Needless to say, selecting a CAC system can be intimidating. The technology reaches deep into systems well beyond billing and coding, making the decision even more nerve-wracking. The wrong choice can result in disgruntled end users, poor adoption rates and, ultimately, a costly mess.
“Do [your] research,” says Heather Eminger, CCA, an AHIMA-approved ICD-10 trainer and CAC product manager for Dolbey Systems. “Know your workflow and involve the various departments that touch the coding department.”
Case Study: Adventist Health System
Eminger recommends forming a multidisciplinary team to handle the selection process including, at minimum, the HIM director, the coding manager, coding team leads, the clinical documentation improvement (CDI) lead or manager, the chief financial and information officers, and the IT team. It’s also important to have the coding team review workflow objectives.
“Discussions should revolve around the facility’s expectations in regard to if the patient chart will be directed to the CDI and/or QA [quality assurance] staff prior to coding to reduce the risk of claims denials, code to the highest level of reimbursement, etc,” Eminger says. “Computer-assisted coding should reduce the number of AR [accounts receivable] days. With productivity gains, your facility may be able to redistribute the work load.”
Bringing together a multifunctional team was the first step in the CAC vendor selection process for Adventist Health System in Altamonte Springs, Florida. Director of health information services Migdalia Seda-Hernandez says HIM, information systems, and administrative teams were involved from the outset.
The evaluation process consisted of interviewing the clients of three prospective vendors and conducting on-site visits. Team leaders discussed the project approach and observed coders in action on the CAC system. They created a findings report, which included their recommendations, and presented it to the senior finance team, which eventually forwarded the final recommendation to senior leadership. Ultimately, Dolbey’s CAC offering emerged as the top preference.
“Our premise was to provide a tool to our coders rather than replace our coding staff,” Seda-Hernandez says. “Coders will always have the final decision on which codes will be posted to the claim. The coding needs to be based on the review of the entire record and should include text and imaged documents. One of our key requirements for the vendor was to provide us with the ability to process all types of patient accounts, not just inpatient.”
Other criteria for acceptance centered on natural language processing and charges. “In addition to leveraging the use of natural language processing for the coding process, we wanted to utilize natural language processing to provide our case managers with a concurrent DRG [diagnosis-related group] on a daily basis,” Seda-Hernandez says. “Finally, we wanted to interface charges to our CAC in order to allow our coders to review charges/codes during the coding process.”
The painstaking selection process already has paid off for Adventist. Previously, HIM ran a daily batch to submit claims to the scrubber software. Coders then had to review accounts again and make any corrections that appeared on the edited report, a process that added an average of two to three days to outpatient coding. Post-CAC, the number of coding-related edits listed on claims scrubber reports has declined by 90%.
During any CAC evaluation, several criteria should be considered, starting with the number of platforms that will be utilizing it. Needs will vary among inpatient, outpatient, ancillary departments, QA, and CDI interaction, so it’s important to gain a clear understanding of all of the technology’s capabilities.
“One of the largest benefits of a CAC application, aside from productivity gains, is for all end users to be able to log in to one platform that houses all of the areas within your facility,” Eminger says. “How do the specific departments interact within the same platform? How are roles and securities defined?”
Adventist facilities utilize CAC for inpatient, observations, emergency department, same-day surgery, and outpatient visits. To help reduce overcoding, the health system worked with Dolbey on identifying documents and document headings that are normally evaluated by a coder during the coding process. The order in which documents are presented has been reorganized based on coder preference rather than the order in which they are displayed in the EMR.
Also identified was the importance of presenting coders with information contained within the nursing documentation. Intubations, bedside procedures, the presence of ulcers, body mass index, and heart rhythm now are organized in tables and presented during the coding session.
“This information enables the coders to optimize the DRG and case mix,” Seda-Hernandez says. “We are currently presenting to the coders over 250 nursing fields within the CAC application. This also expedites the coding process because the coders do not have to spend time looking for this information within the multiple documents.”
She notes that CAC offers several advantages over manual coding, including increased productivity and coder efficiency. It also helps ensure the consistent application of coding rules. To date, Adventist has seen significant returns on its CAC investment in the form of average record productivity improvements in various departments, including the following:
- inpatient: 58%;
- emergency department: 27%;
- outpatient surgery: 42%; and
- outpatient visits: 19%.
“We also experienced a reduction in the number of clicks required to review the electronic health record and to select codes within the encoder,” Seda-Hernandez says. “Currently, overall acceptance of the total codes suggested within the claim is approximately 71%.”
Adventist relies on its CAC system to drive numerous process improvement initiatives, including documentation and case mix improvement, retrospective coding audits, autocoding, and dual coding.
As beneficial as it has been so far, the CAC rollout has not been without challenges. Among the most significant is getting coders to make the transition from coder to auditor. Coders are accustomed to reading the medical record then utilizing the encoder logic for code selection and validation. However, in the new system, natural language processing suggests codes for terms or phrases found within the document but can’t associate them with more specific words or phrases that appear in a handwritten document.
“It relies on the coder to associate these terms and edit the code suggested by the CAC to the correct combination code,” Seda-Hernandez says.
Case Study: Hamilton Medical Center
When Hamilton Medical Center, a 282-bed hospital in Georgia and part of the Hamilton Health Care System, set out to replace its CAC system, it followed many of the same steps as Adventist, starting with a multidisciplinary selection team composed of the medical records director, coding supervisor, HIM analyst from IT, and director of information services, John Forrester. He says the existing system had to be scrapped to avoid “a complete forklift of our current applications suite.”
The objective was straightforward. “For us, the initial selection criteria amounted to a replacement of the current capabilities that we had at the lowest possible cost while providing an integrated system for case management, CDI, coding, and medical records to work within,” Forrester says.
For its first CAC solution, the hospital sought a product that would help bridge the gap to ICD-10. Specifically, it wanted the technology to provide both ICD-9 and ICD-10 codes. Also important was a solution that would not only deliver productivity increases but also help offset any productivity declines after the move to ICD-10 and the accompanying shortage of coding resources.
The latest product search focused on several factors. “We were most interested in the application’s usage and installed base,” Forrester says. “There is comfort in relying on a big company product. We had to get comfortable with just the notion of using a smaller company solution for such a critical part of our operations that would ultimately drive our revenue cycle.”
Hamilton ultimately settled on Precyse’s CAC offering. Even though Precyse is a smaller vendor, Hamilton was comfortable with the CAC solution based on its successful use with the vendor’s outsourced coding clients.
The final decision wasn’t made without first consulting Precyse reference sites as well as local professional networking channels—an important step that was complicated by Hamilton’s status as an early adopter. “Very few even knew what we were talking about at the time,” Forrester says.
Documentation Is Critical
Having been through the CAC selection process twice, Forrester is well equipped to advise others on the keys to success. It all comes down to documentation. “Think first of what you have available in terms of electronic documentation to feed the system. Scanned images of written documentation do not get you very far with a CAC solution,” he says. “The biggest challenge to implementing and ultimately gaining value from a CAC solution is the ability to feed the system with all of the chart components in a readable electronic format where every piece of documentation is identifiable and discernible.”
Mark Hendricks, MBA, RHIA, vice president and commercial general manager at Precyse, expands on the importance of documentation, noting that it is important for any hospital to view CAC as an enterprise decision more so than an HIM or HIM-HIT decision. In particular, hospitals must ensure that the clinician view is represented because clinical documentation can make or break any implementation.
“When you really look at the application of the technology, yes it involves coders and yes there is a significant technology aspect, but you also need to look out into the clinician side,” Hendricks says. “When you really think about the CAC application, it will only be as good as the documentation it can read and interpret. Consider how physicians will interact with the application, especially with queries. It really becomes more than just a departmental solution.”
As such, hospitals should partner with CAC vendors that clearly understand the documentation process and how EMRs feed into the CAC system. Documentation is being created in multiple systems using multiple methodologies, and it all ultimately feeds into the CAC application.
Other vendor criteria include a deep understanding of coding and the coding process, which Hendricks notes will never be completely automated. “It requires human intelligence to make some decisions, so look for someone who views CAC as an enhancement to the coding process and not as taking it over,” he says, “someone who understands how a CDI program interacts with CAC and also how to educate physicians on improving documentation.”
During the selection process, providers should try to obtain answers to several key points. Ask for an estimated completion date. Try to determine whether the vendor truly understands how to implement CAC. Too often, Hendricks says, vendors underestimate both the time and complexity of the process.
Also of note is whether the system uses natural language processing. If so, how is it being applied to CAC? How is the accuracy and quality of the natural language processing engine measured? How does the system assign work to coders to ensure it matches their available skill sets?
Hendricks says the success of CAC will come down to one factor: facility readiness. If a hospital’s documentation is inadequate, it won’t matter which system is deployed.
“The challenge is when you don’t have the appropriate level of specificity in your documentation, you aren’t going to get good results from CAC,” Hendricks says. “CAC can solve productivity problems but only if documentation reflects services provided and severity of patients. If you don’t have that, you need to focus on getting that first and then you can go out and select the technology.”
— Elizabeth S. Roop is a Tampa, Florida-based freelance writer specializing in health care and HIT.
How to Get the Scoop From Reference Sites
When it comes to viewing computer-assisted coding (CAC) in action, there may not be much action at all. Many solutions are not demonstrated live or they’re shown in limited-deployment situations. This makes visiting or talking to a reference site a critical step in the evaluation process.
“It is important to know what is live, beta, pilot, or production and, if not production, when will it be ready for general release,” says Heather Eminger, CAC product manager for Dolbey Systems. “CAC is changing so rapidly, and several vendors have shifted their natural language processing and CAC strategies one or more times, so it is important to validate what is actually live and where.”
Eminger recommends validating the software version being reviewed and comparing it with the deployed version. In addition to quizzing reference sites about specific challenges, concerns, and successes, she suggests asking the following department-specific questions:
- How are you deploying CAC for inpatients?
- Has it increased coder productivity?
- Has it improved coding quality?
- How often do you see information pulled that is not pertinent to coding?
- Have you been able to autobill any accounts?
- Do you have to use the same codes on multiple accounts (eg, physician, anesthesia, hospital)? If so, does it pull accurately?
- Are you able to use CAC to assign evaluation/management levels? If so, how and is it accurate?
Outpatient Surgery/Observation Services
- Has CAC increased coder productivity?
- Has it improved coding quality?
- How often do you see information pulled that is not pertinent to coding?
Outpatient Labs/Radiology/Sleep Studies
- Can you autobill any accounts? If so, what percentage of studies can you autobill?
- How long after install could you autobill?
Questions to Consider
The search for a computer-assisted coding (CAC) solution must be thorough. Heather Eminger, CAC product manager for Dolbey Systems, recommends getting answers to the following questions to start the project on the right foot:
- Which departments will utilize CAC?
- Which coding or clinical documentation improvement (CDI) processes are expected to be enhanced or added?
- How does the CAC system determine the suggested code? Is it through natural language processing, word spotting, structured input, or statistics?
- Should the CAC system suggest codes or highlight them?
- How will the CAC system integrate into current HIM functions (eg, types of documents it accepts)?
- What are the CAC system’s interface and other technical requirements?
- What features are must-haves?
- What type of management-reporting capabilities are desired, and to what extent can coder workflows be created and/or customized?
- Which code sets should be included?
- Is ICD-10 available? If not, when?
- Should ICD-9 be in a separate system from ICD-10?
- How does the vendor roll out CAC (ie, phased or big bang)?
- Should coding occur in a dual environment?
- Should current coding queues be able to be integrated?
- Should CDI be a separate module?