Address Skill Gaps, Boost Coder Productivity
By Elizabeth S. Goar
For The Record
Vol. 32 No. 1 P. 10
While a nationwide coding contest spotlights where coders are falling short, experts offer ideas on how to solve departmental holes.
When coder productivity drops, it poses a real threat to a facility’s bottom line. That is especially true in today’s post–ICD-10 era, where coding has shifted from a largely transactional activity to one that is relational to everything from care quality and risk adjustment to population health initiatives and bundled payment preparation. This has changed not only the ways coder productivity is measured but also the relationship between speed and accuracy.
This has become clear in the four years that Central Learning, a web-based coding assessment and education application that is part of Pena4, has been conducting its nationwide ICD-10 coding contest.
“What we’ve seen is that from 2018 to 2019, on the outpatient side, there was a decrease in accuracy and an increase in the number of cases per hour, so they’re completing more cases, but the accuracy went down,” says Eileen Dano Tkacik, Pena4’s vice president of IT and BPO business development. “The three areas we consistently see as the most challenging are [neoplasm, respiratory, and musculoskeletal] cases, but the common denominators are laterality and specificity. Those combinations keep glaring at us.”
Faster but Less Accurate
The fourth annual national coding contest from Central Learning focused on outpatient coding performance to address the higher rate of coding errors, documentation gaps, and accounts not billed due to claim denials brought about in part by the significant case volume. By focusing on outpatient coding, the goal of the contest was to help HIM, coding, and revenue cycle teams—all of which are experiencing a shift in outpatient reimbursement from fee-for-service to alternative payment methods and quality-based outcomes—pinpoint opportunities for improvement.
The average accuracy rate for the 2019 contest was 40.4% compared with 42.5% in 2018. The average primary diagnosis score was 60.5%, down from 67.8% in 2018, while the average secondary diagnosis score was 38.6%, a miniscule decline from 38.8% in 2018.
The increase in primary diagnosis errors this year was blamed on a lack of detail and specificity in assigning codes, which could be corrected by ensuring that coders have access to complete documentation and perform a detailed review. The rising rate of errors on secondary diagnosis coding was thought to be caused by a failure on the coders’ part to apply correct coding guidelines for code assignment. For example, external cause codes for injuries were missing in many cases. Again, the recommended fix was to ensure that coders are performing a detailed review of all documentation to capture all codes.
“There were no real surprises” in the contest results, Tkacik says. “Skill gaps were pretty consistent with last year.” For example, several of the neoplasm cases had multiple metastatic conditions, which created confusion about what should be coded. For respiratory cases, coders erred on the level of specificity when coding asthma (eg, with or without bronchitis and the level of severity of bronchitis when present), COPD, or pneumonia.
“For musculoskeletal cases, what we saw oftentimes was that the laterality was not coded. They didn’t specify the left or right foot, or they missed coding the external cause of the injury because they were not reading the documentation,” Tkacik says. “The root causes … for a lot of the primary diagnosis errors and the increase in secondary diagnosis errors were the lack of specifics. The coders didn’t review the complete documentation, or they didn’t review the guidelines produced as part of the contest. … It all relates back to those things.”
Improving accuracy rates will require consistently and continuously in-servicing coders to ensure changes to the Coding Clinic are clear and understood, Tkacik says. This can best be done by bringing coders together to work on several complex cases, then reviewing the results to identify and address problem areas.
“Sit down and apply the lessons learned from those cases so coders can see something that is real life to their facility,” Tkacik says. “The other thing to improve coder accuracy is that coders need to query physicians if they have any questions on documentation. That’s easier said than done, though. I can go ask Santa for a puppy at Christmas, but Santa may not deliver.
“Coders are scored on accuracy, but physicians may not respond to their queries,” she adds, noting that coders are still expected to complete cases within a set timeframe whether or not their queries are answered. “There needs to be a coming together of the minds as it relates to coders, documentation, and physicians.”
Where the Problems Lie
While the Central Learning contest identified laterality and specificity as the two most common problems impacting coder accuracy, they are not the only skill gaps threatening accuracy and productivity. In some cases, the weaknesses are localized to specific facilities.
For example, a lack of familiarity with multiple service lines. That’s the case at University of Utah Health, where Data Integrity Supervisor Rachel Pratt, RHIT, CDIP, CCS, says lack of training in all the different service lines coders may encounter can slow productivity.
“The experience side is the biggest [challenge],” she says. The need for cross-line training is “something we’re seeing more of. When we’re talking with other facilities, one of the first things we tell them is that we want all staff to be able to code everything that comes through the door. That’s difficult, because we have four coder levels and service lines are broken up by level, but coders at levels three and four should be able to code anything that comes through the door.”
Meanwhile, Melissa Roberts, MHA, RHIT, systemwide assistant director of coding at UC San Diego Health, says the most common skill gaps she encounters have to do with coders not keeping up on disease processes and anatomy, which impacts their ability to assign correct diagnoses and procedure—and their confidence.
Both of which sap productivity.
“I find coders spend time trying to figure out the correct code with research and/or routing accounts for second-level review for assistance,” Roberts says. “The other issue that feeds into this are coders struggling to determine when a query is necessary and how to write an appropriate query. This can all feed into their self-confidence, which in turn impacts their productivity.”
Coder productivity is also taking a hit from procedure coding under ICD-10. According to Megan DeVoe, CCS, product manager with TruCode, Procedure Coding System coding is challenging for numerous reasons, but the most significant is that “we lost a lot of direction by having the instructional notes removed from the tabular book that previously had helped decide if codes went together or not.”
She adds, “The instruction is conflicting sometimes between the official sources, so it’s really something that requires a lot of thought, which of course is detrimental to productivity.”
Another challenge on the outpatient side is recognizing the importance of the edits that come from the grouper and their exact meaning. “In my experience, I’ve seen that this is something that—once you get the hang of it—is easy. It’s part of your routine. But it’s something you don’t recognize until it’s pointed out,” DeVoe says.
In this post–ICD-10 world, achieving the right mix of accuracy and speed is far more complicated, as evidenced by the sampling of skill gaps and other issues coders and facilities are dealing with. Therefore, it should come as no surprise that establishing an effective method of measuring productivity is equally complex, given the expanded impact coding has on a facility’s revenues, outcomes, and quality scores.
At UC San Diego Health, Roberts is in the process of reviewing how coder productivity is measured. However, in the past she has first carved out a coder’s nonproductive time (breaks, meetings, training, education, etc), which by her calculations is about 18% of an eight-hour day, “making the productive hours for determining production a 6.56-hour workday.”
Productivity is then calculated on a weekly basis to account for those days where complex cases require more time to code than average charts. This helps even out the average for a more accurate reflection of time spent clearing cases.
“I have found [this method of] calculating coder productivity is effective. Coders have found this to be fair, as we are taking into account their nonproductive time,” Roberts says, adding that “I do think we have more work to do on determining a way to effectively and fairly measure coder productivity, as there is always room for improvement.”
Over at the University of Utah Health, Pratt is also in the process of revamping the organization’s productivity reports. Currently, the health system uses 3M’s 360 Encompass system, but it also performs concurrent and discharge coding. Furthermore, while coding is performed in the 3M system, coders must also go into the Epic encoder system to pull additional data, otherwise “we don’t get a true picture of productivity. It’s a lot of manual work,” Pratt says.
In 2018, Pratt’s department conducted a self-study wherein time for concurrent coding was manually tracked on a spreadsheet. She’s now taking the lessons learned from that exercise and applying them to the creation of a new productivity report with 3M that pulls in all the necessary information, including data from Epic and the Kronos time-entry system, which monitors when an account was put on coding status, when it was complete, hours worked, etc.
“We don’t have firm productivity measures, but we are working toward that,” Pratt says. “When all is said and done, we will have four different productivity [measures] based on [complexity] levels. When you’re coding moms and babies, it doesn’t take as long as cardiology. On average, we expect six to eight accounts a day depending on what a coder is doing.”
Making It Count
Given the impact coding has on not only revenues but also quality scores and patient care, it’s important not to lose sight of the fact that accuracy trumps speed. Which is why ensuring coders have the tools and support they need to function at the top of their game is crucial.
Roberts, for example, has “guard rails” in place to support coders and help them gain confidence. These include clear and concise policies and procedures, and guidelines and job aides for coders to follow.
“These ‘guard rails’ can be guidance on workflows to coding sepsis to how to recognize when a query is necessary,” Roberts says, adding that “education should be at the forefront to ensure coders understand the query process, when to query, disease processes for assigning diagnosis codes, and anatomy for assigning procedure codes.”
On the coders’ part, they must take an active role in the process by retaining the education provided and ensuring that they are being proactive about asking questions to improve their skill sets. They should also have a system for keeping any feedback so they can refer to it as necessary.
“Doing all this can assist in building the coder’s confidence, which in turn assists in increasing a coder’s productivity,” Roberts says.
Pratt, who is also a huge proponent of ongoing education, is working to create an environment that isn’t overwhelming, beginning with a change to how they approach discharged not finally coded (DNFC) days. Previously, DNFC was managed on an end-of-the-week basis, meaning, for example, that all accounts discharged by October 8 had to be complete by the 12th.
“In the past, we would just say, ‘These are all the accounts that have to be coded.’ It could be hundreds, and more than half could be high-dollar accounts. That’s pretty overwhelming because [coders] don’t know where to put their focus,” Pratt says.
Now they focus on DNFC daily—when accounts on a certain date need to be complete and how many there are to be done for that date. “By the time Friday gets here, you should have only one day” of cases to code, Pratt says. “It resulted in a huge boost in productivity. It comes across as less overwhelming for the coders—more manageable and easier for them to prioritize.”
Ultimately, the important thing is to stress that while productivity is important and must be monitored, quality is always going to be more important, Pratt says.
“I’d rather take longer on a high-dollar account than rush through it,” she says. “In reality, if you put the focus on your quality, then you are better able to explain why your DNFC is high—if it is—by the end of the week. Leadership is very supportive about what we do, [which is] focus on quality vs quantity.”
— Elizabeth S. Goar is a freelance writer based in Wisconsin.
SOURCES OF INFORMATION AND EDUCATION
For coders who want to up their skills or simply make more sense out of the health care system’s wacky reimbursement system, there are a plethora of sources on- and offline of which they can avail themselves, starting with the Centers for Medicare & Medicaid Services (CMS).
“There are amazing resources on the CMS website,” says TruCode Product Manager Megan DeVoe, who has more than two decades of coding experience under her belt. “Once I got familiar with all the information on the CMS website, it helped me become a better coder because I understood the payment system, coverage determinations, etc. It just gives you a better understanding of the whole picture of the reimbursement system.”
(TruCode has published a CMS website resource guide, which can be accessed at www.trucode.com/resources-2/coding-education-resources.)
The CMS site is home to the Medicare severity diagnosis-related group (MS-DRG) definitions manual, which inpatient coders will find especially useful when trying to figure out why a DRG doesn’t make sense for a specific patient. It also provides useful information regarding the ICD-10-CM and ICD-10-PCS classification systems, including the most recent meeting materials for the ICD Coordination and Maintenance Committee pertaining to ICD-10-PCS.
Other pertinent information includes the following:
• CPT/HCPCS coding materials for reporting outpatient and professional services, including a Medicare physician fee lookup tool, the Medicare Claims Processing Manual, and access to the entire Medicare coverage database; and
• inpatient prospective payment system coding guidance, including MS-DRG guidance.
“One thing that a lot of people aren’t aware of is the cooperating parties, when creating codes, have the ICD Coordination meetings to discuss why we need the new codes, but they also give a lot a lot of background on how to code until the new ones comes into effect,” DeVoe says. “You can get great presentations from physicians who are proposing the codes themselves [and] a lot of very interesting background on why codes are going into effect. If you understand that information, then you understand when and why you need to code it. This helps to really understand what you’re doing.”
Other sources include AHIMA and its plethora of knowledge bases, free webinars, and free educational programs. Another little-known source of knowledge are medical device and drug manufacture websites, as well as YouTube.
“A big part of the health information professional is that we’re really curious people,” DeVoe says. “We want to understand medicine, and it’s always changing.”