Like a detective, Pam Thomson probed the mysteries of CPT coding errors in the pulmonary medicine department at University of Virginia (UVA) Medical Center, looking for hard evidence of what went wrong and why. Were coding errors correlated with the time of day, day of the week, or workload? Was something amiss in the physician/coder interaction that produced the code? Were errors related to some fundamental misunderstanding of a specific type of code that caused consistent overcoding or undercoding?

After a lot of data analysis, Thomson, a nurse-turned-project manager, correlated errors to two main areas: (1) inadequate documentation for the CPT code that doctors use on admission day, and (2) coding for weekend patient encounters. So UVA revamped the pulmonary medicine coding process, increasing coder-physician face time so they can hash over trouble spots and shifting coders’ emphasis from concurrent physician coding to error-riddled physicians. A follow-up audit showed “near-perfect coding,” she says.

Just your everyday compliance-program experience? Not really. The path Thomson and her UVA colleagues took to untangling the riddle of the erroneous codes was paved by an approach to process improvement called Six Sigma. Pioneered by Motorola and also widely adopted at General Electric, Six Sigma is a data-driven method of identifying weaknesses in business functions, devising solutions and monitoring their continued improvement. It was embraced and evangelized at UVA by physician Thomas Massaro, M.D., chief of staff of the UVA Medical Center.

Six Sigma has five steps: define, measure, analyze, improve and control, says Thomson, a Six Sigma expert at UVA. These steps are taken to tackle very discrete issues.

“We have been struck by how small you have to get to attack and improve something,” Massaro says. “Six Sigma is the ultimate reductionist approach to the world. You get down to the genomic data file, the gene of a process, and what nucleic acids are involved in that. If you don’t narrow it to something manageable, you can’t get anything done.”

Massaro says the mechanics of Six Sigma are “not revolutionary.” But it is unique partly because it calls for “a more activist engagement by senior management.” Six Sigma “has been structured in a way that it is owned by executive management. This is the first time we have most senior people in the organization behind this, and we have a commitment to provide rewards and incentives that are consistent with behaviors in these programs. It is rigorous and the tools are powerful, but it works because people believe it will work.”

For example, each project has a team made up of a content expert and a program manager. The Six Sigma expert is called the Black Belt, and the person who is still mastering Six Sigma is the Green Belt. Massaro had to overcome resistance from some quarters because Six Sigma initially is not a money-making proposition.

ER Wait Times Among Other Projects

UVA has Six Sigma projects underway in the following areas so far:

    • Physician coding
    • The understandable bill
    • Discharge cycle time (reducing the amount of time it takes to release patients from the hospital after the discharge decision is made)
    • Appointment availability (reducing the number of days until a patient can get a clinic appointment)
    • ER wait times
    • MRI reports (turnaround time)
    • Appropriateness of case carts in the OR (i.e., when a patient has an operation, is the cart properly stocked?)

Each of these projects is tackled by a team with managers from relevant medical departments and content experts (i.e., internal audit, coding). GE trained UVA staff on implementing Six Sigma. Other medical centers, including Johns Hopkins in Baltimore, are hopping aboard the Six Sigma train, Massaro notes.

Tackling the Six Sigma Steps

The coding Six Sigma project focused on physician fees for inpatient care. Here’s how UVA implemented it:

    1. Define what you are looking at. UVA picked pulmonary medicine mostly because the department chair was receptive to his department becoming a Six Sigma guinea pig. The medical center focused on CPT codes selected by the faculty practice plan physicians and coders for treating inpatients.
    2. Measure. Decide what you are measuring. “Six Sigma is very quantitative,” Massaro notes. Decide how to measure it. For example, should you pick discrete measures (i.e., either yes, the code selected was right, or no, it wasn’t)? Or should you use continuous measures (on a scale of one to 10)? Thomson et al picked discrete measures and kicked off the number crunching by getting a baseline of the percentage of accurate CPT codes.
    3. Analyze. What is causing the errors? “You do a lot of brainstorming,” Thomson says. In the coding project, the team plotted a graph where the Y axis was the outcome and the X axis was the factors that move the outcome to either yes, the code was accurate, or no, the code was inaccurate. Then the team thought deeply about the drivers of the X axis, including the coder-doctor interaction that produced code; time of day; day of week; patient volume; certain type of code always overcoded or undercoded; and high vs. medium level causing problems.”Then there’s another round of data collection to measure outcomes and how each correlates to the outcome,” she says. From the items that correlate, they honed in on what was statistically significant. In this coding project, you run your statistics, and then look to see what is statistically significant. UVA found two things that were statistically significant:
      • The day of stay. CPT codes vary by day, depending on whether the patient was admitted, it was a subsequent care day or it was a discharge day. “We found that the admission day caused a great deal of our problems,” she says. Admission days pay more than subsequent days; and
      • The day of the week. More coding errors occurred when physicians saw patients on Saturday or Sunday than if they treated them on Monday through Friday.

      “The day [Pam] figured out the day of the week issue, there was a euphoria that was infectious,” Massaro recalled. “The difference is you get past the traditional form of anecdotes and speculation, and you have a bunch of numbers that come together using mathematically precise techniques.”

    4. Next up: a root cause analysis to find out why. With admission days, the Six Sigma detectives found that “overall, the physicians didn’t know what they needed to document to get that admission code correct,” Thomson says. This was partly due to inadequate linkage between the residents’ documentation and attending physicians’ documentation. Regarding Monday coding, it turned out that coders have three days worth of coding to accomplish on one day because they don’t work weekends. At UVA Medical Center, physicians select codes themselves, but coders (for internal medicine) check every selection. Because coders are off weekends, they have to figure out what the doctors did after the fact.Essential to this data gathering and analysis is to have the Black Belts “riding herd on the quality and content of the database. This makes it possible to get incremental breakthroughs,” Massaro says. To get to the root cause, Thomson runs a sophisticated mathematical program, called a chi-squared analysis, to match up whether the drivers correlate with the result. “We found that admit days and Saturdays and Sundays were correlated to coding inaccuracies,” she says.
    5. Improve. Once you know where the errors are, you develop a process to improve – in this case – coding based on your concrete data-fueled findings. At UVA, coders and physicians already code together Monday through Friday, so “we tweaked the system,” she says. “We devised a better communication method.” Coders and doctors meet twice weekly in brief mandatory discussion and education sessions. Coders suggest ways to fix problems, either on a chart still in progress or moving forward.Until Six Sigma, coders tracked and checked every single physician code. Now it’s more triage oriented. “Physicians are slowly being weaned from coders,” Massaro says. Coders will only conduct daily reviews for codes selected by physicians who are poor coders, while physicians with a good track record will be reviewed less often (i.e., every couple of days or weeks, depending on the physician’s relative coding aptitude). That way, Mondays aren’t a nightmare for coders, and their propensity for errors should drop.”You are still checking in and auditing. But coders are changing from doing 100% coding to a role more like an auditor and educator,” she says.
    6. Pilot the process. “We did a two-week pilot of formal coder-physician face-to-face communication every day,” with the coders actually coding their charts once a week, Thomson says. And each coder is assigned to one doctor or set of doctors to ensure expertise with a particular set of codes, instead of the previous practice of coders cross-covering for each other. “Codes were audited, and they were almost perfect. It was astonishing,” she says.Physicians have been weaned off coders somewhat, “and by the end of the two-week period they were coding pretty much accurately almost all of the time. Physicians showed dramatic improvement over their own ability to code correctly. Coders still come in as oversight but by the end there is very little the coder has to fix.”
    7. Control. How do you change the systems and structures so benefits are sustained long term? “Quality improvement can backslide over time,” Massaro cautions. “In this area, doctors may get excited for awhile, but how do you get them to keep improving coding?” So UVA created a control plan. The Six Sigma experts will keep track of the pulmonary medicine department coding accuracy in a database. “You graph it and you watch, and if they fall below some defined number – like 90% – then you gear backup,” Thomson says. “At some trigger point, a specific doctor will need the coder back to follow him and investigate why they fell off and what happened. It’s ongoing monitoring.”

Expect Surprises

One thing Massaro and Thomson have learned is that Six Sigma will reveal all sorts of things about people and processes they never anticipated. For example, when they surveyed physicians about improving billing and coding, Massaro and Thomson assumed they would hear physician gripes about not getting paid enough or the burdens of coding and documentation. But instead, physicians focused on making bills more understandable to patients, and the understandable-patient-bill project was born.

“That was a huge revelation to us,” Thomson says. Physicians, they speculate, don’t want to waste time answering patients’ billing questions about their Rube Goldberg-like bills, and perhaps don’t want to appear stupid to patients because they can’t answer them. “So we drilled down further to find out what patients don’t understand about their bills, and had another revelation:” mistakes on the bill, such as an inaccurate diagnosis, prompted most patient calls about billing – not confusion or bafflement. Thomson says it was a welcome insight because they can fix those errors.

“That was a total ‘Aha!’ for me,” Thomson says.

And it “taught us the power of the data to show us the truth,” Massaro notes. “Before we did it on instinct in terms of picking projects and the scope of projects.”

About The Article
This article has been reprinted with permission from the REPORT ON MEDICARE COMPLIANCE, the nation’s leading source of news and strategic information on false claims, overpayments, compliance programs, billing errors, and other Medicare compliance issues. For more information read the REPORT ON MEDICARE COMPLIANCE.

About the Author