Coding and documentation: Not a ‘sexy’ topic—but a crucial one in value-based care

Linda Gates-Striby raised some eyebrows at the American College of Cardiology’s Cardiovascular Summit in February when, in one PowerPoint slide, she showed how much proper documentation and coding could affect a hospital’s quality measures, bottom line—and even a doctor’s reputation. 

Striby used a hypothetical example of a 76-year-old woman with diabetes and heart failure. If clinicians only documented basic information about this patient, her expected cost of care per month would be in the ballpark of $5,000. But by coding more specific designations for the diabetes and heart failure—and noting an interaction between the conditions—the total risk score value would balloon to $13,554 per month.

“Sometimes it’s not a matter of working harder but making sure it’s accurately capturing what you’re already doing,” said Gates-Striby, the director of corporate compliance at St. Vincent Medical Group in Indianapolis who led multiple sessions on coding at the CV Summit.

This issue is only growing in importance with medicine’s shift away from fee-for-service payment models and more toward value-based models. For reimbursement in the value-based models, payers including the Centers for Medicare & Medicaid Services typically use some form of risk-adjustment to come up with a target price.

Facilities are then evaluated based on their quality and cost-effectiveness against that benchmark, so if a hospital or physician commonly undersells the severity of patients’ illness, it’s easy to see how a doctor or facility could be perceived as providing low-value care.

This can have a deeper impact, too, as high-performing centers are eligible for bonus payments but low-performing ones can be hit with penalties.

“The natural inclination for a lot of us who are busy who take care of sick patients, is to say, ‘I take care of the sickest patients,’” said Ty Gluckman, MD, the medical director of the Center for Cardiovascular Analytics, Research and Data Science and a clinical cardiologist at Providence St. Joseph Health in Portland, Oregon.

“Some systems have made substantial investments in analytics to be able to share information back with clinicians to say, ‘I know you may think that you’re taking care of the sickest patients but in fact the data we’re looking at doesn’t suggest you do.’ It is hopefully a nonpunitive sharing of data openly that will drive behavior change to say, ‘Oh my. I never realized how big of an impact what I document is.’”

More important, but also more complex

Documentation and coding is “not necessarily a sexy topic,” Gluckman acknowledges, and it’s sometimes difficult to convince busy physicians it’s worth their time to master it.

An additional hurdle is the growing number of available codes and the struggles some clinicians have in selecting the correct one within the electronic medical record (EMR). The current system, ICD-10, has about 70,000 codes—up from about 14,000 in ICD-9.

“A growing challenge for this is that most clinicians—and I think they’re justified in saying this—didn’t go through their medical training to become a coder,” Gluckman said. “It was challenging enough when we lived in ICD-9 to understand what were the correct codes, how should I best be documenting, and with a significant expansion in the number of codes, you can only imagine it adding insult to injury in terms of people being more perplexed by how best to code overall.”

Gluckman said he first became interested in documentation and coding because he recognized some doctors he regarded highly—whom he would trust to treat his own family members—could easily look like subpar providers based on data analysis. The quality of care wasn’t the problem; it was capturing all the comorbidities of their patients with the proper codes because only specific ones contribute to risk-adjustment.

For instance, “angina” or “unstable angina” adds to the risk-adjustment score, but a simple “chest pain” designation does not. The same applies to “morbid severe obesity” versus unspecified obesity.

Gates-Striby said a lack of education is often to blame.

“As I speak with providers they very quickly grasp the concept of the risk-adjustment and the hierarchical condition coding, but people aren’t getting the message to them. At least that’s what they seem to be telling me is, ‘Why am I just hearing this?’” she said.

Money left on the table

In general, Gates-Striby said everything that factors into a provider’s clinical decision-making is worth documenting. She uses the acronym “MEAT” to remind people she trains about what types of care may be billable and should be documented: anything related to monitoring a disease, evaluating results, assessing or addressing tests or records and treatment.

Clinicians are often surprised to find out a simple three-minute conversation about stopping tobacco use is billable. It might be considered small change at around $14 per case, but add that up over all of the physicians in a large system and it makes a significant difference, Gates-Striby said. And most payers reimburse for that, so providers shouldn’t worry about adding an extra cost for the patient.

Support tools

Health systems have invested in both human and technological support tools to make coding as complete as possible.

Some hospitals have hired nurses and other employees with a strong clinical background to review patients’ charts and double-check with care team members about whether they forgot to list a condition. This is a helpful safeguard, Gluckman said, but the responsibility remains on the physician or advanced practice provider to have complete documentation in the first place.

“At the end of the day—even though there were a lot of people trying to assist the clinician in terms of improving the specificity of their documentation—if their notes don’t reflect it, it’s as if it never existed,” he said.

Other support tools within the EMR have been designed to provide figurative taps on the shoulder if it appears something is going undocumented. For example, natural language processing could recognize when heart failure was mentioned in a note but not documented properly. Other software can scan lab data, vital signs and other information within the EMR and prompt a provider to list a condition which may have been overlooked.

Gluckman said health systems have disproportionately invested in these tools on the hospital side versus the clinic, but accurate risk-adjustment is crucial in both settings. After all, it affects the system’s financial health and how competent its providers look.

“There have been real teeth now with associated penalties that have been borne by hospitals, clinics, etc.,” Gluckman said. “In contrast to in years past where you might get a pass for just reporting quality data or participating in a quality program, now it really comes down to the data and evidence of the care that you provide, so documentation has become much more important as a result.”