Data Banks or Garbage Bins?
By Robert S. Gold, MD
For The Record
Vol. 28 No. 3 P. 6
Recent issues of the Journal of AHIMA have paid a great deal of attention to data, whether it be data entry, data assets, data retrieval, data analytics, or data integrity. The editorial sings the praises of the EHR—despite the fact that interoperability is a myth—computer-assisted coding (CAC)—despite the fact that EHRs destroy any chance of retrieving meaningful codes—and how clinical documentation improvement (CDI) boosts the veracity of patient health information—despite the fact that most CDI programs are driven more by diagnosis-related groups (DRGs) and less by data accuracy.
The health care industry has entered the ICD-10 universe with exactly the same problems it had in ICD-9. Nothing has improved nor will it until those in leadership roles wake up. So long as the industry remains focused on the bottom line, no amount of data on conditions and diseases will lead to a holistic approach to patient care.
We are our own enemies and we're letting others control our destiny.
Sure, an individual physician, hospital, or health system may achieve process improvement, generate cost savings, develop communication to improve patient throughput, and promote purchasing power in certain treatment modalities. That's noteworthy, but it's isolated initiatives that lead to these milestones.
It's the comparative processes that result in skewed severity-adjusted payments, severity-adjusted mortality rates, and complication statistics. The problem lies with patient databases rather than individual initiatives. Data are corrupt, making it nigh impossible to use patient information for any purpose. In other words, data abstraction from ICD databases is a myth.
The industry emphasizes the need for internal and external audits to ensure ICD coding accuracy, but the documentation provided is often trash. Physicians take shortcuts or don't know how to create a reasonable evolution of their disease documentation using the code-finders in the EHRs. Then, coders feed this inadequate information into software that talks a different language.
Whether coding is performed by individuals or computer-assisted software packages, there are major errors in code construction and the systems themselves. So-called "smart tips," stratified questions, and a lack of codes to describe the diseases actually treated lead the coder to assign inadequate, unspecified, or incorrect codes. Take the case of a patient with a prosthesis in place who suffered a fracture due to a fall. The physician identified it as a periprosthetic fracture, mentioning that there was no faulty hardware. The coding software led to documentation of a periprosthetic fracture (the site isn't there in the code) and a second code for the site in the femur. But the case was of a traumatic fracture in a patient who had a prosthesis in place—a totally different scenario. In this case, the coder didn't stand a chance of reporting the correct codes.
Why does the system need retooling? Two reasons.
Speaking the Language
First, code groupers are based on ICD codes, coding rules, includes and excludes notes, Coding Clinic advisories, and some sort of "smart tip" mechanisms that are built into the software packages.
However, physicians don't talk in codes or use coding terminology. Software can misinterpret or ignore physicians' words, or worse yet, take documentation from one part of the medical record and interpret it to be from another section. For example, family history diseases can be coded as being the patient's, and past diseases can be coded as current.
CAC vendors stress that coders must consider all highlighted terms and be aware of colors, bolds, and italics. They say their technology is not designed to replace coders, yet they sell their products as a means to decrease staff.
Another drawback to CAC technology is that it can produce lazy coders. Coders who have left the hospital environment to work at home, where they prepare the kids for school, cook dinner, and take a break to get their favorite cappuccino, seem to take less time per record than if they were still at the hospital coding records with an occasional trip to the coffee pot.
Copy and Paste
The second reason for coding inaccuracy stems from the copy-and-paste mentality, a disease that developed because EHRs were mandated long before the technology was properly tested live by physicians to identify potential roadblocks. Physicians had to learn how to use an EHR, make sense of its relationship to meaningful use, and commit to organizational changes while at the same time seeing patients. As a result, they developed workarounds to get through the day.
Physicians discovered that they could use copy and paste to fill EHR templates—and use the resulting volume to charge a higher evaluation & management level. Physicians also can copy and paste paragraphs from other physician notes and bill as if they had done the work.
Differential diagnoses on admission now last through the entire stay and diagnoses that were once ruled out now are repeated numerous times in a single patient record. The coding software recognizes it, so it must be correct. As a result, databases are filled with junk from the EHR and the coding software.
While the concept of CDI is marvelous, it too often is implemented improperly. In the 1980s and '90s, CDI programs were developed in response to the transition to a DRG-based system in which coders had to derive diagnostic information from clues in the medical record. Most hospital programs focused on extracting additional dollars from Medicare. In this regard, CDI was invaluable.
However, while processes and coders improved, CDI failed to evolve. Despite being forewarned about the arrival of value-based purchasing for 10 years, some CDI vendors still emphasize Medicare DRG enhancement, essentially ignoring the fact that more than 50% of the patient population consists of non-Medicare patients. They seek comorbid conditions (CCs) and major CCs; they manufacture sepsis and acute respiratory failure. They make up rules and mandates, attack their physicians—and the physicians cave in.
To compound matters, hospitalists are taught some of the same material the consultants teach in their "CDI" programs. As a result, patients with alteration of consciousness or stroke have "encephalopathy," and those with a higher creatinine than normal (and even those with normal creatinines) have "acute kidney injury." Nursing home patients have "possible or probable gram-negative pneumonia," and every patient exiting surgery is deemed to have "acute blood loss anemia."
Payers deny those payments and hospitals retaliate by commanding all hands on deck to defend the denial. Recovery audit contractors also demand money back, resulting in more hospital resources being mobilized to defend a position taught to them as righteous when it was actually bogus.
In general, the data in data banks are corrupt, inaccurate, and unethical. Sadly, no one wants to do anything about it. CDI program educators, knowing what's right, still teach the old ways. National coding organizations, knowing how the errors are generated, won't help the coding professionals. The institutions that create the rules are so embedded in their self-protection that they won't take a stand to fix it.
To correct an ICD error takes an act of Congress. On second thought, even a helping hand from Congress may not remedy the situation. I have spoken with several senators and congressmen who say there is so much inertia in the Centers for Medicare & Medicaid Services that they won't waste their time trying to rectify a lost cause.
The health care industry talks about statistics and the meaningful use of medical information being a game-changer. However, it's failing to recognize that too many of those data are flawed, setting up the industry for massive failure. And because there is so much money tied up in a broken system, no one wants to address the problem.
It's time for that to change.
— Robert S. Gold, MD, is cofounder of DCBA.