There are countless factors that contribute to the outcome of a clinical trial. Patient self-reporting relies on honest and accurate data from people outside of the medical field. Physicians can overlook issues or miss symptoms. Both parties can provide inaccurate information if the medical coding is wrong.
Medical codes are meant to standardize symptoms and treatment across the entire country, making data analysis easier and cleaner. However, there are still medical code errors that can cause researchers to present poor data.
Incorrect Clinical Coding Can Put a Trial At Risk
There are tens of thousands of medical codes related to patient care — with more than 70,000 codes in the International Classification of Diseases alone. It’s not uncommon for professional medical coders (who by and large are excellent at their jobs) to mislabel a code or forget to add a modifier. Additionally, if the physician fails to write down accurate information, the coder might not realize the modifier is needed. These small labels can have a significant impact on patient treatment.
“A missing 2-digit modifier for a knee injury could result in having an MRI study conducted on the wrong knee,” the team at Merem Healthcare Solutions explains. “Or, imagine an incorrect diagnosis code causing an obstetrician to withhold pain medicine during a woman’s pregnancy. This would be an extreme case, but altogether terrifying and harmful for the patient.”
While poor clinical coding can have serious consequences for patients, it can also limit the effectiveness of clinical trial data — and may force researchers to rereview the data or even discard incorrect data sets.
“Bad data represent one of the most egregious of themes of errors because there is typically no correct way to analyze bad data, and often no scientifically justifiable conclusions can be reached about the original questions of interest,” write Andrew W. Brown, Kathryn A. Kaiser, and David B. Allison, from the Office of Energetics and Nutrition Obesity Research Center, University of Alabama at Birmingham.
The need for accurate clinical coding is rising as clinical trials are becoming more decentralized. Researchers need to trust the clinical research organizations (CROs) to accurately collect and report medical codes.
“More widespread use of decentralized trials will generate even larger quantities of unstructured data,” says Mary Varghese Presti, global vice president of life sciences at IBM Watson Health. “In an aging population where as many as four in ten adults have two or more chronic diseases, it can be increasingly complex to apply medical codes accurately and efficiently.”
It’s not just the patients or the insurance companies that rely on accurate coding. The future of medical breakthroughs demands accurate data records.
Teams Need to Stay on Top of New Medical Codes
One of the most important strategies for tracking accurate clinical codes is to stay on top of annual code changes.
“The organizations that maintain the three principal medical coding code sets (the WHO for International Classification of Disease, the AMA for Current Procedural Terminology, and the CMS for the Healthcare Common Procedure Coding System) update these manuals yearly,” according to MedicalBillingandCoding.org explains. “It’s up to coders to learn any new or reorganized codes as they come out, and use them correctly.”
Any code changes and new code additions can significantly affect patient records and trial reporting. Some changes might also be farther reaching than you realize.
For example, the American Academy of Pediatrics Division of Health Care Finance created a guide for coding COVID-19, based on codes created by the CDC in early 2020. This allowed physicians to record instances of exposure and whether or not the patient experienced COVID-19 symptoms. The clinical coding also has line items for symptoms, such as pneumonia brought on by COVID-19 and acute bronchitis.
Regardless of whether your clinical trial centers around COVID-19, you will likely want to know if your patient has had it in addition to any underlying symptoms. These codes guide physicians to record the current state of a trial participant’s health.
There Are Already Updated Codes for Telehealth
Medical codes don’t just log patient symptoms. They are also used to track the process of patient care. While many hospitals use these codes for billing, they can also help ensure the accurate nature of clinical trials.
For example, there are already updated guidelines to record telehealth meetings with patients, a practice that skyrocketed at the start of the pandemic and is likely to continue.
“One of the things we did that was very successful in terms of telehealth is, we decided, as a team of coding educators, to develop a grid that had different information about the methods of communications with the patient,” says Ginna Evans, coding educator for IM specialties at Emory Healthcare. Questions on the grid included whether consultations happened over Zoom or by telephone; what the documentation requirements were; and the types of attestations required in the notes.
At an AMA symposium in 2020, David Kanter, vice president of medical coding at Mednax, shared how his organization and many other medical leaders have adopted six different codes in the past year dealing with communication and telemedicine. For example, a medical provider can’t use an e-consult code when the only purpose for the communication is to transfer the care of the patient to another physician.
If clinical researchers are concerned about outliers in data sets, they can turn to clinical coding to better understand the patient interactions with physicians. There may be coding issues related to the physician misrecording information while they take notes or missing information that would otherwise be important.
Researchers Are Turning to Artificial Intelligence for Coding
Another strategy for improving the accuracy of clinical coding is the use of digital tools to assign codes, like the AutoEncoder feature by Anju Software. Digital systems and artificial intelligence tools can “read” physician notes and assign codes while checking for accuracy.
“No human being can remember all the codes for diseases and treatments, especially as the number of codes has climbed over the decades to tens of thousands,” write Thomas Davenport and Steven Miller, cofounder of the International Institute for Analytics, and consultant to Integrated Health Information Systems and to the provost’s office at Singapore Management University, respectively. “And it is not just a matter of finding the right code. There are interpretation issues… There is often more than one way to code a diagnosis or treatment, and the medical coder has to decide on the most appropriate choices.”
There are many benefits to using digital systems to code clinical data. These tools have checks and balances to ensure accuracy and can process several hundred pages of data in a short period of time.
“Beyond processing high volumes of data, the potential of AI to significantly reduce human error is another advantage of implementing AI-driven solutions,” says Raghav Bharadwaj at Emerj Artificial Intelligence Research. “Additionally, reducing the amount of company time required to manually complete the medical coding and billing processes is also important to consider.”
Physicians and researchers have already moved toward digital clinical management, from electronic data capture systems to medical records. However, these existing systems can still be improved, and this is where digital clinical coding comes in.
“Today, 96% of hospitals have adopted [electronic health records], up from just 9% in 2008,” journalists Fred Schulte and Erika Fry write at Fortune Magazine. “But on most other counts, the newly installed technology has fallen well short. Physicians complain about clumsy, unintuitive systems and the number of hours spent clicking, typing and trying to navigate them — which is more than the hours they spend with patients.”
Clinical trials need accurate and updated patient coding systems. Without the latest codes and advanced technology to check for errors, researchers risk presenting poor, inaccurate data that draws the wrong conclusions. A few coding errors can be dismissed as outliers in a large dataset, but in small clinical trials and in early-phase processes, every patient record matters.