Representation of New HCPCS Product Codes in Public Medicare Fee-for-Service Data
This study reveals that only 39.6% of new HCPCS product codes appear in public Medicare fee-for-service datasets within their first two years, demonstrating that code effectiveness does not guarantee data representation and highlighting a critical gap researchers must account for when analyzing claims data.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
In the United States, the vast majority of research into new medical technologies relies on a massive, public ledger of insurance claims. When a doctor treats a patient, they submit a bill that includes a specific code identifying the device or procedure used. Researchers download these public records to count how often new technologies are used, who gets them, and how quickly they spread. This system works well once a technology is established and its code appears in the released files. However, a critical gap exists in the process between a code becoming officially active and that code actually appearing in the public data researchers can see. A code can exist on paper, approved for use, yet fail to show up in the public files for reasons that have nothing to do with whether the medical product was actually used, covered, or paid for.
This gap matters because it creates a blind spot. If a researcher builds a study using only the codes they can find in the public files, they are automatically excluding any new technology that, for administrative reasons, never made it into those specific records. The question is not whether the technology works or if patients received it, but simply whether the public data infrastructure is capable of showing it. Without knowing the size of this missing piece, researchers cannot know how much of the real world they are actually seeing.
A researcher at Rosenheim Technical University of Applied Sciences set out to measure exactly how large this blind spot is. The study focused on a specific group of new medical products that received official codes in the spring of 2021, 2022, and 2023. These were not drugs or complex procedures, but tangible items like wound dressings, prosthetics, and medical devices. The researcher took a list of forty-eight of these new product codes and checked three different public government files to see if any record of them existed. The check covered the year the code became active and the following year, looking for even a single national record of the item being billed.
The results revealed a significant disconnect between the official existence of a code and its presence in public data. Out of the forty-eight new product codes tested, only nineteen appeared in the public files at all. This means that for nearly two-thirds of the new products, the public data showed absolutely no trace of them, even though they were active codes. The absence of a record did not mean the products were unused; it simply meant the public files did not contain them. Of the nineteen that did appear, eleven showed up in the file for the year the code started, while eight only appeared in the file for the next year. One product that appeared in the first year's file had vanished from the second year's file, showing that the data state is not permanent.
The study also found that where the data did appear, it was not evenly distributed. The public files for doctors and practitioners contained records for eleven of the products, while the files for durable medical equipment suppliers contained records for eight. The public files for hospital outpatient departments contained no records for any of the new products, suggesting that for this specific group of items, the hospital data channel was effectively silent. When the researcher looked closer at the types of products, those designed for wound care and skin replacement were much more likely to appear in the public files than standalone devices or their components. Similarly, products that had a specific price listed when they were introduced were more likely to show up than those with an undefined price.
These findings challenge a common assumption in health research: that if a code exists, it will be visible in the data. The study demonstrates that code effectiveness and data representation are two separate administrative events. A code can be fully active and usable by doctors without ever generating a row in the public files. This means that studies which select their subjects based on what is already visible in public claims are inherently biased, as they can never see the products that fall through the cracks of the reporting system. The researcher concludes that future studies must treat the absence of a public record as an unresolved mystery rather than proof that a product was not used. To do otherwise is to mistake the limitations of the data for the reality of medical practice.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.