Elsa Pearson is a senior policy analyst with Boston University School of Public Health. Follow her on Twitter: @epearsonbusph. Research for this piece was supported by the Laura and John Arnold Foundation.
The first post of this series defined upcoding – billing and coding for more intensive and expensive diagnoses and treatments than what was provided or medically necessary – and presented research on its prevalence. If upcoding is fraudulent or, at the very least, not ideal, and as common as the literature suggests, what can be done about it?
Matthew Fiedler of the Brookings Institution put it aptly when he told me via email that “the right policy response depends some on how you conceive of the problem.” Experts have an array of ideas on how to mitigate upcoding and none will defeat it singlehandedly.
A policy brief written for the American Medical Association’s Journal of Ethics suggested two strategies: medical education and front end analytics. The authors argued that medical education and training is the time to familiarize future physicians with upcoding and other fraudulent billing practices. (I would argue this concept should also apply to training for non-physician providers and administrative staff.) Their review of the literature showed that only one third of medical schools currently have any curriculum time dedicated to fraud and abuse. Without teaching physicians how to code and bill properly, it’s hard to expect them to learn these habits on their own.
Front end analytics seeks to catch fraudulent upcoding algorithmically. Medicare’s Fraud Prevention System already employs this approach. In 2014, the Fraud Prevention System saved Medicare over $210 million from inappropriate billing. It’s clear that front end analytics can work. However, it’s also clear that the current system isn’t doing enough. Medicare should expand its use of front end analytics and private insurers should adopt a similar approach.
Another proposal focuses not on auditing but on Medicare Advantage’s (MA)’s “coding intensity adjustment.” This benchmark of sorts allows the Medicare program to financially adjust (i.e., reduce payment) for common coding differences between traditional Medicare and MA. These coding differences are related to patient complexity-based payments: MA pays more than traditional Medicare for more complex patients, incentivizing MA providers to upcode.
Richard Kronick of the University of California San Diego argued in Health Affairs that the current coding intensity adjustment is smaller than what it could and should be and will lead to $200 billion in overpayments in the next decade. As it stands, the coding intensity adjustment is set annually by political appointees at the Centers for Medicare and Medicaid Services (CMS). Kronick contends it should instead be both mandated by law and methodologically determined so as not to be influenced by ever changing political agendas.
Lastly, Paul Van der Water of the Center on Budget and Policy Priorities suggested MA should not include diagnoses collected in health risk assessments in risk score calculations. Because MA bills based on patient complexity, the incentive is to attach more diagnoses to each patient. Health risk assessments, though intended to inform care, often influence the patient’s documented complexity by added new diagnoses. By omitting those diagnoses from risk score calculations, MA would paint a more accurate representation of its enrollees and reduce upcoding associated with patient complexity.
Van der Water notes that the Medicare Payment Advisory Commission (MedPAC) recommended this to CMS in 2016 but CMS has yet to act. MedPAC also recommended CMS use two years of diagnostic data in risk score calculations to ensure a more accurate representation of patient complexity.
While most of the above solutions are presented in the context of traditional Medicare and MA, many of the concepts are applicable to private providers and insurers as well. Moving the needle on upcoding will require a multifaceted approach, dependent on both regulatory oversight and buy-in from payers and providers.
