We recently sat down with Jim Bohnsack, Chief Strategy and Client Officer at Aspirion, to discuss a problem that’s hitting healthcare providers nationwide. Clinical denials—where payers question the medical necessity of care after it’s been delivered—are exploding in both volume and complexity. What makes this particularly challenging is that it’s exacerbated by providers not having enough staff with the right expertise to respond effectively. Here’s our conversation, lightly edited for clarity.
Q: Jim, let’s start with the big picture. What does the denials landscape look like right now for healthcare providers?
Jim Bohnsack: It’s pretty intense, honestly. At Aspirion, we partner with 12 of the 15 largest health systems in the country, plus providers of all sizes nationwide. So, we have a really comprehensive view of what’s happening across the denial landscape—and we’re seeing trends that should concern every CFO and revenue cycle leader in healthcare.
The numbers from recent industry reports back up exactly what our clients are experiencing—things are getting harder, not easier. Initial claim denial rates hit 11.81% in 2024—up 2.4% from the year before. But what really tells the story is the trajectory. In 2022, 30% of providers said at least 10% of their claims were denied. By 2024, that jumped to 38%. Now in 2025? 41% of providers report denial rates over 10%.
What we’re seeing across our nationwide partner base confirms these trends—it doesn’t matter if you’re one of the largest health systems in the country or a community hospital.
And the financial impact is real. Providers are collecting $3 less for every $100 compared to last year. AR days are up more than 5%. Collection rates from insured patients dropped from 37.6% to 34.5%.
Q: Payers say they’re reducing prior authorizations to reduce friction. But something else is happening, isn’t it? What’s really driving these trends?
Jim Bohnsack: Yes, that’s what they’re saying publicly. And the data backs it up—prior authorization denials actually dropped 7.7% in 2024. Sounds great, right?
But here’s what’s really happening: they just moved the scrutiny. Medical necessity denials went up 5%, and requests for more information jumped 5.4%. They shifted from front-end controls to back-end payment integrity processes using large language models (LLMs) to review clinical documentation after the fact.
So instead of fighting for approval before you provide care, now you’re fighting to keep payment after you’ve already delivered the care. Same friction, different timing. And in some ways, it’s worse because you’ve already invested the resources.
There’s always been this information imbalance between payers and providers. Payers have a horizontal view across all claims; providers have deep clinical knowledge about specific patients. Historically, payers managed this through front-end authorizations. Now with LLMs, they can ingest entire medical records on the back end, apply rule sets, and question clinical appropriateness after the fact. They deny the claim, request the records, load them into their AI systems, and send back “we disagree.” Then it’s on you to respond with that same level of sophistication.
Q: How are these clinical denials different from what providers dealt with before, and what does it actually take to respond to one?
Jim Bohnsack: Night and day. The old denials were often administrative—wrong modifier, missing information, codes that don’t go together. Your business office could handle those. Tedious, but solvable.
Clinical denials are a completely different animal. The payer is saying “I don’t agree with you. This wasn’t medically necessary” or “This wasn’t as complex as you claim.” Here’s the problem: your normal business office staff can’t do this work. You need specialized expertise—attorneys, clinicians, certified coders, people who understand medical terminology, care guidelines, coding rules, and can build a clinical argument.
Let’s say you get a denial saying a patient admission wasn’t medically necessary. Your clinical expert has to pull the complete medical record—typically 500+ pages, sometimes thousands for complex cases. Review that entire thing to find every piece of clinical evidence supporting the admission. Match all that evidence against multiple sets of rules—care guidelines, coding guidelines, the payer’s specific policies, your managed care contract terms. Understand the clinical nuances of why this particular patient needed this level of care. Then build a comprehensive, evidence-based argument and write a formal appeal letter that’s clinically sound and compelling.
This takes hours for a complex case. When you’re getting hundreds or thousands of these denials, the math just doesn’t work. You can’t keep up with manual processes.
Q: Talk about the volume you’re seeing and which types of claims are getting hit hardest.
Jim Bohnsack: The volume is exponential. Every provider type, everywhere, is seeing it. Most providers don’t have clinical or legal resources sitting around waiting to write appeals all day. Their nurses are busy with patients. Coders are handling regular workflows. You’d need to hire an army of nurses and coders just to keep up—and good luck with that in today’s job market.
We’re seeing it across the board, but some areas are particularly brutal. DRG downgrades are huge—especially anything related to sepsis. There’s this endless back-and-forth about whether a patient actually had sepsis, and we’re talking about high-dollar claims. Inpatient versus observation status is another big one. Disputes over whether procedures were medically necessary. The complexity is overwhelming because every payer has different rules, different policies, different contract terms.
This is fundamentally not a human-addressable problem anymore at this scale.
Q: Given all this—the volume, complexity, and staffing challenges—how did you approach building a solution?
Jim Bohnsack: Look, AI becomes essential here. Not a nice-to-have, a necessity if you want to stay financially viable. You physically cannot hire enough staff to deal with this volume and complexity.
At Aspirion, we realized this over two years ago. We acquired a company and put about 35 technical experts on this full-time to build, maintain, and expand it. It does what our experienced clinical staff does, just at machine speed and scale. Our attorneys and clinicians have always done this work—pulling records, reviewing them, comparing against guidelines, building appeals. But there just aren’t enough of them to handle the volume.
Our AI, Doc IQ, pulls the clinical evidence from those 500-page, sometimes thousand-page records. Matches that evidence against payer policies, care guidelines, coding guidelines, and managed care contracts. And generates a comprehensive appeal with all the clinical documentation and arguments laid out. The machine doesn’t miss stuff. Even really experienced staff reviewing a 500-page record might miss something buried in there.
We were really deliberate about it. Our baseline was: this has to be as good or better than what a human can do. We rolled pieces out gradually—starting with structuring medical records to make them easier for humans to search, then taught the system to answer standard questions our reviewers always ask. Each step made our staff more efficient while we validated quality.
Q: Can you give us a specific example of how this works and what results you’re seeing?
Jim Bohnsack: Sure. We started with DRG downgrades for sepsis—something that was hitting clients hard everywhere. High dollar denials, happening frequently, really frustrating. We focused our AI on building better appeals specifically for those cases and saw significant improvement in overturn rates. That validation gave us confidence to expand—first to single code DRGs, then all disputed codes, then things like inpatient versus observation status.
The numbers have been impressive. In January, Doc IQ handled about 2,000 appeals out of 20,000 clinical appeals total. Last month? Over 14,000—more than two-thirds of everything we process. The platform handles about 70% of the different clinical denial types we see now, and we’re on track for 90% next quarter.
But here’s what really matters: We’ve cut 38 days out of the process from when we get a case to when the appeal goes out—that’s 2.2x faster. Same staff, way more throughput.
Q: So, you’re essentially solving the volume, complexity, and staffing problems simultaneously. Yet we hear only 14% of providers are using AI for denials. Why the gap?
Jim Bohnsack: Yes, you’re getting the output of multiple experienced clinical staff members without having to recruit, hire, train, and keep those increasingly hard-to-find people. And our clients across the country are seeing these results—fewer appeals required per denial, better success rates, and cash coming in faster.
As for why more aren’t adopting? 62% of providers now understand what AI actually is versus regular automation—that was only 28% a year ago. So, people are learning. And 67% think AI can improve the claims process.
But there’s still uncertainty about implementation, ROI, whether it’ll work for their specific situation. Plus, only 56% think their current technology is adequate—down from 77% three years ago. People are tired of technology that doesn’t deliver. The reality is, providers are already understaffed and drowning in work. They’re hesitant to add another technology project, even when the data says it works.
Q: What would you tell a CFO or VP of Revenue Cycle who’s buried in clinical denials right now?
Jim Bohnsack: First, you’re not alone. Accept that you can’t solve this with people alone. The volume and complexity have blown past what manual processes can handle. You need to match what the payers are doing.
Second, figure out what your specific problem is. Can’t keep up with volume? Can’t find or afford the clinical staff you need? Success rates aren’t good enough? Define it, then measure it.
Third, don’t expect to fix everything overnight. We didn’t automate everything at once. Small wins add up. Every step that makes your existing staff more efficient gives you breathing room.
And finally, do the math. If you’re spending hours of expensive clinical staff time per appeal, times hundreds or thousands of denials, compared to getting 38 days back in your AR cycle with better outcomes? The ROI becomes pretty obvious. Factor in the time and cost to build versus partner, and the decision gets even clearer.
