The single most expensive mistake a large provider makes on a refused claim is starting at the wrong rung. A concurrent medical necessity denial that could have been reversed in a fifteen-minute peer-to-peer conversation on day two becomes, by day forty, a written appeal against a closed record, argued by someone who never met the patient. The clinical facts did not change. The available remedy did. Almost everything that separates a revenue cycle operation that recovers most of its overturnable denials from one that recovers a third of them comes down to routing and timing, not eloquence.
Sort the denial before anyone writes a word
A barely adequate shop treats every denial as an appeal. A good one spends the first hours deciding whether it is an appeal at all. Roughly speaking, refused claims fall into four buckets, and only one of them belongs in the formal appeal pipeline.
- Technical and administrative rejections. Missing modifier, wrong subscriber number, coordination of benefits not on file, taxonomy mismatch. These are corrected claims or reconsideration requests, not appeals. Filing an appeal here burns an appeal level you may need later and adds thirty days.
- Authorization and notification defects. No auth on file, late notification, auth for the wrong site of service. Some are genuinely curable through a retro-auth pathway the payer does not advertise. Others are contractual write-offs. Know which contracts have a cure window before you argue.
- Clinical denials: medical necessity, level of care, length of stay. This is the bucket that needs physician involvement and the bucket where sequence matters most.
- Downgrades and DRG validation. Inpatient reclassified to observation, or a DRG reduced on coding review. These blend clinical and coding argument and need both signatures on the response.
Miscategorization is the quiet killer. A level-of-care denial mislabeled as a coding issue lands on a coder's queue, gets a template letter, and comes back upheld. The overturn was available. Nobody asked for it in the right voice.
The order, in practice
For a commercial or Medicare Advantage inpatient denial, the ladder generally runs like this, and each rung has a shorter clock than the one after it.
- Concurrent reconsideration or peer-to-peer, while the patient is still admitted. The window is often measured in days, sometimes one business day from the adverse determination. This is the cheapest reversal available anywhere in the process, because the treating physician can describe what is happening on the unit right now rather than reconstructing it from a chart.
- First-level appeal, post-discharge. Written, with the record attached, filed inside the contractual deadline. Typically sixty to one hundred eighty days depending on the agreement, and the deadline in your contract governs over whatever the denial letter says.
- Second-level appeal. Some contracts require it before external review, others make it optional. Skipping a required level is a jurisdictional loss, not a substantive one, and it cannot be fixed afterward.
- External or independent review. For Medicare Advantage, this is the independent review entity route with further steps beyond it, including an administrative law judge level. For commercial plans, it depends on whether the coverage is state-regulated or a self-funded ERISA plan, and the two have different reviewers.
- Contractual dispute resolution or arbitration. Where the argument is no longer about the patient's clinical status but about what the contract obliges the plan to pay.
The Centers for Medicare & Medicaid Services oversees the appeal framework that governs Medicare and Medicare Advantage disputes, and that framework is prescriptive about deadlines and levels in a way commercial contracts often are not. A large organization is running both frameworks at once, on the same patient population, with different clocks. That is why a shared calendar beats institutional memory.
Where physician review earns its cost
A written appeal signed only by a nurse reviewer, arguing medical necessity against a payer medical director, is a mismatch. The rebuttal needs to speak to the criteria the payer applied, cite the specific chart findings that satisfy them, and address why the alternative level of care would have been unsafe. That is clinical judgment written for a clinical audience.
At a single small hospital, a part-time medical director can absorb this. At a multi-hospital system or a large multispecialty group, the volume and the timing requirements outstrip what any one physician can cover between rounds, and this is where organizations bring in physician advisor services to carry the peer-to-peer calls and the written clinical rebuttals at scale. The point is not outsourcing the argument. It is making sure the person available for a peer-to-peer at 4 p.m. on a Friday is a physician who has read the chart.
Three tests of whether the clinical review layer is actually working
- What percentage of peer-to-peer opportunities are used, not just logged? A missed call is a forfeited reversal.
- Are overturn rates tracked by payer, by denial reason, and by reviewer? Aggregate overturn rate hides everything useful.
- Do the findings loop back to utilization review and documentation improvement? If the same denial reason recurs quarterly with the same payer, the appeal function is treating symptoms.
What good looks like at scale
The difference between a competent denials operation and an excellent one at a large provider is mostly infrastructure, and it is visible in five places.
A single denial inventory. One work queue, every payer, every facility, with the appeal deadline as a hard date field rather than a note. Systems that run separate denial processes per hospital lose claims in the seams during transitions and staffing gaps.
Deadline math that assumes nothing. Filing deadlines calculated from receipt of the determination, tracked with a buffer, and escalated when the buffer is breached. A strong argument filed late is worth nothing.
Payer-specific playbooks. Which plans reverse on peer-to-peer, which require a second level, which respond to a records-only submission, which criteria set each one applies. This knowledge exists informally in every organization. Writing it down is what makes it survive turnover.
Evidence assembled once. The appeal packet built to carry through to external review: the full record, the criteria applied, the physician rebuttal, proof of timely filing, and the authorization trail. Rebuilding the packet at each level introduces inconsistency that reviewers notice.
Reporting that names the cause. Denial dollars by root cause, with an owner attached. Front-end registration, utilization review, documentation, and coding each fix different problems, and none of them can act on a number that says only "denials up."
Recovered dollars follow from the same discipline every time: classify fast, use the earliest rung available, keep the clock in writing, and put a physician's voice on the clinical questions. Organizations that build that once tend to stop relitigating the same denial types a year later, because the pattern shows up in the reporting before it shows up in the aging.