How should a multi-specialty group route urgent, out-of-network, and peer-to-peer prior authorization exceptions?

A multi-specialty group should route urgent prior authorization exceptions to the fastest-responding clinical owner under a hard clock, out-of-network exceptions to a coordinator who checks benefits and alternatives, and peer-to-peer requests to the ordering physician with a prepared summary and the payer's deadline. Run one prior auth exception routing workflow across every specialty, and let each specialty keep its own payer rules. Measure time to first touch, aged exceptions, and peer-to-peer overturn rate.

Why do prior authorization exceptions need their own routing?

A prior authorization exception is any request that can't ride the normal queue: it's urgent, the rendering provider or facility is out of network, or the payer has denied it and wants a physician conversation before it will reconsider. These cases are a small share of volume, but they carry most of the clinical risk and most of the lost revenue.

The pressure is real. In the AMA's physician survey, 93% of physicians said prior authorization delays care, and 29% said it led to a serious adverse event for a patient. Physicians and staff spend an average of 13 hours a week per physician on it.

In a single-specialty clinic, one lead nurse usually knows who to grab. In a multi-specialty group, that tribal knowledge breaks. Cardiology, orthopedics, GI, and oncology each face different payer rules, different clinical criteria, and different urgency profiles. If every clinic invents its own exception process, a stuck case depends on who happens to be at the desk that day.

The fix is a prior auth exception routing workflow with three defined lanes, one owner per lane, and a clock on each. The rest of this article covers each lane, then the harder question of whether to centralize.

How should you route urgent prior authorization exceptions?

Route urgent exceptions to the fastest-responding clinical owner, and put a hard clock on first touch. Urgent means a delay could seriously jeopardize the patient's life, health, or ability to regain function. It does not mean "the patient is anxious" or "the surgeon wants it Friday."

Payers are being pushed toward faster answers. Under CMS's interoperability and prior authorization final rule, impacted payers must decide expedited requests within 72 hours and standard requests within seven calendar days, with the timeframes starting primarily in 2026. That rule applies to specific payer types, mainly Medicare Advantage, Medicaid, and CHIP plans, so commercial rules still vary. Your own clock has to be tighter than the payer's.

A workable urgent lane looks like this:

• Flag at intake. The scheduler or referral coordinator marks the case urgent using a short written definition, not gut feel. Anything flagged goes to a named urgent queue instead of the general pool.
• Assign a clinical owner, not a department. Pick one nurse or authorization specialist per shift who owns the urgent queue. Name a backup. "Cardiology will get to it" is how a 72-hour clock gets burned in 30 hours of silence.
• Set a first-touch target. Two business hours is a reasonable starting point. First touch means a human has opened the case, confirmed the request is complete, and submitted it as expedited.
• Escalate on a timer. If nothing has moved in four hours, the case pages the practice manager or medical director.

Two failure modes to design against. First, over-flagging: when everything is urgent, nothing is, and the queue turns into a second general pool. Audit the flag monthly. Second, submitting an urgent case through a standard channel because the portal defaults to it. Make the expedited field a required step, not an afterthought.

How should you handle out-of-network prior authorization exceptions?

Send out-of-network cases to a coordinator who can check benefits, verify network status, and line up alternatives before anyone argues for an exception. The goal is to find out in the first hour whether this is a fight worth having.

Out-of-network exceptions are different from urgent ones because the bottleneck is usually information, not speed. A good coordinator answers four questions fast:

1. Is the provider actually out of network for this plan and product? Directory data is stale often enough that a call or portal check is worth it.
2. Is there an in-network option within a reasonable distance and wait time? Payers approve network gaps when no adequate alternative exists, so you need to document that.
3. What does the member's benefit say about out-of-network coverage? Some plans cover it at a higher cost share and need no exception at all.
4. Is there a continuity-of-care argument? A patient mid-treatment with a specialist often qualifies for a transition period.

Then the coordinator builds the exception packet: the clinical reason this provider is needed, the documented search for alternatives, and the plan's own gap-exception language if you can find it. That packet goes to the payer with a single point of contact named.

Your coordinator should also have the authority to say no. If the in-network alternative is adequate and the patient is willing, redirecting saves weeks. A coordinator who can only submit and wait will submit everything.

Pair this lane with benefits verification at scheduling. Catching network status before the appointment turns many exceptions into simple redirects, and it keeps your physicians out of conversations they never needed to have.

How should you handle peer-to-peer requests?

Route peer-to-peer requests to the ordering physician, with a prepared summary and the payer's deadline attached. The physician is the only person the payer will talk to, so the rest of the workflow exists to make that call short and winnable.

Peer-to-peer is expensive physician time. The AMA reports that 65% of physicians at least sometimes take part in peer-to-peer reviews, and 56% of those say the volume has grown over five years. Only 16% say the plan's reviewer often or always has appropriate qualifications, which is why prep matters: you're often explaining a specialty to a reviewer outside it.

The routing steps are mechanical:

• Capture the deadline the day the denial arrives. Many payers allow a peer-to-peer window of just a few business days after a denial. Miss it and you're into a formal appeal. Put the date in the case header.
• Build a one-page summary. Diagnosis, what's been tried, why the plan's criteria are met or don't fit, the specific policy section cited in the denial, and the single ask. A physician should be able to read it in 90 seconds.
• Schedule the call, don't leave it to chance. Book a 15-minute block on the physician's calendar before the case leaves the coordinator's hands, and include the payer's call-in details.
• Log the outcome in a structured way. Overturned, upheld, or partially approved, plus the reviewer's stated reason. That data is how you find payer patterns later.

When the peer-to-peer fails, the same summary seeds the written appeal. CMS also requires impacted payers to give a specific reason for denial, which helps you decide whether to resubmit or appeal.

Should you centralize exception handling or keep it inside each specialty?

Centralize the routing and the tracking. Keep the payer-specific and clinical judgment local. That split gives you most of the benefit of both models.

Here's how the tradeoffs actually fall out:

• Consistency. A central team applies one definition of urgent, one packet format, and one escalation path. Decentralized clinics drift. Three sites with three definitions of "urgent" means you can't compare anything.
• Turnaround. Central queues smooth out workload: when orthopedics spikes after a payer policy change, the pool absorbs it. But a central team that doesn't know the specialty will send back more incomplete packets. Local teams are faster on familiar cases and slower when someone is out.
• Cross-training. Central teams need staff who can handle GI one hour and neurology the next. That takes playbooks per specialty and real training time. Local teams need less training but are one resignation away from a coverage gap.
• Physician trust. Specialists are more willing to hand off work to a team they know. A fully central model can feel like sending cases into a void.

A hybrid works for most groups. A central desk owns intake, flagging, the three lanes, the clocks, and the dashboard. A named specialty liaison in each service line owns payer rules, clinical criteria, and the physician relationship. The central desk routes; the liaison answers "does this payer need a step-therapy failure documented for biologics?"

If you're a PE-backed MSO adding practices through acquisition, this hybrid also gives you a standard you can plug new sites into without rewriting their payer knowledge on day one.

What should you measure to know the routing works?

Measure three numbers weekly, by lane and by specialty: time to first touch, aged exceptions, and peer-to-peer overturn rate. Everything else is secondary until these are stable.

• Time to first touch. The hours from when an exception is created to when a human first acts on it. This is the metric that exposes silent queues. Targets by lane: urgent in hours, out-of-network and peer-to-peer within one business day.
• Aged exceptions. The count of open exceptions older than a threshold, such as 3 or 5 business days. Report it by owner, since an aging pile usually points to one overloaded person or one confusing payer.
• Peer-to-peer overturn rate. The share of peer-to-peer calls that reverse the denial. A low rate by payer may mean your summaries are weak, the wrong physician is calling, or that payer isn't worth the time. A high rate means you should be requesting these calls more often.

Add two supporting numbers once the basics hold: the share of exceptions that were avoidable (for example, out-of-network cases that a benefits check would have caught) and the cost per case. The 2024 CAQH Index puts a manual medical prior authorization at $16.29 per transaction industry-wide, against $5.43 fully electronic. Exceptions cost more than the average, because they involve physician time.

Review the numbers in a 20-minute weekly huddle with the specialty liaisons. If a metric moves, someone should be able to say why.

How does an AI agent help apply one routing standard across specialties?

An AI agent applies the same routing rules to every case at intake, while each specialty keeps its own payer logic. That's the problem multi-specialty groups struggle to solve with spreadsheets and email.

Honey Health's Prior Authorization agent reads incoming orders and referrals, pulls the clinical documentation, and checks the request against the payer's requirements for that specialty. It flags cases that look urgent or out-of-network, routes them into the right lane, and starts the clock. Cardiology's payer rules and orthopedics' payer rules live side by side; the routing standard sits above both.

In practice, that means a few things:

• The coordinator opens a case that already has the clinical summary and the cited policy attached.
• Deadlines, including peer-to-peer windows, are captured from the denial instead of re-keyed.
• Time to first touch and aged exceptions are reported across all sites without anyone building a pivot table.

We'd be straight about the limits. An agent doesn't replace the physician on a peer-to-peer call, and it won't know your local payer relationships. It handles the preparation and the tracking so humans spend their time on the judgment calls. For groups that want to start small, the usual path is to run it on one lane at a time, usually urgent first, and compare turnaround against the manual baseline.

Frequently Asked Questions

What is a prior authorization exception?

A prior authorization exception is any request that needs special handling outside the standard queue. The common types are urgent or expedited requests, out-of-network requests, and denials that call for a peer-to-peer review or appeal. Each needs a different owner and a different clock.

How fast should urgent prior authorization requests be handled?

Aim for human first touch within two business hours, then submit as expedited the same day. Payers subject to the CMS rule must decide expedited requests within 72 hours, so your internal time should be a fraction of that. Escalate automatically if nothing has moved within a few hours.

Who should handle peer-to-peer requests?

The ordering physician takes the call, since payers generally won't accept anyone else. A coordinator or authorization specialist should prepare a one-page summary, capture the payer's deadline, and book the call on the physician's calendar so the physician only has to show up and argue the case.

Should a multi-specialty group centralize prior authorization exceptions?

Centralize routing, tracking, and clocks, and keep payer rules and clinical judgment with specialty liaisons. Full centralization risks sending incomplete packets for unfamiliar specialties; full decentralization risks inconsistency and coverage gaps. The hybrid gives you one standard without discarding specialty knowledge.

What metrics matter most for prior authorization exception handling?

Track time to first touch, aged exceptions, and peer-to-peer overturn rate, each by lane and specialty. Together they show whether cases are being picked up, whether any are stalling, and whether your physician time is actually winning reversals.