Prior authorization is the tax nobody put on the invoice. You don't see it on the P&L as a line item, but it's there, sitting inside payroll, inside the medical assistant's day, inside the gap between a patient's visit and their first dose. If you run a practice that writes anything more complicated than amoxicillin, you already know this. What you may not have done is actually measured what PA is costing you in hours, dollars, and revenue you never collected because the patient gave up.

I want to walk through where the time actually goes, because the headline number ("PAs take a long time") hides a lot of operational rot underneath. Then I'll talk about what you can do about it without pretending there's a magic button.

The real anatomy of a prior auth

Most operators think of a PA as a single task: fill out the form, send it, wait. In practice it's eight to twelve discrete steps, and each one has its own failure mode.

  • Identifying that a PA is even required for this drug, this plan, this dose
  • Finding the right form (payer portal, fax form, CoverMyMeds, or a phone tree)
  • Pulling the clinical justification from the chart, sometimes from multiple notes
  • Gathering supporting labs, prior therapies, ICD-10s, and dates
  • Submitting and confirming receipt
  • Tracking status (which usually means calling)
  • Responding to peer-to-peer requests within whatever narrow window the payer set
  • Handling denials and appeals
  • Communicating with the patient at every step so they don't churn
  • Updating the pharmacy once approval lands

The MGMA survey work that gets quoted everywhere puts the average practice at around 45 PAs per physician per week, with roughly two business days of staff time spent on them. Compounding and concierge practices skew higher because the medications are non-formulary by default. If you're prescribing GLP-1s, testosterone, LDN, peptides, or anything that triggers a step-therapy review, you're not at the average. You're well above it.

Where the hours actually disappear

When we've audited PA workflows in clinics, the time loss almost never lives where people think it does. The form itself is fast. What kills you is everything around the form.

Context-switching

A medical assistant handling PAs is also handling rooming, refill requests, message triage, and the occasional walk-in. Every interruption costs about 23 minutes of refocus time, per the Gloria Mark research that's been replicated to death. If your MA is touching a single PA five separate times across two days because each attempt gets interrupted, you're paying for the same task five times.

Hunting for clinical detail

The provider wrote a great note. The PA form wants something slightly different: a specific failed therapy with dates, a specific lab value within the last 90 days, a BMI documented at a particular visit. The MA opens the chart, searches, pings the provider, waits. Repeat for the next PA.

Payer portals that log you out

I'm not going to name names. You know the ones. Sessions expire, two-factor codes go to a phone nobody has anymore, formulary lookups give different answers depending on which portal you check. Staff develop folk wisdom about which payer is "good on Tuesdays."

Status checks

The single biggest hidden cost. Nobody tells you the PA was approved. Nobody tells you it was denied either. Your staff has to call, hold, navigate IVR, and ask. Multiply by 45 a week.

Patient communication

The patient texts on day three asking if their medication is ready. The MA stops what they're doing, looks up the PA, sees it's pending, drafts a reassuring message, sends it. Day five, same patient, same question. By day seven the patient is frustrated and your front desk is in the loop too.

What this actually costs you

Let's put real numbers on it. An MA at $24/hour fully loaded is roughly $50,000/year. If 30% of their week goes to PA work (low, in concierge and compounding), that's $15,000 per MA per year, just on prior auth. A two-provider practice running two MAs is burning $30K annually before you count provider time on peer-to-peers, front desk time on patient inquiries, or the revenue you lost when 18% of those PAs got abandoned because the patient walked.

That last number is the one that should keep you up. CoverMyMeds' own data on abandonment puts non-completion of authorized prescriptions somewhere between 15% and 25% depending on category. For a cash-pay concierge model, abandonment maps directly to lifetime value lost. For an insurance-billing practice, it maps to visits you can't bill follow-ups on because the patient never started therapy.

What actually moves the needle

I've watched practices try to solve PA with a wall of sticky notes, with a dedicated PA coordinator, with outsourced services, and with software. Here's what I've seen work, in rough order of leverage.

1. Build a payer-and-drug matrix and keep it current

A simple spreadsheet (or a structured table in your EMR) that says: for this payer, this drug, here is the form, here are the required criteria, here is the typical turnaround, here is the appeal pathway. Whoever does the first PA for a new combination updates the matrix. Within six months you have institutional memory that doesn't walk out the door when staff turns over.

2. Move clinical justification into a template at the point of prescribing

The provider should be capturing PA-relevant detail during the visit, not the MA digging for it three days later. Build SmartPhrases or note templates for your top 15 medications that prompt for the exact criteria payers ask for. This single change cuts the back-and-forth in half.

3. Batch the work

PAs are bad context-switch tasks. Block two 90-minute windows a day for PA work. No messages, no rooming, no interruptions. Staff hate it for a week and then love it.

4. Automate the status loop

This is where AI agents earn their keep. The "did the PA get approved yet" check is structured, repetitive, and 80% of it can be handled without a human picking up a phone. Same with the initial patient-facing status update. An agent that checks portals, parses responses, and updates both the chart and the patient in plain language gives you back the single largest chunk of MA time.

5. Track abandonment as a KPI

If you can't tell me what percentage of your PAs result in the patient actually starting therapy, you don't know what PA is costing you. Measure it weekly. It will change how you triage.

Where infrastructure fits in

The reason we built Qintara the way we did is that PA workflow is not a bolt-on. It touches prescribing, charting, messaging, the pharmacy interface, and the patient portal. If those systems don't talk to each other, you end up with the exact problem above: humans as the integration layer, paid by the hour, switching context every nine minutes.

An EMR that knows which medications trigger PA at which payers, prompts the provider for the right criteria at the right moment, hands off submission and status-tracking to an agent, and keeps the patient quietly informed the whole way through is not a luxury for a concierge or compounding practice. It's the difference between an MA who can take on a third provider and an MA who's already at capacity. If you want to see what that looks like running in a real clinic, talk to our team and we'll walk you through it.

Frequently Asked Questions

How many hours per week does manual PA work actually consume in a typical specialty practice?

In the practices we've audited, it ranges from 12 to 20 hours per provider per week of combined MA, front desk, and provider time. Compounding-heavy and concierge practices tend toward the upper end because more of their prescribing falls outside standard formularies.

Is outsourcing PA work to a third party cheaper than handling it in-house?

Sometimes, on a pure hourly basis. But outsourced PA services rarely close the loop on patient communication, and they don't fix the upstream problem of clinical detail being missing from the note. You can end up paying for the same task twice: once to the vendor, once to the staff who still has to handle the patient side.

What's the single highest-leverage change a practice can make this quarter?

Templating the clinical justification at the point of prescribing for your top 10 to 15 medications. It costs you a weekend of template-building and saves hours every week from that point forward. Everything else (batching, automation, matrices) compounds on top of that foundation.

How do we measure whether changes to our PA process are working?

Track four numbers weekly: median time from prescription to approval, percentage of PAs requiring a peer-to-peer, percentage of approved PAs where the patient actually filled, and staff hours logged against PA tasks. If all four trend the right direction, you're winning. If only one or two improve, you've shifted the problem rather than solved it.