Beating Prior Authorization Denials in Psych (The Clinician’s Survival Guide)

Why prior auth exists, and why your psych patients get hit twice as hard Prior authorization (PA) is the process where, before an insurance company will…

Sections

1. Why prior auth exists, and why your psych patients get hit twice as hard
2. MHPAEA, the 2024 final rule, and the NQTL framework (your highest-leverage argument)
3. The first move on every denial: peer-to-peer review
4. Medical necessity documentation: the magic phrases
5. Step therapy: how to skip step 1 without losing the appeal
6. High-frequency denials and how to win them
7. The escalation ladder: P2P, internal appeal, external review, parity escalation
8. The administrative tax math, and how to build a system so you stop paying it

Why prior auth exists, and why your psych patients get hit twice as hard

Prior authorization (PA) is the process where, before an insurance company will pay for a medication or procedure you’ve prescribed, they make you prove it’s necessary. On paper, the rationale is cost containment and formulary tier enforcement. In practice, it’s a wall between you and the treatment plan you already decided was clinically appropriate, staffed by people who in many cases are not in your specialty and sometimes not even in clinical practice anymore.

Every clinician deals with this. But if you work in psychiatry, you deal with it more. The reason isn’t a mystery, and the data isn’t ambiguous. Behavioral health treatments get denied at a higher rate than equivalent physical health treatments. Psych medications get pushed onto step therapy protocols (where you must “fail” a cheaper drug before insurance covers the one you actually want) more aggressively than cardiology or endocrine medications, even when the clinical evidence base for the targeted drug is just as strong. Procedural treatments like TMS (transcranial magnetic stimulation, a non-invasive brain stimulation therapy used for depression) and Spravato (esketamine nasal spray, FDA-approved for treatment-resistant depression) get put behind documentation hurdles that, if you applied them to a cardiac stent, would be considered malpractice on the payer’s side.

This isn’t just frustrating. It’s illegal in a specific, actionable way, which we will get to. But before we get there, the first thing to internalize is that the PA system is not designed to be fair, and the medical director on the other end of the line is not in a position to advocate for your patient. You are. The whole posture of this article is built around that fact.

MHPAEA, the 2024 final rule, and the NQTL framework (your highest-leverage argument)

The Mental Health Parity and Addiction Equity Act of 2008, usually shortened to MHPAEA, says that if a health plan covers mental health and substance use disorder services, the financial requirements and treatment limitations on those services have to be no more restrictive than what the plan applies to medical and surgical benefits. The 2024 final rule, issued by the Departments of Labor, Health and Human Services, and Treasury, sharpened that significantly. It required plans to actually collect and evaluate outcomes data showing parity, and to fix anything that wasn’t.

The crucial term inside MHPAEA is non-quantitative treatment limitation, or NQTL. A quantitative limitation is something you can count: visit caps, dollar caps, copays. A non-quantitative limitation is everything else. It includes prior authorization protocols, step therapy requirements, medical necessity criteria, formulary tiering, and network admission standards. The 2024 rule was very clear that NQTLs applied to mental health benefits cannot be more stringent, as written or as applied, than the comparable NQTLs on the medical side.

Here’s why this matters in a denial letter. If a plan requires you to document treatment-resistant depression with two failed adequate trials before approving Spravato, but does not require equivalent documentation of failed first-line therapy before approving, say, a second-line biologic for inflammatory bowel disease, that’s an NQTL parity problem. If a plan applies step therapy to brand-name stimulants but does not apply step therapy to comparable specialty cardiology drugs, that’s an NQTL parity problem. You don’t have to prove the plan is doing this on purpose. You only have to point out that the limitation is more stringent in practice on the mental health side.

Most plan medical directors do not want to fight this argument on the record. They will deny on first pass because most clinicians don’t know to invoke it. The second you put “this denial appears to constitute a non-quantitative treatment limitation in violation of MHPAEA” in writing, your file moves to a different desk. Sometimes that’s enough to flip the decision before you even get to formal appeal.

The first move on every denial: peer-to-peer review

You get a denial. Your first move, before anything else, is to call and request a peer-to-peer review (P2P). This is a phone conversation between you, the prescribing clinician, and a medical director at the insurance company. By federal and most state regulations, you have a right to this conversation, and you should use it on basically every denial that matters, because it is by a wide margin the highest yield single intervention in the appeals process.

A few practical things. First, you usually have to request the P2P within a short window (often 24 to 72 hours of the denial, sometimes up to 7 days). Read the denial letter for the deadline and treat it as real. Second, the insurance medical director on the other end is often not a psychiatrist. Sometimes they’re family medicine, sometimes occupational, sometimes a specialty entirely unrelated. This is itself a parity issue worth noting in your file if it becomes a pattern, but in the moment it means you should explain your reasoning in clear, citation-anchored terms, the way you would to a smart medical student rather than to a peer.

Third, and this is the thing nobody tells you in residency: the P2P is not a debate, it’s a documentation event. You are creating a record. Have your notes in front of you, have the relevant guideline ready to cite (APA practice guidelines, the Texas Medication Algorithm Project, VA/DoD clinical practice guidelines, NICE if you want to gild it), and speak in the same vocabulary the medical director’s checklist uses.

Peer-to-peer cheat sheet

• State the patient’s diagnosis with the specific ICD-10 code, not the general category. “Major depressive disorder, recurrent, severe, F33.2” beats “depression.”
• State what’s been tried, when, at what dose, for how long, and what happened. Dates matter. “Sertraline 200 mg daily for 8 weeks, achieved PHQ-9 reduction from 22 to 18, did not meet response criteria” beats “SSRI didn’t work.”
• State the evidence-based rationale for the requested treatment, citing the guideline by name.
• If step therapy is the issue, state the specific reason you’re skipping a step (allergy, contraindication, prior failure with documentation, severity warranting faster escalation).
• State the clinical risk of not approving. “Patient has active passive suicidal ideation and a prior hospitalization, delaying treatment escalation poses meaningful safety risk” is the kind of sentence that ends arguments.
• If anything in the denial pattern smells like a parity issue, name it. Just saying the letters MHPAEA out loud changes the conversation.

Medical necessity documentation: the magic phrases

If you write your notes in normal clinical prose, you will get denied more often than the clinician one floor down who writes their notes in the exact vocabulary the PA reviewers are scanning for. This is unfair, it shouldn’t work this way, and it does work this way. Adjust accordingly.

The PA reviewer is reading your note for keywords, not narrative. If the keywords aren’t there, the denial isn’t a comment on your clinical reasoning, it’s a comment on your documentation. Fix the documentation, the denial usually goes away.

Magic phrases that move medical necessity reviews

• “First-line treatment failed” with the specific drug, dose, duration, and outcome metric.
• “Treatment-resistant” with the criteria for resistance spelled out (typically two failed adequate trials at therapeutic dose and duration).
• “Previously stable on” when you’re continuing or restarting a medication that worked before.
• “Evidence-based” followed by the actual guideline citation, not just the phrase.
• “Adequate trial” with the standard of 6 to 8 weeks at therapeutic dose, documented.
• “Contraindicated” with the specific contraindication (allergy, drug interaction, comorbid condition).
• “Documented intolerance” with the adverse effect named.
• “Functional impairment” tied to a specific domain (occupational, academic, social, ADLs).
• “Acute safety concern” tied to documented risk (SI, recent hospitalization, prior self-harm).

Build these into your note templates. Your EHR almost certainly supports macros. The five minutes you spend building the macro once buys back hours per month in PA work.

Step therapy: how to skip step 1 without losing the appeal

Step therapy is the requirement that you try and fail a cheaper option before insurance covers your preferred one. It is the single most common PA mechanism for psychiatric medications. The good news is that step therapy almost always has exceptions written into the plan’s own policy, and you can invoke them directly.

The standard exceptions, which apply across most major payers, are:

• Allergy or documented adverse reaction to the step 1 drug or a related compound.
• Contraindication based on a comorbid medical condition, drug interaction, or organ function.
• Prior treatment failure at therapeutic dose and adequate duration, documented in your notes or in records from a prior prescriber. (Records from a prior prescriber count, but you have to actually get them and reference them.)
• Severity that warrants faster escalation, framed as risk of harm from delay.
• Patient currently stable on the requested medication, with documented worsening on prior trials of the step 1 option.

For every patient you’re submitting a PA for, at least one of these is usually documentable. The hard part isn’t the clinical reasoning, it’s pulling the records and writing it in the format the reviewer is scanning for. Build that into intake.

High-frequency denials and how to win them

Stimulants

Stimulant PAs (for medications like methylphenidate, amphetamine, and lisdexamfetamine, used in ADHD) are heavily scrutinized because of DEA scheduling and cost. The two documentation hits that get stimulant approvals are diagnosis specificity and cardiac workup.

Use the specific ICD-10 code (F90.0, F90.1, F90.2, F90.8, F90.9 for the various ADHD presentations) and document the criteria you used to reach it. “ADHD per DSM-5 criteria, with onset before age 12, symptoms across two or more settings, functional impairment in occupational and academic domains, screening positive on ASRS and confirmed by clinical interview” is the kind of sentence that ends ADHD stimulant denials before they start.

For cardiac workup, document that you screened for personal and family history of cardiac disease, syncope, arrhythmia, and sudden death. An ECG isn’t required by guideline for most adult patients without risk factors, but the documentation that you considered it and concluded it wasn’t indicated (or did it) needs to be in the note. If you skip this, even when it’s not clinically required, the PA reviewer will deny on it.

Second-generation antipsychotics in non-psychotic populations

Second-generation antipsychotics, or SGAs (drugs like quetiapine, aripiprazole, lurasidone, olanzapine), are increasingly used off-label or on-label for mood stabilization, augmentation in depression, anxiety, sleep, and irritability in autism. PA reviewers are trained to flag any SGA prescription that isn’t paired with a psychotic diagnosis.

The fix is straightforward documentation: state the on-label or evidence-supported indication explicitly, cite the guideline (APA guidelines for bipolar disorder, augmentation for treatment-resistant depression, etc.), and note prior trials. “Aripiprazole 5 mg as augmentation for treatment-resistant major depressive disorder, two prior antidepressant trials failed, evidence-based per APA practice guideline” is the rationale that wins.

Spravato (esketamine)

Spravato PAs are documentation-heavy because the drug is expensive and the indication is narrow. You need formal documentation of treatment-resistant depression, which standardly means two failed adequate antidepressant trials at therapeutic dose for at least 6 weeks each, in the current episode. You need a quantified depression severity measure, typically PHQ-9 or Hamilton Depression Rating Scale (HAM-D), at baseline and ideally at the end of each prior trial.

The PA-winning language for Spravato looks like this: “Patient meets criteria for treatment-resistant major depressive disorder, with two adequate antidepressant trials failed in the current episode. Trial 1: sertraline titrated to 200 mg daily for 8 weeks, PHQ-9 22 to 19, did not meet response criteria. Trial 2: venlafaxine XR titrated to 225 mg daily for 7 weeks, PHQ-9 21 to 20, no response. Current PHQ-9 23, HAM-D 24. Requesting esketamine per FDA-approved indication for treatment-resistant depression, REMS-enrolled site, monitoring plan in place.” That paragraph, dropped into the PA submission, is harder to deny than to approve.

TMS

TMS denials follow the same logic as Spravato. Document the treatment-resistant depression criteria with the same specificity. Document that you’ve considered ECT and either ruled it out (patient preference, contraindication, prior failure) or are escalating in a stepped fashion. Cite the guideline. The clinical evidence for TMS is solid, and the denial rate drops sharply when the paperwork is right.

Benzodiazepines

Benzo PAs get tighter every year, and for population-level reasons that the data actually supports. The winning approach is to write the PA the way you’d write the chart for a careful clinician: narrow indication, time-limited plan, taper schedule already on paper.

“Lorazepam 0.5 mg as needed up to twice daily for acute panic disorder with significant functional impairment, SSRI being initiated concurrently, planned taper at 6 to 8 weeks pending SSRI response, no history of substance use disorder, no concurrent opioid prescription” is the kind of submission that gets through. Open-ended chronic benzo prescriptions without a taper plan won’t, and shouldn’t.

The escalation ladder: P2P, internal appeal, external review, parity escalation

If the P2P doesn’t flip the denial, you escalate. The order matters.

First is the internal appeal. This is a written submission to the insurance company’s internal review process. You include the full clinical record, the relevant guidelines, your letter of medical necessity, and any new information that’s emerged. Most plans have two levels of internal appeal. Use both if needed.

Second is the external review. After internal appeals are exhausted, federal law (and every state) gives the patient the right to an external independent review by a reviewer not employed by the insurance company. This is binding on the insurer. External review approval rates for psychiatric treatments are meaningfully higher than initial PA approval rates, because the reviewer is actually a clinician in your specialty and isn’t paid based on denial volume. Patients often don’t know this option exists. Tell them.

Third is the parity escalation. If the denial pattern looks like an NQTL violation, you (or the patient, or both) can file a parity complaint with the Department of Labor (for ERISA plans, which covers most employer-sponsored insurance), the state insurance commissioner (for fully-insured plans), or CMS (for Medicare/Medicaid managed care). These complaints get attention. Plans that get repeated parity complaints get audited. Audits are expensive. The threat is sometimes more useful than the filing, but the filing is real and you should use it when the case is clear.

The administrative tax math, and how to build a system so you stop paying it

Here’s the part nobody runs the numbers on. If you’re spending 5 to 10 hours a week on PA work for a single payer, and that payer’s reimbursement is at the bottom of your panel, you are paying for the privilege of being in-network with them. The math:

Take your hourly billable rate for clinical work. Multiply by the hours per week you spend on PA work for that payer. That’s the gross revenue you’re forgoing. Compare it to the net revenue that payer generates after their reimbursement rate, denial rate, and clawback rate. If the PA tax is bigger than the net revenue, you are losing money on that contract.

Most clinicians don’t run this calculation, which is how panels end up dominated by the worst-behaved payers. Run it once a year for every contract. If a payer is denying disproportionately and reimbursing under market, drop them. Your patients on that plan can use out-of-network benefits or transition to a panel that respects clinician time. The threat of clinicians leaving the panel is one of the few things that actually changes payer behavior at scale.

Build the system once, run it on every patient

The clinicians who win at PA work aren’t smarter or more aggressive. They’ve just built the system once and they run it on every patient.

Build EHR macros for the magic phrases. Build a PA documentation template by drug class (stimulants, SGAs, Spravato, TMS, benzodiazepines, controlled-substance combinations) that pre-fills the diagnosis specificity, prior trials section, guideline citation, and risk-of-delay language. Build a form letter for each common denial type that you can adapt in 5 minutes instead of writing from scratch.

For peer-to-peer calls, keep a one-page script on your desk with the patient’s diagnosis, ICD-10 code, prior trials with dates and doses, current severity scores, and the guideline citation. Five minutes of prep, ten minutes on the phone, and you’ve recovered hours of downstream work.

For appeals, save your best appeal letters and reuse the structure. The first one takes 90 minutes. The hundredth takes 15.

None of this is glamorous. None of it is what you went to school for. And you should not have to do any of it, because most of what we’re working around is functionally an illegal restraint on care that has been normalized by a generation of clinicians too tired to push back. The good news is that the leverage is real, the parity law has teeth now, and the payers are vulnerable in ways they were not five years ago. Document well, escalate fast, and stop apologizing for prescribing the right treatment.