Prior authorization documentation in psychiatry: what reviewers look for

By Selam Koch, PMHNP-BC, founder of OneStep Scribe and practicing psychiatric nurse practitioner. Published August 2026.

A prior authorization is the moment your documentation gets read by a stranger. The reviewer on the other side has never met the patient and works from written criteria; any follow-up question arrives as a records request or a peer-to-peer call, each of which costs days. Whether the request goes in as a records packet or an electronic questionnaire, every answer has to be defensible from your chart, and when the plan asks for records, the chart is the packet. When a psychiatric PA fails, the medicine is usually sound. What failed is the record's ability to demonstrate, on paper, what the prescriber already knows.

This guide covers what utilization reviewers commonly look for in psychiatric requests, why these requests get denied, and how routine progress notes either carry the load or leave you rebuilding history from memory. One caution up front: criteria differ by payer, plan, and state, and they change. Nothing here guarantees an approval. The aim is a chart that answers the questions reviewers most often ask, so the request stands on evidence rather than assertion.

Why psychiatric prior authorizations get denied

Setting aside pure administrative errors, wrong form, missing NPI, lapsed submission window, the clinical denials in psychiatry commonly trace to one of three documentation gaps.

• The chart asserts medical necessity without demonstrating it. The request says the patient needs the agent or service. The notes behind it describe a stable patient, or describe severity in conclusory terms with no symptoms, scores, or functional detail a reviewer can match against criteria.
• The step-therapy history is missing or vague. "Failed multiple antidepressants" appears somewhere in the record, with no agent names, no doses, no durations, and no outcomes. To a reviewer applying a step-therapy rule, an undocumented trial commonly counts as no trial.
• The diagnosis on the request is not supported in the chart. The PA form names a diagnosis that appears only on a problem list, or that conflicts with the diagnosis in recent notes. Reviewers routinely check whether the condition the criteria are written for is actually documented, with supporting findings, in the record submitted.

All three are fixable at the note level, before any request exists.

The core elements reviewers commonly look for

Across payers and across request types, medication or service, the same evidence set comes up again and again. Think of it as five questions the packet has to answer.

A specific diagnosis the chart actually supports

Plan criteria are written per condition, so the reviewer's first move is to confirm the condition. That means the diagnosis on the request matches the diagnosis in recent notes, and the notes contain the findings that support it: the symptoms that meet DSM-5-TR criteria, their duration, and the pertinent negatives that rule out the nearby alternatives. A diagnosis that lives only in a problem list, with no supporting symptom documentation anywhere in the visits submitted, invites either a denial or a records request that stalls the case.

Where the diagnostic picture is genuinely evolving, document that too. A brief formulation explaining why the working diagnosis is what it is reads far better to a reviewer than a bare label, and it is the same formulation a thorough intake produces anyway. If your initial evaluations already do this work, the PA inherits it; the intake evaluation guide covers what a complete diagnostic evaluation documents.

Documented trials of formulary alternatives, with doses, durations, and outcomes

This is the center of gravity for medication requests, because step therapy is among the most common criteria applied to them. The plan wants evidence that preferred alternatives were tried and did not work, and "did not work" has to be legible. For each prior trial, the strongest documentation states:

• The agent, by name.
• The dose, specifically the highest dose reached, since a subtherapeutic trial may not satisfy the criterion.
• The duration at that dose, because response takes time and reviewers know it.
• Adherence, since a trial the patient did not take is a different fact than a trial that failed.
• The outcome, split into its two distinct types: inadequate response despite an adequate trial, or intolerance, with the specific adverse effect named.

Class-level summaries have their place. "Two adequate SSRI trials with inadequate response" is a legitimate way to open the story, as long as the specifics sit underneath it in the chart. What does not survive review is the summary with nothing underneath. Note also that a documented contraindication to the formulary alternative can substitute for a trial under many plans' criteria, but only if the contraindication itself is documented, not implied.

One documentation habit pays for itself here: when a dose trajectory has several steps, write them in chronological order. A note that reads "titrated to the target dose, developed insomnia, reduced, now re-escalating with the adverse effect resolved" gives a reviewer the entire arc in one place. A chart where each visit mentions only the current dose forces someone, later, to reconstruct the arc across six notes.

Symptom severity and functional impairment

Criteria for costlier agents and services usually include a severity threshold, and severity has to be shown rather than announced. Two kinds of evidence do the work. The first is standardized measurement: a validated symptom scale, scored, interpreted, and repeated over time, which is exactly the practice described in the measurement-based care guide. Serial scores are especially persuasive because they document trajectory, not just a snapshot. The second is functional detail: missed work or school, strained caregiving, impaired self-care, withdrawal from relationships, emergency visits or hospitalizations. "Moderate to severe depression" is a label; "has missed eleven workdays this quarter and stopped preparing meals" is evidence a reviewer can weigh against a criterion.

Where safety history is relevant to the request, a current, plainly stated risk assessment belongs in the picture as well.

A clinical rationale for the requested agent or service

The history explains what has not worked. The rationale explains why the requested option is the reasonable next step for this particular patient. That may rest on the pattern of prior response, an interaction or comorbidity that rules out alternatives, a partial response worth building on, or the clinical logic of moving to a different mechanism or a higher level of care. The form matters less than the presence: a reviewer who can see a reasoned "therefore" connecting the history to the request is reading a different packet than one who sees a request with no stated reasoning. The chart's job is to show that the judgment happened.

A monitoring plan

Requests read stronger when they show what happens after approval: the follow-up interval, what will be measured, any labs or vitals the agent class calls for, and what would prompt a change in course. Continuation and reauthorization reviews make this concrete, because they ask a simple question: did it work? A chart with documented response, ideally the same scale trending down and function recovering, answers it. A chart that never revisits the outcome leaves the renewal exposed. Treatment goals written in measurable terms make this nearly automatic; the measurable goals guide covers how to write them.

When the trials happened under a previous prescriber

Step-therapy history generally does not reset when the patient changes prescribers, though some plans apply lookback windows, limit how old a trial can be, or credit only trials they can verify, so confirm with the specific plan. Either way, the evidence problem becomes yours. Three sources, in descending order of strength:

1. Prior treatment records. Request them, and document that you did. Even a single medication list from the previous practice can anchor the history.
2. Pharmacy fill history. Fill records establish agents, doses, and rough durations, and they often recover details the patient cannot. Many payers can see claims history on their side as well, but do not assume the reviewer will go looking.
3. Patient report, documented as patient report. When records are unavailable, write what the patient recalls and label it: "Patient reports a trial of an SNRI for approximately four months under a prior prescriber, dose not recalled, discontinued due to blood pressure elevation." The labeling is the point: a patient-reported history presented as verified fact is a credibility problem, while the same history labeled accurately is usable documentation, and whether it satisfies a given plan's criterion is the payer's call.

Capture this history at intake, when the patient is already telling you the story. A medication history section built once, with doses, durations, and outcomes as the patient recalls them, becomes the trial documentation for every future request.

Where prior authorization concentrates in psychiatry

Some corners of psychiatric practice generate far more PA work than others. Branded antidepressants and antipsychotics in classes with generic alternatives sit behind step-therapy rules almost by default. Long-acting injectable antipsychotics frequently carry criteria about the oral trial that preceded them and the clinical reason an injectable is indicated. Esketamine and other REMS-governed treatments layer program requirements on top of plan criteria. TMS and other interventional services typically ask for a documented depth of medication and therapy history. Higher levels of care, intensive outpatient through residential, are reviewed against medical necessity criteria for the level itself, often with concurrent review during the stay.

The criteria differ by product, service, payer, and plan, and this guide deliberately offers no per-product playbook, because any specific threshold quoted here would be wrong for someone's plan by next quarter. The durable observation is that every one of these reviews draws on the same five elements above. A chart that routinely documents diagnosis support, trial history, severity, rationale, and monitoring is prepared for whichever of these requests the practice generates next.

When the request is denied: appeals and peer-to-peer

A denial letter states a reason, and the single most useful appeal habit is to answer that reason, specifically, rather than resubmitting the original packet with more adjectives. If the stated gap is trial duration, the appeal points to the dates in the chart that establish duration. If the stated gap is severity, the appeal points to the scores and the functional documentation. Quote your own record by date. An appeal that maps chart evidence onto the denial reason gives the next reviewer a path to a different answer; an appeal that argues in general terms usually gets the general answer again.

Peer-to-peer review is the same exercise, spoken. Before the call, have three things in front of you: the diagnosis with its supporting findings, the complete trial history with agents, doses, durations, and outcomes, and the rationale for the request. Know the denial reason and address it in the first minute. The prescribers who find peer-to-peers productive are, almost uniformly, the ones who can cite the chart from the page rather than from memory. Appeal levels, deadlines, and external review rights vary by plan and state, so read the letter's fine print early rather than late.

The golden thread: strong routine notes make prior authorization requests fast to assemble

Here is the practical payoff. A PA packet is not a new document; it is an extract of documents you already wrote. When every follow-up note documents the interval response in the patient's own words, records each dose step in chronological order, restates the diagnosis with the findings that support it, ties today's decision to the treatment goals, and closes with a plan and a medical necessity justification, the request becomes assembly rather than authorship.

Clinicians sometimes call this the golden thread: chief complaint, diagnosis, progress toward goals, today's session, and plan, each traceable to the last, visit after visit. A record with that thread intact lets a reviewer, or a future you preparing an appeal, follow the entire treatment logic without inference. A record without it forces reconstruction, and reconstruction is where PA hours go to die. The annotated progress note example shows what a follow-up note with the thread intact looks like in practice, section by section.

This is where documentation habit compounds. Today's ordinary follow-up note is the evidence exhibit for a request you cannot yet predict. Prescribers who document trials, severity, and rationale as a habit rarely experience prior authorization as an archaeology project. They experience it as paperwork, which is the best it can be.

Frequently asked questions

What makes a medication trial adequate for step therapy?

The threshold is payer-defined, but reviewers commonly look for a therapeutic dose held for a clinically meaningful duration, with adherence addressed. What you can control is the record: document the agent, the highest dose reached, how long the patient stayed at that dose, whether they actually took it, and what happened. A trial cut short by an adverse effect is still a completed data point when the note says which effect and at what dose.

Do trials from a previous prescriber count toward step therapy?

Many payers accept prior trials regardless of who prescribed them, but they want evidence. Prior treatment records and pharmacy fill history are the strongest support. When neither is available, document the history as patient report, with as much specificity as the patient can provide, and note your attempt to obtain records. Acceptance of patient-reported history alone varies by payer and plan.

What if the patient cannot recall the doses of past medications?

Request prior records and pharmacy fill history, which often recover the missing detail. In the meantime, document what the patient does recall and label the uncertainty honestly, for example a trial of an SSRI for roughly six months under a prior prescriber, dose not recalled, discontinued for nausea. A partial history stated precisely is more useful to a reviewer than a confident summary with no specifics.

Does a denial mean the medication or service is not covered?

Not necessarily. Many denials state that the submitted documentation did not demonstrate the plan's criteria, which is a different problem from an exclusion. The denial letter identifies the stated reason and the appeal path. Read the reason carefully, because an appeal that answers the actual question asked, with chart evidence, is handled differently from one that resubmits the original packet. Appeal rights and timelines vary by plan and state.

How should I prepare for a peer-to-peer review?

Have the chart open and three things in one place before the call: the diagnosis with its supporting findings, the trial history with agents, doses, durations, and outcomes, and the clinical rationale for the requested agent or service. Know the stated denial reason and answer it directly. A peer-to-peer where the prescriber can cite dates and doses from the record is a conversation; one where the history is reconstructed from memory usually is not.

Do rating scales strengthen a prior authorization request?

Commonly, yes, as objective severity evidence, and some plan criteria reference standardized measures explicitly. A score works hardest when the note interprets it and ties it to function, for example a severe-range score alongside missed work and impaired self-care. Serial scores across visits also document trajectory, which supports both initial requests and continuation reviews. Whether a specific measure is required varies by payer.

Related guides

• Psychiatric progress note example, annotated section by section
• Writing treatment plans with measurable goals
• Measurement-based care: choosing, using, and documenting scales
• 90791 vs 90792: the psychiatric intake evaluation codes
• Billing 99214 with 90833: the rules that make it defensible
• 96127: billing brief standardized assessments

This article is educational and reflects one clinician's understanding of payer review practices at the time of writing. It is not legal, billing, payer-specific, or medical advice, and it does not guarantee authorization, reimbursement, or appeal outcomes. Utilization management criteria and payer policies vary and change; always verify current requirements with the specific plan, your coding resources, and compliance advisors. CPT is a registered trademark of the American Medical Association. DSM and DSM-5-TR are registered trademarks of the American Psychiatric Association.