ONESTEP SCRIBE

Choosing the E/M level in psychiatry: 99213, 99214 and 99215

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

99213 vs 99214 vs 99215: choosing the E/M level - OneStep Scribe

Every established outpatient psychiatric visit gets a level. In the charts I review it is picked on instinct, running one direction: downward. The 2021 office and outpatient rules changed what the level rests on, and much psychiatric documentation never caught up.

This guide covers 99213, 99214 and 99215 for established patients. Below them, one self-limited or minor problem with straightforward decision making is a 99212, and 99211 needs no prescriber at all. If you pair the visit with a psychotherapy add-on, the E/M level still follows these rules, and the pairing mechanics live in billing 99214 with 90833.

CPT is a registered trademark of the American Medical Association; the criteria below are summarized in my own words, and the current manual is authoritative. Payers publish their own interpretations and audit standards, and those govern your claims. Level selection is the billing clinician's responsibility, so check your payer's policy and compliance resources before changing how you code.

The level is a conclusion, not a word count

You select an established visit two ways: by medical decision making, or by total time on the date of the encounter. History and examination no longer drive the level: you still document what is medically appropriate, but its volume does not move a 99213 to a 99214.

That change explains most psychiatric miscoding I see. Three pages with a full review of systems can still be a 99213; a tight note about one hard decision can be a 99214. Length is not the currency; the thinking is.

The three elements are how many problems you addressed and how complex they were, how much information you gathered from outside the interview, and how much harm your decision could expose the patient to.

Take the highest level that at least two of the three elements meet or exceed. Problems moderate plus risk moderate is a 99214 even if you reviewed nothing outside the room. One high element alone does not set the level: high with two lows is a 99213, and high with one moderate is a 99214.

Element one: problems addressed

A problem is addressed when you evaluate or manage it. Listing a diagnosis does not address it, and neither does noting a hypothyroidism diagnosis managed by primary care.

The undercoding error I see most often is "stable" in a note whose own body shows partial response, impairment or side effects being managed. If you are still titrating, the patient is not at goal.

Element two: data reviewed and analyzed

Data falls into three categories, plus one movable item.

Limited data (99213) is two category one items, or an assessment that required an independent historian. Moderate data (99214) is three category one items, one independent interpretation, or one qualifying discussion. Extensive data (99215) is two of the three categories, and category one still needs three items when it is one of them, so extensive usually means three items plus an external discussion.

In practice: the prior prescriber's records and an outside therapist's summary are two unique sources. External applies here too, so the note by the therapist down the hall in your own practice is not countable. A lithium level, a TSH and a metabolic panel are three items, though ordering and later reviewing the same test counts once, and each needs a documented clinical reason. Order tests to answer a clinical question, never to reach a data threshold.

Collateral from a spouse, parent or group home staff when the patient cannot give an adequate history is the independent historian: document who, why it was needed, and what they said. Obtaining collateral and disclosing back to a collateral source are different acts, so record the authorization you relied on. State mental health confidentiality law, 42 CFR Part 2 and minor consent rules can be stricter than HIPAA. Collateral contact documentation covers the mechanics.

Do not double count rating scales. An instrument you separately report, for example under 96127, cannot also count as data toward the E/M, though a scale you are not separately reporting can count as a unique test. And "outside records reviewed" with no source named is an uncountable source.

Element three: risk of management

Risk is what your management decision exposes the patient to, including the options you rejected. It is where psychiatric visits most often qualify at moderate, and where notes most often fail to show it.

Moderate risk is anchored by prescription drug management: starting, stopping, changing, titrating, tapering or cross-tapering. The AMA position is that a documented decision to continue a medication, with risks and benefits weighed on the page, also counts, but some payer auditors read that narrowly and want a change. If continuation is the only thing carrying risk, expect it to be tested, and a bare refill with no reasoning reads as low risk. Moderate also covers care constrained by social circumstances: cost, no transport to the lab, housing that makes a divided-dose regimen unworkable. Name the constraint and how it changed the plan.

High risk has two realistic anchors. The first is a decision about hospitalization or hospital-level care, counted when you genuinely consider it, admitted or not. Note the qualifier: intensifying outpatient treatment or shortening the interval is not by itself this element. Where you judge that a patient with acute suicidal ideation can be managed as an outpatient with a safety plan, means restriction and short-interval follow-up, that weighing is the hospitalization decision for coding. The coding rule never drives the disposition: document what you judged clinically appropriate, and record the crisis resources you gave, including 988. Suicide risk assessment documentation goes deeper.

The second anchor is drug therapy requiring intensive monitoring for toxicity, under three constraints: the drug must be capable of causing serious morbidity or death; the monitoring must be by laboratory, physiologic or imaging test, not by asking about symptoms; and long-term monitoring has to run at least quarterly. Clozapine with absolute neutrophil count monitoring qualifies. Short-term monitoring can, but it must target toxicity: repeating a PHQ-9 to see whether an antidepressant works is efficacy monitoring, as is a level drawn to confirm therapeutic range.

Controlled substances do not automatically mean high risk. A stimulant, or a continued benzodiazepine, is prescription drug management, which is moderate. It can rise when the picture supports it, for example a taper in a patient with prior withdrawal seizure. Controlled prescribing demands documented risk and benefit weighing, prescription monitoring program review where required, and a rationale; the schedule alone does not meet the high-risk element.

The three levels side by side

Element 99213 (low) 99214 (moderate) 99215 (high)
Problems addressed One stable chronic illness at goal, or a single uncomplicated acute problem A chronic illness that worsened, progressed or is causing side effects; two or more stable chronic illnesses; or an undiagnosed new problem Chronic illness in severe exacerbation, or an illness threatening life or bodily function
Data reviewed Two category one items, or an independent historian Three category one items, an independent interpretation, or a qualifying external discussion Two of the three categories, category one still needing three items
Risk of management Over-the-counter agents, supportive measures, no decision shown Prescription drug management; care constrained by cost, transport or housing Admission or hospital-level care genuinely considered; drug therapy needing frequent lab monitoring for toxicity
Total time threshold if selecting by time 20 minutes or more 30 minutes or more 40 minutes or more

What the documentation has to show

The level is carried by the assessment and plan. A reader who never met the patient should be able to name the two elements that carried it. Both excerpts below are fictional.

FICTIONAL EXAMPLE: 99214 ASSESSMENT AND PLAN EXCERPT

ASSESSMENT

1. Major depressive disorder, recurrent, moderate, partial response. PHQ-9 today 14, down from 19 six weeks ago. Still waking at 0430, still missing two shifts a month. Partially responding chronic illness, not a stable one.

2. Sexual side effects on current SSRI, four weeks, the reason for two missed doses weekly.

PLAN

Options discussed: raise sertraline to 150 mg, add bupropion XL 150 mg, or cross-taper. Patient prefers augmentation given partial benefit already obtained.

Start bupropion XL 150 mg every morning, continue sertraline 100 mg daily. Seizure threshold reviewed, no seizure or eating disorder history. Activation and insomnia discussed as expected early effects, morning dosing advised. Discussed the risk of emerging or worsening suicidal thoughts during initiation and dose change, what to watch for and what to do; patient agreed.

Prior prescriber's records obtained this week: adequate escitalopram trial, 20 mg, ten weeks, no response.

Return visit four weeks, telephone check at two weeks. Contact the office for worsening mood or agitation; call or text 988 (Suicide and Crisis Lifeline) or go to the emergency department for thoughts of self-harm.

Level selected by medical decision making.

Why this matters. Two elements carry this to moderate. Problems: a chronic illness with partial response plus a side effect being managed. Risk: prescription drug management, with the options weighed on the page. Data is minimal: one unique external source, one category one item, which does not reach limited and does not need to. The PHQ-9 is not counted as data, because an instrument separately reported under 96127 cannot also count toward the E/M. See medication informed consent documentation.

FICTIONAL EXAMPLE: 99215 ASSESSMENT AND PLAN EXCERPT

ASSESSMENT

Bipolar I disorder, current episode manic, severe, without psychotic features. Four nights of two to three hours of sleep, pressured speech, approximately $3,000 spent in five days, self-reported speeding. Severe exacerbation with impaired judgment.

PLAN

Admission considered and discussed with the patient, who stated the options and risks in her own words. Capacity to authorize disclosure assessed and documented; she authorized speaking with her sister, present today. For admission: severity, driving history, sleep deprivation. Against: no suicidal or homicidal ideation, intact reality testing, agreement to stop driving and hand over keys, sister in the home, daily contact agreed. Decision is outpatient management tonight with in-person reassessment tomorrow, and a stated threshold for emergency evaluation if sleep does not improve, driving resumes, or the sister cannot remain. A family member is not a substitute for a controlled setting, and admission remains on the table.

Lithium yesterday 0.6 mEq/L, creatinine and TSH normal. Increasing lithium from 600 mg to 900 mg nightly, with repeat level and renal function in five days. Quetiapine 100 mg nightly for sleep, titrating toward an antimanic dose over several days. Sedation, orthostasis and metabolic monitoring discussed.

Telephone contact with the patient's therapist today regarding the safety plan and session frequency.

Level selected by medical decision making.

Why this matters. Problems reach high on severe exacerbation. Risk reaches high on one ground: the documented hospitalization decision, with both sides of the weighing shown. Notice what is not claimed: the five-day recheck during a lithium increase is largely a therapeutic range check, and the renal review is short-term, so intensive monitoring is not counted. Data reaches moderate on the therapist call, not the conversation with the sister, because family members are not an appropriate source for category three.
These excerpts illustrate how the elements appear on the page. They are not a recommended disposition. Whether a person in acute mania or with acute suicidal ideation can be managed outside hospital is an individual clinical judgment, and reporting duties for medically impaired drivers vary by state. Check your state law and licensing board.

Selecting the level by total time instead

You may select by the total time you personally spend on the date of the encounter: the face-to-face portion plus pre-visit record review, ordering, documenting and coordinating care. It excludes clinical staff time, time on a different day, and time already counted toward a separately reported service.

CPT removed the upper bounds in 2023, so these are thresholds, not ranges. Total time has to meet or exceed 20 minutes for 99213, 30 for 99214 and 40 for 99215. A 58 minute visit is a 99215 on time; there is no band it falls outside. Document the minutes and activities: "total time 34 minutes: pre-visit record review, the visit, medication ordering and documentation" is what the rule asks for, and a bare number is not. Either method is permitted, provided the documentation supports the one you rely on.

Prolonged service may only be added when you selected the E/M by total time and reached the top level, and it is reported in full 15 minute increments. The CPT add-on begins at 55 minutes for an established patient; Medicare's code begins higher, so confirm your payer's threshold.

Time-based selection is unavailable when you report a psychotherapy add-on. If you bill 90833, 90836 or 90838 alongside the E/M, the level has to be selected by medical decision making, because the psychotherapy time must be separate and distinct from the E/M work. You cannot count psychotherapy minutes toward the E/M, and you cannot report prolonged service on a date you bill an add-on. See choosing the psychotherapy add-on tier for how psychotherapy time is measured.

Where prescribers get this wrong

In the charts I review, 99213 is the reflex level, because it feels safer. It is not neutral: a level pattern that does not track the record misdescribes the care delivered. Select the level the documentation supports, up or down. If you changed a medication and documented the weighing, you have moderate risk, and that is one element; check whether a second reaches moderate, usually the problem itself, before calling the visit a 99214. The mirror image is a template that carries last visit's assessment, plan and level forward when nothing was reconsidered. Tells: "continue current medications" while the level claims prescription drug management, a data line claiming record review with no source named, five diagnoses listed where one was addressed.

Frequently asked questions

Can a stable refill visit ever be a 99214?

Sometimes, but not by default. If the patient is at treatment goal on one medication and you made no decision beyond continuing it, you are at one stable chronic illness and low risk, which is a 99213. It becomes moderate when a second element rises: another chronic condition managed, a side effect you are addressing, or a data count reaching three category one items.

Does checking the prescription monitoring program count as data?

Treat it cautiously. A database query is not a test result, and is not a clinical note from a unique external source, so counting it as a data item is a stretch I would not rely on. It earns its keep on the risk axis instead: document what you found and how it changed your controlled substance decision.

Does a telehealth visit change how I pick the level?

The medical decision making rules carry over unchanged. What varies by payer is place of service, whether the encounter was audio and video or audio only, which modifier applies, and whether audio only is payable at all. Confirm current policy per payer.

Related guides

OneStep Scribe drafts this documentation automatically.

OneStep Scribe is an AI scribe built for psychiatric prescribers. It listens to the visit, in person or over telehealth, and drafts the complete note from what occurred in the visit, including the telehealth details and a billing section, for your review and signature. Every account is NPI-verified.

Start a 14-day free trial

This article is educational and reflects one clinician's understanding of coding rules at the time of writing. It is not legal, billing, or payer-specific advice, and it does not guarantee reimbursement or audit outcomes. Coding rules, payer policies, and state and federal laws change; always verify current requirements with your payers, coding resources, and compliance advisors, and consult your licensing board(s), current DEA guidance, and a healthcare attorney for licensure, consent, and controlled-substance questions. CPT is a registered trademark of the American Medical Association.