PT and OT Evaluation Codes 97161-97168: Worked Cases

Eight CPT codes, 97161 through 97168, cover PT and OT evaluations and re-evaluations under Medicare Part B. Per MLN Matters MM9782, the CPT Editorial Panel created these codes for CY2017, replacing 97001-97004 effective January 1, 2017; all eight are “always therapy” codes billed with the GP (PT) or GO (OT) modifier.

Each discipline has three evaluation levels, low, moderate, and high complexity, plus one re-evaluation code. The level you choose doesn’t change what Medicare pays: 97161, 97162, and 97163 all carry a total non-facility value of 2.93 RVUs in the January 2026 national relative value file, an unadjusted illustration of about $97.86 at the CY2026 conversion factor. CMS valued the three levels as a group on purpose in the CY2017 final rule, to limit upcoding while collecting utilization data, and that group valuation still holds in the January 2026 RVU file.

Parity doesn’t mean coding doesn’t matter: the level on a claim is a statement about the note, other payers aren’t bound to Medicare’s parity, and re-evaluation has its own triggers unrelated to any calendar. This post quotes the CMS descriptors, walks four PT and three OT evaluations through every component, and covers when 97164 or 97168 is billable. Current as of September 2026.

97161 vs 97162 vs 97163: the four components

CMS’s MM9782 reproduces the CPT long descriptors that officially define these codes, noting “the short descriptors cannot be used in place of” them. The four required PT components:

All three require “a standardized patient assessment instrument and/or measurable assessment of functional outcome.” The exam-element ranges also overlap at 4, so a 4-element exam alone never proves high complexity, this post’s own reading, not CMS’s.

The lowest component sets the level

Every descriptor opens with “requiring these components” and lists all four joined by “and.” Read literally, a level is supportable only when every component is met at that level, so the evaluation is billed at the level where all four are satisfied, usually the level of whichever component is lowest. That’s this post’s reading of the text, and it matches what CMS told commenters directly: on evaluations that fall between two levels, CMS wrote in the CY2017 final rule that “the therapist should select the evaluation complexity level that best represents the furnished service and for which the medical necessity is clearly documented.”

Typical times are not thresholds

The 20/30/45-minute language is not a minimum. CMS said so directly in the CY2017 final rule: “these ‘typical times’ are included as a frame of reference and do not represent a minimum coding requirement.” Document actual minutes for the record, but don’t hold a level back, or push one up, to match the typical time.

How to choose the physical therapy evaluation complexity level

CMS itself asked, in the rulemaking creating these codes, “how the physical therapist differentiates the number of personal factors that actually affect the plan of care.” That question points to the counting discipline the descriptors require:

• Personal factors/comorbidities count only if they change the plan (a precaution, a slower progression, a different goal or discharge target). One that doesn’t change the plan isn’t countable.
• Exam elements span three categories: body structures/functions, activity limitations, and participation restrictions, each backed by a standardized test or measure, not an informal observation.
• Clinical presentation is one of three states: stable/uncomplicated, evolving/changing, or unstable/unpredictable, tied to a documented finding.
• Decision making must name a standardized instrument or measurable functional outcome at every level.

Four PT evaluation cases, worked

The cases below are hypothetical composites built to illustrate the framework, not records of actual patients.

Case 1, 97161. 34, ankle sprain grade I, 10 days ago, no plan-changing comorbidities (low). Exam: ankle ROM and a gait/hop assessment, two elements (low). Presentation stable, improving (low). Decision making uses the Lower Extremity Functional Scale, predictable course (low). All low: 97161.

Case 2, 97162 (exam alone would support high). 58, three weeks post total knee arthroplasty. History: diabetes (wound precautions) and BMI over 40 (loading progression), two factors (moderate). Exam: knee ROM/quad strength, gait with device and stairs, a work-tolerance measure: four or more elements (high). Presentation evolving, swelling and ROM changing on an expected trajectory (moderate). Decision making moderate. History and presentation are lowest, so 97162: exam breadth would support high, but two comorbidities and a predictable, evolving course cap the level.

Case 3, 97163. 72, Parkinson’s disease, two falls this month, heart failure, peripheral neuropathy, living alone: three or more factors (high). Exam: strength/rigidity, Berg Balance Scale, gait speed, community-ambulation measure: four or more elements (high). Presentation unstable, fluctuations tied to medication timing plus a recent fall (high). Decision making high: fall risk, cardiac precautions, equipment/caregiver decisions. All high: 97163.

Case 4, 97162 (history and exam alone would support high). 45, chronic low back pain, two years, unchanged for six months. History: depression, obesity, a physically demanding job, three factors (high). Exam: lumbar ROM/hip strength, lifting/sitting tolerance, an Oswestry-scored participation measure, four or more elements (high). But presentation is stable, long-standing, predictable with activity, the low-level language, not “evolving” or “unstable” (low). Decision making moderate. Read strictly, a stable presentation is the low-level component and the lowest component would point to 97161, while history, exam, and decision making all exceed low. This post bills 97162 as the level that best represents the service (CMS’s instruction for evaluations between levels), documents why the presentation is not evolving or unstable, and does not bill 97163 despite three factors and a broad exam. Edge case: check this specific call against your own documentation policy rather than treat it as settled.

97165 vs 97166 vs 97167: counting performance deficits

The OT descriptors, also from MM9782, are built around performance deficits, not exam elements:

The deficit-count ranges overlap at 3 and at 5, so a boundary count is decided by the other components, not the number alone:

• A performance deficit is a skill impairment (physical, cognitive, or psychosocial) producing an activity limitation or participation restriction, not the diagnosis or the activity itself: “can’t dress independently” is the limitation; shoulder ROM and sequencing are the deficits behind it.
• A comorbidity counts only if it affects occupational performance: low requires none, moderate says the patient “may” have one, high says the patient “presents with” one.
• Assistance/modification is a required component, running “not necessary” (low) to “minimal to moderate” (moderate) to “significant” (high).
• History depth runs brief (low) to expanded (moderate) to extensive (high).

Three OT evaluation cases, worked

As with the PT cases, these are hypothetical composites, not records of actual patients.

Case 1, 97165. 40, distal radius fracture, cast removed yesterday. Brief history, orthopedic note review (low). Deficits: wrist ROM and grip strength limiting dressing fasteners and keyboard work, two physical deficits (low). No performance-affecting comorbidities (low). Assessment completed without modification or assistance (low). Decision making low, limited options. All low: 97165.

Case 2, 97166. 67, four weeks post right-hemisphere stroke, mild left hemiparesis, home with a spouse. Expanded review of hospital/rehab records plus cognitive/psychosocial history (moderate). Deficits: left UE motor control/coordination (physical, 2), attention/short-term memory (cognitive, 2), anxiety about cooking (psychosocial, 1), five total, each tied to a limitation like meal prep or medication management. Diabetes and hypertension with fatigue are performance-affecting comorbidities. Assessments required verbal cueing and task simplification (moderate assistance). Decision making moderate. Five deficits sit exactly on the moderate/high boundary; assistance level and history depth (expanded, not extensive) resolve it to 97166.

Case 3, 97167. 78, two weeks post hip fracture repair, moderate dementia, depression, prior independent living alone. Extensive review of hospital, rehab, primary-care, and collateral family history (high). Deficits: LE strength/balance for transfers and UE endurance (physical, 2), memory/sequencing/safety awareness (cognitive, 3), depression with low initiation (psychosocial, 1), six total, producing limitations in bathing/toileting and a participation restriction around living alone. Comorbidities affect performance (high). Significant modification plus physical/verbal assistance were needed (high). Decision making high: discharge placement, caregiver training, multiple options. All high: 97167.

Does Medicare pay more for 97163 than 97161?

No. All three PT levels, and all three OT levels, carry the same RVU total per the January 2026 national relative value file (RVU26A):

*This post’s own illustration (total RVU x the CY2026 non-QPP conversion factor of $33.4009, GPCIs at 1.0), not a CMS-published payment amount. Local MAC amounts differ, and the multiple procedure payment reduction, sequestration, and coinsurance apply on top; facility PE RVUs equal non-facility for all eight codes.

This parity isn’t an accident. The specialty societies had projected moderate PT would be billed 50 percent of the time (low and high each 25 percent) and OT would run 50/40/10 percent low/moderate/high, per the CY2017 final rule. CMS worried that the coding stratification “may result in upcoding incentives” and that stratified payment rates “may provide, in some cases, a payment incentive to therapists to upcode to a higher complexity level than was actually furnished to receive a higher payment.” Using its Protecting Access to Medicare Act authority to value services as groups, CMS instead finalized one work RVU per family (1.20 in CY2017): it “limits the incentives for and consequences of upcoding” while “allowing us to collect and analyze utilization data of the complexity levels for possible future rulemaking.” CMS agreed that “an entire year of data is likely needed” to analyze utilization, and noted that “other payers may have direct interest” in paying by complexity.

Why evaluation-level distribution still matters for audits

No MAC, CERT, or OIG source on evaluation-level distribution exists in this post’s sources, so treat this as this post’s own reasoning, not a CMS finding. CMS’s stated goal of removing the upcoding incentive and gathering distribution data implies CMS is watching it; the societies’ own projections put high-complexity at roughly a quarter of PT evaluations and a tenth of OT evaluations, so a practice far outside those ranges is an outlier. Commercial and Medicare Advantage payers aren’t bound to Medicare’s parity, and CMS itself flagged that other payers may pay differently by level, so a 97163-heavy pattern can matter outside Medicare even where it’s a non-issue inside it. And the level on the claim is a statement about the record regardless of what it pays: a 97163 note showing one comorbidity and a stable presentation is inconsistent on its own terms.

When to bill 97164 or 97168 re-evaluation

Per MM9782, 97164 requires “an examination including a review of history and use of standardized tests and measures” and a “revised plan of care”; 97168 adds an updated profile.

Never on a calendar. The Benefit Policy Manual (Chapter 15, Rev. 13889, issued July 30, 2026) defines re-evaluation as “periodically indicated… when the professional assessment of a clinician indicates a significant improvement, or decline, or change in the patient’s condition or functional status that was not anticipated in the plan of care,” per Chapter 15, warning that “routine weekly assessments of expected progression… are not payable as re-evaluations.” If a contract or practice act requires a fixed-schedule re-assessment, that governs documentation for that payer or board, not Medicare billing, which still requires “new clinical findings, a significant change in the patient’s condition, or failure to respond to the therapeutic interventions,” or “prior to planned discharge for the purposes of determining whether goals have been met.”

NCCI rules. Per the 2026 NCCI Policy Manual (Chapter XI, revised January 1, 2026): only one evaluation or re-evaluation per discipline per date of service, and one practitioner can’t bill both a PT and OT evaluation/re-evaluation the same day. Evaluation codes get one unit per episode of care; re-evaluation codes, one unit per date of service. Re-evaluations “shall not be routinely reported during a planned course of physical or occupational therapy,” but a medically necessary status change can be reported same-day with modifier 59 or XU (NCCI edits and modifier 59).

Who may perform it, and how time counts. “Only a clinician may perform an initial examination, evaluation, re-evaluation and assessment or establish a diagnosis or a plan of care.” An assistant can contribute measurements within scope, but “the clinician must actively and personally participate” (see PTA/OTA supervision requirements). And “evaluation minutes are untimed… but minutes of evaluation shall not be included in the minutes for timed codes,” so keep that separation clean against the 8-minute rule unit count.

Re-evaluation vs progress report vs recertification

The two clocks run separately: “the dates for recertification of plans of care do not affect the dates for required progress reports.” Recertification is a physician signature, not a clinical trigger, so it doesn’t by itself justify 97164 or 97168; only the triggers above do. See Medicare PT documentation requirements and plan of care certification.

Evaluation header documentation template

A note carrying these fields makes the level defensible without extra narrative:

PT evaluation header

Diagnosis and body part evaluated:
Evaluation level billed: 9716_ (GP)
History: [factor]: [plan change it drives]
  Count: __ (low 0 / mod 1-2 / high 3+)
Examination: [test]: [structure/function | activity limitation | participation]
  Elements: __ (low 1-2 / mod 3+ / high 4+)
Presentation: stable / evolving / unstable, supported by: [finding]
Decision making: low / mod / high, outcome instrument: [name, score]
Face-to-face time: __ min (a reference, not a minimum)
Level rationale: [controlling, lowest component]

OT evaluation header

Diagnosis and body part / performance area:
Evaluation level billed: 9716_ (GO)
Profile/history: brief / expanded / extensive, records: [list]
Deficits by domain: physical [ ] cognitive [ ] psychosocial [ ]
  - [deficit]: [limitation or restriction it produces]
  Count: __ (low 1-3 / mod 3-5 / high 5+)
Comorbidities affecting performance: none / [list with effect]
Assistance to complete assessments: none / min-to-mod / significant
Decision making: low / mod / high, options: limited / several / multiple
Face-to-face time: __ minutes
Level rationale: [controlling component]

The evaluation level on a claim is only as defensible as the note behind it, and this counting is exactly the arithmetic that’s easy to skip at the end of a long day. PrismEHR lets organizations define their own billing rules, including minute and unit restrictions by code, enforced at documentation and reinforced at claim creation, so a second evaluation unit in the same episode or an undocumented re-evaluation can be blocked before it reaches a claim; units also default to zero rather than prefilling from the prior visit. Build your template to force the component counts, not just the diagnosis. See how this fits physical therapy practices, and pair it with Medicare PT documentation requirements.