Physical Therapy Visit Caps and Authorization

A visit cap and a prior authorization limit physical therapy coverage through two different mechanisms, and confusing them costs practices money in opposite ways. A visit cap is benefit design: the plan covers a fixed number of therapy visits per year, and visits beyond it are not a covered benefit, which makes them patient responsibility that no authorization request can restore. Prior authorization is utilization review: visits are covered only if approved, usually in batches, and more can be requested with clinical justification. Medicare uses neither a cap nor routine prior authorization for outpatient therapy; it uses an annual threshold with the KX modifier and targeted medical review above it. Many commercial plans layer a cap and authorization together, so the front desk needs to know which limit governs each patient before the plan of care is built.

What is the difference between a visit cap and an authorization?

The distinction determines what the practice can do when visits run out. A visit cap lives in the benefit contract: 20 or 30 covered therapy visits per year is a common shape, and once they are used, further visits are non-covered rather than denied, meaning the plan never owed them and an appeal has nothing to overturn. The correct workflow at a cap is financial counseling: tell the patient where they stand and what continued care costs. Authorization is the opposite: coverage exists, but each batch of visits must be approved, and a denial of additional visits is a utilization decision the practice can contest with progress documentation or a peer-to-peer call. The failure mode is treating one like the other, appealing into a benefit cap that cannot move, or collecting cash from a patient whose plan only needed a continued-visit request. Verifying benefits at evaluation should capture both numbers: the annual visit maximum, if any, and the authorization pattern, including how many visits the initial approval covers.

Does Medicare cap physical therapy visits?

No, and the mechanism that replaced the cap is worth understanding precisely. The Bipartisan Budget Act of 2018 removed the hard therapy cap and preserved the former cap amounts as thresholds. For CY 2026 the KX modifier threshold is $2,480 for PT and speech-language pathology services combined, and separately $2,480 for occupational therapy, per CMS. Below the threshold, claims process normally. Above it, each claim line must carry the KX modifier, the therapist's attestation that continued treatment is medically necessary and that the medical record documents it; claims over the threshold without the KX modifier are denied. This is not an authorization: nothing is submitted for approval and there is no waiting period. The control sits behind the claim instead, through a targeted medical review process that applies above $3,000 and selects providers based on billing patterns rather than reviewing every claim. The operational consequence is that Medicare therapy compliance is a documentation discipline, not a request discipline: track each patient's accrued amount, append KX deliberately rather than by default, and make sure the progress notes above the threshold would survive a records request.

Keeping count across caps, authorizations, and recertification dates is bookkeeping suited to an agent. Flexbone tracks each patient's remaining visits and flags recertifications ahead of the deadline. See therapy authorization.

How does plan-of-care recertification work?

Therapy coverage rests on a certified plan of care, and certification has its own clock separate from any authorization. Under Medicare, the physician or other qualified practitioner certifies the initial plan when treatment begins, and must recertify it at least every 90 days, or sooner when the plan changes significantly in duration, frequency, or goals. Commercial plans impose their own certification rules, often shorter. A lapsed certification is quiet: visits continue, the schedule looks normal, and the defect surfaces only when claims for the uncertified period deny or fail an audit. The recurring operational problem is that the recertification depends on someone outside the practice, the referring physician, signing on time, which turns a compliance requirement into a chasing-signatures workflow. The fix is to treat certification dates like authorization end dates: log them at evaluation, surface them ahead of expiration, and start the signature chase early enough that a slow physician office does not open a coverage gap. Where therapy EHRs put these dates and how the tracking is automated is covered on our prior authorization for therapy and behavioral health EHRs page.

How does concurrent authorization for continued visits work?

Plans that authorize therapy rarely approve a whole episode at once. The common pattern is concurrent review: an initial batch of visits is approved around the evaluation, and continued visits require a new request supported by progress documentation before the batch runs out. The request typically needs objective measures showing progress, the functional deficits that remain, and the expected remaining duration; a plan reviewer is deciding whether continued skilled therapy is still justified, so notes that show measurable change argue better than notes that show attendance. The mechanics are where practices lose coverage: the request has a turnaround time, so it must be submitted while approved visits remain, and a patient who keeps their scheduled visits during a gap is being treated without coverage. Tracking visits-used against visits-approved per patient, and triggering the continued-visit request at a set remaining-visit count rather than at zero, is the whole game. When a continued-visit request is denied despite documented progress, a peer-to-peer call with the reviewer is usually faster than a written appeal, and the denial reason tells you which documentation element was missing.

How can a practice keep authorization from interrupting care?

The therapy-specific risk is that authorization failures interrupt an active plan of care, which harms outcomes as well as revenue, because a two-week coverage gap in a progressive loading program is a clinical setback, not just a billing one. Keeping the machine running comes down to four tracked numbers per patient: visits used against visits approved, the authorization end date, the plan-of-care recertification date, and, for Medicare, the accrued amount against the KX threshold. Each has a trigger point that starts a workflow before the number hits its limit. The constraint is that most of these workflows run through payer phone lines and portals, and phone work is exactly what therapy front desks do not have slack for; MGMA has documented that phones remain a backlog that costs practices staff time. Practices that keep episodes uninterrupted are the ones that made these four numbers visible on one worklist instead of scattered across the EHR, the payer portal, and a spreadsheet. The revenue side of the same discipline, clean claims and denial follow-up, is covered in our guide to physical therapy billing.

How does Flexbone help physical therapy practices with authorization?

Flexbone runs voice and browser agents, backed by US-based staff, that carry the tracking and chasing work in this post. An agent can verify benefits at evaluation and capture both the visit maximum and the authorization pattern, submit continued-visit requests with the progress documentation attached before the approved batch runs out, track plan-of-care recertification dates and chase the referring office for signatures, monitor Medicare accruals against the KX threshold, and hold on payer phone queues to resolve pending requests, writing every status back to the EHR. We do not make clinical judgments about medical necessity; the agents move information and deadlines, and the therapist's documentation carries the case. What this looks like inside a PT practice's systems is covered on our AI for physical therapy page. To map it against your payer mix and episode volume, book a call with Flexbone.