BRENNEN

Industries / Health / Healthcare / Physical therapy

Scheduling and billing

Scheduling and billing is how patients reach a therapist, with or without a doctor's referral, and how each visit is recorded, coded and paid.

6

Milestones achieved, from 1999 to 2024.

2

To watch for: announced, not yet confirmed outside.

2

Open deliverables. Nobody has shipped them.

The road

AchievedWatch forOpen

Achieved

  1. 1999

    A yearly cap on therapy

    Section 4541 of the US Balanced Budget Act of 1997 put a limit on what Medicare would pay each year for one person's outpatient therapy: from 1999, $1,500 for physical therapy and speech therapy together, and a separate $1,500 for occupational therapy. Congress then passed repeated moratoria that suspended the cap, and from 2006 an exceptions process for medically necessary care. The Bipartisan Budget Act of 2018, signed on 9 February 2018, repealed the cap but kept its amount as a threshold above which claims must carry a modifier, the KX modifier, confirming that the records show the care was needed; for 2026 that threshold is $2,480.

    Lessona hard dollar limit on care proved hard to enforce and gave way to a check on documentation.

    cms.govcms.gov

    Credit

    US Congress; the Centers for Medicare & Medicaid Services (CMS)

  2. 2006

    Dutch patients refer themselves

    Before 2006, people in the Netherlands could see a physical therapist only on a doctor's referral; that year direct access was introduced. In a national network of 93 therapists in 43 private practices, 28% of patients came without a referral in the first year, those patients had fewer sessions, and the total number of patients did not rise from 2005. For low back pain, the share of care episodes started by direct access rose from 28.9% in 2006 to 52.1% in 2009.

    europepmc.orgeuropepmc.org

    Credit

    Dutch government; C. J. Leemrijse, I. C. Swinkels and C. Veenhof (Physical Therapy, 2008); J. Scheele and colleagues (Physical Therapy, 2014)

  3. 2013

    Function codes, then dropped

    From 1 January 2013, US law required every Medicare outpatient therapy claim to report the patient's main limitation in function, using 42 G-codes that paid nothing and seven severity modifiers, at the start of care, at set points during it and at discharge. CMS ended the requirement from 1 January 2019, concluding that collecting the data in the same format "would not yield additional information" to guide reform of therapy payment.

    Lessonsix years of function data were collected on every claim and then judged not useful for the payment reform they were meant to support.

    cms.gov

    Credit

    US Congress (Middle Class Tax Relief and Jobs Creation Act); CMS

  4. 2019

    Paying for patients, not minutes

    Medicare used to pay skilled nursing facilities more per day the more minutes of therapy a patient received. The government's Office of Inspector General found that facilities billed the top level, ultra high therapy, for 49% of therapy days in fiscal year 2011 and 57% in 2013 though patients' characteristics stayed largely the same, and gave exactly the 720-minute weekly minimum for 21% and then 34% of ultra high stays. From 1 October 2019 the Patient Driven Payment Model pays by the patient's diagnoses and needs instead. Average therapy fell from 122.2 to 96.5 minutes a day right after the change, by a 2026 study of 3.5 million stays, which found the drop in therapy strongly associated with fewer successful discharges home and more returns to hospital.

    Lessonpaying for minutes bought minutes, and taking the payment away also took away therapy.

    oig.hhs.govcms.govpmc.ncbi.nlm.nih.gov

    Credit

    HHS Office of Inspector General (2015); CMS; R. A. Prusynski, T. M. Mroz and colleagues (Health Affairs Scholar, 2026)

  5. 2022

    Paid to monitor home exercise

    Remote therapeutic monitoring (RTM) is a set of five billing codes, created by the CPT Editorial Panel, which writes the standard US codes for medical services, in October 2020, for collecting data such as musculoskeletal status, therapy adherence and therapy response from a patient's device or app between visits. In its proposed rule CMS said physical therapists could not bill the codes; after many comments it decided in its final rule of November 2021 that they were meant for therapists and paid them from 2022. For 2026 CMS added three more RTM codes to the therapy list.

    federalregister.govcms.gov

    Credit

    CMS; the CPT Editorial Panel

  6. 2024

    Every US state opens access

    By APTA's account, when Alabama moved to provisional direct access in early 2024, after Mississippi earlier that year and Missouri in 2023, no US state was left with "limited" direct access, under which patients could see a physical therapist without referral only in narrow cases. Twenty-one states allow unrestricted direct access; the rest set limits such as a number of visits before a referral is needed. Worldwide, in World Physiotherapy's 2024 census, 47% of member organizations said direct access is limited to certain settings, 31% reported limits tied to payment, 20% limits on which people or conditions may be treated and 13% limits on treatment time.

    apta.orgworld.physio

    Credit

    APTA and its state chapters; state legislatures; World Physiotherapy

Watch for

  1. AI scribes for therapy notes

    Raintree, which sells record-keeping software to therapy clinics, says its AI scribe, ScribeIQ, cut the time to write an evaluation note at Physical Rehabilitation Network, which runs more than 280 clinics in 16 US states, from about 17 minutes to under 90 seconds, and that therapists seeing 16 evaluations a week save about four hours a week.

    Achieved whenan independent study times therapists' notes with and without the scribe and checks the notes for errors and missing items.

    raintreeinc.com

    Claimed by

    Raintree, January 2026

  2. Phone triage for physio waits

    In the WAIT-less trial, six public hospitals in Sydney set out to assign at random 368 adults referred to outpatient physiotherapy for musculoskeletal pain either to usual care or to a phone call from a physiotherapist who sorts them by risk of a poor outcome with the Keele STarT MSK tool. Low-risk patients get advice during the call and medium-risk patients a course of video physiotherapy. The investigators expect shorter waits for treatment with function at 6 months no worse than usual care; recruitment ran from September 2024 to December 2025, with follow-up due by December 2026.

    Achieved whenthe trial's results are published and show shorter waits without worse function.

    pmc.ncbi.nlm.nih.gov

    Claimed by

    J. M. Hutton, J. R. Zadro and colleagues, Institute for Musculoskeletal Health, Sydney, January 2025

Open

  1. 01

    Minutes the scribe heard

    A public test of whether AI scribes record therapy time correctly. Film staged outpatient sessions, played from scripts by a volunteer physical therapist and healthy adult actors, with a stopwatch log of each activity: therapeutic exercise, manual therapy, gait training and balance work. Score each scribe's note on the minutes it gives each activity and on the billing units those minutes produce under Medicare's rule for timed codes.

    The brief
    Credit · unclaimed
  2. 02

    Time therapy approvals

    A public, service-level record of prior authorization, the insurer's approval before treatment, for outpatient physical therapy in the US. First, a table of the published therapy approval rules of the largest health plans in each market: whether approval is needed, from which visit, through which outside review company, and for how many visits at a time, recoded every six months. Second, a log kept by volunteer clinics of each request's submission time, decision time and visits approved, with no patient details.

    The brief
    Credit · unclaimed

Open 01 · the brief

Minutes the scribe heard

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A public test of whether AI scribes record therapy time correctly. Film staged outpatient sessions, played from scripts by a volunteer physical therapist and healthy adult actors, with a stopwatch log of each activity: therapeutic exercise, manual therapy, gait training and balance work. Score each scribe's note on the minutes it gives each activity and on the billing units those minutes produce under Medicare's rule for timed codes.

Why it is still open

Makers publish time saved, not accuracy (see Watch for). A buyer's guide from Physitrack, which sells exercise software to clinics, found in August 2026 no independent evidence of any scribe's accuracy in physical therapy and advised clinics to test timed codes themselves. Medicare bills timed therapy in 15-minute units, one unit for 8 to 22 minutes, three for 38 to 52, so an error of a few minutes can change a claim. Open-source unit calculators are on GitHub, but they take minutes typed in by a person, and a search of Europe PMC found no test of scribes against a timed record of a therapy session. The staged-visit scribe test proposed in healthcare.md scores note quality, not time.

physitrack.comcms.govgithub.com

Built with what exists

scripts with planned minutes per activity; two cameras and the free annotation tool ELAN for the timed log; AI scribes sold to therapy clinics, under trial access; ElevenLabs Scribe v3 for speech-to-text (see the Note "Fewer word errors") and a language model such as Claude as an open baseline; the timed-code rules in chapter 5, section 20.2 of CMS's claims manual.

  1. First step

    Write five scripts and film one with a stopwatch log, then run one scribe on it.

  2. Done looks like

    40 staged sessions of 30 to 60 minutes; for each tool, the mean error in minutes per activity, the share of sessions in which billed units match the video, and overbilled and underbilled units counted separately; scripts, videos and scores published.

Open 02 · the brief

Time therapy approvals

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A public, service-level record of prior authorization, the insurer's approval before treatment, for outpatient physical therapy in the US. First, a table of the published therapy approval rules of the largest health plans in each market: whether approval is needed, from which visit, through which outside review company, and for how many visits at a time, recoded every six months. Second, a log kept by volunteer clinics of each request's submission time, decision time and visits approved, with no patient details.

Why it is still open

Since March 2026, plans under the CMS rule must post prior authorization statistics, but KFF found they are added up across all medical services, so they cannot show which services are denied, and plans need not report response times by service; the median response was about 1 day. In APTA's survey of 856 members in summer 2025, 80.2% said they waited 3 days or more for a decision, and therapists must often ask again for more visits. Silna, a company that handles approval requests for clinics, told Axios in July 2026 that about 15% of its requests for physical, occupational and speech therapy were only partly approved, but its data are not public and give no times.

axios.comkff.orgcms.govapta.org

Built with what exists

plans' public medical policies and provider manuals; a shared form such as REDCap or a spreadsheet for the clinic logs; an AI model such as Claude to draft codes from policy text, each checked by two people; GitHub for versions.

  1. First step

    Code the therapy approval rules of the six Medicare Advantage insurers in KFF's analysis.

  2. Done looks like

    Rules for at least 60 plans coded by two people who agree on at least 90% of items before reconciling; at least 10 clinics logging at least 1,000 requests over six months; median and 90th-percentile decision time per plan set beside the plan's posted all-service median; an update after January 2027, when US health plans have pledged to answer most complete electronic requests in real time.