Home Exercise Program Software: A Buyer's Framework

Non-adherence to home-based physical therapy exercise can reach 70 percent, and physiotherapists’ own perception of how well patients follow through averages only 67 percent (Essery et al. 2017; Peek et al. 2017). In musculoskeletal cohorts specifically, non-compliance runs between 30 and 50 percent, with average adherence of 67 percent across 12 studies and no gold-standard measure, since most methods rely on self-report prone to bias (a viewpoint article; Argent et al. 2018).

A disclosure before we go further: we build Rehab Prescribed, Ready2Exercise, and PrismEHR, so read what follows as the framework of an interested party, not a neutral one. Every figure below carries its own source and date, and nothing here ranks a specific vendor against another.

This is a buyer’s framework, not a vendor list: what the adherence trials actually tested, how many exercises to prescribe, why CPG alignment protects a plan of care under Medicare’s manual, whether adherence data reaches the note or just sits in a chart as a PDF, what CY 2027 proposes for remote therapeutic monitoring, and a scoring worksheet to run against any demo. Current as of September 2026.

How much does home exercise adherence actually matter?

A systematic review of 20 musculoskeletal outpatient studies found strong evidence linking poor adherence to low baseline activity, low self-efficacy, depression, anxiety, more perceived barriers, and higher exercise pain (Jack et al. 2010). It matters beyond compliance for its own sake: in a 60-month follow-up of 150 hip and knee osteoarthritis patients, adherence to home exercise was significantly associated with better pain, function, and self-perceived outcomes, consistently across all five years, though adherence itself declines over time (Pisters et al. 2010). The predictors with relatively strong evidence are intention to engage, self-motivation, self-efficacy, prior adherence, and social support (Essery et al. 2017), none of which an app manufactures alone.

Does app or video delivery improve adherence?

Delivery format is where vendor marketing tends to outrun the trial data:

The pattern: features layering human contact onto a program (calls, booster visits) show modest but real evidence; features that only remind or display (automated texts, cues) mostly don’t move adherence alone, and an app did not beat direct supervision in the osteoarthritis literature. Ask which category a vendor’s “improves adherence” claim actually falls into.

How many exercises should an HEP prescribe?

More exercises is not a feature. In adults over 65, a group prescribed 2 exercises performed the exercises better (by performance score) than a group prescribed 8, a group prescribed 5 did not differ from either, and self-reported compliance did not differ among the three groups (Henry et al. 1999). In neck and low back pain patients across 8 primary care centers, prescribing more than 6 exercises cut the odds of frequency adherence to a fifth of what it was for 3 or fewer (OR 0.2, 95% CI 0.1 to 0.9, p < 0.05); and patients whose doubts were clarified adhered better on frequency (adjusted OR 4.1), while patients supervised while learning the exercises adhered better on duration (adjusted OR 3.3) (Medina-Mirapeix et al. 2009).

In knee osteoarthritis, a review of 48 trials found single-aim programs outperformed mixed programs (SMD 0.61 versus 0.16, p < 0.001), and supervised exercise at least three times a week beat less frequent supervision (Juhl et al. 2014). A 2024 Cochrane review of 139 knee osteoarthritis trials found no relationship between outcomes and the total number of sessions prescribed (Lawford et al. 2024). The implication: a library that makes it easy to hand a patient 10 or 12 exercises works against this evidence, not with it.

Why clinical practice guideline alignment makes the plan of care defensible

Medicare’s manual defines a plan of care’s minimum contents: diagnoses, long-term goals, and the type, amount, duration, and frequency of therapy, consistent with the evaluation and stated as measurable functional goals (Medicare Benefit Policy Manual, Pub. 100-02, Ch. 15, §220.1.2(B); see our plan of care certification guide). Reviewers judge medical necessity by whether services are skilled, safe, and effective, meaning progress indicates the care is working (§220.3(A)). A program built from a current guideline gives a reviewer an obvious answer to “why this exercise, at this dose.”

That library is current. As of September 2026, APTA’s CPG page lists 2026 and 2025 entries on total knee arthroplasty (revision 2026), hip osteoarthritis (revision 2025), and carpal tunnel syndrome, plus the ACSM resistance training position stand and female athlete injury-prevention recommendations (APTA CPG library), and the still-current low back pain (2021) and neck pain (2017) guidelines remain the reference points for the two most common outpatient diagnoses (George, Fritz et al. 2021; Blanpied et al. 2017). The page notes guidelines “assist clinical decision-making but do not take the place of clinician judgment,” the right caveat for software too: a CPG-linked tool should surface a recommendation, not silently auto-select it. Rehab Prescribed builds programs from recommendations in currently published guidelines, the kind of feature this section describes.

Does adherence data flow into the chart and the note?

Documentation “must be legible, relevant and sufficient to justify the services billed” (§220.3(A)), but the treatment note has a narrower job than the plan of care: exercise specifics “need not be repeated in the treatment notes unless they are changed from the plan,” and “new exercises added or changes made to the exercise program help justify that the services are skilled” (§220.3(E), Medicare Benefit Policy Manual; our PT documentation requirements post covers the note rules in full). Required elements are the date, each billed intervention, total timed-code and treatment minutes, and a signed identification. A “notation of non-skilled treatment or report of activities performed by the patient” is allowed but not required as voluntary supplementary information: an HEP’s progressions, regressions, and swaps belong in the record as dated changes because they justify a skilled visit, while adherence logs are voluntary, not a billing basis for the therapy code itself. RTM is the actual billing path for monitoring home activity, covered next.

That’s why “the plan gets stored in the chart” isn’t the same claim as “the software gives structured adherence data.” As of September 2026, the incumbent benchmark, WebPT, generates a plan, automatically stores a PDF in the EMR chart, and separately shows clinicians patient activity, workouts completed, and an NPS score inside its own platform (WebPT). A PDF in the chart is a document attachment, not data a note can cite by exception. Ask any vendor, ours included, whether adherence lands as text a therapist can reference, or only as a dashboard. PrismEHR keeps exercise programs and compliance tracking inside the same record as the note, not a separate system.

Can the software support RTM billing (and what CY 2027 proposes to change)?

For RTM, the device used “must meet the FDA definition of a device” under FFDCA section 201(h), and unlike Remote Physiologic Monitoring, RTM data can be patient-reported rather than only digitally captured (CY 2022 PFS final rule, 86 FR 64996). Ask any vendor whether its app meets that definition. The CY 2026 code set added three RTM device supply codes covering 2 to 15 and 16 to 30 days per 30-day period, plus revised management codes for the first 10, first 20, and each additional 20 minutes, requiring a therapy modifier for PT/OT and generally CQ or CO for PTA/OTA (CY 2026 PFS final rule, 90 FR 49394 to 49398); see our RTM billing breakdown for the full code table.

Nothing below is final. CMS’s proposed rule (CMS-1848-P, 91 FR 43892 to 43895, published July 16, 2026, comments closed September 14, 2026; nothing takes effect unless finalized, with a proposed effective date of January 1, 2027) would change RTM this way:

The staffing line is worth reading twice: a vendor-supplied “monitoring team” would not generate billable RTM time if finalized, since that time would have to come from staff who are direct employees of the billing practice, not a contractor, beginning January 1, 2027 under the proposal (CY 2027 fact sheet). Software still needs to produce day counts and a management-time log; the monitoring itself would shift to the clinic’s own staff. See our CY 2027 fee schedule overview for the rest.

What do HEP software pricing models and contracts look like?

Two contract terms matter more than a demo shows. First, language access: HHS’s rule requires language assistance be “provided free of charge, be accurate and timely,” and that machine translation of complex or rights-related content “be reviewed by a qualified human translator” (45 CFR 92.201, 2024 edition). Ask whether translated exercise instructions are human-reviewed or generated on the fly.

Second, the BAA termination clause: at contract end, a business associate must, “if feasible,” return or destroy all PHI it holds and retain no copies, or, if that isn’t feasible, extend the contract’s protections to the information and limit further uses and disclosures (45 CFR 164.504(e)(2)(ii)(J), 2024 edition). That decides what happens to a year of adherence logs the day a clinic switches software; ask what format it returns in, and whether export is self-serve or a paid service. (Our HIPAA-compliant form builder post covers the same BAA mechanics from the intake side.)

Expect quote-based pricing. As of September 2026, WebPT does not publish prices, describing pricing as configurable “per provider per month or per visit” (WebPT pricing). Per-clinician, per-clinic, and per-patient models each shift risk differently: per-clinician fees rise with hiring, per-clinic fees favor growing rosters, and per-patient fees track volume. Ask what happens to the fee if volume or headcount changes mid-contract, and whether a plan created before termination stays accessible to the patient afterward.

Exercise prescription software for sports medicine

Return-to-sport programming needs load that progresses across a program, and substitutions that account for whatever equipment is on hand. As of September 2026, APTA’s CPG library includes the ACSM resistance training position stand and female athlete injury-prevention recommendations, alongside the orthopaedic guidelines above (APTA CPG library). The dosage evidence generalizes too: single-aim programs outperformed mixed, grab-bag programs in the knee osteoarthritis literature (Juhl et al. 2014), the same argument for one clear performance goal over an undifferentiated list of strength, mobility, and conditioning work.

Rehab Prescribed draws on a library spanning rehabilitation and performance exercises, so a program can move from a rehab goal to a return-to-sport goal without switching systems, using AI to update progressions, regressions, and swaps as status changes. Ready2Exercise covers the cash-pay side of that population, with workouts adapted to available equipment, included with PrismEHR as a revenue stream separate from insurance on our physical therapy page.

HEP software scoring worksheet

Score any vendor, including ours, against the evidence above rather than a features checklist. Weights sum to 100.

None of the evidence above says a home exercise program has to be complicated to work. It has to be aimed at one goal, linked to a guideline, small enough that a patient actually does it, and built so a therapist can change it and prove that in the note. Rehab Prescribed builds CPG-linked programs with progressions, regressions, and exercise swaps, and tracks patient response and compliance so that data is there when you need it. PrismEHR keeps exercise programs and compliance tracking inside the same chart as the treatment note, for physical therapy and other specialties alike. If Remote Therapeutic Monitoring is part of your evaluation, read our breakdown of the current RTM code set alongside the CY 2027 proposals above before you sign anything.