Home exercise program adherence: the prescription evidence

Fifteen adults over 65, randomly prescribed two, five, or eight strengthening exercises, reassessed a week later. That 1999 experiment is close to the entire trial literature on how many exercises a home program should contain. Its finding is easy to compress into “fewer exercises, better adherence.” What it actually found is more specific, and more useful.

One boundary before anything else: why patients quit therapy altogether, and who is most at risk, is a different question with its own evidence, covered in our patient retention guide. This piece stays on the prescription side, the part you control at the plan of care: what the program looks like, how it reaches the patient, and how you find out whether it is happening.

How many exercises should a home program have?

No trial has established an optimal number. The study behind the rule of thumb is Henry, Rosemond and Eckert in Physical Therapy (1999). Fifteen community-dwelling adults aged 67 to 82 were randomly given 2, 5, or 8 general strengthening exercises, taught once, and scored 7 to 10 days later. The two-exercise group performed their exercises better than the eight-exercise group: less cueing, better alignment, better movement control. On self-reported compliance, the outcome the advice is usually borrowed for, the groups did not differ. The authors closed by writing that the optimal number “warrants further study.” It still does. The adherence reviews cited through this piece pool trials by technique and by delivery format; none of them analyzes the number of exercises prescribed, because the trials to pool are not there.

So the honest reading is narrower than the rule. Fewer exercises has evidence for being performed correctly, in one small study of older adults, and none for being done more often. The direction still earns respect, because it lines up with what is known about information load. Kessels’ review of patients’ memory for medical information, in the Journal of the Royal Society of Medicine, is blunt: the greater the amount of information presented, the lower the proportion recalled correctly. An eight-exercise program is a lot of information.

Do HEP apps beat paper handouts?

By a little, in the short term, and not for the reason an app vendor would prefer. The head-to-head trial to know is Lambert and colleagues in the Journal of Physiotherapy (2017). Eighty adults with upper or lower limb conditions were prescribed a four-week program and randomized to a paper handout or to an app. Adherence came out 1.3 points higher on an 11-point scale in the app group, and the authors wrote that “the clinical importance of this added adherence is unclear.” Look closer at the arms, though. The app group also received supplementary phone calls and motivational text messages, so strictly the comparison was an app plus a human paying attention, against paper alone.

The review evidence keeps that shape. A systematic review in Archives of Physiotherapy (Lang et al., 2022) found ten randomized trials of digital adjuncts to home exercise, 1,117 participants in all. Seven favored the digital arm on adherence. The three that found no difference were the three measuring longer-term outcomes, and the authors’ verdict was that digital additions “can likely increase exercise adherence in the short term, with longer term effects less certain.” Strip out the support and compare the bare formats, and the difference tends to vanish. A randomized trial in Sensors (2023) delivered the same four-week balance program to older adults by smartphone or on paper. No group difference appeared on any outcome, self-reported enjoyment included.

The delivery format is not the active ingredient. The attention wrapped around it is. Worth knowing before a vendor invoice promises the opposite.

Which behavior-change techniques have trial evidence?

Five, on moderate evidence in musculoskeletal populations: social support, goal setting, instruction of the behavior, demonstration of the behavior, and practice or rehearsal. That list comes from a systematic review in the British Journal of Health Psychology (Meade et al., 2019) of randomized trials in persistent musculoskeletal pain. Notice that the last three items amount to teaching the exercise properly and having the patient perform it in front of you.

A meta-analysis in Patient Education and Counseling (Eisele et al., 2019) sized the effect across 22 trials in chronic musculoskeletal conditions, where the common techniques were graded tasks, goal setting, self-monitoring, problem solving and feedback. The pooled effect on adherence at three to six months was small, a standardized mean difference of 0.20, and by seven to twelve months there was no significant effect at all. The two reviews even disagree about dosage of technique: Eisele’s subgroup analysis favored interventions stacking more techniques, while Meade found interventions using seven or fewer worked best. That is the actual state of the science: real effects, small and short-lived, with open arguments about how to combine them. Any pitch quoting a large permanent adherence gain is quoting something other than these trials.

Will patients remember what you showed them?

Mostly no, and part of what they remember will be wrong. Kessels’ review carries the two numbers worth keeping: 40 to 80% of medical information provided by practitioners is forgotten immediately, and almost half of what is remembered is remembered incorrectly.

The exercise-specific version is older and sharper. Friedrich, Cermak and Maderbacher in Physical Therapy (1996) followed 87 patients with neck or low back pain who learned their exercises either under a physical therapist’s supervision or from a brochure alone. In the brochure group, only about half performed the exercises properly at follow-up. The supervised group performed better, and the quality of performance correlated strongly with the decrease in pain. That last correlation is the part to sit with: in their data, performance quality traveled with the pain result, and the brochure-only group ended with fewer improvements in impairment.

This converges with the technique evidence above. Demonstration, instruction and rehearsal are both the best-evidenced behavior-change techniques and the answer to the recall problem. Teach the program in the clinic and watch the patient perform it before it goes home. The handout, the video, the app: memory support for teaching that already happened.

How do you know whether the program is happening?

With the tools most clinics use, you cannot reliably tell. The measurement literature says so without embarrassment. A systematic review in BMJ Open (Bollen et al., 2014) collected 61 self-report adherence measures across 58 studies: 29 questionnaires, 29 logs and diaries, two visual analogue scales and one tally counter. Two of the 61 scored positively on even one psychometric property. The researchers who later built the Exercise Adherence Rating Scale (Newman-Beinart et al., Physiotherapy, 2017) opened from the same place: no gold standard exists for measuring adherence to prescribed home exercise, and diaries are limited by inaccurate recall and self-presentation bias. The patient who likes you fills in a kinder diary.

Profiling at the evaluation does not rescue this. A longitudinal analysis in Physical Therapy (Arensman et al., 2023) tracked adherence session by session in 208 low back pain patients across 58 primary care practices and found three distinct trajectories: declining (12%), stable (45%) and increasing (43%). No baseline characteristic distinguished the groups. The authors’ advice to therapists: do not try to guess a patient’s trajectory at the start of treatment; monitor adherence as treatment progresses and support it when it slips.

Between-visit monitoring is also the one piece of this Medicare has decided to pay for. The remote therapeutic monitoring codes have covered physical therapists since 2022, and CMS says the monitored data can be patient reported rather than device captured. Whether the arithmetic works for your caseload is its own question, worked through code by code in our RTM guide.

The prescription the evidence actually supports

Assembled from the trials instead of the tips, the defensible home program is short. Short is performed better and remembered better, even though nobody has proven it is done more often. It is taught by demonstration and rehearsed in front of the therapist before it goes home, because that is where the technique evidence and the recall evidence agree. It leaves the clinic in writing or on video, as a reminder of what was taught. It has goals set with the patient and, where possible, someone at home enlisted. And it gets checked on while the episode is running, because no baseline profile predicts who will drift, and the diary will be polite about it.

The software share of that list is smaller than the HEP industry implies. In Orion, the plan of care holds goals and interventions on one case record, and the patient portal carries secure messages and documents to the patient. The program you sent home and the conversation about it live in the same chart as the visits. The teaching, the rehearsal and the check-in remain clinical work, which is the point.

Henry’s fifteen volunteers were never going to settle how a profession prescribes. Until a bigger trial does, prescribe the way the study actually read: fewer things, taught well, checked on.