How Much Is Enough — An Introduction to Dose-Response in Rehabilitation

We cannot yet answer with a single number how much rehabilitation treatment is enough. This is because there is no single yardstick to measure "how much," and there are not many data tracking how much is actually done in a real therapy room. This article looks at those two things in turn.

The primer "Neuroplasticity — A Brain That Builds New Roads Beside the Broken Ones" defined dose like this: in rehabilitation, dose is how much you practice, commonly measured by practice time or repetition count, while also looking at how frequently and intensely you did it. That piece went as far as to say that the amount of practice matters, but simply increasing the amount does not make the effect grow endlessly. Dose-response is the relationship of how the effect changes when you change the dose. This article starts with the problem that must be solved before examining that relationship: namely, what we measure dose with.

The place of this article within the series is as follows. The preceding piece, "Why Therapy Is Designed That Way — An Introduction to Motor Learning Principles," covered how to arrange practice and what feedback to return, in other words, the "how." This piece is the "how much." But there is a stumbling block right at the starting line. At least three things are lumped together inside the single word "dose."

Why You Need to Know — Three Axes in One Word

The three axes that make up dose are time, intensity, and scheduling. Time is how long the practice lasted. Intensity is how much was done per unit of time, measured by the number of repetitions or repetitions per minute. Scheduling is how practice was distributed and arranged—how many times a day and across how many days. Even with the same total practice time, the dose differs depending on whether it was crammed into one session or spread across multiple sessions, and how many movements were made within that time.

Different axes can even point toward different directions of effect. AVERT, which we saw in "When Does Recovery Happen — What Time Did, and What Treatment Did", is an example. AVERT was a trial that got stroke patients out of bed and moving at a very early stage right after a stroke. In that data, a higher frequency of out-of-bed sessions per day went together with better outcomes, while a greater volume of movement per day went together with worse outcomes. Frequency is the scheduling axis, and volume is the time axis. Under the same word "dose," the directions were opposite.

Researchers point out this very problem. Although the quality of stroke recovery clinical trials has improved, the treatments have failed to clearly alter the path that recovery traces over time—its recovery trajectory. Researchers suggest that this failure may stem from the complexity of dose and the failure to specify it (PMID 31781018). Dose articulation is the process of specifically defining dose along each axis in a study, systematically varying it, and recording those values.

In practice, articulation is rarely done well. A scoping review is a review that maps out and summarizes what research has been done in a field and in what manner. A scoping review that surveyed studies testing altered doses in physical therapy for motor recovery after stroke included 41 papers (PMID 33429089).

• All studies took a one-dimensional approach. This means they only varied one of the three axes.
• The most frequently varied axis was time (53%), followed by intensity (29%), and finally scheduling (18%).
• 95% reported efficacy outcomes showing whether the treatment worked. However, 65% reported the dose, and 45% reported safety.
• 61% had a high risk of bias. Risk of bias is the same concept of bias risk we saw in the primer "Rehabilitation — Starting to Plan a Livable Home as the Fire Is Brought Under Control"—the risk that results skewed in one direction due to flaws in study design.

This means that while almost all studies recorded whether there was an effect, the dose that produced that effect went unrecorded in roughly one out of three. Consequently, readers of a paper know that an effect occurred, but cannot tell how much practice it took to get there. This mirrors what we saw in the previous piece: the conditions of practice are poorly recorded. Effects are reported, and the conditions that produced those effects are underreported.

The Big Picture — Early Studies That Narrow Down Dosage Are Rare

The same review sorted these studies by each step of the discovery pipeline (PMID 33429089). The discovery pipeline is the sequence of research that starts a new treatment in animal experiments and expands it to human trials, narrowing down the questions at each step. In dosage studies, the preclinical stage is the step where the dosage is prepared in animals before it is used in humans. Phase 1 sets the range of doses to try in a small number of people, Phase 2A screens candidate doses, and Phase 2B decides which of those doses to use.

The front end is thin. There are only 3 preclinical and 2 Phase 1 studies that systematically narrow down the dosage, and 27 Phase 2B studies are running on top of that foundation. The authors concluded that this field needs more high-quality early-stage research that systematically examines dosage across multiple axes, and wrote that to achieve this, Phase 1 trial designs in particular must be rewritten (PMID 33429089). This means that the stage of setting the range is almost skipped, and research is concentrated on the stage of deciding which dose to use.

⚠️ The contrast "animals do hundreds of repetitions a day, clinics do dozens," which is often cited in this topic, was not used in this article. Among the abstracts we secured, none stated how many times rodents repeated per day. What we secured is the actual measurement that dosage-controlled research itself is thin at the front end, and this article states only that.

Reading the Three Axes Separately — There Are Also Two Inside Time

Of the three axes, the most frequently used time axis is divided into two again. Time in therapy is the time the patient was within the treatment schedule, and time on task is the time spent actually moving while practicing tasks during that period. Treatment time passes while the therapist explains, adjusts posture, and moves to the next equipment. The difference between these two is clearly revealed in the actual measurements below.

Multidimensionality is manipulating and reporting all three axes together. Only when all three axes are recorded together can we speak of a dose-response showing that "doing this much changed things by that much." All 41 papers seen earlier changed only a single axis (PMID 33429089).

What It Looks Like in Practice — Those Who Counted

If research narrowing down the dose is thin, we at least need to know how much is actually being done in therapy rooms right now. There are three sets of data that have counted this directly.

① Actual Repetition Counts in Outpatient Therapy Rooms

When you count directly in outpatient therapy rooms, the number of repetitions is not high, excluding step counts. The subjects are patients with hemiplegia. Hemiplegia, as seen in the primer "What a Stroke Actually Is — Blocked, Burst, and the Clock", is a condition where one side of the body's limbs lose strength or cannot move at all. Outpatient care means visiting the hospital to receive treatment without being admitted. The starting point of this observation was that if clinicians believe "the more practice, the better," we first need to know how much practice actually happens right now. They observed 36 outpatient physical and occupational therapy sessions and counted the types of activities and their repetitions (PMID 17419883).

Before reading the results, three types of movement are distinguished. Active movement is movement where the patient uses their own strength, and passive movement is when the therapist moves the patient's limbs on their behalf. Purposeful movement is movement directed toward an actual task, such as grasping and moving a cup. Upper limbs refer to the arms and hands, and lower limbs refer to the legs.

• Therapy time averaged 36 minutes per session.
• For sessions dealing with upper limbs, the average repetitions per session were 39 for active movement, 34 for passive movement, and 12 for purposeful movement.
• For lower limb sessions, there were 33 active, 6 passive, and 8 purposeful movements.
• In gait sessions, patients walked an average of 292 steps.
• Transfer sessions, sitting down and standing up to move, averaged 11 times.
• Variation in repetition counts was considerable across most categories.

The authors concluded that, aside from step counts, the number of repetitions is relatively low. That purposeful arm movements were fewer than active and passive movements was unexpected even to the authors. This was because it did not align with the teaching at the time that practicing purposeful movement is essential for boosting function (PMID 17419883).

Once you see these numbers, the starting point for discussions on "increasing dosage" changes. Purposeful upper limb movements average 12 per session.

② The Dosage of Usual Care — The Upper Extremity

Usual care is therapy provided as normally given in a hospital, without a protocol set for research purposes. A systematic review is a study that uses a defined method to find, gather, and organize all relevant research without omission. The dosage of arm exercise therapy that stroke inpatients normally receive is as follows. These values come from a systematic review that searched literature from January 2015 to February 2023 and gathered 8 papers from 4 countries, with the data mostly from inpatient rehabilitation wards (PMID 37151039).

• Time in therapy was 23 to 121 minutes per day.
• Time on task was 8 to 44 minutes per day.
• The number of repetitions was 36 to 57 per session, and 15 to 282 per day.
• The lowest value for time on task came from the group with severe upper extremity impairment, at 8 minutes per day. The upper bound for this stratum was 41.5 minutes.
• Reporting on the content of usual care—meaning what therapy was actually done—was minimal across all studies.

The authors wrote that this dosage appears to have increased compared to past reports (an average of 21 minutes per day and 23 to 32 repetitions per session). They also stated that the dosage appeared lowest in patients with severe arm impairment, and concluded that multiple axes of dosage and therapy content should be consistently reported (PMID 37151039). The number of repetitions in inpatient rehabilitation seen in the primer "Neuroplasticity — A Brain That Builds New Roads Beside the Broken Ones" was also in the same range as these past reports.

Two things stand out here. The first is the gap between the time spent in the therapy room and the time spent doing tasks. Time in therapy is 23 to 121 minutes a day, and time on task is 8 to 44 minutes a day, so the two ranges are entirely offset. However, this is a comparison of two separate ranges across 8 papers, not values paired and reported within the same study. The second is that the most severely impaired people do the least. The time spent in the therapy room and the time actually spent doing tasks are not the same thing.

③ Does Changing Places Increase Activity?

Patients in rehabilitation wards did not spend more time using their bodies than patients in acute care wards. The researchers hypothesized that the rehabilitation ward would drive up physical activity more than the acute care ward, but the results differed from that hypothesis. Moderate-to-vigorous physical activity is an activity that requires relatively high bodily exertion, falling in the upper half when activity levels are divided from low to high. The time spent on this activity was not higher in the rehabilitation ward than in the acute care hospital (rehabilitation 24% vs. acute 23%, P=.74). This P-value means that a difference of that magnitude could easily arise by chance alone.

Some things did change. Rehabilitation ward patients spent less time lying in bed, and more time sitting outside the bed, outside the hospital room, and with therapists (all adjusted P<.001) (PMID 26851969). Adjustment means that other differences between the two groups were accounted for in the calculations, and these P-values mean the differences are unlikely to be due to chance. The observation method involved watching 190 patients (104 in rehabilitation, 86 in acute) across rehabilitation wards in four Swedish hospitals and the Karolinska acute stroke unit every 10 minutes for 1 minute from 8 a.m. to 5 p.m.

The authors concluded that an inpatient rehabilitation setting does not appear to boost physical activity more than an acute care hospital following a stroke (PMID 26851969). The rehabilitation side had more time out of bed and more time meeting with therapists, but not more time engaged in high-exertion physical activity.

Common Misconceptions

"Trials with increased doses have already given an answer."

There are not many trials that isolate whether doing more of the same kind of exercise therapy improves recovery. When searching for randomized trials where the experimental and control groups differed only in dose while everything else was the same, 7 studies (9 papers) were gathered (PMID 20942915). A randomized trial is a study that compares treatments by dividing participants randomly like drawing lots, and this review even conducted a meta-analysis pooling and calculating those results together.

• Only 3 out of the 7 were evaluated as having a low risk of bias across all design elements.
• The therapy received by the control group averaged 9 to 28 hours over up to 20 weeks, and the experimental group received 14 to 92 hours.
• The included studies were heterogeneous, differing in therapy type, outcome measures, and measurement time points. Therefore, effect sizes mostly come from single studies.

The results are as follows. Looking at the effect sizes of individual studies, the high-dose side tended to show better results at the end of treatment, which was less distinct at the follow-up time point. Among the indicators that could be pooled, comfortable walking speed (0.3 [0.1, 0.5]) and Motricity Arm at follow-up (10.7 [1.7, 19.8]) showed positive values. On the other hand, ARAT was 0.1 [-5.7, 6.0], spanning zero, and grip strength at the end of treatment was -10.1 [-19.1, -1.2], with the entire range being negative values. The abstract did not provide a separate explanation for this grip strength result. The Motricity Arm is a scale scoring arm muscle strength, and ARAT is a test measuring arm function through tasks of gripping and moving objects with the arm. The values inside the brackets are confidence intervals, meaning the likely range where the true value lies. If the entire range is positive, it points toward the high-dose side being better; if the range spans both negative and positive values, the possibility of no difference cannot be ruled out.

The authors concluded that current evidence "limitedly yet partially" supports the hypothesis that doing more of the same kind of exercise therapy improves recovery. They also noted that prospective dose-finding studies that predetermine doses and investigate them are needed (PMID 20942915).

The reason this conclusion is cautious lies in the shape of the evidence. The evidence base for "does increasing the dose make people better?" has a small number of studies and disparate comparison targets. We cannot put a trial with a 9-hour control group and one with a 28-hour control group in the same table and read "high dose" as meaning the same thing.

"Increasing therapy time increases the dose."

Therapy time and task performance time can move independently. The gap in ② above shakes this assumption. Between the upper limit of therapy time at 121 minutes and the upper limit of task performance time at 44 minutes, time spent not performing tasks intervenes. This means we must check separately whether what increases when therapy time is increased is the actual task performance time or other time in between. Point ③ points in the same direction as well. Simply changing the environment did not increase moderate-to-high-intensity activity. However, in the same data, the time spent sitting out of bed and the time spent with therapists were greater on the rehabilitation ward side.

To Go Deeper

• When reading a paper — If it says "dose," look first at which axis it is. Whether it is time, intensity, or schedule. And look at what the dose of the control group is. Without that value, "high dose" is just a relative expression.
• When looking at time — Separate the time spent in the treatment room from the time spent performing tasks. Typically, in actual measurements of treated upper limbs, the ranges for those two were 23 to 121 minutes and 8 to 44 minutes per day (the values were not reported as pairs within the same study).
• When looking from the treatment side — Counting is the fastest way. The fact that an outpatient actual measurement was 12 goal-directed upper limb movements per session was something no one knew before counting.
• Next piece — If we have looked at the principles (how) and the dose (how much) so far, what remains is an example where these two are bundled into a single treatment method. "Able But Unused — An Introduction to Learned Non-Use and CIMT" sits in that spot. The basic definition of dose and the limits of "more is better" are in the primer "Neuroplasticity — A Brain That Builds New Roads Beside the Broken Ones".

Source Papers

• Counting repetitions: an observational study of outpatient therapy for people with hemiparesis post-stroke. PMID 17419883 (2007)
• Dose, Content, and Context of Usual Care in Stroke Upper Limb Motor Interventions: A Systematic Review. PMID 37151039 (2023)
• Poststroke Physical Activity Levels No Higher in Rehabilitation than in the Acute Hospital. PMID 26851969 (2016)
• Early-phase dose articulation trials are underutilized for post-stroke motor recovery: A systematic scoping review. PMID 33429089 (2022)
• Dose Articulation in Preclinical and Clinical Stroke Recovery: Refining a Discovery Research Pipeline and Presenting a Scoping Review Protocol. PMID 31781018 (2019)
• The effects of increased dose of exercise-based therapies to enhance motor recovery after stroke: a systematic review and meta-analysis. PMID 20942915 (2010)