Published by Unseen Progress, an independent publisher of caregiver research. Last reviewed 2026-05-10. Part of the stroke caregiver research overview.
Short answer. Some stroke recovery plateaus are real ceilings. Most are not. The research literature distinguishes true plateau (no further improvement despite adequate dose, intensity, and modality variation) from apparent plateau (no further improvement on the current protocol, which often responds to a change). The EXCITE trial (Wolf et al., 2006), the modified CIMT work (Page et al., 2012), and Krakauer's reviews of neurorehabilitation (Krakauer et al., 2012) all show meaningful gains in chronic-phase survivors when intensity or modality changes — gains that, by definition, were not the ceiling.
Kwakkel's longitudinal recovery curves (Kwakkel et al., 2006) are the source of the popular "recovery flattens at six months" narrative. The curves do flatten — but flattening at the population level is not the same thing as flatlining at the individual level, and is not the same thing as being unable to improve further. The curves describe what tends to happen on standard rehab protocols at standard doses, not what is biologically possible.
The Constraint-Induced Movement Therapy literature is the cleanest counter-example. The EXCITE randomised trial (Wolf et al., 2006) enrolled stroke survivors 3–9 months post-stroke — the phase usually described as plateau territory — and demonstrated meaningful, durable upper-limb gains from a two-week intensive CIMT block. The implication is direct: if survivors at month 6 can gain measurably from intensity and modality change, then "plateau at month 6" was not a ceiling for them; it was a current-protocol artefact.
Page and colleagues (Page et al., 2012) showed similar results from modified CIMT — lower-intensity, more home-feasible versions — extending the finding to families and survivors who could not commit to the EXCITE protocol's massed-practice schedule.
The 2016 AHA/ASA guidelines (Winstein et al., 2016) explicitly recommend that plateaus be evaluated as opportunities to change therapy intensity, modality, or focus, not as discharge cues.
Insurance ends. Outpatient sessions taper. The home exercise program drifts. The survivor stops improving — because the dose stopped, not because the brain stopped. This is the most common type, and is misnamed as a clinical plateau. The fix is to convert what was being done in formal therapy into a sustained home protocol.
The survivor has reached the limit of the current protocol. Continued practice of the same exercises at the same intensity produces no further measurable gain. This often responds to a structured change — CIMT for upper-limb non-use, intensive aphasia therapy block, treadmill training, dual-task practice. The research suggests changing one variable and giving it 60–90 days before concluding anything.
The metrics being used are too coarse to detect the gains that are happening. Gait speed measured to the nearest 0.1 m/s may show no change while a stride-by-stride analysis shows gradual improvement. Words per dinner may be flat while sentence complexity is increasing. The plateau is in the measurement, not in the survivor.
The survivor has reached an actual biological limit on the current task — typically because the lesion damaged a structure essential to the function in question, and compensation has been maximised. True ceilings exist. They are less common than the other three, and the research suggests they should be the last explanation considered, not the first.
| Sign | True ceiling | Protocol plateau | Service-system | Measurement |
|---|---|---|---|---|
| Has any modality been changed? | Yes, at least 2 | No | N/A | N/A |
| Is current dose adequate? | Yes | Possibly | No | N/A |
| Are markers granular enough? | Yes | Yes | Yes | No |
| Has 60–90 days passed on current change? | Yes | No | N/A | N/A |
A "true ceiling" conclusion is only defensible after the other three have been ruled out.
1. Ask which kind of plateau. Specifically: has the modality been varied? Has dose been measured? 2. Audit the current dose. Many home protocols drift to 10–15 minutes a day when the literature supports much higher repetition counts for motor tasks (Lang et al., 2009 documented average reps per outpatient session at well below the dose research suggests is needed). 3. Pick one structured change. CIMT or modified CIMT for an affected upper limb. Intensive aphasia therapy block. Treadmill training for gait. A salience change — practising the actual task the survivor wants back, not a generic exercise. 4. Define a marker. Wait 60–90 days. Re-evaluate.
Most plateaus that families had accepted as permanent move at this point.
---
Unseen Progress publishes long-form caregiver research and builds research-backed daily trackers for the families covered. See the full stroke caregiver research overview for the complete framework.