Background and objectives
Whilst rehabilitation professionals acknowledge the benefits of rehabilitation late after stroke, received wisdom in the literature still suggests that the best outcomes follow early rehabilitation and that most patients make little functional gain beyond 3 months.

The objectives of this study were to evaluate functional gain and cost-efficiency of specialist inpatient rehabilitation in patients with severe stroke admitted more than 3 months after stroke, in the context of real life clinical practice.

Methods

Design subjects and setting: 
The UK Rehabilitation Outcomes Collaborative (UKROC) database collates data on needs, inputs and outcomes for all episodes of specialist inpatient rehabilitation in England. This was a retrospective analysis of prospectively-collected routine clinical data from the UKROC database. 

Data were extracted for all registered episodes of patients with severe stroke (n=2174), admitted for in-patient rehabilitation between 2010-2017 in a Level 1 or 2 specialist rehabilitation service (76 centres in all), for whom the time since onset was over 90 days at the time of admission.

Methods:
Recorded on admission and discharge, the UK FIM+FAM measures change in motor and cognitive function. The Northwick Park Dependency and Care Needs Assessment (NPDS/NPCNA) calculates cost efficiency as the time for savings in ongoing care-costs in the community to offset the initial costs of rehabilitation. 

Data were analysed in three groups of dependency based on NPDS scores on admission: ‘High’ (NPDS>25); ‘Medium’ (NPDS=10-25): Low (NPDS<10). 
Summary statistics are reported using median (IQR) for ordinal data, or mean (95%CI) for interval data, with bootstrapped samples of 1000 to account for skewed data.

Results: 
The sample comprised 58% Males, mean age 55.2 years (95%CI 54.6, 55.8) and mean length of stay 112 (109, 115) days. Stroke aetiology was infarct (35%), haemorrhagic (36%), sub-arachnoid haemorrhage (19%). Left/ right hemisphere stokes were equally represented.

The mean time since onset of stroke at admission was 403 days (95%CI 365, 450) or median 140 days (IQR 108-232). 
The Median total UK FIM+FAM score on admission was 86 (IQR 57-125): median change score from admission to discharge 24 (IQR 9-42). 

The mean episode cost was £38,546 (95%CI £36,333, 40,894) with mean reduction in care costs/week of £334 (95%CI £250, £376) or £16,328 per annum. The cost of rehabilitation was offset in 28 months. 

When analysed in groups of dependency, the time to offset the initial costs of rehabilitation was shortest in the high-dependency group (27 months) compared with the medium (46 months) and low groups (101 months).


Conclusions:
This analysis of real-life data collected in routine clinical practice demonstrates that some patients continue to make functional gains from specialist rehabilitation even more than one year after stroke. Moreover, this intervention was highly cost-efficient, especially for the most dependent patients.

Benefits of Rehabilitation Late After Stroke: A Multi-Centre Cohort Analysis From the UK Rehabilitation Outcomes Collaborative (UKROC) Database