Results

Of 45 patients referred 42 were recruited and entered the study in just over a year (table 4). One patient declined further assessments after a single treatment period, another after recruitment but before treatment; thus 40 patients formed the basis of the analysis, of whom 39 underwent all assessments. Slightly more treatments were missed in hospital than at home (table 3). Total therapy delivered at home was slightly greater than in hospital but time for other tasks was slightly higher in hospital; achievement of treatment targets was similar at home and hospital. The assessor was aware of being unmasked to active treatment (but not venue) in 28 instances of 283 home visits.

The results in tables 5 and 6 refer to an analysis of all 40 patients treated and assessed: the results shown include the primary outcome measure and a selection of secondary outcome measures shown in table 1(other data analysed similarly (except video data) are available on request from the authors but did not change the essential outcome of the study). For the post-treatment Rivermead mobility index (primary outcome) ANCOVA showed that there was a highly significant difference (p<0.001) between the three treatments but no clear evidence of any difference between the three periods (p=0.216). There was significant variation between the patient's response on this measure but no significant effect of the pretreatment score. Pretreatment scores and differences in post-treatment scores after adjustment for patient, period, and baseline are shown in tables 5 and 6. Thus for the Rivermead scale the advantage of hospital over no treatment was a mean of 1.4 units (95% CI 0.6 to 2.1) and was similar for home compared with no treatment so the null hypothesis of no difference was clearly rejected (p<0.001): there was no significant difference between home and hospital treatment. Estimates of differences in treatment effect by period hinted at a reduced effect in period 3 but this was not significant.

Findings on the Rivermead index were essentially corroborated by other measures. Less significant findings on the Barthel index probably reflected a ceiling effect with scores clustered at the upper end of the scale. Balance score and time improved with active treatment as did measures of walking. The short test of concentration, orientation, and memory was not influenced by treatment but anxiety and depression scores improved after home and hospital treatment.

The assessor's global view of “mobility change” used a visual analogue scale where 50 represented no change from the previous assessment, 100 maximum improvement, and 0 maximum deterioration. Effect sizes for each treatment period are shown in table 5. Post-treatment values (visits 2, 4, 6) were all over 50 on average for the active treatment periods but not for controls whereas the follow up scores (visits 3, 5, 7) showed a falling away of benefit in the two active treatment groups to levels no different from the control group. Patients perceived better mobility and a reduced tendency to fall after both active treatments and this was confirmed by their carers (table6). Patients slightly preferred home treatment; carers assessed benefit as slightly greater at home but had a strong preference for home treatment. Physiotherapists judged that home treatment was more often beneficial than hospital treatment, but that in about half the patients venue made no difference to benefit, and that about 10% showed no benefit in either (table 2).

Costs of a therapy session were calculated from the employment costs of the physiotherapists, duration of sessions including extra time (table3), travel time, and mileage costs. On average a home therapy session cost £25 whereas a hospital session cost £11: the second excludes patient travel costs, which were estimated as £7 per visit, and time.

Discussion

This randomised crossover study of the effects of physiotherapy in multiple sclerosis has, unlike our previous inpatient study, clearly shown that mobility can be improved to a clinically relevant extent as assessed at home by an independent observer. There was no substantial difference between the benefits of therapy delivered in the patient's home or as a hospital outpatient although carers preferred home treatment. Improvement amounted to an average of 1.4–1.5 units on the primary outcome criterion, the Rivermead mobility index; more than the minimum clinically relevant improvement used for the study power calculation. The Rivermead index has not been widely used in multiple sclerosis but has been found valid and sensitive to changes and functional gains in a rehabilitation setting over time, as well as being rapid and simple to use in the patient's environment.21 22 Improvements found with the Rivermead index are supported by changes in other disability scales, measures of impairment, and subjective views of patients, carers, and assessor. There is an impression that net benefit results, in part, from prevention of deterioration in “no treatment” periods (tables 5 and6). As the physiotherapy approaches differed to some extent at home and in hospital we cannot state whether venue or approach, both, or neither are relevant factors in the benefit shown. A study constraining the therapy approaches to be identical in these different environments would be difficult to construct although different approaches in the same environment can be tested (see below).7

The present study was not designed to test a general multidisciplinary rehabilitation package but to investigate the effect of physiotherapy on ambulation—although physiotherapists offered general advice and made referrals to various agencies when appropriate (not differing significantly between home and hospital venues), their therapy role was specifically directed to predetermined targets based on their own initial assessments of mobility. A study of “physical rehabilitation” for 3 weeks as an inpatient was shown to have a benefit on the motor domain of the functional independence measure and the mental (but not physical) component of health related quality of life profile which persisted (though fading) 6 weeks later.8 A general inpatient rehabilitation package was shown in a randomised controlled trial23 to benefit patients and a follow up study of such a package, on more disabled patients with multiple sclerosis (preassessed as “suitable” for inpatient rehabilitation) assessed openly without a control group in later assessments, supports the view that benefits may carry over into the community although they decline with time24: however, the component of improvement related to walking is not clear. The findings contrast with the previously reported inpatient physiotherapy trial5 in which assessments undertaken in the patients' home failed to show a statistically significant benefit. We suspect that the greater duration of therapy in the present study and the reduced level of necessary activity while a hospital inpatient in the previous study may have been important factors in this difference. Therapy (outpatient or home) may also have been more focused on activities relevant to the patient and the home environment in the present study compared with the inpatient study.

We cannot state what “dose” of treatment was necessary: optimum number, duration and interval of treatment sessions all need further study but affect cost and practicality. Furthermore, the specific content of physiotherapy might be important although a pilot study showed no difference in benefit between task oriented (disability focused) and facilitation (impairment based) approaches.7The tendency of benefit from active treatment to wane during follow up with little carry over from one treatment period to the next suggests a short lived effect, unlikely to result in long term benefit unless repeated, continuous, or supported by other interventions or change in motor behaviour. If major benefit from general education and advice given by a physiotherapist carried through from one period to the next a positive period effect and less falling away of perceived benefit by the assessing therapist across follow up periods might be anticipated (before next treatment session) but neither were seen. These findings also support our supposition that a crossover study can be utilised for evaluation of short term therapy interventions in multiple sclerosis.

Benefits found might result from specific factors (for example, change in posture or tone) or secondary factors such as altered mood, effects (on motivation and confidence) of increased exercise,6 or discussions about mobility. Upper limb function improved as well as mobility. Several therapy techniques () were directed at improved trunk control and head, neck, and trunk posture so potentially influencing arm function. We tested the effect of pretreatment mood on the Rivermead mobility index by examining treatment effect against dichotomised (<8, ⩾8) mood scores and examining the rank correlation (Spearman) of change in Rivermead score against pretreatment anxiety or depression subscore. Hospital treatment data suggested a tendency (non-significant) for greater benefit in those who were initially depressed but this was less evident for home treatment. Benefit from home treatment was, if anything, less marked in those who were anxious. Further studies could target issues of specific versus non-specific treatment, of mood enhancement on mobility, and the neuropsychological characteristics of patients expected to benefit.

Home based therapy was more costly to the health service in our simple analysis due to the substantial time spent travelling by the physiotherapist. This element could be reduced by utilising more locally based therapists but they may have less specific expertise in multiple sclerosis. It should also be emphasised that, in this study of efficacy, outpatient physiotherapy was optimally arranged using private transport or taxi and avoiding hospital transport systems which are a potential source of both missed appointments or late arrival for therapy and hence increased cost.

We conclude that, compared with no treatment, an 8 week course of physiotherapy (twice a week) results in significant improvements in mobility, subjective wellbeing, and mood in patients with chronic multiple sclerosis whether it is provided at home or in a hospital outpatient physiotherapy department (the second being associated with less direct cost to the NHS). These benefits, although clinically significant and relatively low in cost, are shortlived, suggesting that ongoing physiotherapy input might be necessary for sustained benefit whether this is defined as improvement in mobility or prevention of deterioration.