Primary and secondary outcomes
The primary outcome was not significantly different between the randomised groups. The CGI-I (patient-rated perception of improvement of motor symptoms) was significantly different in favour of specialist physiotherapy. Based on this measure, participants randomised to specialist physiotherapy were 4.7 times more likely to rate their motor symptoms as improved at 6 months compared to TAU, and 2.3 times more likely to rate their motor symptoms as improved at 12 months. This finding is of clinical importance, as the patient-rated CGI-I was the main recommended outcome measure for FND research in a recent systematic review and expert consensus recommendation of outcome measurement in FND.31
In contrast to the feasibility study, the current study found that both randomised groups improved in the primary outcome (SF-36 Physical Functioning) from baseline to 12 months. In the feasibility study, only the intervention group improved on this measure.20 The improved performance of physical function in the TAU group may point to a national improvement in physiotherapists’ knowledge and skills over the past few years and, in part, account for the lack of significant difference between groups at 12 months.
In post hoc analysis, we found that 38% of participants in the TAU group made a 10-point improvement on the primary outcome (SF-36 Physical Functioning), which matched the proportion of participants who rated their symptoms as improved on the CGI-I, also 38%. For specialist physiotherapy, there was a discrepancy between these values, 49% made a 10-point improvement on the primary outcome, yet 59% rated their symptoms as improved. We hypothesised that the greater perception of improvement in specialist physiotherapy may be related to the parallel findings in this group, of higher confidence that their diagnosis was correct, perceived better understanding of their symptoms, perceived better control over their symptoms, and better scores of mental health (less anxiety at 6 months, higher SF-36 Mental Health at 12 months). These parallel findings may be associated with developing self-efficacy.
In terms of the serious adverse events (SAEs), we found that physiotherapy was safe, with no SAEs related to treatment. However, an important finding was a high rate of hospitalisation among trial participants in both arms over the 12-month follow-up period. Seventeen per cent of participants from both treatment arms who had timely treatment had at least one hospital admission. There were 77 events leading to hospitalisation. The most common reason for hospital admission related to gastroenterological symptoms (21% of events), followed by neurological symptoms (19%), respiratory symptoms/illness (12%), cardiological (9%), urological (8%), oncological (7%), gynaecological (6%), psychiatric (6%), orthopaedic (4%) and other reasons (8%). Hospitalisations were reported as SAEs in the trial outcomes, and all were judged to be unrelated to physiotherapy treatment. The high rate of hospitalisation relates to the health complexity of people with FMD, and the highlights the need for multidisciplinary care.
Health economic analysis
At 6 months, specialist physiotherapy had significantly higher EQ-5D-5L utility scores compared to TAU. The difference was not significant at 12 months (adjusted difference, 0.040, 95% CI −0.014 to 0.095). This corresponded to a difference in QALYs at 12 months of 0.030 (95% CI −0.007 to 0.067). After accounting for the cost of training and delivering the respective interventions, the mean incremental cost per QALY was £4133. The research was not specifically powered for the health economic analysis, and greater subject numbers would be needed to be assured of the significance of this finding. Sensitivity analysis revealed that those recruited and treated in the trial extension after COVID-19 lockdowns saw the greatest savings, influencing the cost-effectiveness of specialist physiotherapy. The full health economic analysis, including data on employment and costs from a societal perspective, is published separately.28
The impact of the COVID-19 pandemic
As described above, recruitment and physiotherapy treatment were suspended during the peak of the COVID-19 pandemic. Return to clinical and research activity occurred after 6–9 months and varied by trial site, depending on local policies. The immediate impact of the pandemic on the trial was the interruption to the delivery of the physiotherapy treatment. We categorised the trial participants into four groups, according to how their trial journey interacted with the pandemic as follows:
- Group A (n = 25): received treatment and completed their 12-month follow-up before 23 March 2020 (the date when national lockdown restrictions were imposed in the UK).
- Group B (n = 134): received treatment before 23 March 2020 but were followed up after 23 March 2020.
- Group C (n = 89): recruited prior to but did not receive treatment before 23 March 2020, followed up as per protocol.
- Group D (n = 88): participants recruited after July 2021 as part of the trial extension.
In prespecified analysis plans,24 we excluded participants in group C from the primary analysis. In the specialist physiotherapy group, 27 participants met the criteria for group C; among them, only 8 (26.6%) received physiotherapy within the trial follow-up period. This treatment was delayed by a median of 253 days (compared to a median treatment start delay of 36 days for COVID groups A, B and D). In the TAU group, 62 participants met the criteria for group C, 22 (out of 60 with available data, 36.7%) received physiotherapy within the trial follow-up period. This was delayed by a median of 174 days (compared to a median treatment start delay of 97 days for COVID groups A, B and D). These findings support the decision to exclude group C from the primary analysis.
Sensitivity analyses found that excluding group C from the primary analysis had no impact on the significance of the result for the primary outcome. Adding a supplementary fixed effect of group indicator and its interaction with the assigned treatment to the primary analysis model did not find that the intervention differed by COVID group indicator (COVID group C vs. groups A, B and D) (p = 0.825).
How the randomised treatments differed
We explored the composition of physiotherapy received in both randomised groups by participant report in a telephone interview conducted at approximately 6 months after randomisation. The results of this interview are reported in the supplementary appendix of the main trial outcome paper.27
Treatment as usual comprised fewer sessions (median 4, IQR 2–7, compared to specialist physiotherapy median number of sessions 9, IQR 8–9), which were scheduled less frequently (75% had sessions that were scheduled fortnightly or less frequently, while specialist physiotherapy scheduled more than one session per week).
Specialist physiotherapy was centred around a printed, interactive workbook, whereas there was little evidence of a similar device used in TAU. One of the primary objectives of specialist physiotherapy was to help participants understand their movement problem and the role of attention in driving symptoms. Ninety-nine per cent of participants in specialist physiotherapy agreed that their physiotherapist tried to help them understand their movement problem, and 98% agreed that their physiotherapist talked to them about how attending to movement can make movement worse. In participants allocated to TAU, these figures were 66% and 61%, respectively.
The specialist physiotherapy intervention protocol included education and development of management plans for persistent pain and fatigue when present. Among participants who experienced persistent pain (82%), 91% in specialist physiotherapy agreed their physiotherapist tried to help them with this, compared to 48% in TAU. Among participants who experienced fatigue (93%), 98% in specialist physiotherapy agreed their physiotherapist tried to help with this, compared to 59% of TAU.
The specialist physiotherapy intervention protocol included education about the potential negative side effects of medications, such as opiates and muscle relaxants. Fifty-six per cent of participants reported taking medication for their movement problem. Among this group, 84% of participants in specialist physiotherapy reported discussing medication side effects with their physiotherapist, compared to 33% in TAU.
The specialist physiotherapy intervention included information about the symptoms of memory and concentration problems. Of those with memory and concentration problems (88% specialist physiotherapy, 71% TAU), 91% in specialist physiotherapy reported discussing this with their physiotherapist, compared to 40% of TAU.
Finally, 98% in specialist physiotherapy reported that they made a self-management plan to help continue to improve following treatment, compared to 66% of TAU.
We used the data from the telephone interview to look for evidence of contamination between randomised groups, that is, if participants randomised to TAU received elements of specialist physiotherapy. We found one case of clear contamination, where a participant allocated to TAU received the specialist physiotherapy workbook and the core treatment components. This was recorded as a deviation from the protocol.
There are some clear differences between the randomised treatments. Specialist physiotherapy included more treatment sessions that were scheduled with higher frequency. The workbook is unique to the trial intervention, but 69% of TAU received some form of printed material, most commonly a list of exercises. The components of specialist physiotherapy are based on published consensus recommendations, and it appears that many of these were included in TAU (e.g. 61% learnt about the role of attention in symptoms, 49% received pain management, 59% received fatigue management and 66% developed a self-management plan). Overall, one might conclude that TAU for most participants was delivered to a high standard. This supports our suggestion that the quality of TAU improved from the feasibility study to the current trial.
Recommendations for implementation: the physiotherapist’s perspective
Following completion of the intervention phase of the study, the physiotherapists delivering specialist physiotherapy attended a meeting to debrief and consider factors relating to implementation (paper under review). Nominal Group Technique32 was used to identify and rank in order of importance, facilitators and potential barriers to implementing the Physio4FMD treatment protocol. Twenty-three facilitators and 26 potential barriers were identified. The top two facilitators were the intervention workbook and physiotherapist training to deliver specialist treatment for FMD. These factors reflect the novelty of the intervention. The top potential barrier was inappropriate referral/selection of patients. In line with the eligibility criteria for the study, we acknowledge that the trial intervention is not suitable for all people with FMD, and in previous work found that 32% of patients with functional motor symptoms met the eligibility criteria for specialist physiotherapy.20 The availability of alternative treatment pathways for those not meeting eligibility for specialist physiotherapy was considered to be important. Alternative treatments include specialist FND multidisciplinary programmes and psychological therapy. We reported a list of recommendations for implementation of specialist physiotherapy based on content analysis of the final facilitators and potential barriers.
In light of the perceived importance of training, we explored how the training could be adapted to aid implementation, using an online questionnaire completed by study physiotherapists. All reported that the training programme introduced them to concepts or ways of treating FMD that they were not previously using, and that the training had changed their ongoing practice. The training was delivered over 5 full days, which has implications for cost. In the training questionnaire, we explored the opinions of the physiotherapists about whether the training could be reduced by condensing the content over fewer days. Most agreed that training could be reduced from 5 to 4 days, and that further cuts could be made, but this would impact the effectiveness of training. The findings indicate the specialist physiotherapy for FMD is a complex intervention that requires experience, training and some support from a multidisciplinary team.
Contributions to existing knowledge
The Physio4FMD is the first published, powered, multicentre RCT of a physical-based intervention for FMD. There was no difference in SF-36 Physical Functioning scores between specialist physiotherapy and TAU; both groups improved over 12 months for this outcome measure. SF-36 Physical Functioning is a self-report measure that takes into account ability to complete vigorous activities, moderate activities, stair climbing, walking various distances, carrying groceries, and washing and dressing. In secondary outcomes, participants receiving specialist physiotherapy were more likely to report their motor symptoms as improved at 6 months (OR 4.7, 95% CI 2.6 to 8.6) and 12 months (OR 2.3, 95% CI 1.4 to 3.9). Additionally, specialist physiotherapy was associated greater scores for mental health, and other outcomes relating to understanding and self-efficacy. Improvements occurred despite high levels of complexity at baseline, as indicated by long symptom durations (mean 4.8 years) and high rates of coexisting health problems, such as anxiety, pain and fatigue.
The Physio4FMD trial demonstrated that rehabilitation for FMD can occur outside specialist centres. We set up 11 trial sites, of which most did not previously have a dedicated service for FMD/FND. The neurologists who diagnosed and screened potential participants and the physiotherapists delivering treatment completed training for their respective roles, and all had some previous experience with FMD.
We found a high cost of illness, with hospital admissions and hospital appointments accounting for the greatest proportion of costs. Specialist physiotherapy showed a high probability of cost-effectiveness, resulting in more QALYs and lower costs than TAU, although the difference did not reach statistical significance. The probability of cost-effectiveness at the £20,000 cost per QALY threshold was 86%.
Strengths and weakness of the Physio4FMD trial: reflections on the project
Strengths of this research include the protocolised study intervention, which proved capable of being delivered with fidelity. Specialist physiotherapy was highly rated by the trial participants in the satisfaction survey, which was backed up by high levels of attendance and low attrition. The trial physiotherapists also rated the intervention highly during the debrief meeting.
Strengths in the research design include the powered sample size, 12-month follow-up and blinding of those who collected and analysed the trial outcomes. A strength of the conduct of the trial was the high retention and low levels of missing data for the primary outcome.
There are important weaknesses to acknowledge. With the pragmatic design, waiting times to start treatment were different between randomised groups, with specialist physiotherapy starting at a median of 36 (IQR 25–58) days after randomisation and TAU 97 (IQR 60–176) days. Added to this, the duration of treatment was longer for TAU, with the result that specialist physiotherapy had completed treatment a median of 310 days (IQR 282–323) prior to completing the primary outcome and TAU 179 (IQR 123–238) days. However, the treatment effect did not appear to reduce over time. To address this issue in future research, the timing and nature of the comparator intervention could be more strictly defined. However, this would come with additional costs and the loss of the pragmatic ‘real-world’ effects. Alternatively, if the trial were to be repeated, the primary outcome could be defined as 12 months post-intervention commencement.
We found that both randomised treatments improved on scores of physical function from baseline to 12 months. Without a ‘no treatment’ or ‘waiting list control’ arm in the trial design, it is unclear how much these changes relate to the efficacy of both interventions or natural recovery.
The study relied on subjective, patient-reported outcomes. There are currently few FND-specific, validated, objective outcome measures.31 Developing and choosing an appropriate assessment tool for FMD is complicated by factors such as the potential for variability of FMD symptoms over short periods of time. Additionally, patients typically experience multiple symptoms, and each may contribute to the illness burden, so measuring only one domain may miss clinically important changes.33 A recent consensus statement for a core outcome measurement set for FND highlighted the value of patient-reported outcomes over objective or single time-point measurements and recommended the patient-reported CGI-I as the most useful measure of symptom severity.31 With this measure, we found a statistically significant difference in favour of specialist physiotherapy.
We were unable to record the number of patients screened for eligibility. This limits our ability to comment on generalisability. This was related to the method of recruitment from multiple sources and sites, and limitation of resources. We did record this data in the feasibility study from a single site and found that 32% met selection criteria. While this may be viewed as a limitation in generalisability, we argue that given the heterogeneity of people with FMD, different treatment approaches are necessary, and a one-size-fits-all approach to treatment is a dated view. Additionally, there are alternative existing treatments for those whom specialist physiotherapy is deemed unsuitable, including multidisciplinary rehabilitation, and psychotherapy.34–36
Another limitation was that determining participant eligibility required interpretation by the recruiting neurologist rather than being operationalised, for example, with a cut-off score from a validated assessment tool. We felt that there were no suitable assessment tools for this purpose, and this may be a focus for future research.
The COVID-19 pandemic and associated ‘stay at home’ orders (lockdown) occurred during the trial data collection. We took a number of steps to minimise the impact of these events on the trial results, which included excluding participants whose treatment was interrupted by the pandemic from the primary analysis. We explored the impact of this decision with a series of sensitivity analyses. Although the sensitivity analyses did not show differences in the primary outcome depending on whether participants were recruited before or after the pandemic, it is impossible to rule out that confounding factors may have influenced the primary and secondary outcomes. In the economic evaluation, we found a higher probability that specialist physiotherapy was cost-effective compared to TAU in participants recruited after the pandemic compared to those recruited before. We hypothesised that this may be related to improved skill within the specialist physiotherapy group with practice; however, other confounding factors related to the pandemic may have influenced this finding.
Take-home messages
People with FMD often have complex health problems that can be associated with persistent pain, fatigue, anxiety and other coexisting diagnoses. This complexity is associated with high rates of healthcare utilisation, including hospitalisation, and a high cost of illness. It follows that access to multidisciplinary care is needed.
Despite this complexity, physiotherapy within neurology/neurorehabilitation services is a valuable treatment for people with FMD. Specialist FMD physiotherapy appeared to be more cost-effective than TAU (non-specialist physiotherapy). The additional value of specialist physiotherapy over TAU was greater participant perception of improvement, improved mental health, greater confidence in the correctness of the diagnosis, greater understanding of symptoms and greater perception of control over symptoms.