Key findings

The REFORM trial is the largest study of a podiatric programme that includes a foot and ankle exercise programme to reduce the risk of falling. A total of 1010 participants were randomised. Our sample size allowed for a 10% loss to follow-up. The actual overall loss to follow-up observed at 12 months was 12.4% [in total, a 12-month questionnaire was returned for 885/1010 (87.6%) randomised participants]. Although this loss was higher than expected, we still had sufficient numbers relative to the target sample size of 890, as the trial over-recruited to 1010 participants. The primary clinical outcome for the trial was the incidence rate of falls reported on monthly falls calendars in the 12 months following randomisation. In practice, it is difficult to calculate the required sample size for a regression model, such as a Poisson or negative binomial regression model, to analyse count data. This requires an estimate of the measure of overdispersion and a justifiable treatment effect to detect. There were a limited number of data on which to base these parameters and so the decision was made to power the trial to detect a difference in the percentage of participants who reported at least one fall over the 12-month follow-up.

In total, 992 (98.2%) trial participants returned at least one falls calendar following randomisation, with similar proportions across the two groups [484 (98.2%) participants in the intervention group and 508 (98.3%) participants in the usual-care group]. We found a reduction in the rate of falls per person-year (IRR 0.88, 95% CI 0.73 to 1.05) and in the proportion of participants who had one or more falls over the 12 months from randomisation (OR 0.78, 95% CI 0.60 to 1.00). The difference was not statistically significant in our prespecified primary outcome of rate of falls (p = 0.16); however, the difference in the proportion of participants who had at least one fall (54.9% and 49.7% for usual care and intervention groups, respectively), a key secondary outcome, was of borderline statistical significance (p = 0.05). In our sample size calculation, we assumed that 50% of the usual-care group would fall during the 12-month follow-up, and we powered to detect a fall to 40% in the intervention group. In fact, 55% of the usual-care group experienced a fall. With the numbers recruited, we had 80% power to detect a fall to 46%, and approximately 36% power to detect the difference of 5% observed. Although a 5% decrease in the number of participants falling is of borderline statistical significance, it is difficult to say whether or not it is clinically meaningful. The estimated number of participants to whom we would need to offer the intervention to prevent one person from experiencing a fall is 20, which is relatively low.

A small, and similar, proportion of participants reported at baseline that they had been referred to a falls clinic or service in the previous 12 months in the two groups. At the end of the 12-month follow-up, we asked this question again: 30 out of 416 (7.2%) intervention participants and 22 out of 452 (4.9%) usual care participants said that they had been referred to a falls clinic or service in the previous 12 months. It is possible that participants in the intervention group interpreted this question as referring to their trial appointments at the podiatry clinic. If participants in the usual-care group received some form of intervention shortly before or during the trial follow-up, this could potentially have diluted the treatment effect. However, with only small numbers reporting this, we do not believe that this could have significantly influenced the results, and in any case we ran this as a pragmatic trial and so the results will reflect usual practice.

Time to first fall was reduced in the intervention group but this was not statistically significantly (hazard ratio 0.88, 95% CI 0.74 to 1.04; p = 0.14). No statistically significant differences between the two groups were observed at 6 or 12 months in the fear of falling question, the Short Falls Efficacy Scale-International, the FAI, the GDS or the CD-RISC2. The intervention group did, however, report higher levels of foot pain at 12 months on a 10-cm visual analogue scale from 0 (no pain) to 10 (worst pain possible). The mean pain in the intervention group was 3.1, compared with 2.6 in the usual-care group (adjusted mean difference 0.43, 95% CI 0.06, 0.80; p = 0.02); however, although statistically significant, a difference of 4.3 mm may not be clinically meaningful. It is unclear why participants in the intervention group reported higher pain scores. Evidence from the qualitative study suggests that, in some cases, increased foot pain could have been a result of insoles reducing the space in footwear. In other cases it may be that intervention participants were simply more aware and more critical of (problems with) their feet, or they were using their feet more while performing the exercises. Alternatively, this could be a chance finding.

Cost-effectiveness

The results of the economic evaluation conducted alongside the REFORM trial suggest that the multifaceted intervention could be a cost-effective option for falls prevention in terms of QALYs gained calculated using the EQ-5D. The ICER for the ITT approach in the imputed data set ranged between £19,494 and £20,593 per additional QALY. The probability of being cost-effective for the base-case analysis is > 60%. The results are robust to the sensitivity analyses testing the assumptions regarding resource use, perspective of analysis and level of imputation regarding missing data on HRQoL. With the one exception of when the missing data mechanism is tested, the complete-case analysis suggests that the multifaceted podiatry intervention is expected to be more costly and slightly less beneficial than usual care. However, the complete case in REFORM is not without limitations. In addition to the much reduced sample size of the original data (28.3%), missing data patterns showed that incomplete data followed a non-monotonic pattern, which suggests that the complete-case assessment would be inefficient, as it would discard observed data from individuals who have some missing outcomes. A logistic regression analysis showed that advancing age and lower EQ-5D at baseline are associated with missing QALY data. This suggests that the data are unlikely to be MCAR; consequently, the results from the multiple imputed data set are likely to be more accurate and more reliable than complete care results.

The main limitation of this economic evaluation, conducted alongside the REFORM trial, is that it does not account for any differences in costs and QALYs that may be expected over the longer term (> 12 months post randomisation). The HRQoL data showed that the reduction in the number of participants in perfect health in the intervention group is lower than that in the usual-care group (17.7%); the increase in the number of participants having problems is also lower in the intervention group. The effectiveness analysis also indicated a reduction, albeit a non-statistically significant one, in fall rate in the intervention group relative to usual care. Cost-effectiveness did not noticeably differ when we projected HRQoL beyond the trial duration (up to 5 years). However, we consider this exploratory projection likely to be conservative, and it would be important to explore the long-term impact of reducing the number of falls, as this might also lead to a reduction in the number of fractures, which in turn will make it more likely that the intervention yields long-term cost savings in the NHS.

Qualitative findings

The qualitative study explored issues of acceptability and implementation from the perspectives of both patients and podiatrists. It found that most podiatrists could implement some elements of the programme, such as the footwear advice and the provision of the orthotic, as part of their normal clinic practice, with some podiatrists continuing to offer the intervention outside the trial. Some concerns were raised about the ability of podiatrists to effectively deliver the exercise component within the time constraints of a routine clinical appointment. Although the podiatrists generally felt confident in doing so, time (and equipment) would have to be allocated for this purpose, alongside any necessary follow-up appointments. Given the way in which most falls prevention services are set up, some podiatrists felt that the intervention may be well suited to a multidisciplinary falls service, which would include podiatry alongside physiotherapy input for the exercise intervention, particularly in a group setting.

The trial participants were largely content with the intervention, and adherence was generally good. Some trial participants, especially those with comorbidities, found some of the exercises challenging; however, generally, both podiatrists and participants were able to adapt the exercises to suit individual circumstances. Some participants noticed a benefit of the exercise training after several weeks and felt more confident as a result. The trial participants found pragmatic ways to incorporate wearing an orthotic, when it was comfortable to do so. Some participants, however, were not able to adhere to the footwear advice/orthotic, as they were unable to achieve a good fit owing to existing foot problems or they were resistant to wearing the footwear options available.

Comparison with other studies

Our results to some extent support the earlier findings by Spink et al.²³ In this Australian trial, among 305 community-dwelling men and women (mean age 74 years) who were suffering from disabling foot pain and who had an elevated risk of falling, a reduction in the incidence rate of falls was observed (IRR 0.64, 95% CI 0.45 to 0.91). The Australian population was similar to ours in that they were all receiving routine podiatry care and were recruited from podiatry patient lists. However, participants had to be suffering from disabling foot pain, which was not the case for our population; patients may have had foot pathology but they did not necessarily have significant foot pain. Our population had a higher risk of falling; the usual-care group sustained an average of 1.5 falls per year, compared with 1.06 for the Australian patient group. Similarly, 55% of our usual care participants sustained one or more falls, compared with 49% in the Spink et al.²³ study.

The key elements of the interventions were similar, comprising foot and ankle exercises, an orthosis and an assessment for poor footwear. Both studies were carried out among patients who were receiving ‘standard’ podiatry. However, there were some differences. We did not use exactly the same orthosis as that used in the Australian study, and the foot and ankle exercises were modified partly in light of lessons learned from the Australian study. In our study, when possible, new footwear was provided to participants in the intervention group whose own current footwear was inappropriate. In the Spink et al.²³ trial, participants were provided with a subsidy for new footwear in the form of a voucher. Furthermore, the participants in our study did not need to have ‘disabling foot pain’, as was the case in the Australian study. Forest plots to compare the results of the two studies graphically are presented in Figures 17 and 18. An analysis of the REFORM data was repeated including only treatment groups in the models for comparability with the Spink et al.²³ trial. Individual patient data were provided by the authors of the Spink et al. trial, and so, whereas results for the proportion of fallers are presented as a risk ratio in the publication, here we were able to present these as an OR.

FIGURE 17

Impact of multifaceted podiatry intervention on the incidence rate of falls over 12 months in older adults. Reproduced from Spink et al. This is an open-access article distributed under the terms of the Creative Commons Attribution Non-commercial License (more...)

FIGURE 18

Impact of multifaceted podiatry intervention on proportion of participants who fall at least once over 12 months. Reproduced from Spink et al. This is an open-access article distributed under the terms of the Creative Commons Attribution Non-commercial (more...)

Strengths and limitations of the study

This was a large pragmatic trial, and we used a novel design, namely a cohort randomised trial, to evaluate this podiatric intervention. The design had several strengths: the use of a run-in period with outcome data collection could have reduced the incidence of post-randomisation attrition; those in the usual-care group were unaware of the exact time at which they were randomised, and, in theory, this should have limited resentful demoralisation. The design also allowed us to recruit participants who initially were ineligible because they had not fallen but later became eligible because they had fallen while part of the observational cohort. The initial engagement of participants with the intervention was high; 84% of intervention participants attended a trial appointment. Compliance with the exercise component was reasonable (at 12 months, 29% of intervention participants reported performing the exercises at least three times per week and 75% reported doing them at least once per week). However, in the qualitative interviews, some podiatrists stated that they felt that this could have been higher if they had had additional contact with the participants. Another limitation of the study is that the sample size was based on detecting a difference not in the primary outcome of incidence rate of falls but in the proportion of participants reporting at least one fall in 12 months. This was because of the difficulty in calculating a sample size for a count outcome, as discussed in Key findings. It is not possible, therefore, to confirm that the trial was sufficiently powered for the primary outcome. In addition, participants were recruited from podiatry clinics; therefore, the estimated impact of the intervention among people who do not regularly see a NHS podiatrist or who receive care from a private podiatrist may be different. Using a run-in period may also have biased the sample towards volunteers with a heightened interest and commitment to the intervention. Furthermore, the intervention is a ‘complex’ one, and our design does not allow us to estimate the different contributions of changes in footwear, the addition of an orthotic insole or the undertaking of foot and ankle exercises to the observed effect. It may well be that one or more of the interventions included in the ‘package of care’ is ineffective. There is also the possibility that some participants in the usual-care group had enrolled in another falls prevention programme as part of their NHS care, which could have diluted the treatment effect. This dilution effect is likely to be minimal, however, given that only a small proportion of participants in the usual-care group reported being referred to a falls clinic or service during the trial.

Generalisability of the results

The REFORM intervention was a pragmatic RCT across nine sites in the UK and one site in Ireland. All participants were recruited from podiatry clinic lists. This was to ensure that we could identify an additional effect of the intervention not confounded by routine podiatric care. Consequently, the trial cannot answer the question of whether or not the intervention is effective among patients who do not have routine podiatry care. However, approximately one in six people aged > 65 years receives NHS podiatry care and, therefore, our results are applicable to a significant proportion of the older population.

The trial results may also not be generalisable to patients who would not fulfil the eligibility criteria, that is, those with lower limb amputations, neuropathy, dementia or other neurological conditions; those unable to walk household distances without the help of a walking aid; those living in residential or nursing care homes; and those aged < 65 years. The views of the podiatrists interviewed in the qualitative part of the study were mixed on whether or not people with neuropathy or amputations could have benefited from the intervention, and the majority agreed that a more intensive follow-up would have been required in order to ensure patient safety.

Implications for health care

Our results suggest that there is a role for NHS podiatrists in reducing the risk of falling among their patients. Although cost-effectiveness was demonstrated based on QALYs gained calculated via the EQ-5D and not necessarily on reducing falls, falls could potentially have a negative effect of patients’ quality of life and any intervention to improve this is valid. However, in terms of the current intervention, some of the podiatrists felt that additional podiatry contact was required to maximise compliance with the individual intervention components. There is the potential for the cost of the intervention to be further reduced if a podiatry assistant rather than the podiatrist undertook the assessment of participants’ footwear and the measuring, ordering and fitting of new footwear.

Implications for research

The impact of falls risk among these patients was relatively modest. As falls are a major source of morbidity in an older population, research into combining different interventions to develop a more effective overall strategy might be worth pursuing. Further research could also examine the risk and cost of falls in other populations or settings (e.g. people with neuropathy or residential aged care facilities). Additionally, the intervention could be tested in populations deemed to be at high risk of falling.

There is evidence to suggest that exercise is an effective falls prevention strategy, and it may be the case that it is equally, or possibly more, effective when demonstrated to patients in group sessions, as opposed to one on one. This would have the additional benefit of being cheaper to deliver and, therefore, being more cost-effective. Further research could be undertaken to test the clinical effectiveness and cost-effectiveness of a group exercise programme, which could also investigate whether or not the intervention could be delivered equally effectively across the professional boundaries of podiatry and physiotherapy. Alternatively, further research into the intensity of the exercise could be undertaken to see how much is actually needed.