key: cord-0316223-q4c4ai0z
authors: Oddsson, L. I. E.; Bisson, T.; Cohen, H.; Iloputaife, I.; Jacobs, L.; Kung, D.; Lipsitz, L. A.; Manor, B.; McCracken, P.; Rumsey, Y.; Wrisley, D.; Koehler-McNicholas, S.
title: Extended Effects of a Wearable Sensory Prosthesis on Gait, Balance Function and Falls After 26 Weeks of Use in Persons with Peripheral Neuropathy and High Fall Risk: The walk2Wellness Trial
date: 2022-04-28
journal: nan
DOI: 10.1101/2022.04.28.22274328
sha: cb497ca8efe96902205d0d6c58abe5141a36acda
doc_id: 316223
cord_uid: q4c4ai0z

Background: We recently reported that individuals with impaired plantar sensation and high fall risk due to sensory peripheral neuropathy (PN) improved gait and balance function following 10 weeks of use of Walkasins, a wearable lower limb sensory prosthesis that provides directional tactile stimuli related to plantar pressure measurements during standing and walking (RxFunction Inc., MN, USA). Here, we report 26-week outcomes and compare pre- and in-study fall rates. We expected improvements in outcomes and reduced fall rates reported after 10 weeks of use to be sustained. Methods: Participants had clinically diagnosed PN with impaired plantar sensation, high fall risk (Functional Gait Assessment, FGA score <23) and ability to sense tactile stimuli above the ankle at the location of the device. Additional outcomes included 10m Gait Speed, Timed Up&Go (TUG), Four-Stage Balance Test, and self-reported outcomes, including Activities-Specific Balance Confidence scale and Vestibular Disorders Activities of Daily Living Scale. Participants tracked falls using a calendar. Results: We assessed falls and self-reported outcomes from 44 individuals after 26 weeks of device use; 30 of them conducted in-person testing of clinical outcomes. Overall, improvements in clinical outcomes seen at 10 weeks of use remained sustained at 26 weeks with statistically significant increases compared to baseline seen in FGA scores (from 15.0 to 19.2), self-selected gait speed (from 0.89 m/s to 0.97 m/s), and 4-Stage Balance Test (from 25.6s to 28.4s), indicating a decrease in fall risk. Non-significant improvements were observed in TUG and fast gait speed. Overall, 39 falls were reported; 31 of them did not require medical treatment and four caused severe injury. Participants who reported falls over 6 months prior to the study had a 43% decrease in fall rate during the study as compared to self-report 6-month pre-study (11.8 vs. 6.7 falls/1000 patient days, respectively, p<0.004), similar to the 46% decrease reported after 10 weeks of use. Conclusion: A wearable sensory prosthesis can improve outcomes of gait and balance function and substantially decreases incidence of falls during long-term use. The sustained long-term benefits in clinical outcomes reported here lessen the likelihood that improvements are placebo effects. Trial registration: ClinicalTrials.gov (#NCT03538756)

Detailed descriptions of materials and methods have been reported previously (Oddsson, Bisson et al. 144 2020), and are summarized below. The measures we use have been validated (Oddsson, Bisson et al. 145 2020) and our primary outcome measure, the FGA, is recommended for assessment of walking 146 balance in adults with neurologic conditions by current physical therapy Clinical Practice Guidelines 147 (Moore, Potter et al. 2018) . The walk2Wellness study is registered at ClinicalTrials.gov 148 (#NCT03538756). Figure 1 shows a picture of the Walkasins device components with a brief 149 description of its functionality. 150 151

2.1 Participants 152

The walk2Wellness clinical trial received approval for human subject research from Advarra IRB 153 (formerly Quorum Review) for four sites and from the IRB Subcommittee, the Subcommittee on 154

Research Safety, and the Research and Development Committee of the Minneapolis VA Health Care 155

System (MVAHCS). Recruitment for the walk2Wellness trial began in October 2018 and ended at 156 the last site in April 2021. Because of COVID-19 disruptions, the study ended earlier than planned at 157 four sites; the fifth site completed the trial but with a reduced enrollment goal. Study site team 158 members posted fliers at community businesses and clinics to advertise the trial; in addition, some of 159 the sites received referrals from physicians who were informed about the study. 160 161

To participate in the trial, individuals met the following inclusion criteria: male or female; ages 21-162 90; a formal diagnosis of sensory PN prior to participating in the study as indicated by their medical 163 record or a letter by a physician; self-reported problems with balance; ability for transfers or 164 ambulation on level surfaces at fixed cadence as assessed by trained study personnel; FGA score <23, 165 the cut-off score for high fall risk (Wrisley and Kumar 2010); ability to understand and provide 166 informed consent; foot size to allow the Walkasins device to function properly, and ability to 167 complete all functional outcome measures without the use of an assistive device in order to ensure 168 sufficient motor function. Participants could use an assistive device at their discretion during their 169 daily activities. 170 171

Individuals were excluded if they were unable to perceive the tactile stimuli from the Walkasins Leg 172

Unit (Figure 1 ) or if they used an ankle-foot orthosis for ambulation that prevented donning the 173 device. In addition, individuals having any of the following conditions were excluded from 174 participation: acute thrombophlebitis; deep vein thrombosis; untreated lymphedema; a lesion of any 175 kind, swelling, infection, inflamed area of skin, or eruptions on the lower leg near placement of the 176 device; foot or ankle fractures; severe peripheral vascular disease; or any musculoskeletal or 177 neurological condition that would prohibit use of the device, as determined by a clinician. Because of 178 a potential to overload the pressure sensors in the foot pad, individuals weighing over 136 kg (300 179 lbs) were excluded from participation. The criteria to participate in the study were similar to 180 receiving a prescription for the device in a clinic. Potential participants also agreed not to initiate any 181 balance training (e.g., Tai-Chi, etc.) or balance-related therapy during the first 10 weeks of the trial 182 (Oddsson, Bisson et al. 2020 ). Furthermore, participants were blinded to their outcomes during the 183 trial and not shared by study personnel. 184 185

Before undergoing any study-related procedures, individuals signed the IRB-approved consent form 187

and HIPAA authorization. After enrollment, site staff tested whether participants were able to 188 perceive the tactile stimuli from the Walkasins Leg Unit. Anyone unable to feel the tactile stimuli 189 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 28, 2022. ; https://doi.org/10.1101/2022.04.28.22274328 doi: medRxiv preprint Extended Effects of a Wearable Sensory Prosthesis 5 was excluded from continued participation in the study. Individuals who were able to feel the tactile 190 stimuli from the Walkasins Leg Unit bilaterally then completed a medical history form to assess 191 common health issues and systemic diseases. They also provided information on falls they had 192 experienced in the previous six and twelve months, including the number of falls and any injuries 193 they sustained as a result. In addition, participants provided a list of their medications (medication 194 name, indication, dose, and frequency), which was updated over the course of the study. 195 196 Participants then completed the Activities-Specific Balance Confidence (ABC) Questionnaire 197 (Powell and Myers 1995) to assess levels of balance self-confidence and the Vestibular Activities of 198

Daily Living Scale (VADL), (Cohen, Kimball et al. 2000) to evaluate the effects of vertigo and 199 balance disorders on their ability to perform everyday activities independently. At baseline, week 10, 200 and week 26, site personnel also assessed participants using the Weinstein Enhanced Sensory Test 201

(WEST) monofilament foot test (0.5g, 2g, 10g, 50g, and 200g), as described previously (Oddsson, 202 Bisson et al. 2020 CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted April 28, 2022. ; https://doi.org/10.1101/2022.04.28.22274328 doi: medRxiv preprint 6 assessment to measure the degree of internal self-perception of balance capability (ABC score) in 238 relation to the externally observed walking balance performance (FGA score). We noted that a ratio 239 of around 3.3 would be "expected" based on the maximum scores of the two scales. This construct 240 aligns with a multifactorial causation model for falls emphasizing the importance of a "Realistic 241

Appraisal of One's Own Abilities", (Hadjistavropoulos, Delbaere et al. 2011 ). 242 243

Between the in-person visits at weeks 10 and 26, study sites contacted participants via telephone at 244 weeks 14, 18, and 22 to remind them of study requirements and to collect follow-up information 245 regarding health changes, falls, adverse events, pain scores, device usage, and device functioning, as 246 well as whether they had begun any physical therapy for their balance problems. If participants 247

reported adverse events and/or falls during these contacts, site personnel recorded the details on the 248 appropriate case report forms in REDCap Cloud, the electronic data capture system used in the study 249

(https://www.redcapcloud.com/, Encinitas, CA, United States). 250

An analysis of outcomes after 2, 6, and 10 weeks, with 10 weeks being the primary endpoint, was 252

presented earlier (Oddsson, Bisson et al. 2020 ). Here we focus on outcomes after extended long-term 253 device use and compare outcomes to baseline for individuals participating in 26 weeks of device use. 254

Graphical data from 2, 6 and 10 weeks are provided below for illustration. We calculated descriptive 255 statistics (mean and standard deviation or median as appropriate) for all variables and tested for 256 normality with the Shapiro-Wilk's test. We applied the two-proportion Z-test to compare proportion-257 based measures. We conducted a post-hoc analysis to compare participants at baseline who reported 258 falls in the previous six months (Pre-Fallers, n=25) to those who did not (Pre-NonFallers, n=19). 259

Their baseline characteristics were compared using a t-test for independent samples or a . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted April 28, 2022. ; https://doi.org/10.1101/2022.04.28.22274328 doi: medRxiv preprint 7 3 Results 280

Our previous report with 45 participants at the 10 week primary endpoint (Oddsson, Bisson et al. 282 2020) was published early due to the unknown circumstances at the time related to COVID-19, 283

which had caused recruitment holds and suspension of in-person testing activities at the research 284 sites. Over time, the trial continued with the enrollment of additional participants who are part of the 285 current report as illustrated in the flow chart in Figure 2 . Figure 2 shows the flow chart for the study 286

as it was continued from the 10-week primary endpoint assessment that we have previously reported 287 (Oddsson, Bisson et al. 2020 ). Text boxes with dashed lines in Figure 2 indicate enrollment numbers 288

for the initial 10-week report. 289 290

Of the 85 individuals who were assessed for eligibility, a total of 69 participants were enrolled in the 291 trial (17 more than in previous report, cf. Figure 2 ). Sixty participants took part in the 2-week follow 292 up visit, 55 in the 6-week follow-up visit, and 51 in the 10-week follow-up visit (six more than in the 293 previous report). Eleven of the 18 participants who discontinued interventions through the 10-week 294

follow-up visit stopped participation due to circumstances related to COVID-19 (cf. Figure 2 ). (Note: 295

due to the timing of study initiation at Johns Hopkins and the onset of the pandemic, the three 296 participants enrolled at the site were unable to complete most in-person assessments. When the 297 situation did not improve, the sponsor decided to terminate Johns Hopkins as a research site. None of 298 the participants had completed follow-up datasets through week 10, so their results were not included 299 in this analysis.) 300 301

Forty-four of the 51 participants who reached the 10-week follow-up assessment were analyzed after 302 their 26-week assessment. Due to circumstances related to COVID-19, including lockdowns at the 303 research sites, 14 of the 44 participants were unable to participate in the 26-week in-person outcomes 304 testing (FGA, 10MWT, TUG, and 4-Stage Balance Test). All 44 participants, however, were 305 assessed on self-reported outcomes over the phone, and all 44 participants provided reports of falls 306 throughout the 26-week period. Consequently, we report baseline assessment of clinical outcomes for 307 all 44 participants, comparing those who reported falling in the prior six months (Pre-Fallers) to those 308 who did not (Pre-NonFallers). We also present data over time for the 30 participants who completed 309 the 26-week in-person assessment (weeks 0, 2, 6, 10 and 26, Figure 2 ), comparing the baseline 310 assessments of those 30 participants to their 26-week assessment. 311 312

Seven of the 44 participants discontinued the intervention after the 10-week assessment (cf. Figure  313 2); one passed away; two had been enrolled in solely the 10-week protocol; three expressed device 314 dissatisfaction, and one experienced an unrelated adverse event. 315

Baseline characteristics for the 44 individuals who were evaluated at 26 weeks are shown in Table 1  317 and were substantially similar to those reported previously (Oddsson, Bisson et al. 2020) . Participants 318 across all sites were pooled for the continued analysis. 319 320

The baseline characteristics of the 44 participants are shown in Table 2 . Characteristics are also 321 reported separately for participants who reported falling in the six months preceding the study (Pre-F, 322 n=25) and participants who did not report a fall (Pre-NF, n=19). Overall, characteristics reported 323 previously for participants entering the trial were maintained (Oddsson, Bisson et al. 2020 ) except for 324 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 28, 2022. ; https://doi.org/10.1101/2022.04.28.22274328 doi: medRxiv preprint 8 some substantiated differences between the Pre-F and Pre-NF participants that were now statistically 325 significant (Table 2 ). In addition to previous observations, baseline self-selected gait speed and ABC 326 scores were significantly different between the Pre-F and Pre-NF groups, 0.83 m/s vs. 0.97 m/s and 327 57.8% vs. 69.6%, respectively (Table 2) . A higher observed mean value in PHQ-9 score for the Pre-F 328 group was nearly statistically significant (5.7 vs. 3.3, respectively, p=0.052). Mean values for all 329 PROMIS measures were near 50, and any differences were well within 10 ( 

Improvements in FGA scores were seen across all individuals, irrespective of their baseline FGA 334 scores and appeared similar after 2-, 6-, 10-and 26-weeks of device use ( Figure 3 ). This finding is 335 indicated by the regression lines in Figure 3 being shifted above the line of equality. Figure 3 did not reach statistical significance (p=0.12 and p=0.07, respectively). The Pre-F group showed a 353 decrease in vibration sensation at the first MTP joint (mean 2.9 to 1.8, p<0.005) and at the lateral 354 malleolus that did not reach statistical significance (mean 3.75 to 3.25, p=0.07). 355 356

The Pre-NF group (n=12) increased their mean FGA score from 17.0 at Baseline to 20.8 at 26 weeks 357 (p<0.004, Cohen's drm=1.23, Table 3 ). A small increase in self-selected gait speed from 1.00 m/s to 358 1.08 m/s did not reach statistical significance (p=0.06). Other clinical outcomes, including Rydel-359

Seiffer tuning fork sensitivity testing scores, remained unchanged at 26 weeks compared to baseline 360 (Table 3) . 361 362

The mean values of all in-person outcomes across the 26 weeks of device use for all participants are 363 depicted in Figure 4 and are shown separately for the Pre-F and the Pre-NF groups. In general, the 364 behavior over time of the two subgroups was qualitatively similar, although the Pre-NF group 365 consistently performed better than the Pre-F cohort. 366

3.4 Self-reported Outcomes, Device Use and ABC to FGA ratios 367 Self-reported outcomes for the 44 participants who completed assessments through 26 weeks are 368

shown in Table 4 . Across all participants, a slight but statistically significant increase was seen in 369 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 28, 2022. days. Participants reporting that they used the device weekly either "Every Day" or "At least 5 Days" 376 was 71.8±10.5%. An average of 94.8±3.3% of reporting participants stated they used the device "1-2 377

Days" or more per week. 378 379 We also sought to monitor participants' self-perceived balance confidence (Lajoie and Gallagher  380 2004) in relation to their assessed gait function performance over time in the trial by measuring 381 changes in the ABC to FGA ratio, which is shown in Figure 5 . Figure 5 illustrates two observations. 382

First, the ABC/FGA ratio for the overly confident individuals gradually decreased from high values 383 at baseline to week 6 when the ratio essentially aligned with the low self-confidence participants, 384

who maintained a consistent ABC/FGA ratio of about 3.5 throughout the 26 week period of the trial 385 ( Figure 5 , left y-axis). Second, both groups of participants increased their FGA scores in a similar 386 fashion although the overly confident participants showed higher levels of improvement in their FGA 387 scores ( Figure 5 , right y-axis). 388

Falls were reported throughout the 26-week period for all 44 participants and separately for the Pre-F 390 (n=25) and Pre-NF (n=19) groups, respectively (Table 5 ). The 44 participants reported a total of 53 391 falls over 6 months prior to participating in the trial while 39 falls were documented during the 26 392 weeks of the trial, corresponding to a pre-study mean fall rate of 6.7 falls/1000 patient days 393 (median=5.6 falls/1000 patient days) and a post study mean fall rate of 4.8 falls/1000 patient days 394 (median=0 falls/1000 patient days). Wilcoxon Signed Rank test was used to compare pre-and post-395 study fall rates since the data was not normally distributed (Shapiro-Wilk W-statistic=0.82, 396

p<0.0001). Across all participants, the median of the post-study fall rate was lower than the pre-study 397 fall rate ( CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 28, 2022. ; https://doi.org/10.1101/2022.04.28.22274328 doi: medRxiv preprint Figure 6 illustrates the cumulative sum of falls based on six-month pre-study fall reports from the 417 participants and the falls recorded by the Pre-F cohort during the trial. The 53 pre-study falls were 418 randomly distributed across the six-month pre-study period for illustration purposes since the exact 419 time of their occurrence prior to the trial were not known (see Figure 6 ). Note that 53 falls over a 420 period of six months corresponds to 53/180~0.29 falls/day, which becomes the slope of the blue 421 regression line in Figure 6 , irrespective of the random distribution of the pre-study falls. The in-study 422 falls are presented on the day they occurred according to participant reports (see Figure 6 ). The in-423 study decrease of falls quantified in Table 5 can be visually observed over time in Figure 6 as a lower 424 slope for the best fit regression line (blue-dotted line pre-study vs. orange-dotted line in-study). It 425

should be noted that the lower slope appears to begin after about 20 days of device use. 426 427 428

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 28, 2022. Participants had a decrease in sensitivity to vibration, an overall small decrease at the lateral 455 malleolus and a larger change at the MTP joint in the Pre-Faller group. Such changes did not occur at 456 the 10-week assessment (Oddsson, Bisson et al. 2020 ). This decrease in tuning fork vibration 457

perception may indicate a progression of some participants' PN and increased sensory loss. In spite 458 of such changes, none of the participants stated that they were unable to perceive the stimuli provided 459 by the device, suggesting they were still benefiting from its use as confirmed by improved clinical 460 outcomes. 461

We can compare our findings to interventions using different forms of physical exercise in patients 463 with PN since we see similar improvements in outcomes from simply wearing the device in the 464 current trial. Several review studies conclude that balance training is "the most effective exercise is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 28, 2022. ; https://doi.org/10.1101/2022.04.28.22274328 doi: medRxiv preprint lose it" principle (Hart 2021 total length of treatment either because they feel they have reached their maximum level of 518 improvement, because they dislike the hassle of taking the time for the therapy visit, or because they 519 cannot get transportation. In addition, some insurance programs may refuse to pay for more than a 520 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 28, 2022. ; https://doi.org/10.1101/2022.04.28.22274328 doi: medRxiv preprint 13 limited number of therapy visits. In the case of exposure to balance stimuli from a sensory prosthesis, 521 however, participants used the device on average 36.1 hours/week with more than 80% using the 522 device over 21 hours/week (Oddsson, Bisson et al. 2020) . Over 26 weeks this use would amount to a 523 range from around 500 to almost 1000 hours of specific gait and balance stimulation, essentially an 524 order of magnitude higher than a regular balance training intervention. Furthermore, the device is 525 intended to be used on a continuous basis, which furthers compliance. 526 527

Participants may not be physically active and receiving sensory stimuli the entire abovementioned 528 time. However, they will regularly experience hundreds of specific balance stimuli/hour that are 529

intimately integrated into their regular standing and walking activities and that are likely to cause 530 related changes in balance behavior. Over time, this feedback may lead to a reweighting of sensory 531 information that is relevant for balance function ( 

On the ABC, 67% is the cut-off for high-fall risk (Lajoie and Gallagher 2004) . We previously 552

reported that participants with baseline ABC scores below 67% improved their ABC scores after 10 553 weeks of device use, whereas those with ABC scores above 67% did not, although both subgroups 554 improved their FGA scores (Oddsson, Bisson et al. 2020 ). Since these two categories of participants 555 had similar baseline FGA scores, the discrepancy in ABC scores led to seemingly high baseline ABC 556

to FGA ratios in participants with high ABC scores (Oddsson, Bisson et al. 2020) ( Figure 5) . We 557 posited that the high ratio individuals were either too confident or simply unaware of their abilities or 558

impairments, a construct that aligns with a proposed multifactorial causation model for falls 559 emphasizing the importance of a "Realistic Appraisal of One's Own Abilities", (Hadjistavropoulos, 560 Delbaere et al. 2011). This ratio indicates the degree of internal self-perception of balance capability 561 (ABC score) in relation to the externally observed performance (FGA score). A ratio of around 3.3 562 would be "expected" based on the maximum scores of the two scales (Powell and Myers 1995, 563 Wrisley and Kumar 2010). We further noted that after six weeks of device use, the overly confident 564 participants appeared to "normalize" their self-perception indicating a more sensible perception of 565 their abilities and align with those holding more expected ABC scores. This change was mainly due 566

to an improvement in their FGA score and a maintained or slightly decreased ABC score. 567 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 28, 2022. ; https://doi.org/10.1101/2022.04.28.22274328 doi: medRxiv preprint

We show here that this observation is preserved after 26 weeks of device use ( Figure 5 ). This change 568 in self confidence is important because it is a fall risk factor. As the data suggest, this ratio is a 569 modifiable risk factor. Therefore, the ratio can and should be measured during studies of balance 570

interventions. 571 572

Furthermore, the current data ( Figure 5) shows that the high-confidence individuals appear to 573 improve their FGA scores more than the low confidence individuals. Since the baseline FGA scores 574

were similar for these two cohorts, it could be argued that the higher balance-confidence individuals 575 may have "naturally" challenged themselves more in their daily activities, triggering a larger increase 576 in FGA score. Correspondingly, the low balance-confidence individuals may have "explored" their 577 improved functional capabilities less than the high confidence group leading to a less beneficial long-578 term effect. We postulate that individuals with low balance confidence would potentially benefit 579

further from a guided interaction through physical and occupational therapy or other means, whereas 580 the high confidence individuals may be better at "self-coaching" their renewed capabilities. 581

Consequently, to provide an individually optimized rehabilitation strategy for falls prevention and 582

self-efficacy, clinicians should monitor outcomes that measure both balance outcomes and balance 583

confidence, similar to the ratio we have used to allow this construct to be investigated further 584 ( level of device use, on average more than five days a week and over 70% of participants using it 591 every day or at least 5 days/week, suggests that users find the device helpful. 592

One of the most important findings from this trial is the 43% decrease in fall rate for participants with 594 a fall history following 26 weeks of device use. We reported an encouraging 46% decrease already 595 after 10 weeks of use (Oddsson, Bisson et al. 2020) , while appreciating that 10 weeks would not be 596 long enough for a relevant comparison to six months of pre-study data. Consequently, the similar 597 statistically significant decrease after 26 weeks provides a stronger argument for a meaningful 598 decrease in falls. In fact, even including the pre-study non-fallers in the analysis showed a 599 statistically significant decrease in fall rate over 26 weeks of device use (Table 5) . Note, however, 600 that fall rate was not a primary outcome in this trial. In fact, we gathered pre-study falls data 601

primarily as a means of describing our study population, and we monitored falls during the trial as a 602

subset of adverse events. Thus, we enrolled participants at high fall risk based on their FGA scores, 603

irrespective of whether or not they had fallen in the past. This procedure allowed us to conduct a 604 post-hoc analysis and investigate any trend in the data indicating an actual decrease in falls. A 605 comparison with historical self-report data on prior falls is far from ideal as compared to a parallel 606 arm trial with a control group, but other investigators have found that asking individuals to recall past 607 falls commonly leads to an underreporting of actual falls; even recalling the past 3 months led to 25% 608 underreporting of falls (Hannan, Gagnon et al. 2010) , and 26% did not recall falls over the prior six 609 months (Cummings, Nevitt et al. 1988 ). Under such circumstances, the 43% decrease in falls 610

reported here probably underestimates the real effect. 611 612

The data on injuries reported from falls are probably clinically significant. Previous studies have 613

reported that 65% of older individuals with PN fell over a year with 30% reporting an injury from a 614 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 28, 2022. ; https://doi.org/10.1101/2022.04.28.22274328 doi: medRxiv preprint fall (DeMott, Richardson et al. 2007 This individual continues to use the device daily, now for more than four years. Also, blinding 630 participants in a study using this kind of intervention is challenging and may not really be viable 631 unless some form of deception is used since a requirement to use the device is perceiving the tactile 632 stimuli (Oddsson, Bisson et al. 2020 ). Consequently, the most feasible placebo treatment for a control 633 group would likely be wearing a device that is non-functional. 634 635

An important strength of the trial includes our real-life inclusion/exclusion criteria that were 636 essentially the same as the requirements for receiving a prescription for the device outside of a 637 research trial. These requirements are aligned with the intentions of a pragmatic clinical trial to 638 advance applicability of study findings (Patsopoulos 2011 , Gill, Bhasin et al. 2021 ). Only 7% of 639 participants (6 of 85) were excluded due to medical circumstances (Figure 2) . We wanted the 640 participants to realistically reflect patients who are seen in the clinic, and we did not try to screen for 641 only those individuals who may be the best responders. In fact, our participants reported an average 642 of 8.3 chronic conditions, had multiple fall-risk factors, and polypharmacy (mean of 8 subscription 643 medications). 644 645

Furthermore, the long-term benefits we report here combined with the participants not receiving any 646 useful feedback about their outcomes or systematic encouragement during their assessments and 647

knowing that only one individual decided to participate in a balance therapy intervention, decreases 648 the chance these effects are placebo ( 

Patients with peripheral neuropathy who have gait and balance problems with a high risk of falls 656 including a history of falls improve their walking balance and decrease their fall rates from long-term 657 use of a wearable sensory prosthesis. 658 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 28, 2022. leading to nerve damage, is recognized as an independent risk factor for falls and is often related to 671 diabetes or chemotherapy treatment or is of unknown etiology. A common symptom involves 672 numbness and sensory loss in the feet that can directly cause falls. Here we report data after extended 673 long-term daily use of a wearable sensory prosthesis (Walkasins ® , RxFunction Inc., MN, USA) from 674 the multi-site clinical trial, walk2Wellness. The device provides directional tactile stimuli around the 675 ankle during standing and walking activities, reflecting changes in foot pressure related to balance. 676

Patients can then "feel" their feet in contact with the ground and notice body sway during standing 677 because the device replaces related aspects of their impaired afferent nerve function. Data after 26 678

weeks of use shows that previously reported improvements in clinical outcomes after 10 weeks of 679 use are sustained. In addition, we can now report a decrease in fall rate in individuals who reported 680 falling six months prior to participating in the study. 681

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 28, 2022. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 28, 2022. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 28, 2022. ; https://doi.org/10.1101/2022.04.28.22274328 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 28, 2022. ; https://doi.org/10.1101/2022.04.28.22274328 doi: medRxiv preprint Table 3 . Clinical outcomes and Rydel-Seiffer vibration sensation screening for the 30 individuals who were able to 1008 complete all in-person assessments at baseline and after 26 weeks as well as the subgroups of Pre-Fallers and Pre-1009 NonFallers. Statistical significance is indicated in bold. Bonferroni's adjustment of significance levels for correlated 1010 measures was applied (Uitenbroek 1997 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted April 28, 2022. ; https://doi.org/10.1101/2022.04.28.22274328 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted April 28, 2022. ; https://doi.org/10.1101/2022.04.28.22274328 doi: medRxiv preprint 

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted April 28, 2022. ; https://doi.org/10.1101/2022.04.28.22274328 doi: medRxiv preprint

Predictors of compliance with 683 short-term treatment among patients with back pain

Effect Sizes and their Intervals: The Two-685 Level Repeated Measures Case

Evidence from diverse 687 clinical populations supported clinical validity of PROMIS pain interference and pain behavior

Effects of long-term balance training with vibrotactile sensory 691 augmentation among community-dwelling healthy older adults: a randomized preliminary study

Retention Effects of Long-Term Balance Training with 695

Vibrotactile Sensory Augmentation in Healthy Older Adults

Minimal clinically important difference of 697 the functional gait assessment in older adults

Falls and Fall Injuries Among Adults Aged ≥65 699 Years -United States

The trajectory of gait 701 speed preceding mild cognitive impairment

Problems with gait 703 and posture in neuropathic patients with insulin-dependent diabetes mellitus

STEADI -Older Adult Fall Prevention

Older Adult Fall Prevention

Prognostic value of usual gait speed in well-functioning older people

Aging and Body Composition Study

Automaticity of walking: functional significance, mechanisms, measurement 714 and rehabilitation strategies

Enhanced 716 somatosensory feedback reduces prefrontal cortical activity during walking in older adults

Application of the vestibular disorders 719 activities of daily living scale

Statistical Power Analysis for the Behavioral Sciences

Prevention Program Lowered 723 The Risk Of Falls And Decreased Claims For Long-Term Services Among Elder Participants

Placebo-related effects: a meta-narrative review of 726 conceptualization, mechanisms and their relevance in rheumatology

Forgetting falls. The limited accuracy of recall 729 of falls in the elderly

Falls and gait 731 characteristics among older persons with peripheral neuropathy

Clinical measures of balance in 734 people with type two diabetes: A systematic literature review

Slow 736 walking speed and cardiovascular death in well functioning older adults: prospective cohort study

The placebo response in medicine: 739 minimize, maximize or personalize?

Evidence for strong synaptic 741 coupling between single tactile afferents from the sole of the foot and motoneurons supplying leg 742 muscles

Biological, clinical, and ethical 744 advances of placebo effects

Prevention of Falls in Community-Dwelling Older Adults

Effect of a Multifactorial Fall Injury Prevention Intervention on Patient Well-754 Being: The STRIDE Study

Prevalence of lower-extremity disease in the 757 US adult population >=40 years of age with and without diabetes: 1999-2000 national health and 758 nutrition examination survey

Central pattern generator for locomotion: anatomical, physiological, and 760 pathophysiological considerations

Reconceptualizing the role of fear 762 of falling and balance confidence in fall risk

The PROMIS satisfaction with social participation measures demonstrated responsiveness in 765 diverse clinical populations

New English and Spanish social health measures will facilitate evaluating health 768 determinants

Measuring social function in diverse cancer populations: Evaluation of 771 measurement equivalence of the Patient Reported Outcomes Measurement Information System

Long-term 774 effects of new progressive group balance training for elderly people with increased risk of falling -a 775 randomized controlled trial

Polyneuropathy relates to impairment in daily 778 activities, worse gait, and fall-related injuries

Optimizing the tracking of falls in studies of older participants: 781 comparison of quarterly telephone recall with monthly falls calendars in the MOBILIZE Boston 782 Study

Improvement in 784 usual gait speed predicts better survival in older adults

Learning From Human Responses to Deconditioning Environments: Improved 786 Understanding of the "Use It or Lose It" Principle

Prevalence of 788 peripheral neuropathy defined by monofilament insensitivity in middle-aged and older adults in two 789 US cohorts

791 "Impairments and comorbidities of polyneuropathy revealed by population-based analyses

Six-Month Lower-Leg 794 Sensory Stimulation Augments Neural Network Connectivity Associated With Improved Gait

Balance 797 interventions for diabetic peripheral neuropathy: a systematic review

Validation of a Claims-Based Frailty Index Against Physical Performance and Adverse Health 801 Outcomes in the Health and Retirement Study

Neuromodulation 803 to improve gait and balance function using a sensory neuroprosthesis in people who report insensate 804 feet -A randomized control cross-over study

Falls and Functional Impairments in Breast Cancer Patients with Chemotherapy-807 Induced Peripheral Neuropathy

Risk factors for major injurious falls 809 among the home-dwelling elderly by functional abilities. A prospective population-based study

The PHQ-9: validity of a brief depression 812 severity measure

Fall and balance outcomes after an 814 intervention to promote leg strength, balance, and walking in people with diabetic peripheral 815 neuropathy: "feet first" randomized controlled trial

The value of the 817

Rydel-Seiffer tuning fork as a predictor of diabetic polyneuropathy compared with a 818 neurothesiometer

Predicting falls within the elderly community: comparison of 820 postural sway, reaction time, the Berg balance scale and the Activities-specific Balance Confidence 821 (ABC) scale for comparing fallers and non-fallers

Calculating and reporting effect sizes to facilitate cumulative science: a practical 823 primer for t-tests and ANOVAs

Extended Effects of a Wearable Sensory Prosthesis 20

Tai Chi and fall reductions in older adults: a randomized controlled trial

Long term Tai Chi exercise improves physical performance among 828 people with peripheral neuropathy

A Cluster Randomized Trial of Tai Chi vs Health Education in Subsidized 831 Housing: The MI-WiSH Study

The 833 pace and prognosis of peripheral sensory loss in advanced age: association with gait speed and falls

Functional benefits of tai chi training in senior housing facilities

Balance in elderly patients: the "get-up and go" test

Improving balance control and self-reported lower extremity 841 function in community-dwelling older adults: a randomized control trial

Walking stability and 844 sensorimotor function in older people with diabetic peripheral neuropathy

Reduced plantar sensitivity alters postural 847 responses to lateral perturbations of balance

The role of plantar cutaneous sensation in 849 unperturbed stance

Gait velocity as a single predictor of adverse events in healthy seniors aged 75 852 years and older

Evaluation of Clinical Practice Guidelines on Fall Prevention and 857 Management for Older Adults: A Systematic Review

A 859 Core Set of Outcome Measures for Adults With Neurologic Conditions Undergoing Rehabilitation: A 860 CLINICAL PRACTICE GUIDELINE

Trends in Nonfatal Falls and Fall-Related Injuries 862 Among Adults Aged ≥65 Years -United States

Combining effect size estimates in meta-analysis with 865 repeated measures and independent-groups designs

Supervised Balance 867 Training and Wii Fit-Based Exercises Lower Falls Risk in Older Adults With Type 2 Diabetes

Postural Control and 870 Gait Performance in the Diabetic Peripheral Neuropathy: A Systematic Review

Functional Brain Activation in Response to a Clinical Vestibular Test Correlates 874 with Balance

The 877 Effects of a Wearable Sensory Prosthesis on Gait and Balance Function After 10 Weeks of Use in 878 Persons With Peripheral Neuropathy and High Fall Risk -The walk2Wellness Trial

How to improve gait and balance function in 881 elderly individuals-compliance with principles of training

A pragmatic view on pragmatic trials

Automaticity of walking -implications for 886 physiotherapy practice

Meaningful change and 888 responsiveness in common physical performance measures in older adults

The Activities-specific Balance Confidence (ABC) Scale

Exercise interventions, gait, and balance in older subjects with distal symmetric 894 polyneuropathy: a three-group randomized clinical trial

Sensory-Motor Mechanisms Increasing Falls 897 Risk in Diabetic Peripheral Neuropathy

Peripheral neuropathy: a true risk factor for falls

A focused exercise regimen improves clinical 901 measures of balance in patients with peripheral neuropathy

903 "Idiopathic peripheral neuropathy increases fall risk in a population-based cohort study of older 904 adults

Some aspects of old age

The effect of multidimensional 907 exercises on balance, mobility, and fall risk in community-dwelling older adults

Potential Mechanisms of Sensory Augmentation Systems on Human Balance Control

Correlates of exercise compliance in physical 913 therapy

Falls efficacy: 915 Extending the understanding of self-efficacy in older adults towards managing falls

Agonist-antagonist myoneural interface amputation preserves proprioceptive sensorimotor 919 neurophysiology in lower limbs

Extended Effects of a Wearable Sensory Prosthesis 22

STEADI-Older Adult Fall Prevention

Development of STEADI: a fall prevention resource for 923 health care providers

Exercise intervention studies in patients with peripheral neuropathy: a 926 systematic review

Cutaneous afferent 928 innervation of the human foot sole: what can we learn from single-unit recordings?

Gait speed and survival in older adults

Physical performance measures in the clinical setting

It's always a trade-off

Strength and balance training for adults with 940 peripheral neuropathy and high risk of fall: current evidence and implications for future research

SISA-Binomial

Effect of Low-Intensity Physical Activity and Moderate-to

High-Intensity Physical Exercise During Adjuvant Chemotherapy on Physical Fitness, Fatigue, and 949 Chemotherapy Completion Rates: Results of the PACES Randomized Clinical Trial

The roles of mechanoreceptors in muscle and skin in human 952 proprioception

AGING, DIABETES, AND FALLS

Ankle-Foot Orthoses for Fall Prevention in Older Adults: A Randomized Controlled Trial

Falls, Functioning, and Disability Among Women With Persistent Symptoms of 960 Chemotherapy-Induced Peripheral Neuropathy

Effect of a 962 physical therapeutic intervention for balance problems in the elderly: a single-blind, randomized, 963 controlled multicentre trial

Intense 965 tai chi exercise training and fall occurrences in older, transitionally frail adults: a randomized, 966 controlled trial

The two components of the Walkasins device, the pressure sensitive flexible foot pad that is placed in the 1042 shoe and connects to the leg unit that contains a rechargeable battery, a microprocessor, supporting electronics, and four 1043 tactile stimulators. The embedded software algorithm evaluates pressure data and activates tactile stimulators at relevant 1044 times during standing and walking to signal balance-related information to the afferent nervous system

Walkasins user wearing the device in the process of turning it on

2, 6, 10, and 26 weeks for the 1087 30 participants (blue lines and symbols) who were able to be tested in clinic after 26 weeks of device use. Grey symbols 1088 and lines represent the pre-study non-fallers (n=12 of 30,) and orange lines and symbols represent pre-study fallers 1089 (n=18 of 30). Notice that most improvements appeared to peak after 6-10 weeks of device use

Illustration of changes in ABC/FGA ratio (left axis, solid lines), an indication of self-perceived balance 1099 confidence in relation to externally observed gait function performance, for participants with baseline ABC scores >67% 1100 (high balance-confidence, orange triangles) and those with ABC scores ≤67%

The associated changes over time in FGA scores for the same two groups are also shown (right axis

Over time, the ratio for the high balance-confidence participants gradually decreased and aligned with the low 1104 balance-confidence group at ~3.5, mainly due to an increased FGA score and a maintained ABC score. The low balance-1105 confidence group maintained an ABC/FGA ratio ~3.5 throughout the 26 weeks associated with a proportional increase in 1106 ABC and FGA scores. Interestingly, the high balance-confidence group showed a higher increase in FGA score than the 1107 low balance-confidence group