Budget impact analysis of robotic exoskeleton use for locomotor training following spinal cord injury in four SCI Model Systems
We know little about the budget impact of integrating robotic exoskeleton over-ground training into therapy services for locomotor training. The purpose of this study was to estimate the budget impact of adding robotic exoskeleton over-ground training to existing locomotor training strategies in the rehabilitation of people with spinal cord injury.
A Budget Impact Analysis (BIA) was conducted using data provided by four Spinal Cord Injury (SCI) Model Systems rehabilitation hospitals. Hospitals provided estimates of therapy utilization and costs about people with spinal cord injury who participated in locomotor training in the calendar year 2017. Interventions were standard of care walking training including body-weight supported treadmill training, overground training, stationary robotic systems (i.e., treadmill-based robotic gait orthoses), and overground robotic exoskeleton training. The main outcome measures included device costs, training costs for personnel to use the device, human capital costs of locomotor training, device demand, and the number of training sessions per person with SCI.
Robotic exoskeletons for over-ground training decreased hospital costs associated with delivering locomotor training in the base case analysis. This analysis assumed no difference in intervention effectiveness across locomotor training strategies. Providing robotic exoskeleton overground training for 10% of locomotor training sessions over the course of the year (range 226–397 sessions) results in decreased annual locomotor training costs (i.e., net savings) between $1114 to $4784 per annum. The base case shows small savings that are sensitive to parameters of the BIA model which were tested in one-way sensitivity analyses, scenarios analyses, and probability sensitivity analyses. The base case scenario was more sensitive to clinical utilization parameters (e.g., how often devices sit idle and the substitution of high cost training) than device-specific parameters (e.g., robotic exoskeleton device cost or device life). Probabilistic sensitivity analysis simultaneously considered human capital cost, device cost, and locomotor device substitution. With probabilistic sensitivity analysis, the introduction of a robotic exoskeleton only remained cost saving for one facility.
Providing robotic exoskeleton for over-ground training was associated with lower costs for the locomotor training of people with SCI in the base case analyses. The analysis was sensitive to parameter assumptions.
KeywordsEconomic Budget impact analysis Spinal cord injury Robotics Locomotor training
Budget Impact Analysis
Body-weight supported treadmill training
Inpatient Rehabilitation Facility
Long Term Acute Care Hospital
Spinal Cord Injury
Between 249,000 to 363,000 persons live with disabilities due to spinal cord injuries (SCI) in the United States, and approximately 17,730 individuals experience new SCI each year . The average age of SCI onset is 43 years, affording the opportunity to resume active community involvement and employment. Despite strong incentives to resume activities, much depends on the restoration of upper and lower limb function. The inability to stand and walk not only limits community involvement and employment but these functional limitations also impose significant secondary health conditions. These conditions include depression, pressure ulcers, severe spasticity, pain, limited joint range of motion, contractures, muscle atrophy, bone loss, and impaired digestive, respiratory, renal, and cardiovascular function . Each condition markedly reduces health and quality of life of persons with SCI. In addition to the physical burden, spinal cord injury accounts for significant health care costs ranging from $368,562 to $1,129,302 in first year costs and $44,766 to $196,107 in annual health care costs depending on the severity of injury .
For most individuals, some form of functional limitation persists; however, improvement in mobility and function is possible following the injury. The preservation of volitional motor or sensory function below the lesion level provides an opportunity for motor recovery from intensive training during rehabilitation. This population has great potential for motor recovery and walking. Locomotor training is a standard of care for the rehabilitation of the individual with incomplete SCI. Likewise, walking is usually one of the primary goals of rehabilitation due to physiological and social benefits [3, 4].
Conventional locomotor training strategies include those in which body-weight is supported while clinicians may manually facilitate stepping motions. This training requires multiple personnel to assist with support or lower extremity movement. Although body-weight supported treadmill training (BWSTT) and overground training (OGT) remain common locomotor training strategies, alternative strategies using robotic therapy (RT) devices require less human capital and allow greater repetition than typically is delivered in personnel-intensive approaches . RT may enhance therapists’ productivity and effectiveness while facilitating neurological recovery . RT devices have several advantages, including a structure that provides stability, programmable force-production that facilitates lower extremity movement, [6, 7] precise measurement of forces, [5, 7, 8] and the ability to perform repetitive, labor-intensive tasks [9, 10]. A limitation of many RT devices is their limited interaction with the external environment. Many are stationary and only serve as a tool for the promotion of neural recovery or exercise.
Robotic exoskeletons (RT-exo) are prescription devices comprising an external, motorized orthosis that is placed over paralyzed or weakened limbs for the purpose of facilitating standing, walking, climbing stairs, and performing activities of daily living [11, 12]. Exoskeletons can be used at home, at work, and during rehabilitation . Thus, RT-exo may serve the therapeutic role proposed by stationary robotic devices  while allowing mobility and more active community involvement.
Multiple randomized trials have found outcomes associated with RT to be no different from conventional therapy when similar doses are compared [15, 16, 17, 18, 19, 20];; however, a recent meta-analysis of studies investigating RT in individuals with SCI found greater improvement for walking independence (3.73; 95% confidence interval [CI], − 4.92 to − 2.53; P < .00001) and endurance (53.32 m; 95% CI, − 73.15 to − 33.48; P < .00001) in robot-assisted training groups . Additional systematic reviews found improvement in locomotor recovery using robot-assisted therapy following SCI . Tefertiller and colleagues reported that RT is likely to be at least as effective as therapy alternatives for locomotor training for individuals with neurological injuries and diseases [14, 22]. Whereas RT-exo has less evidence supporting its use versus treadmill-based RT, several pilot studies of RT-exo have recently been published with preliminary evidence of efficacy [10, 23]. Sale et al. used a prospective, quasi-experimental study, pre-post design and found that RT-exo training resulted in significant improvements in timed up-and-go (TUG), six minute walk test (6MWT), and ten meter walk test (10MWT) . Chang et al. randomized patients to RT-exo versus conventional physical therapy and found the RT-exo group to increase significantly in 6MWT, stride length, and step length after the intervention .
RT-exo shares many of the features of treadmill-based RT devices while allowing greater user autonomy and fewer environmental restrictions than treadmill-based RT devices . A recent qualitative study of gym-based RT-exo users identified themes supporting improved community integration ; however in a recent narrative overview of RT-exo, the authors critiqued four RT-exo devices, suggesting that they fall short of promises to reengage in activities set in real-world contexts . Furthermore, the authors suggest that RT-exo is better suited for rehabilitation settings than in home settings .
Cost considerations of new technology
On average new technologies improve health outcomes
Many new technologies do not improve health outcomes and it is not always easy to discriminate between effective and ineffective technologies.
On average new technologies add to health care costs
Diffusion of new technology often is inefficient
Demand for new technology almost always is high.
The rapid growth and development of robotic devices is an excellent example of such a technology in rehabilitation. The costs of robotic exoskeletons include a high initial purchase price (relative to competing rehabilitation technology), annual maintenance costs, and training costs per physical therapist user. Whereas many new technologies can be cost-inefficient, others have been shown to be cost-neutral or cost-saving . It has been suggested that there is potential for new robotic technology in rehabilitation to be cost saving if the cost of a robotic system could be offset by a decreased need for rehabilitation personnel to conduct the labor-intensive demands of rehabilitation . Whereas robotic exoskeletons initially have been considered a research tool and a device for individual users intent on community use, the robotic exoskeleton is less expensive than stationary robotic devices and may be an attractive option for rehabilitation hospitals as a device for training. An unfortunate reality of technology is the delay between the rising clinical evidence and the rate at which reimbursement changes to account for the costs of technology .
Budget impact analysis – framework
We know little about the budget impact of integrating RT-exo devices into routine therapy services. Carpino et al. conducted a review and model-based cost-effectiveness analysis of lower extremity robotics, showing RT-exo to be cost-effective relative to conventional care and stationary robotics; however, model assumptions have been criticized . Morrison performed an analysis of the financial feasibility of lower extremity robotics, but did not address exoskeletal devices . No budget impact analysis was performed for RT-exo in these economic studies. A budget impact analysis (BIA) provides a framework for synthesizing knowledge when healthcare systems wish to estimate the financial consequences of purchasing new technology . A BIA is not designed to provide a single estimate applicable to all decision-makers, but instead reflects local health systems’ environment and decision-makers’ varying perspectives . For the purpose of this study, we evaluate a model that reflects the introduction or expansion of RT-exo on the total health system budget at four model health systems with different case mixes.
BIA framework and conceptual model for SCI
The perspective of the analysis is from the rehabilitation facilities, specifically Inpatient Rehabilitation Facilities and Long Term Care Hospitals, as the purchasers of technology. A retrospective pre-post study design was implemented to estimate the treatment cost of locomotor strategies. The budget impact of RT-exo was evaluated over one year. The calculation was replicated in a Microsoft Excel-based spreadsheet. Probabilistic sensitivity analysis was conducted using R Statistical Software .
Four SCI Model Systems agreed to collaborate on this project:  The Shirley Ryan AbilityLab (formerly the Rehabilitation Institute of Chicago),  Craig Hospital,  Shepherd Center, and  TIRR Memorial Hermann. Participating model system sites served as the primary data source for estimates of resource utilization.
Number of individuals with SCI/year eligible for locomotor training
Number of sessions offered per user
Number of locomotor training sessions per year for individuals with SCI
Locomotor strategy cost components in base case analysis
OGT- low cost
OGT– high cost
$8500/yr × 5 years
$15,000/yr maintenance × 5 years
$7500/yr × 5 years
1 physical therapist (PT), 1 Exercise specialist†, 2 aides
1 PT, 1 aide
1 PT, 1 aide
1 PT, 1 aide
Current and projected proportion of locomotor training in base case analysis
Current locomotor training offerings
Current Market Share per locomotor strategy/number of locomotor sessions*
Future Market Share per locomotor strategy /number of locomotor sessions
Facility A, 2752 locomotor sessions/year
33% / 908
31% / 881
16% / 440
11% / 303
51% / 1404
48% / 1293
10% / 275
Facility B, 3968 locomotor sessions/year
61% / 2420
56% / 2063
5% / 198
3% / 119
34% / 1349
31% / 1309
10% / 397
Facility C, 2256 locomotor sessions/year
34% / 767
30% / 677
33% / 744
30% / 677
33% / 744
30% / 677
10% / 226
Facility D, 3105 locomotor sessions/year
39% / 1211
10% / 311
51% / 1584
Table 2 reports BIA inputs and key assumptions concerning the device costs, training costs, and human capital costs associated with locomotor training in inpatient rehabilitation facilities. Where facilities provided more specific data for the analysis, we adapted the inputs accordingly. The duration of all training sessions was assumed to be one hour, including set-up time. We assumed all devices had an adoption rate of 50% across clinicians (i.e., the robotic device sits idle 50% of the time). The adoption rate represents the rate at which the new technology is accepted and the demand for the locomotor training device. We are using assumptions for training session time and adoption rate because RT-exo currently is not used for treatment across facilities.
Table 3 reports the changes in locomotor training strategies between the current and future market share of locomotor strategies within the respective hospital systems. The Appendix provides greater detail of BIA inputs with costs in 2017 USD.
Several parameters are likely to influence the RT-exo BIA, including device cost, the choice of locomotor strategy for substitution, and the efficiency of the use of RT-exo. One way sensitivity analyses will vary the following parameters: cost of robotic exoskeleton (50, 200%), adoption rate of robotic exoskeleton device (100–10%), training strategy substituted (highest cost training substituted, lowest cost training substituted), and exoskeleton device life (8 year device life – 3 year device life). A probabilistic sensitivity analysis was performed where uncertainty across all parameters was considered simultaneously. For the probabilistic sensitivity analysis, we considered the uncertainty around human capital cost. Finally, we also tested the assumption of varying conventional locomotor training strategies by 10% (without the inclusion of RT-exo) to assess the influence of varying conventional locomotor strategies on budget impact as scenario analysis.
Current and projected market share per locomotor strategy
Current locomotor training offerings
Current Market Share: costs* per locomotor strategy
Future Market Share: costs per locomotor strategy
A: 2752 locomotor sessions/ year
Net difference = $294,818 (future costs) - $297,643 (current costs), Savings $2825
B: 3968 locomotor sessions/ year
Net difference = $497,133 (future costs) - $501,916 (current costs), Savings $4784
C: 2256 locomotor sessions/year
Net difference = $295,754 (future costs) - $298,297 (current costs), Savings $2543
D: 3105 locomotor sessions/year
Net difference = $338,993 (future costs) - $340,107 (current costs), Savings $1114
We explored additional scenario analyses where, instead of choosing to introduce RT-exo, we shift market share to each of the conventional locomotor training strategies by 10%, e.g., in Facility A we assess how locomotor training costs change after we shift from 33 to 43% BWSTT. When stationary robotic and BWSTT are each increased by 10% locomotor training costs increase at all facilities. This is because of the higher personnel cost in BWSTT and higher device cost of stationary robotics relative to the overground training cost. When conventional over-ground training is increased by 10% it lowers costs at all facilities. Full scenario analyses are presented in Appendix 3. For all facilities, the savings from increasing market share to conventional overground training are greater than the savings from shifting market share to RT-exo.
The base case analysis shows small savings that are sensitive to cost or structural components of RT-exo. Despite the expense of RT-exo devices, savings related to a decrease in staffing needs and training for the therapeutic device exist. For hospital systems considering the addition of RT-exo as a locomotor training strategy, those with a higher percentage of human capital intensive strategies such as BWSTT may experience reduced costs if the introduction of RT-exo substitutes for these locomotor training interventions. However, the savings are very small with large degrees of uncertainty.
In the base case scenario, we assumed the robotic overground exoskeleton adoption rate would be 50%. This estimate may be too generous given two factors, the population eligible for RT-exo and clinician acceptance of new technology. RT-exo is a new technology that clinicians could choose not to use for locomotor training despite hospital purchasing decisions. Interestingly, the base case scenario was more sensitive to assumptions on clinical utilization parameters than device-specific parameters. For example, limiting the RT-exo useful life to 3 years (versus 5) only resulted in an increase in locomotor training cost per annum of between $3847 and $7611 across facilities. However, if RT-exo was adopted at a rate of 10%, i.e., the device is idle 90% of the time (as opposed to the 50% used in our model), annual costs of locomotor training would increase between $32,096 and $52,934 across facilities. When considering all sensitivity and scenario analyses the least costly alternative is when over-ground training sessions are increased by 10%. This makes intuitive sense because device costs and personnel costs are lowest. Over-ground training is a well developed and clinically integrated technology, therefore few efficiencies can be expected from changes in technological adoption and further integration into clinical practice. Conversely, further technological development stands to improve the business case for RT-exo but barriers in adoption and integration can impair its efficiency. Issues of technological adoption warrant additional attention.
Turchetti et al. provide a comprehensive overview of technology adoption with application to robotic devices in rehabilitation . The authors discuss barriers organized around technological, behavioral, organizational, and economic factors. Technological factors include ease of use, level of training necessary for caregivers to master the technology, and the anthropomorphic design of the product . Of particular note with technological barriers is the shift from roles once played by the clinician to the device that may create behavioral barriers to adoption. Learning new technologies requires effort on the part of clinicians, patient and caregivers, and resistance to change is a common phenomenon. Organization factors may be manifested in the need for redistribution of responsibilities and establishing a new organizational balance. Economic barriers include the dearth of cost-effectiveness data and the discordance between the timing required for purchasing decisions and the rate of change concerning reimbursement.
Another barrier to adoption is the different perspectives on the value added by the new technology suggesting the need for further economic studies that consider multiple perspectives in order to address the interests of all stakeholders . Swank and colleagues report on the feasibility of integrating RT-exo as a part of inpatient rehabilitation for an individual with SCI and illuminate on many aspects of RT-exo adoption identified above . Of note was the need for additional training for intra- and inter-physical therapist consistency with use of RT-exo despite being certified in use of the device. Working with hospital administrators and support staff was needed to address workflow issues such as optimizing device sizing and adjust settings prior to the rehabilitation visit.
We did not quantify several key considerations regarding RT-exo in this analysis. Hospital systems may invest in RT-exo as a means of building or maintaining their reputation as a leader in the use of cutting-edge technology. In this model, RT-exo is considered only as a substitute for locomotor training in the rehabilitation setting; however, there is potential for RT-exo to augment some community mobility. Devices such as RT-exo are likely to decrease in cost with greater competition in the market . Numerous robotic device companies are entering the market with several at substantially lower price points than the price weight used for this investigation, e.g., $30,000 instead of $150,000 . One device manufacturer anticipates being able to produce RT-exos at $10,000 – $15,000,  which would be comparable with the price weight used for our low cost OGT device. With the use of new medical devices, there is a device-operator interaction known as the ‘learning curve” effect [40, 41] during which inefficiencies are more likely; this is arguably the state that we are currently in with respect to RT-exo use in rehabilitation and we anticipate an improved business case in the future.
This project did not take into account differences in locomotor training effectiveness between alternatives. There is no strong evidence showing the superiority or non-inferiority of RT-exo. The evidence for the comparative effectiveness of robotic overground exoskeleton for locomotor training in SCI is limited, but several pilot studies show promising results [10, 23], and multiple clinical studies are underway or in development that will add to the evidence base (ClinicalTrials.gov Identifier: NCT03340792, NCT03477123, NCT02322125, NCT03443700). Therefore, the findings of this BIA must be interpreted with caution because we assume non-inferiority. In addition, there is limited data on the risks to users with SCI and personnel associated with locomotor training for RT-exo relative to other locomotor training strategies . The retrospective study design provides estimates for the number of individuals with SCI who received locomotor training therapies in 2017. Individuals with high cervical incomplete injuries who do not have upper extremity strength are not always eligible to use all RT-exo systems, but they are eligible to receive locomotor training with one RT-exo device and with other interventions. Therefore the volume of eligible patients may be lower than our estimates suggest; however, devices under development are designed to be used with individuals who have higher levels of SCI . We may have overestimated eligibility across therapy types due to varying eligibility criteria specific to each device (e.g., robotic exoskeleton device fit versus BWSTT fit criteria).
The assessment of economic efficiency-with robotic exoskeleton is in its infancy. Our model provides a framework upon which economic studies can build. Utilizing RT-exo for locomotor training may be an efficient use of hospital resources if key assumptions concerning efficacy hold true and the technology is adopted within the health system; however, a large degree of uncertainty surrounds our estimates as evidence for differences in the effectiveness of among approaches is limited.
DP, AJ, AWH conceived of the study, JB, SHC, SC, EFF, CF, CT, CKM, HT, AJ provided site specific data on utilization and costs, DP and MG performed all analyses and data interpretation, DP wrote initial draft of manuscript, All authors provided critical feedback on the manuscript. All authors read and provided approval of the final manuscript.
The National Institute on Disability, Independent Living, and Rehabilitation Research provided funding through the Midwest Regional SCI Model System (90SI5009), the Rocky Mountain Regional Spinal Injury System (90SI5015), the Southeastern Regional Spinal Cord Injury Model System at Shepherd Center (90SI5016), and the Texas Model Spinal Cord Injury System at TIRR Memorial Hermann (90SI5027).
Ethics approval and consent to participate
Consent for publication
The authors declare that they have no competing interests.
- 1.National Spinal Cord Injury Statistical Center. Spinal cord injury (SCI) facts and figures at a glance. Birmingham, Alabama: University of Alabama at Birmingham; 2019.Google Scholar
- 11.Food and Drug Administration. PART 890—PHYSICAL MEDICINE DEVICES Food and Drug Administration, HHS.; 2015 Tuesday, February 24, 2015.Google Scholar
- 12.Miller LE, Zimmermann AK, Herbert WG. Clinical effectiveness and safety of powered exoskeleton-assisted walking in patients with spinal cord injury: systematic review with meta-analysis. Med Devices (Auckl). 2016;9:455–66.Google Scholar
- 23.Sale P, Russo EF, Scarton A, Calabro RS, Masiero S, Filoni S. Training for mobility with exoskeleton robot in person with spinal cord injury: a pilot study. Eur J Phys Rehabil Med. 2018;54(5):745-51.Google Scholar
- 29.Neumannm PJ, Weinstein MC. The Diffusion of New Technology: Costs and Benefits to Health Care. In: Gelijns AC, Halm EA, editors. The Changing Economics of Medical Technology. Washington, DC: National Academies Press (US); 1991. p. 21.Google Scholar
- 35.R Core Team. A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2013.Google Scholar
- 36.Consortium YHE. Base Case Analysis [online] 2016 [Available from: https://www.yhec.co.uk/glossary/base-case-analysis/.
- 37.Swank C, Sikka S, Driver S, Bennett M, Callender L. Feasibility of integrating robotic exoskeleton gait training in inpatient rehabilitation. Disabil Rehabil Assist Technol. 2019;19:1-9.Google Scholar
- 39.Wilson M. A Budget Exoskeleton Allows Paraplegics To Walk–For The Price Of A Car2016 2/1/2016. Available from: https://www.fastcompany.com/3056049/a-budget-exoskeleton-allows-parapalegics-to-walk-for-the-price-of-a-car.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.