As health care expenditures continue to increase, health care providers are under increasing pressure to justify the value they provide with their interventions. This emphasis on value is seen in the transition of modern health care reimbursement strategies from fee-for-service health care models that were largely cost insensitive to fee-for-value models in which costs are heavily considered and compared against the associated value of the services. In the fee-for-service model, the provision of a service was justification for reimbursement. In the fee-for-value model, evidence justifying the need, potential benefits, and anticipated outcomes resulting from the service are key factors when reimbursement is considered.
Within this new standard of value-based health care, clinicians must quickly accept and understand that the value of their interventions is a product of their measured benefits divided by the cost of intervention (Figure 1).
Modern health care systems demand an objective demonstration of value. In a health care economy where reimbursement dollars are confined, they also foster an environment in which those interventions with the highest demonstrated values are reimbursed at the expense of those interventions with lower or poorly defined values.
A common means of increasing value in modern health care has been reducing its associated costs. A more daunting but ultimately more palatable means of augmenting the value of prosthetic care is to establish its measured benefits. Until this occurs, competitive increases in value can only occur with progressive decreases in cost, a pattern that continues to threaten the viability of providing consumers (i.e., patients) with the most appropriate prosthetic rehabilitation. The purpose of this manuscript is to present the basic concepts, definitions, and evaluation designs that constitute value-based health care economics.
Assessing the value of a health care intervention requires weighing its cost against some form of objective measurement of its benefit. Economic scientists have developed tools to better understand the importance of various domains of health (i.e., mental health, mobility, pain) to the individual or groups of people. This facilitates an understanding of the impact of an intervention, both in specific health domains and upon the overall health of an individual or group of persons. In other cases, the objective measurement of an intervention’s benefit may be made using outcome measures more familiar to clinicians. Many of the aforementioned elements may be included in an economic evaluation and facilitate shifting the discussion from cost to value.
The units of measure of an economic evaluation depend upon the type of analysis being performed. In cost-benefit analyses (CBAs), for example, the value of the intervention is measured solely in monetary units. In cost-consequence analyses (CCAs), value is assessed across a range of potential consequences potentially influenced by the intervention. In cost-effectiveness analyses (CEAs), benefit is measured in terms of the intervention’s efficacy (i.e., data from controlled clinical trials) or effectiveness (i.e., real-world data) on some target of shared interest such as fall risk or other determinants of health. A common mathematical analysis from CEA is the incremental cost-effectiveness ratio (ICER), which will be discussed in greater depth later in the manuscript. In cost-utility analyses (CUAs), benefit is measured against some measure of broad utility that is eventually converted to the accepted standard of quality-adjusted life years (QALYs), another metric that will be described in greater detail. Analyses of this type lend themselves to the comparison of different health care benefits associated with different interventions. These approaches, along with examples in prosthetic research, are described below and presented in Table 1. Although other forms of economic evaluations exist, their detailed description is beyond the scope of this manuscript. One example includes the cost minimization analysis (CMA) where intervention efficacy is considered equivalent and cost is the only difference. One obvious limitation to CMA is that cost is the key factor that disregards intervention efficacy, especially in nuanced situations.
ASSESSING VALUE THROUGH COST-BENEFIT ANALYSIS
One means of assessing value is through CBA. This type of analysis is a strict assessment of the money spent on a given intervention relative to the monetary value of the benefits accrued through the program, treatment, or intervention. This approach is relatively straightforward, but severely limited by its exclusive reliance on fiscal metrics and considerations. In a strict CBA, no attempt is made to place a value on the nonfinancial benefits experienced by the patient. This approach is not common in the evaluation of health care interventions outside of economics. The approach has been criticized within the health care literature due to this feature of assigning a monetary value to the quantity and quality of life.
ASESSING VALUE THROUGH COST-CONSEQUENCE ANALYSIS
As described previously, CCA may include analyses across an array of outcome measures and costs, but does so in disaggregated ways that preclude the use of ratio analyses that will be described in subsequent techniques. An example of a CCA is seen in the recent works of Dobson-Davanzo and their analyses of health care expenditures among matched cohorts of Medicare beneficiaries with lower-limb amputation who either did or did not receive a lower-limb prosthesis.1,2 A core consideration in their analyses suggests that the initial upfront costs of a lower-limb prosthesis are ultimately amortized over the ensuing 12 to 15 months due to the savings in health care costs associated with reduced utilization of acute care hospitalization and long-term hospital care.1,2 However, although this analysis suggests long-term cost savings through a reduction in certain elements of subsequent health care utilization, the additional value of the mobility associated with prosthesis use is not represented in this retrospective data analysis. The work also considers fractures, falls, and number of health care claims to further demonstrate differences between groups.
Other CCA examples in prosthetics were studied in a recent systematic review and included comparisons between limb salvage to amputation, transtibial socket types, and the provision of care in the public and private sectors.3
ASSESSING VALUE THROUGH COST-EFFECTIVENESS ANALYSIS
In CEAs, the value of an intervention is determined by weighing its costs against a natural unit associated with the intervention. Examples may include avoidance of injurious falls, altered biomechanics, and patient-reported quality of life.
In CEAs, outcome measures must be carefully selected to ensure that their use culminates in an accurate assessment of benefits that are meaningful to both the consumer and payer of the health care services. Avedis Donabedian, a pioneer in health care quality improvement, suggested that the measured benefits of health care can be well captured by examining the five Ds, namely, death, disease (with its signs, symptoms), discomfort (or pain), disability, and dissatisfaction.4 Within this general context, each field must examine their patient population and interventions to determine how their benefits will be best measured. In the case of prosthetic rehabilitation, the RAND organization recently suggested scrutiny of physical function, quality of life, and health outcomes including falls, musculoskeletal disease, and chronic illness among individuals with transfemoral amputation (Figure 2).5
In assessing the general domains of effectiveness, the field must identify common outcome currencies that are relevant to all of the economic stakeholders in prosthetic rehabilitation. In health care economics, these stakeholders include patients, providers, and payers. The value of an outcome measure is limited to the value each stakeholder places on the outcome relative to their perspective. Although this consideration begins with the identification of measures and benefits that are meaningful to the patient, such measures must also be valued by the payers and providers to ultimately influence health care decision making and policy. For example, in its comparative analysis of prosthetic knee types, the RAND organization considered the variables of reduced fall risks and fall rates, osteoarthritis of the sound side knee joint, and reduced health care expenditures,5 currencies that are accessible and valued to each of the three stakeholders in the provision of lower-limb prostheses.
Although it is the responsibility of providers to understand the impact of various prosthetic components and socket designs on the biomechanical functions of the individual patient, clinicians must recognize that many of the currencies used to assess biomechanics may not be accessible nor valued by payers or patients. Here, the organizational framework of the International Classification of Function (Figure 3) may prove helpful. Within this model, disability can be seen as affecting individuals at the levels of body functions and structures, activity, and participation. Within this framework, clinicians must recognize that the benefits that are the most accessible and valuable to patients and payers are often observed within the domains of activity (transfers, walking, and mobility) and participation (in work, home, and community settings). By contrast, the beneficial impacts at the level of body functions and structures that are not experienced or measurable at the activity and participation levels may not be understood or valued by the other necessary stakeholders of patients and payers.
For example, a clinical prosthetist and referring physician may determine that a patient will benefit from a prosthetic foot and ankle mechanism that expands the available range of motion at the ankle. However, if the impact of this component change at the level of body function and structure fails to create meaningful benefit to the patient’s day-to-day function and participation, its value may not be recognized by the payer or the patient. However, if the associated biomechanical benefits allow the patient to expand his or her mobility across previously inaccessible environmental barriers (i.e., activity) and participate in previously inaccessible roles and settings (i.e., participation), and if these improvements can be objectively measured through patient performance or patient report, the benefit of the intervention can begin to be weighed against its cost to determine its ultimate value to this patient.
Many of the basic quality measures that are being actively monitored in health care environments, such as length of hospital stay, hospital readmission rates, and mortality, are unlikely to be affected in a meaningful way by prosthetic management plans. It then falls to the field to either identify or develop those quality measures that are sensitive enough to measure the benefits experienced by the patient with variations in prosthetic design and function. Without such quality measures, the value of prosthetic care cannot be adequately considered. Should the field fail to identify or develop these quality measures, outside entities will select substitute measures based on experience with other practice areas that will likely lack the sensitivity required to measure the benefits of the intervention. In the absence of measured improvements in patient outcomes, reduced reimbursement remains the only means of increasing the value of prosthetic interventions.
ASSESSING VALUE THROUGH COST-UTILITY ANALYSIS
In CUAs, the benefits of a given intervention are reported by using one of several global quality measures. Rihn et al.7 defined ideal quality measures as those that are 1) patient centered, 2) that can be used in economic evaluation, 3) that can be communicated effectively among the stakeholders in health care delivery (i.e., payers, patients, and providers), and that 4) can be compared across other disease states. Measures that meet this description are often referred to as utility measures. These utility measures are then converted to QALYs, which can be used and considered across very different interventions.
The Euroquol 5-D (EQ-5D) is an instrument frequently used to generate utilities and QALYs.8 This patient-reported health outcomes questionnaire contains the five domains of mobility, self-care, daily activities, pain and discomfort, and anxiety or depression. Depending upon the version of the EQ-5D, the questions in each of these five domains have three possible responses of “no problems,” “some problems,” and “extreme problems” (weighted as 1, 2, and 3, respectively), or five possible responses of “no problems,” “slight problems,” “moderate problems,” “severe problems,” and “extreme problems” (weighted as 1, 2, 3, 4, and 5, respectively). In addition, participants may be asked to rate their overall health on a visual analog scale, although this component of the EQ-5D is separate from the five-item descriptive portion and is not needed to generate QALYs.
In addition to the EQ-5D, the SF-6D is a second commonly used utility instrument based on the more familiar SF-36. The SF-6D assesses the six health domains of pain, physical functioning, role limitations, social functioning, mental health, and vitality, each with four to six possible health states.9 The Health Utility Index (HUI) represents a third commonly used instrument to generate utility weights and QALYs.10 Alternatively, more familiar global health instruments such as the PROMIS measures and the SF-36 can be converted into the utility weights and QALYs defined previously.11
The use of CUA and quality measures in prosthetic rehabilitation has been infrequent, but such measures have been used in studies related to the provision of microprocessor-regulated hydraulic knee units (MPKs). Brodtkorb et al.,12 Gerzeli et al.,13 and Cutti et al.14 have reported upon the impact of MPKs on EQ-5D scores, reporting a cumulative mean improvement of 21%.5 Separately, Seelen et al.15 used the SF-36 utility measure to assess the utility of MPKs over non-MPKs, permitting a conversion into EQ-5D scores.5
Importantly, such global quality measures may not be sensitive enough to fully inform the impact of variants in prosthetic design on patient experience. As such, the profession may need to explore population-specific quality measures that are sensitive enough to measure the impact of prosthetic treatment variations that can be converted to more generic health utility scores. This could facilitate the execution of economic evaluations and compared across other disease states.
Quality-Adjusted Life Years
Utility measures are useful in informing the value of health care interventions because they can be used to calculate the QALYs associated with a given health care intervention. The QALY measures both the duration and quality of life experienced. Although historically used less in the health economics of the United States, it is a frequently used unit in health economic decisions in other developed countries. In cost-effectiveness research, the value of a given health care intervention is seen as the cost per QALY gained. QALYs are determined as the areas under the curve in a graph of utility (y-axis) against time (x-axis; Figure 4).
In QALY calculations, utility is measured on a scale from 0 (death) to 1 (perfect health). In the graphic example in Figure 4, the person has a health condition, such as a lower-limb amputation leading to a hypothetical decreased life utility of 0.6 as measured by one of the utility measures described previously, with a gradual decline in utility over time associated with aging.
An intervention that increases the life utility or increases the life expectancy associated with a given health condition may increase the QALYs for that individual, as seen by the increased shaded area in the resultant graph (Figure 5).
The work of the RAND group on transfemoral prosthetic rehabilitation includes an application of QALY analysis in prosthetic science.5 Analyzing relative fall risks, fall-related mortality, and osteoarthritis, the authors determined that in a simulation of 100 MPK users and 100 nonmicroprocessor hydraulic knee (NMPK) users over 10 years, the MPK users would experience 8.8 more years of life when compared with NMPK users, or about 0.09 life years per person over a 10-year period compared with NMPK users5 (i.e., extending the x-axis in Figure 5). In addition, when quality of life is considered, as measured by the EQ-5D and SF-36, further increases in utility (i.e., extending the y-axis in Figure 5), collectively yielded 0.91 more QALYs over a 10-year period for MPK users when compared with NMPK users.5
LIMITATIONS OF QUALITY AND UTILITY ASSESSMENTS
As the use of effectiveness and indirect utility assessments becomes more commonplace in clinical settings, the limitations of such scores should be understood. Meaningful effectiveness and utility measures are often assessed through the filter of the patient’s experience. Thus, the patient’s relationship with his or her provider will ultimately affect these assessments. Because of this relationship, both effectiveness and utility measures reflect the impact of the prosthetic service experience rather than an isolated component or socket design. Thus, the accessibility and responsiveness of providers will ultimately affect the outcome measures assigned to the patient experience.
Additional limitations and considerations include the Hawthorn effect and timing of assessment. The Hawthorn effect simply recognizes that variables that are measured are affected by the measurement process.16 Patients who are queried on their daily mobility, for example, become more aware of the day-to-day mobility and may, in turn, alter their mobility. Similarly, patients who are asked to score or describe their pain levels, socket comfort, or quality of life may become more aware of these elements in their personal experiences.
The timing of an effectiveness or utility assessment will also impact the ultimate responses. Quality assessments are often subjective in nature and reflect the patient’s prior experiences and current expectations. For example, Hafner et al.17 reported upon the affects observed when the quality and satisfaction observed with an NMPK before and after exposure to the functionality of an MPK. Their results represent the impact that exposures to new technologies can have on the performance standards and expectations harbored by prostheses users, as the satisfaction with the same NPMKs declined significantly after a period of sustained use of MPKs.
In addition to the complexities associated with defining the quality of a given intervention are the challenges of determining its cost. Direct costs are always more accessible and in the realm of lower-limb prosthetic rehabilitation and intuitively include the costs associated with the provision of a lower-limb prosthesis. However, additional costs are found in the physician visits, outpatient care, skilled nursing facility stays, therapy visits, home health visits, and acute hospitalizations that an individual with a lower-limb amputation might experience.1 Thus, the prosthesis represents a small portion of the overall direct health care expenditures experienced by an individual with a recent lower-limb amputation. As described earlier, the average costs associated with these additional health care services among persons with lower-limb amputation that do not receive a prosthesis ultimately exceed the initial procurement costs associated with the prosthesis.1 These findings underscore the principle that direct costs should be monitored over time to fully determine the impact of health care interventions.
A more comprehensive consideration of direct health care costs can be obtained through modeling. For example, the RAND effort calculated the relative risks of falls and injurious falls as well as their associated costs, citing an average direct cost of nearly $25,000 with major fall-related injuries and $1,300 with minor fall-related injuries. When the average number of falls experienced within a given treatment arm is multiplied across these figures, the direct costs associated with the use or disuse of a health care intervention can be reasonably projected. In the case of MPKs, the projected reduction in direct health care costs due to falls was determined to be $3,576 per person per year.
Indirect costs include those costs borne by the individual and society at large and are generally very difficult to determine. Lost wages, caregiving expenses, personally purchased assistive technologies, and transportation expenses represent commonly cited indirect costs. Using the earlier work of Gerzeli et al., the RAND effort determined an additional reduction of $909 in annual indirect costs with the use of an MPK.5
Total costs represent the summation of direct costs and indirect costs. In the case of persons with transfemoral amputation, RAND reports a reduced overall cost of over $4,500 when patients are fit with an MPK rather than an MPK.5
INCREMENTAL COST-EFFECTIVENESS RATIOS
Ultimately, economic evaluation involves a comparison of two or more courses of action in terms of both costs and consequences.7 Questions of effectiveness are informed by ratios comparing the cost of the defined treatment and the resultant benefits. Although benefits can be reported across a number of quality outcomes, including the common currencies described earlier, the most broadly accepted standard is the QALY. When the cost of two interventions is established, along with their projected effect on the QALYs of a target patient population or episode of care, an ICER can be determined using the following equation (Figure 6).
The resultant figure represents the relative financial cost per QALY gained with one treatment over another. In the case of persons with transfemoral amputation, when the increased procurement costs of MPKs are combined with the subsequent projected declines in health care utilization costs and compared against the augmented QALYs due to increased quality of life, the ICER of receiving an MPK instead of an NMPK was determined to be $11,606. Restated, it takes $11,606 to generate an additional QALY for an MPK candidate with a transfemoral amputation.5
Although ICERS are frequently used in CUAs as described previously, and expressed in terms of QALYs, ICERs can also be used in CEAs, in which the differences in experienced cost is divided by the differences expected in therapeutic effects between the interventions. An example is presented in Figure 7.
An example of this approach is seen in the recent comparison of two MPK systems to each other using the Physical Functional Performance 10 (PFP-10) assessment as a measure of the treatment effect.18
There are benefits and limitations to this approach. A benefit is that a clinical research team has the ability to use any outcome measure available to it to establish the difference in effect between two intervention approaches. A limitation is, however, that the outcome used may not be universally adopted. This could complicate the validation of findings by other study teams and also the comparison across populations if the outcome measure selected is population specific. This limitation is minimized through the use of utility measures that have been used across a broad range of treatment populations. For example, in the MPK example described previously, although the use of the PFP-10 represents the first application in individuals with transfemoral amputation, the measure has been used in multiple other diagnostic groups such as poststroke, frail elderly, cardiovascular compromised, and others.19
The use of ICERs allows some policy makers to make difficult decisions about what health care interventions should be provided. Those with the lowest ICERs (or lowest cost per QALY) are generally seen as the best investment of health care dollars. Although this model is rarely used in health care policy development in the United States, it is increasingly used in other developed nations including much of Europe. Health care systems may determine minimal ICER thresholds, or a range of acceptable ICER thresholds, to facilitate health care policy decisions.
Once calculated, ICERs can be visualized according to their slope on a coordinate scale in which utility (or quality) is measured along the x-axis and cost is measured along the y-axis (Figure 7). Within this scale, treatments in the upper left-hand corner, which cost more with decreased benefit, would never be seen as cost-effective. Conversely, treatments in the lower right-hand corner that cost less but provide increased utility would always be seen as cost-effective. More difficult decisions are made in those cases where increased utility is associated with increased health care cost in the upper right-hand corner of the scale (Figure 8).
Coverage of those health care interventions in the upper right-hand corner is ultimately determined by the ratio between the increased utility and increased cost as they relate to the payer’s willingness-to-pay threshold. Figure 8 hypothesizes two willingness-to-pay thresholds at $50,000/QALY and the comparatively steeper slope of a willingness-to-pay threshold of $100,000/QALY. Interventions plotted above a given willingness-to-pay threshold are unlikely to be covered, whereas interventions plotted below an organization’s willingness-to-pay threshold have an improved likelihood of coverage.
Although the upper right-hand corner in Figure 8 represents those interventions whose increased procurement costs are associated with increased benefit to the patient, the ultimate question of coverage is based on the relative ratios of cost to benefit and the slope of the payer’s individual willingness-to-pay threshold, both of which are illustrated in Figure 9. Experience dictates that different payers have different tolerances toward incurring additional health care costs for added health care benefits. The Department of Defense and certain Workers Compensation Health Plans have historically been more willing to pay for more costly, more promising interventions with emerging or less-established clinical benefits, a disposition characterized by a fairly steep willingness-to-pay threshold. By contrast, many private third-party health plans and State Medicaid policies are founded in cost-containment and are only willing to incur modest cost increases for clearly defined, superior clinical benefits, characterized by a much more gradual willingness-to-pay threshold.
Within this construct, manufacturers and providers alike must consider and ultimately establish the increased utility associated with promising interventions and weigh these against their likely cost increases. Treatments associated with modest increases in utility and extremely high procurement costs will exceed most if not all payers’ willingness-to-pay thresholds. By contrast, interventions associated with substantial increases in utility and modestly higher procurement costs will fall below the willingness-to-pay thresholds of more payers.
MERIT-BASED INCENTIVE PAYMENT SYSTEM
Many of the principles described throughout this article are evident in the federal government’s new Merit-Based Incentive Payment System (MIPS). Mandated for physicians in 2017 and extending to rehabilitation therapists in 2019, MIPS obligates clinicians who bill Medicare to use the regular collection of quality measures. These measures are collected at the beginning and end of treatment to determine the impact of the provided health care services on the patients treated. Once the changes in quality measures for each patient are risk adjusted against known predictive variables, participating clinicians or clinics are assigned a quality score. The aggregate quality scores of a given provider or clinic are included in that entity’s ultimate MIPS score.
Once Centers for Medicare & Medicaid Services has aggregated these MIPS scores from across its providers, a threshold performance score is established. Importantly, those clinicians whose MIPS score exceeds this threshold will be paid a bonus, whereas those whose MIPS scores falling short of this threshold will be fiscally penalized, with incentives and penalties ultimately reaching 9% of billed Medicare revenue.
In addition, these quality scores will be made accessible to payers and patients alike, facilitating a more transparent selection of health care providers based on the quality of their outcomes and costs of their services. Although the ultimate rollout and extent of MIPS across allied health is still largely undefined, the program in its current state exemplifies the increasing need for documenting the value of provided health care services.
ADDITIONAL CONSIDERATIONS AND RECOMMENDATIONS
Thus far, numerous economic study design models have been presented along with available prosthetic examples. This section serves to present selected additional factors to consider if contemplating the addition of economic analyses in tandem with other clinical research.
Clinical research in prosthetics has not commonly included the collection of costs as part of routine data collection. However, as outlined thus far in certain economic analytic designs, future prosthetic research should give strong consideration to collecting cost data to allow for more explicit economic analyses in prosthetic rehabilitation.
Similarly, the elongation of time horizons in clinical trials is necessary to more fully assess the value associated with prosthetic intervention. The recent review of transtibial economic studies shows that existing time horizons in the reviewed body of literature were on the order of days up to a year.3 Although optimal timelines for accommodation and training are the subject of debate, researchers should be mindful that if economic study is to be included, a defensible time horizon should be also be included.
Another issue in economic science is the changing costs of intervention over time and discounting. Discounting in many cases is situationally dependent by geographic region, population, individual practice groups, and other numerous other factors. Further, the impact of inflation (i.e., rise in price and fall in purchasing power of currency) must be accounted for depending upon the time horizon of the study period as needed to allow for optimal therapeutic effect.
Other selected factors to consider in the conduct of economic science in prosthetics include a description of assumptions and perspective(s). For instance, it may be assumed in a particular study that a subject is accommodated fully with an experimental prosthetic component at a certain time point and this accommodation may be used to justify a particular time horizon for the clinical trial. Further, it may be assumed that the etiology of a particular group of persons with amputation are clinically appropriate to receive the experimental intervention and that the intervention may continue to be in functional condition over a certain duration. This could be the basis for amortizing costs over a certain period or justification for added future costs for maintenance or repair to the device as outlined in study costs.
Regarding perspectives, it is important to study multiple stakeholders’ perspectives. For example, the providers will justifiably be concerned with upfront component costs, overhead costs, visit counts and duration, and other clinically related factors. The payer may, however, be concerned with the magnitude of treatment (i.e., therapeutic) effect relative to alternatives along with the duration of service life of the intervention options as well. Finally, the consumer may for instance be concerned with lost time from work, child and self-care costs, costs of worn-out or damaged clothing from devices, vehicle and home modifications, copayments, deductible payments, and other personal out-of-pocket expenses. In summary, it is prudent to review rating tools for economic research in advance of adding such a component to clinical outcomes.
Within modern health care climates of objectively verified quality outcomes and strict cost containment, those involved in prosthetic rehabilitation must begin to embrace a culture of quality measurement and improvement. Given that quality measures will ultimately be established to determine the value of prosthetic services, the field is encouraged to identify or develop quality measures that are sensitive enough to detect difference in performance across prosthetic interventions and to meet the common currency standards of being meaningful to providers, patients, and payers. In addition, global utility measures must be integrated into current and future prosthetic research to facilitate CUAs.
Moreover, treating clinicians will need to balance patient-centered care with efficient use of resources. Clinicians must expect more scrutiny of their resource utilization, anticipating that providers who supply high-cost services without improved outcomes will ultimately be penalized by providers and patients.