Cost-Effectiveness Studies in Diabetes-Related Foot Ulceration: To Model or Not to Model, That Is the Question

Foot ulcers and amputations are common in patients with diabetes-related foot disease and/or chronic limb-threatening ischemia (CLTI).¹ These ulcers and amputations cause a large burden for patients in the loss of quality of life, and for society in associated health care costs.^2,3 To reduce this burden, insight into interventions that help increase quality of life and/or decrease costs is needed. Such insights can be obtained from health-economic studies. Unfortunately, primary health-economic studies (where data are prospectively collected in participants) in the field of diabetes-related foot disease are scarce. As an alternative, health-economic modeling studies are performed. However, these come with challenges that are not always met.⁴ 

While Markov modeling in health-economic studies is a valid and widely used method with clear practical benefits over having to prospectively collect all data, there are some important drawbacks. These drawbacks are universal, but some are especially important in the field of diabetes-related foot ulceration. Health-economic studies are a combination of either prospectively collected data or assumptions based on existing data from other studies related to costs, clinical effectiveness, and quality of life outcomes;⁵ all 3 will be discussed here. The aim of this paper is to showcase potential shortcomings in health-economic modeling studies in diabetes-related foot disease, and to provide the reader with the knowledge to better assess and understand those studies.

Costs in healthcare are nonlinear, and this also applies to the treatment of foot ulcers in diabetes. The mean cost for 1 episode of care for 1 foot ulcer has been shown to be around €10,000 (US $11,523).^3,6,7 However, all studies show large variations around this mean. These differences are caused by various factors:

• A foot ulcer can be superficial and small or deep and large; while both are defined as a “foot ulcer”, the differences in treatment costs are large (for example, the Eurodiale study showed mean treatment costs of €4500 [US $5185] for ulcers without infection or ischemia, €9500 [US $10,947) for those with either infection or ischemia, and €17,000 [US $19,590] for those with both infection and ischemia).⁸
• Hospitalization and amputation may be required during treatment after a foot ulcer develops, which have major effects on costs. Hospitalization (without amputation) costs already exceed costs of almost all other aspects of foot ulcer treatments, and hospitalization with amputation is even more costly. For example, when an amputation is needed in the treatment of a foot ulcer, cost increases from €10,000 (US $11,523) per episode to more than €25,000 (US $28,808) per episode, without considering the costs after amputation, such as prosthesis prescription and extensive rehabilitation.³ 
• Indirect costs are also important but not always included. For example, most patients with a foot ulcer are age 65 and older, and retired or unemployed. A smaller but still significant population is in the workforce; this population may be growing in numbers, with epidemiologic studies showing an increase in incidence of foot disease in younger men.⁹ When a person is unable to work because of a foot ulcer, costs related to productivity loss can easily surpass thousands of Euros.⁷ Furthermore, productivity losses are just one type of indirect costs, which may include family or caregivers’ productivity losses, time and expenses, and out-of-pocket expenses. The large extent of these indirect costs was recently shown in a 2025 report on CLTI.¹⁰

These 3 factors show why reliably modeling the costs of foot ulcer treatment is extremely difficult: baseline ulcer characteristics have an impact on outcomes but are often unknown in modeling studies, hospitalization and amputation are elective choices in treatment that are hard to predict, and most modeling studies exclude societal costs (such as productivity loss) to avoid the uncertainty, thereby simplifying their models and missing important cost-related factors (Example 1).

In modeling studies, 2 types of clinical effectiveness are seen: assumed or known from a randomized controlled trial (RCT). A pitfall when assuming clinical effectiveness is an overly optimistic assumption, such as those outlined in Examples 2 and 3. To avoid this, many modeling studies use data from an RCT. While that may include various metrics relevant for cost calculations, RCTs do not inherently capture all treatments. For example, emergency department presentation, hospitalization (without amputation), and amputation are not automatically captured within an RCT. Unless all treatment parameters are prospectively included in the trial design, using data from an existing RCT still requires making (inherently uncertain) assumptions about clinical outcomes. 

Studies have consistently shown that people with a diabetes-related foot ulcer have a reduced quality of life compared to those who have never had one and compared to those who have healed from a foot ulcer.²² These findings have formed the fundamental basis for many modeling studies. However, surprisingly few RCTs have actually measured quality of life prospectively. Of 4 prospective health-economic studies identified, 2 (50%) found quality-of-life effectiveness to be different from the hypothesis (Examples 4 and 5). 

Regardless of the findings in Examples 4 and 5 being an exception or not, one can learn from both examples that this would not have been known had quality of life not been measured prospectively. Therefore, when modeling quality of life, less plausible scenarios should also be considered, or threshold scenarios to determine the boundaries of cost-effectiveness. Reality is more complex than a simple model can capture.

Additionally, the differences in quality of life between the various states (eg, noninfected, infected, hospitalized without amputation, minor amputation, major amputation) vary widely between and within studies.^7,26,27 These uncertainties showcase that effectiveness on quality of life should preferably be prospectively measured. If modeled, it should be done with caution, taking different scenarios into account.

Where possible, in our opinion the answer to the Shakespearean question is “not to model” but to measure prospectively in a primary study. Setting up a prospective health-economic study reduces bias, reduces the uncertainty related to the myriad factors that play a role in the treatment outcomes of diabetes-related foot ulcers, and provides the benefits of prospective registration.²⁸ However, modeling studies may still be helpful, for example, when prospective data collection is impossible, when performed adjunctive to such data collection, when performed with multiple scenarios (including those less plausible), or when performed as hypothesis-generating. If you do come across such a study, it can be suggested to consider asking yourself some of the 16 questions specified in the Figure. For more extensive critical appraisal, we suggest readers familiarize themselves with available tools, such as those from NICE²⁹ and the Joanna Briggs Institute,³⁰ or with any of the reporting tools available on the Equator Network.³¹

Health-economic studies are an important tool in the assessment of the effectiveness and implementation of interventions in the management of diabetes-related foot disease. However, this disease is characterized by large variations between patients, treatments, and outcomes, along with large uncertainty in estimates of costs and effects. Primary studies are therefore preferred over modeling studies when assessing the cost-effectiveness of interventions in this field. When modeling studies are performed, readers should remain critical and carefully assess all assumptions before conclusions on cost-effectiveness can be accepted. It is our hope that this paper will motivate researchers to perform primary health-economic studies, improve their health-economic modeling when such a design remains needed, and help readers to better assess those studies when published.

© 2026 HMP Global. All Rights Reserved.

Any views and opinions expressed are those of the author(s) and/or participants and do not necessarily reflect the views, policy, or position of the Journal of Critical Limb Ischemia or HMP Global, their employees, and affiliates.