A Cost Analysis of a Living Skin Equivalent in the Treatment of Diabetic Foot Ulcers

Introduction It is estimated that 15 percent of patients with diabetes will develop foot ulcers and that 15 to 20 percent of those will progress to lower-extremity amputation.1,2 Indeed, it has been reported that foot ulcers precede 85 percent of all nontraumatic, lower-extremity amputations.3 The cost of treating diabetic foot ulcers (DFUs) is reported to range between $4,000.00 to $8,000.00 per ulcer episode and almost $28,000.00 over the first two years after diagnosis.1,3 The attributable cost of amputations is estimated at between $20,000.00 to $60,000.00.2,3 These figures motivate current efforts to develop cost-effective wound healing treatments, such as a living skin equivalent (LSE)* for DFUs. The Apligraf Diabetic Foot Ulcer Study (ADFUS) demonstrated the efficacy of LSE as an adjunctive therapy for DFUs that are resistant to conventional therapy.4 LSE is a bioengineered living skin equivalent that consists of dermal and epidermal layers containing human keratinocytes and fibroblasts. The purpose of the current study was to evaluate the cost effectiveness of a standard dressing regimen plus LSE versus a standard dressing regimen alone for treatment of DFUs using ADFUS pivotal trial data. Because both treatment groups in the pivotal trial were treated with similar dressing regimens, the addition of LSE was cost additive. However, it was hypothesized that when the clinical benefit associated with LSE was incorporated into the cost-effectiveness equation (the incremental cost per ulcer-free month gained and the incremental cost per amputation/resection avoided), the resulting cost-effectiveness ratios would fall within an acceptable range of cost effectiveness relative to other medical interventions. Methods Using the ADFUS data, we computed incremental cost-effectiveness ratios of the differences in the mean costs of medical resources associated with LSE versus standard dressing care to differences in selected clinical outcomes between the treatment groups. Specifically, we examined the incremental cost per amputation/resection avoided for LSE plus standard dressing care relative to standard dressing care alone, and the incremental cost per additional ulcer-free month for LSE relative to standard dressing care. The time horizon for the cost-effectiveness analysis covered the period of the clinical trial from randomization (baseline) to six months follow up. Costs were assigned to all nonprotocol-driven, ulcer-related clinical events, regardless of whether they differed significantly in occurrence between groups. These were identified as economically important clinical events. These included ulcer-related adverse events (AE), nonprotocol-scheduled office visits, and debridement procedures performed during office visits. We also included nonulcer-related adverse events if they differed significantly between groups. In the pivotal trial, diarrhea was the only nonulcer-related adverse event that differed in frequency between the treatment groups (p 5 Because many private payers in the US look to Medicare in establishing their own cost schedules, this approach should be relevant to their concerns. We assigned costs to the AEs based on the corresponding procedures and treatment regimens recorded. With respect to the medications, costs were estimated from Drug Topics’ Redbook, published by the Medical Economics Company. The average wholesale price (AWP) of the medications as well as LSE treatment were discounted by 20 percent to reflect a more accurate estimate of actual drug and product costs.6 Statistical Approach Patients randomized into the LSE or control group in the pivotal clinical trial served as the primary population in the analysis. Baseline demographic and clinical characteristics, the mean number of clinical events that were assigned costs, and total costs were compared between the LSE and control groups (Tables 1 and 2). The Mann-Whitney U test and t-test were used when appropriate to compare continuous variables. Categorical variables were compared using the Fisher’s Exact test. Variables that differed significantly at baseline between treatment groups were controlled for in a linear regression model regressing total costs (transformed using the natural log) on treatment group. Clinical outcomes compared between groups included the percentages of patients undergoing amputation/resection, the mean number of amputations/resections per patient, and the duration of ulcer-free months. These outcomes were compared between groups using the Fisher’s exact chi-square test and the Mann-Whitney U test for categorical and continuous variables, respectively. The clinical analysis identified 22 patients (15 control and 7 LSE patients) who underwent amputations or resections. The amputations incorporated in our analysis included only those that were indicated for the ulcer site. The duration of ulcer-free months was measured based on whether the ulcer was open or closed at each study visit and whether the ulcer had recurred. Information about ulcer closure was incomplete for patients who did not return for all scheduled visits. When calculating ulcer-free time over the six-month study period for these patients, we assumed that if the patient’s ulcer was open at the patient’s last protocol visit, then it remained open from this date to six months follow up. If the patient’s ulcer was closed at the last protocol visit, then we assumed it remained closed from this date to six months follow up unless an ulcer-related AE occurred during this period. If an ulcer-related AE occurred after the patient’s last visit date, it was assumed that the ulcer was open from the last visit date to six months follow up. Cost-Effectiveness Ratios The cost-effectiveness analysis examined the incremental cost per amputation/resection avoided in the LSE group relative to the control group. The numerator of this equation was the difference in mean total costs between treatment groups reflected by the incremental costs associated with the study treatment (LSE) and nonprotocol scheduled services occurring during the clinical trial. The denominator was the difference in the mean number of amputations/resections per patient. We also examined the incremental cost per percentage decrease of patients undergoing amputations/resections as observed in the trial. Lastly, we evaluated the incremental cost per additional ulcer-free month in the LSE group relative to the control group. The difference in mean total costs between treatment groups served as the numerator in the equation, and the denominator was the difference in the duration of mean ulcer-free months between treatment groups. Utilizing the differences in mean total costs and mean outcomes is consistent with the economic theory underlying cost-effectiveness ratios.7 Sensitivity analyses were performed to assess uncertainty about the external validity and generalizability of our findings. Key variables were varied to evaluate the impact on the incremental cost-effectiveness ratios. Specifically, the number of LSE applications and the use of median vs. mean ulcer-free time were varied. We also performed subgroup analyses of the baseline ulcer duration ( 2 months) and the presence of Charcot’s disease. Results All patients who were randomized and exposed to study treatment were included in the cost-effectiveness analysis. Thus, the study population included 112 LSE patients and 96 active control patients. The mean age of the population was 56.8 years, 69.2 percent were Caucasian, and 18.3 percent had clinical characteristics that were consistent with Charcot’s disease. The patients’ study ulcers had a mean ulcer area of 2.9cm2 and were open for an average of 11.3 months. Differences in baseline demographic and clinical characteristics were not statistically significant between the LSE and control groups with the exception of body mass index (BMI) (30.9 versus 33.1, respectively; p = 0.026) (Table 2). Because the trial was a randomized study, we expected that the control and LSE ulcers would have similar characteristics. Our baseline comparisons between groups helped to confirm that randomization was effective. Economically important clinical events. Almost twice as many control patients experienced a severe AE compared to LSE patients (19.8% vs. 10.7%, respectively; p = 0.080). Differences in the mean number of AEs per patient and AEs between the LSE and control groups in which a trend was observed (p Cost-effectiveness findings. The total mean costs per patient were significantly higher in the LSE group compared to the control group ($7,366.00 vs. $2,020.00, respectively; p 1,2,4,9–11 Our findings from the sensitivity/subgroup analyses suggest that the cost effectiveness of LSE may improve in routine practice and among patients without Charcot’s disease. The substantial difference in the median number of ulcer-free months observed among ulcers less than two months old resulted in more favorable cost-effectiveness ratios, decreasing more than 50 percent in the clinical trial and the routine daily practice settings. This early application benefit may play a larger role in avoiding amputations/resections thus offsetting the cost associated with using LSE.2,12 For the linear regression model evaluating total costs, we transformed the costs, which typically are skewed, using the natural log. This is a commonly used approach so that the cost data will have a more normal distribution and can be used in linear regression. The primary independent variable in the model was treatment group (LSE vs. control), which is a categorical variable. Because the patients were randomized into one of these two groups, it was expected that the baseline characteristics would be similar between groups and would not have to be controlled for in the model. Our baseline testing showed that only body mass index was significantly different and thus was added to the model but did not significantly influence the difference in total costs observed between groups. This analysis was limited to the six-month trial period and did not include quality-of-life assessments. It would be useful to evaluate the projected costs of treatment and associated quality-of-life issues with poor wound healing and subsequent amputation/resection.3,13,14 Other research has shown the beneficial impact of prevention and wound care programs that may impact on the observed similarities in ulcer recurrence.2 There was a high number of applications observed in the trial relative to current observation in clinical practice and in some cases an entire unit of LSE may not be used for an application but cut and the pieces used for different applications. In addition, many of the study sites may have been introduced to LSE for the first time when the pivotal trial started and, thus, these findings in part may reflect the learning curve for physicians. The findings from this study demonstrated that treatment of DFUs with LSE might be cost effective especially if outcomes observed in the clinical trial are similar to those observed in routine clinical practice. As experience with LSE increases, it would be useful to perform further economic research exploring a longer time horizon and routine clinical practice outcomes associated with LSE. *Apligraf (Novartis Pharmaceuticals, East Hanover, New Jersey) **Dermagraft (Smith & Nephew Inc., Largo, Florida)