Cost-effectiveness of revascularization for limb preservation in patients with end-stage renal disease

Abstract

Background

Limb revascularization in patients with end-stage renal disease (ESRD) has been criticized because of the low rates of limb preservation and overall survival that characterize this patient population. We undertook a formal cost-utility analysis to evaluate the role of revascularization in the ESRD population.

Methods

A probabilistic Markov model was used to simulate the clinical outcomes and long-term outcomes after six different strategies for the management of nonhealing foot wounds in patients with critical limb ischemia and ESRD. All scenarios considered all-cause mortality and major amputation for failure of limb salvage. Parameter estimates of the costs, clinical events, and functional outcomes used in the model were derived from primary data or published literature. Costs are reported in 2011 U.S. dollars.

Results

Local wound care alone had the lowest long-term total cost of the management strategies evaluated; primary amputation had the highest. Purely endovascular intervention yielded the highest limb salvage rates. Endovascular intervention had a cost of $15,403 per additional year of ambulation beyond that by local wound care alone. Endo-vascular intervention had the potential for cost-savings (ie, better health benefits at lower cost) only with very high 1-year wound healing rates. The 5-year survival rates ranged from 17% to 34% in all management strategies.

Conclusions

Endovascular intervention may be a cost-effective alternative to local wound care alone for patients with ESRD and ischemic foot wounds, but with small marginal health benefits at considerable cost. Local wound care alone may be preferable to primary amputation.

Patients with end-stage renal disease (ESRD) and peripheral arterial disease (PAD) are at high risk for development of nonhealing foot wounds.^1,2 Two particular characteristics of the ESRD patient population make attempts at limb preservation through revascularization and subsequent wound healing very challenging, however. First, limb salvage attempts fail more frequently in the ESRD population than in the non-ESRD critical limb ischemia (CLI) patient population, often despite a patent bypass graft.^3–5 Second, the perioperative survival rate and the long-term survival rate of patients with ESRD are both significantly lower than those of the non-ESRD CLI population.^6,7 These challenges have led to some uncertainty about whether efforts to achieve limb preservation are worthwhile or cost-effective in the ESRD patient population.^3,8–11

Our goal for this study was to assess the health benefits and the total costs of various strategies used to manage PAD associated with nonhealing foot wounds (Rutherford category 5 chronic limb ischemia¹²) in patients with ESRD. Herein we describe this assessment, done in the form of a formal cost-utility analysis.

METHODS

Overall study design and definitions

The objective of this study was a formal assessment of the total costs and health benefits (ie, a cost-utility analysis) of strategies for management of nonhealing ulcers associated with severe PAD (ie, Rutherford category 5 ischemia) among patients with ESRD who are ambulatory and independently living at baseline. The following management strategies were considered: (1) local wound care, with selective major amputation as indicated; (2) primary major amputation; (3) revascularization with infrainguinal surgical bypass using an autologous vein conduit and subsequent endovascular intervention as needed to maintain or to restore patency of the bypass graft; (4) revascularization with infrainguinal surgical bypass using an autologous vein conduit and open surgical intervention as needed to maintain or to restore patency of the bypass graft; (5) initial revascularization with endovascular intervention with surgical bypass for failure of wound healing; possible endovascular revisions as needed for failure of the initial endovascular intervention; and (6) revascularization achieved purely through an initial endovascular intervention, then subsequent additional endovascular reinterventions as needed.

Clinical parameters used

The clinical parameters used in this study were based primarily on those identified in our background research and in the initial Model to Optimize Value in Ischemic Extremities (MOVIE) study.^13,14 Certain clinical parameter estimates were then modified to simulate outcomes specific to a patient population with PAD and ESRD. In general, estimates were obtained from studies specifically of ESRD patients with PAD and Rutherford category 5 limb ischemia whenever possible. When these were not available, studies of ESRD patients with PAD undergoing revascularization for CLI (either ischemic rest pain [Rutherford category 4] or nonhealing wounds [Rutherford category 5]) were used. No randomized studies specific to the ESRD/PAD population were identified; we therefore relied on meta-analyses of single-center studies, multi-institution observational studies, and data from large clinical databases (including data from the Vascular Study Group of New England [VSGNE] and the American College of Surgeons National Surgical Quality Improvement Project) to the extent possible. Data from single-institution observational studies were used only when estimates were not available from meta-analyses or multi-institution studies.

Numerous changes were made to the clinical estimates used in the general Rutherford category 5 patient population in the initial MOVIE study to reflect the clinical outcomes of a patient population with ESRD, severe PAD, and nonhealing foot wounds (Table I). All of the revised clinical outcome parameters were based on previously published studies of revascularization or foot ulcer outcomes in patients with ESRD and CLI. Specifically:

Table I.

Comparison of important parameters that were modified from the initial model of the general critical limb ischemia (CLI) patient population (see review¹³) to simulate an end-stage renal disease (ESRD) patient population

Parameter	General CLI population, %	ESRD population, %
Clinical events
Annual (baseline) mortality	11.8	21.0
Additional perioperative mortality rate associated with
Major amputation	3.9	16.4
Surgical bypass	2.6	10.8
Endovascular intervention	2.6	8.1
One-year limb salvage rate associated with
Local wound care	62.9	54.5
Surgical bypass	89.2	78.6
Endovascular intervention	89.2	78.6
Annual rate of major amputation during first year after
Local wound care	38.0	27.9–48.1
Surgical bypass	2.6	2.0–3.0
Endovascular intervention	2.6	2.0–3.0
Functional outcomes
Proportion remaining ambulatory at 1 year after
Major amputation	55.5	37.9
Surgical bypass	97.1	75.7
Endovascular intervention	97.1	75.7
Proportion remaining independent at 1 year after
Major amputation	92.0	63.8
Surgical bypass	98.6	80.4
Endovascular intervention	98.6	80.4

The full range of values used in the probabilistic model was determined by various distributions, which included beta, gamma, and triangular.

• Annual baseline mortality rate was increased from 12% to 21% to simulate the survival rates of 53% at 2 years and 28% at 5 years reported by Albers et al.⁶

• The perioperative mortality rate for surgical revascularization was increased from the 2.6% reported in the Bypass vs Angioplasty in Severe Ischaemia of the Leg (BASIL) trial to 10.8% as reported by Gajdos et al.⁷ The periprocedural mortality rate for endovascular intervention was estimated to be 75% of this, or 8.1%. The perioperative mortality rate of major amputation was estimated as 16.1% on the basis of estimates from a risk-adjusted study of the perioperative mortality for surgical revascularization and primary amputation.¹⁵

• The 1-year limb salvage rate for endovascular intervention and surgical revascularization was decreased from 90% to 79% as reported by Albers et al.⁶ When patients who died within the first year after having undergone revascularization and subsequent major amputation were included, this reflects a total major amputation rate of 33% during the initial year, increased from the 11% rate used in the initial MOVIE study. The major amputation rate beyond the first year after revascularization was increased from 3% to 6% per year.⁶

• The probability of remaining ambulatory after revascularization was decreased from 97%^16,17 to 76% on the basis of data of ESRD patients undergoing revascularization from the VSGNE.¹⁸

• The probability of independence (ie, maintaining the ability to live independently) after revascularization was decreased from 98%¹⁷ to 64% on the basis of the VSGNE data,¹⁸ and the proportion of patients needing a temporary stay in a skilled nursing facility after revascularization was increased from 29% to 38% on the basis of the VSGNE data.¹⁸

• The 1-year wound healing rate after a successful (patent) surgical and endovascular revascularization was decreased from 95% and 60%, respectively, to 50% and 50% on the basis of single-center reports by Johnson³ and Aulivola.¹⁹

• The probability of remaining ambulatory after major (transtibial or transfemoral) amputation was decreased from 55%¹³ to 38%.²⁰ An additional deterministic sensitivity analysis was performed in which this probability was maintained at 55%.

• The 1-year wound healing rate with local wound care alone was decreased from 41% to 20%. The 1-year wound recurrence rate with local wound care was increased from 61% to 80%, and the major amputation rate at 1 year was increased from 38% to 50% on the basis of a large review by Gershater.²¹

The total (direct and indirect) inpatient costs used in the model were based on previous work by our group.¹⁴ We used previously published outpatient cost estimates.¹³ All cost estimates (obtained in 2009 U.S. dollars) were converted to 2011 U.S. dollars on the basis of inflation data from the U.S. Bureau of Labor Statistics.

Model design and analysis

A probabilistic Markov model was used to formally assess the relationship between long-term total costs and health benefits. A 10-year time horizon was used to balance the lifetime time horizon typically recommended for cost-utility analyses²² with both the limited survival that characterizes this patient population and the dearth of clinical parameter estimates extending past 5 years. The analysis was performed with 1000 trials of cohorts having 1000 hypothetical patients. In brief, hypothetical patients with ESRD and Rutherford category 5 CLI (foot wounds associated with significant PAD) entered the model at time zero and underwent one of the management strategies mentioned before. All patients were at risk for limb loss and all-cause mortality at any point in time. Patients progressed through various clinical states for 10 years unless death occurred before this time point. Differential costs and health benefits were tallied throughout this time period (Supplementary Fig, online only). Costs and utilities were discounted at a standard 3.5% annual rate.²² The cost (in 2011 U.S. dollars) of providing one additional year of ambulatory ability (ie, 1 year of ambulation) beyond that provided by local wound care alone (strategy 1) was the primary cost-utility analysis in this study; this was measured by incremental cost-effectiveness ratios (ICERs).²³ Median survival times and total costs were calculated secondary end points. Quality-adjusted life-years was not an end point used in analyses of this study. Deterministic sensitivity analyses—sequential analyses in which a parameter is varied over a set (determined) range—were used to investigate the influence of certain variables on overall costs and health benefits.²³

RESULTS

Base case scenario: clinical outcomes, health benefits, costs, and ICERs

The median 5-year survival rates for the six management strategies were 24.4% to 26.4% over the 1000 trials, with an interquartile range of 22.7% to 27.9% among the strategies (Table II). This corresponded to median survival times of 2.4 years for the bypass and primary amputation management strategies, 2.5 years for the endovascular-first management strategies, and 2.7 years for the wound care strategy. The median 5-year limb salvage rates ranged from 64.8% to 65.2% for the revascularization strategies (strategies 3–6) over the 1000 trials, with interquartile ranges of 57.9% to 75.6% for these strategies. The median 5-year limb salvage rate for local wound care (strategy 1) was 14.1% with an interquartile range of 12.6% to 15.2%.

Table II.

Projected clinical outcomes, health benefits, and costs of various strategies used to manage peripheral arterial disease (PAD) with nonhealing foot wounds (Rutherford category 5 ischemia) in a population with end-stage renal disease (ESRD)

Strategy	Median 5-year survival, %	Median 5-year limb salvage, %	Median years of ambulation	Median total costs, 2011 U.S. dollars
Wound care alone (strategy 1)	26.4	14.1	1.71	118,086
Primary amputation (strategy 2)	24.4	0	1.19	152,426
Initial endovascular intervention; repeated interventions as needed (strategy 5)	25.4	65.0	1.93	121,478
Initial endovascular intervention; surgical bypass ± revisions as needed for failure (strategy 6)	24.9	65.2	1.87	124,696
Initial surgical bypass; surgical revisions as needed (strategy 3)	24.4	65.0	1.82	128,517
Initial surgical bypass; endovascular revisions as needed (strategy 4)	24.4	65.0	1.82	126,487

The median number of ambulatory years is listed in Table II. The endovascular strategies (strategies 3 and 4) had the highest median ambulatory years, followed by the surgical revascularization strategies (strategies 5 and 6) and local wound care (strategy 1). Primary amputation (strategy 2) had the lowest median ambulatory years (Table II). Thus, in a cost-utility analysis of cost per additional year of ambulation, purely endovascular intervention (strategy 5) had an ICER of $15,403 per additional year of ambulation over local wound care alone (strategy 1) (Fig). Initial endovascular intervention with surgical bypass for failures (strategy 6) had an ICER of $40,594 per additional year of ambulation over local wound care alone (strategy 1). The surgical bypass strategies (strategies 3 and 4) produced very small marginal benefits above local wound care at much higher costs, resulting in ICERs for each of these strategies that exceeded $70,000 per additional year of ambulation. Primary amputation (strategy 2) was both more costly and less beneficial than local wound care (“dominated” in the typical terminology of cost-effectiveness analyses).

Fig.

Scatterplot comparing the differences in long-term total costs (x-axis) and differences in median ambulatory years (y-axis) between purely endovascular intervention (strategy 5) and local wound care alone (strategy 1) in the base case scenario.

Deterministic sensitivity analyses

The base case scenario assumed the probability of ambulation after a major (above-ankle) amputation to be lower in the ESRD population than in the general CLI population (see Methods). A deterministic sensitivity analysis was done to assess the impact of assuming the probability of ambulation to be no different in the ESRD and general CLI populations. In this analysis, local wound care (strategy 1) and endovascular intervention (strategy 5) both resulted in 2.0 median years of ambulatory ability, higher than all remaining strategies. As local wound care had a lower total cost ($50,807 vs $93,278) in this scenario, the management strategy of local wound care appears to dominate (ie, provide at least as much health benefits at lower total costs) all other management strategies as the probability of successful ambulation with a limb prosthesis of this patient population approaches that of the general PAD/CLI patient population.

An additional deterministic sensitivity analysis was performed to evaluate the impact of improved wound healing rates after surgical revascularization. Surgical revascularization did not approach cost-effectiveness or cost-savings at any wound healing rate. The ICER of endovascular intervention decreased as the 1-year wound healing rate increased; cost-savings (ie, improved health benefits at lower cost) was achieved when the wound healing rate exceeded 65% at 1 year.

DISCUSSION

In a previous study, we found surgical bypass to be the most cost-effective alternative to local wound care for patients with PAD and nonhealing wounds of the foot (Rutherford category 5 chronic limb ischemia). As this previous analysis was based on a general CLI population similar to that in the Project of Ex-Vivo vein graft Engineering via Transfection III (PREVENT III) trial, it assumed a population with an approximately 12% incidence of ESRD. Compared with the typical CLI patient, however, ESRD patients with nonhealing foot wounds have long been recognized as having a clinical problem that is especially challenging to resolve. Indeed, the higher perioperative mortality rate, the low long-term survival rate, and the high rate of progression to amputation—often despite a patent bypass graft—have led some previous authors to suggest primary amputation as the best management option for patients with ESRD, PAD, and nonhealing foot wounds or gangrene.^3,8–10

The results of the current analysis suggest that local wound care and endovascular intervention would be options that are more cost-effective than primary amputation. Consistent with findings from the BASIL trial,^24,25 endovascular intervention may be preferable to surgical revascularization because of the limited survival seen in this patient population. In a report of their experiences with endovascular intervention for patients with ESRD, Aulivola et al reported results comparable to those projected by our model. In particular, although the overall technical success rate was high, four of 15 (27%) required repeated intervention. Only four of 14 (29%) receiving purely endovascular intervention healed, whereas three (21%) remained unhealed and seven (50%) eventually proceeded to major amputation. Indeed, the one previous cost-effectiveness analysis and previous studies or reviews with commentary all appear to be consistent with the results of our study in suggesting that the outcomes of revascularization in this patient population are poor but that revascularization seems to provide benefit over primary amputation.^9–11,26

In our analysis, endovascular intervention was the most cost-effective alternative to local wound care. Although it may be an acceptable alternative, we note that this strategy for revascularization is much more expensive and much less beneficial in the ESRD patient population than in the general CLI population. Specifically, the median projected long-term (10-year) total cost of endovascular intervention was $89,040 in the general CLI population¹⁴ and $121,478 in the ESRD population of this study. In the general CLI population, endovascular intervention produced a median of 4.62 years of ambulatory ability, about 2.9 additional years more than the strategy of local wound care alone.¹⁴ In the ESRD population of the current model, endovascular intervention produced 1.92 years of ambulatory ability, only about 0.22 additional years (just more than 2 additional months) more than the strategy of local wound care alone. Therefore, compared with the strategy of local wound care alone, endovascular intervention cost $6960 per additional year of ambulation in the general CLI population but $15,403 per additional year of ambulation in the ESRD population.

Local wound care alone (“wound palliation”) with major amputation as needed may be the next best option to consider. There may be several potential benefits of this expectant form of management over primary amputation. First among the benefits might be the possibility of avoiding the perioperative risks associated with ESRD. The perioperative mortality rate for patients with ESRD and CLI undergoing surgical revascularization was 10.8% in a recent American College of Surgeons National Surgical Quality Improvement Project series,⁷ a strikingly high rate but consistent with several single-center studies of this patient population.^3,8,9 In addition, local wound care may avoid or at least delay the costs of limb loss, including an impaired quality of life, functional limitations, and monetary costs of long-term nursing care. The projections of our model suggest that the strategy of wound palliation may be associated with significantly decreased total monetary costs and improved functional status compared with primary amputation; thus this strategy may be an option to discuss in situations in which endovascular intervention is not feasible.

There are several potential limitations to this analysis. Whereas many studies have reported survival, limb salvage, and patency rates for ESRD patients undergoing revascularization, there are few data reported describing wound healing rates specific to the ESRD patient population. Although limb salvage rates are commonly reported, wound healing is an important cost driver after revascularization.¹⁴ Varying the post-revascularization wound healing rates that were assumed in this model did have an impact on the results. At the assumed 1-year wound healing rate of 50% used in the base case scenario, endovascular intervention had a reasonable cost-effectiveness ratio. Below this rate, endovascular intervention would be much less appealing, whereas the possibility of cost-savings (equal or better health outcomes at lower cost) was not seen unless the 1-year wound healing rate exceeded 65%, a scenario that we might submit is unlikely in clinical practice in the ESRD patient population. More data, however, are very much needed to further clarify this point as very little currently exists.

Likewise, functional outcomes (namely, probability of ambulating with a limb prosthesis) specific to an ESRD patient population after major amputation have not been described. Considering the high incidence of severe comorbidities in this patient population, it may be safe to assume that functional outcomes after major amputation would be worse in the ESRD population than in the general CLI population. As the postamputation functional outcomes of the ESRD population approach those of the general CLI population, however, the benefits of limb salvage attempts decrease. This appears to be the reason that the strategy of local wound care dominates endovascular intervention (ie, becomes as beneficial but at lower costs) as the functional outcomes of the ESRD population approach those assumed for the general CLI population in the first deterministic sensitivity analysis of the current study.

There are additional limitations posed by our methodology. As in the model of the general CLI population,¹⁴ the current model assumes a patient population that is independently living and ambulatory at baseline and has arterial disease that is amenable to effective treatment with either endovascular or surgical revascularization. The proportion of patients with ESRD, PAD, and nonhealing foot wounds meeting these baseline assumptions is not known. The ICERs in this study are done in a pairwise fashion with local wound care as the comparator, however, so eliminating some of the management strategies (for a patient who does not have endovascular options, for example) would not affect the estimates of ICERs. Finally, as previously discussed,¹⁴ the clinical outcomes, functional outcomes, and outpatient cost parameters used in this model were based on a thorough review of previous literature.¹³ The inpatient cost estimates were obtained from our single-center analysis using a transaction cost accounting system.¹⁴ Similar to generalizing the results of a randomized trial or observational study, the extent to which the findings of this model are generalizable at any given medical center are therefore dependent on the extent to which the given medical center has comparable clinical outcomes, functional outcomes, and total costs (not charges²³).

Legislation in the United States has prohibited the insurers from making reimbursement decisions on the basis of cost-utility analyses that assess cost per additional quality-adjusted life-year as the primary measure of cost-effectiveness.²⁷ The current analysis has focused on cost per additional year of ambulatory ability to use a patient-centered functional outcome rather than quality-adjusted life-years. At this time, there are no established or recommended “thresholds” of cost per year of ambulation to suggest when various interventions should or should not be pursued, however. The results of this analysis may demonstrate some noteworthy themes that perhaps should be considered in clinical practice, but we emphasize that the decision of whether to proceed with any management strategy should continue to be made on a patient-by-patient basis. The decision should include discussion with the patient and be inclusive of the preferences of the patient for various outcomes as well as his or her acceptance of various risks.

CONCLUSIONS

Endovascular intervention appears to be a cost-effective alternative to local wound care alone for patients with ESRD, PAD, and nonhealing foot wounds. Primary amputation does not appear to be a cost-effective option in this patient population. More data are needed on the functional outcomes and wound healing rates in patients with ESRD and nonhealing foot wounds undergoing revascularization, local wound care alone, and major amputations.