Cost-effectiveness of Becaplermin for Nonhealing Neuropathic Diabetic Foot Ulcers
Foot ulcers are a common complication of diabetes. People with diabetes have a 15% lifetime ulcer risk1 and an estimated community prevalence of 2% to 5% of the diabetic population.2-5 Peripheral neuropathy and vascular disease are the primary factors contributing to ulcer development,6,7 which along with infection contribute to difficulty in healing. Ulcers failing to heal require daily dressing changes; involve functional limitations, inconvenience, and worry; and may result in costly and occasionally devastating consequences such as gangrene, infection, and amputation.1
Appropriate management can prevent or heal foot ulcers; thereby, reducing the risk of amputation.8 Best clinical care includes treating the ulcer cause and providing local ulcer care.6,9,10 Vascular insufficiency, if present, should be corrected with bypass or angioplasty. Patients should be followed to ensure proper diabetic control, systemic treatment of infection, proper foot care, and local pressure offloading with appropriate orthotic, casting, or non-weight bearing regimens. Local ulcer care includes active surgical debridement at frequent intervals (this alone has been shown to increase healing rates11) and provision of a moist interactive wound-healing environment.10,12,13
Even with appropriate management, ulcers may become chronic and fail to heal. Becaplermin, a platelet-derived growth factor in a hydrogel vehicle, stimulates cellular growth and migration of granulation tissue, thereby promoting healing. In a 20-week, double-blind multicenter phase III clinical trial, patients with diabetes, adequate vasculature, and infection-free chronic ulcers that did not probe to bone were randomized to receive best clinical care with once-daily applications of 100 microgram/g becaplermin gel (N=124 patients) or placebo vehicle gel (N= 127) for 20 weeks or until healing. Patients using becaplermin had a greater chance of 100% ulcer closure by 20 weeks (62 of 124 patients, 50%) than patients receiving vehicle gel alone (44 of 127 patients, 35%, 43% improvement; P = 0.007). For patients whose ulcers did close, time to healing was shorter with becaplermin (35th percentile 86 days versus 127 days, P = 0.013), while groups did not differ significantly in adverse events or recurrence.11,14 In a post-hoc analysis, patients who responded to treatment (30% ulcer closure) after 2 months of treatment showed a much higher probability of healing by 20 weeks (62% of responders healed versus 1% of non-responders).
In the current healthcare environment, decision-makers need clinical effectiveness and cost information. The objective of the current study, based on economic evaluation guidelines,15,16 was to estimate incremental cost-effectiveness (additional cost for each ulcer day averted) associated with adding becaplermin to best clinical care for the treatment of non-healing ulcers in patients with adequately controlled diabetes, adequate oxygen supply to surrounding tissue, and no infection or nonvitalized tissue in the ulcer.
A 1-year modeled evaluation was conducted using a decision-tree framework based on data from the 20-week, phase III randomized clinical trial11,14 and follow-up data on healing past the treatment phase. Results of best clinical care practices alone were compared to best clinical care with the addition of once-daily applications of 100 microgram/g becaplermin. Best clinical care of an ulcer included treatment of the ulcer cause and its associations, as well as local ulcer care (involving frequent active debridement and daily dressing changes). In the best clinical care alone segment, hydrogel was used as the primary wound dressing. In the best clinical care plus becaplermin segment, patients responding at 10 weeks (30% closure of the ulcer) continued use of becaplermin until healing or 20 weeks, whichever occurred first. For non-responding patients, becaplermin was discontinued after 10 weeks and replaced by the use of hydrogel. Hydrogel was used as the comparator in the model because it is a commonly used primary wound contact layer for granulating wounds free of infection and was the control dressing in the clinical trial.
The evaluation, based on the clinical trial population, is relevant to adult patients with diabetes mellitus type 1 or type 2 and a foot ulcer which: 1) has been present for 8 weeks or longer, 2) is full-thickness (to subcutaneous fat), 3) is free of local wound infection, cellulitis, or osteomyelitis, and 4) has been debrided to remove callus, necrotic debris, and slough. Arterial circulation in the leg with the ulcer was assumed sufficient to support healing. Such patients would typically have had diabetes mellitus for approximately 15 years, with neuropathy as a contributing cause to development of the ulcer.
This evaluation was structured using a decision-tree framework. Information on effectiveness and safety (days with an unhealed ulcer and risk of infection) from the 20-week randomized clinical trial of becaplermin was modeled to extend to 52 weeks to capture relevant clinical and economic benefits. Clinical practice for treatment of an ulcer or infection and associated health resource use was estimated by a Delphi panel of health professionals actively involved with continuing education in this area and representing the range of disciplines providing such care. Health resource costs were calculated assuming a societal perspective as recommended by international guidelines.15,17,18 The decision analytic model (see Figure 1) was used to structure the estimation of the average number of ulcer-days per patient averted by adding use of becaplermin gel to best clinical care. The model is divided into two major time periods, 0 to 20 weeks and 21 to 52 weeks. To estimate the average number of days with an unhealed ulcer up to 20 weeks for each treatment group, reported proportions of patients showing 100% ulcer closure and the average time to achieve healing during the 20-week clinical trial were used.
Although the clinical trial evaluated use of becaplermin for 20 weeks or until healing, in usual care it is expected that becaplermin would be discontinued in patients who are not responding. Prescribing information for becaplermin includes a recommendation that if the wound has not decreased in size by approximately 30% after 10 weeks of treatment or complete healing has not occurred in 20 weeks, continued use of becaplermin should be reassessed. The current evaluation is based on a treatment recommendation that non-responding patients stop becaplermin after 10 weeks of treatment, and costs and effectiveness in the current evaluation were adjusted for reduced treatment.
For the 21 to 52 week time period, it was assumed that unhealed patients in both treatment segments received the same treatment (moist wound care using a hydrogel) and achieved the same healing rate (55% of patients with open wounds after 20 weeks healed after 1 year of care). This rate was estimated using ulcer survival data from the 40-week, open-label extension study of the clinical trial,14 which, in turn, was extended to 52 weeks by fitting a Weibull curve, the appropriate statistical distribution to model survival data. In addition, ulcers were allowed to recur during this time period and the data were estimated based on recurrence data from the clinical trial,14 supplemented by the expert panel estimates that a recurrent ulcer would have duration of 12 weeks. The incidence of infection in the model was based on the proportion of patients with infection in each treatment group in the 20-week clinical trial (5.7% in the becaplermin and 7.1% in the best clinical care group had severe or potentially limb-threatening infection and 23.6% of patients in groups had a localized infection). Although the difference in risk of severe infection was not statistically significant, it was used in the model as the best available estimate for the first 20 weeks. Data were extended to 1 year, assuming that the risk of infection continued for the remainder of the time that patients had an unhealed ulcer and, therefore, were at risk of infection. Becaplermin is associated with lowering of infection risk by reducing the number of days with an ulcer.
Treatment and health resource use. Because no health resource utilization information related to the use of becaplermin has been collected prospectively, a modified Delphi method was used to collect information on treatment practice and health resource utilization. The panel of health professionals currently providing care for patients with diabetic foot ulcers in Ontario included six registered nurses (including two enterostomal therapists with wound care expertise and one nurse practitioner), two chiropodists, two dermatologists, two infectious disease specialists, one endocrinologist, and one family practitioner. A surgeon was not included since patients in the original study did not have osteomyelitis or a Charcot deformity that might require surgical management.
Panel members completed a questionnaire regarding their individual practice in caring for patients with foot ulcers in their area of expertise - ie, health professionals designated as primarily responsible for care responded to questions on assessment of ulcer etiology, frequency of wound care visits for consultation and follow-up, assessment/debridement, referral to other specialties, and weight offloading; while those providing moist wound care completed relevant components of the questionnaire, and physicians treating infection estimated the infection-related components of care. A neutral facilitator experienced in this therapeutic area guided a central meeting, where members reviewed a summary of all responses, reached a consensus as to the meaning of each question, and then provided a final response about their treatment of patients with foot ulcers. Resource use was estimated using the average of responses by members providing each component of care.
The panel population is representative of the healthcare providers likely to prescribe and use becaplermin. The care described by the panel may be better than typical wound care because panel members treat a high proportion of patients with diabetic foot ulcers and are well-trained in this area. However, selecting a panel composed of opinion leaders was deemed appropriate to ensure that modeled resource use matched the high quality care provided in the study that supplied the clinical data for the analysis.
Cost estimation. Cost per patient for care of unhealed ulcers was estimated as the cost per ulcer day multiplied by the number of days with an unhealed ulcer plus the probability of localized infection and severe or potentially limb-threatening infection multiplied by the cost of treating each infection type. A fixed cost per patient was added for special devices and weight offloading equipment.
Societal perspective. The societal perspective includes services of physicians and other health professionals, hospitalization, community supportive services, supplies and devices, as well as patient out-of-pocket costs for travel to receive healthcare and time lost from paid employment. A sensitivity analysis, in addition, included societal cost for time lost from all usual activities.
Unit prices. Unit prices were obtained from standard sources, including the Ontario Schedule of Benefits for physician services and fully allocated costs of the Hamilton Health Sciences Corporation for hospital services. Prescription pharmaceuticals were priced according to the Ontario Drug Benefit (ODB) formulary with appropriate mark-ups and dispensing fees. Becaplermin price was obtained from Janssen-Ortho Inc. and, for this analysis, a conservative estimate of average waste of one-half tube per patient was assumed. The Canadian average industrial aggregate wage was used to calculate the value of lost time.
Costs were estimated in 1998 Canadian dollars (1998 mean exchange rate, $1.00 Cdn = $0.67 U.S.) using prices from a single province (Ontario) and updated to 2002 costs using the Canadian Consumer Price Index for Personal and Health Care.
Sensitivity analyses. Sensitivity analyses were conducted by varying some of the parameters in the model to address the impact of possible uncertainty in the best information available on clinical benefits and treatment costs. One key parameter was the reported proportion of patients healed within 20 weeks. Because this parameter was obtained within a clinical trial setting, the proportion of patients to heal in clinical practice might be different than that observed. The proportion of patients who healed with becaplermin by 20 weeks (49% in the base case) ranged between 40% and 60%. For the proportion of patients who healed between 20 weeks and 1 year, the base case estimated that 55% of patients unhealed at 20 weeks would heal by 1 year based on open label follow-up data. Alternative values explored were that the proportion of those unhealed at 20 weeks to heal by 1 year might be as low as 45% or as high as 60%.
Costs of health resource use were increased by 50% and decreased by 50% to investigate the impact of health resources being either more or less costly than in the base case. Because home care visits to provide dressing changes formed a large component of the relevant cost, the impact of the cost of these visits was explored by increasing the cost by 25% and decreasing the cost of such a visit by 25%.
In the base case, the cost of time lost from work was included as a societal cost, and for a sensitivity analysis, the cost of time associated with not being able to perform usual unpaid activities was also included as a relevant cost to society.
The reported clinical study results and projected clinical consequences over 1 year (see Table 1) show, on average, 26 fewer ulcer-days and a small reduction in the proportion of infected wounds in patients treated with best clinical care plus becaplermin.
Based on the treatment and health resource use estimated by each Delphi panel member, unit prices were applied and a cost for components of the care were estimated. The societal viewpoint (see Table 2) showed increased wound interface material costs and cost savings for local wound care, treatment of ulcer cause, and weight offloading associated with reduction in ulcer treatment days required. Reduction in ulcer days also resulted in a reduced risk of infection and savings related to infection treatment. The final incremental annual cost when using becaplermin is $167 per patient. Dividing this incremental cost by the 26 ulcer-days averted results in an incremental cost-effectiveness ratio of $6 per ulcer-day averted (see Table 3).
Key cost and effectiveness components were varied in sensitivity analyses to test model robustness (see Table 3). Becaplermin efficacy (14% above placebo in the base case) varied from 5% above placebo (judged to be the minimum clinically important difference) to 25% above placebo (60% healing observed in an open-label, uncontrolled trial). Results were sensitive to becaplermin efficacy and to the rate of healing with best clinical care alone after 20 weeks. Results were also sensitive to an increase in the cost of home care visits and to the inclusion of a societal cost value for all time lost from usual activities.
The decision-analytic model estimated the 1-year effectiveness and cost-effectiveness of adding becaplermin to best clinical care. From the societal perspective, best clinical care with becaplermin was more effective and slightly more costly than best clinical care alone. Over the year following the start of becaplermin treatment, its usage resulted in 26 fewer days with an open ulcer at a cost to society of $6 for each ulcer day prevented. This cost for an ulcer day averted should be helpful to decision makers who want to incorporate cost-effectiveness findings into their decision making.
These results apply to patients like those treated in the clinical trial used for this model - ie, patients with adequate diabetic control who are presenting with an ulcer that extends to the layer of subcutaneous tissue that is free from infection and nonvitalized tissue and has adequate oxygen supply to surrounding tissue. Furthermore, the results apply to becaplermin used with regular surgical sharp debridement, moist wound healing, and regimens provided to reduce pressure and mechanical stress on healing tissue. Where patients and treatment do not meet these criteria, costs and outcomes may not equal those reported.
The clinical trial data used included patients with ulcer duration of 8 weeks who had not necessarily received best clinical care before the start of the study. Becaplermin may be expected to be even more cost-effective where best clinical care has been tried for at least 8 weeks before treatment and where the patient is motivated to adhere to pressure offloading regimens and dressing changes. For patients whose foot ulcers have been thoroughly assessed to ensure they do not probe to bone, the probability of osteomyelitis is greatly reduced19,20 and the effectiveness and cost-effectiveness of becaplermin is expected to be improved.
The decision-analytic model was based on data from a randomized, controlled 20-week, clinical study which provided evidence for the efficacy of becaplermin and a non-controlled study in which healing rates with becaplermin were as high as 60%.21 After becaplermin treatment, and considering the common healing rate of 55% was derived from treated patients, this may be an overestimate of the healing in usual care over 1 year. However, this is a conservative estimate; when a lower common rate of healing after 20 weeks was assumed, becaplermin was more cost-effective.
This analysis is based on a 10-week becaplermin treatment recommendation for patients who are not responding to treatment because the expert panel estimated that for patients who did not heal, treatment would continue for an average of 50 days and because the clinical study data showed that patients who had not responded had a very low chance of healing within 20 weeks. If treatment with becaplermin were to be continued for all patients until healing or 1 year, the cost per ulcer day averted is greatly increased ($85 per ulcer day averted).
The possibility that treatment response after 2 months of becaplermin is less predictive of healing by 20 weeks was also investigated. In the clinical trial, an optimum criteria for terminating becaplermin was identified to be less than 30% wound closure at 10 weeks (non-healing ulcer). Using this threshold, 62% of patients who met the criteria were healed by 20 weeks while only 1% of the others were healed. These healing rates for responders and non-responders were used in the model base case. If, in general practice, only 56% of responding patients heal by 20 weeks (halfway between 50% healing for entire becaplermin group and 62% for responding patients), 14 ulcer days would be averted and the cost per ulcer-day averted would be $72. If becaplermin also were used on recurrent ulcers after 8 weeks unhealed, the incremental cost per ulcer-day averted would be $12 due to increased treatment cost.
When infection probability was estimated from clinical trial data (pooling treatment groups rather than using the rate of infection in each group separately) and reduced infection risk over the remainder of the year was estimated based on fewer ulcer days alone, the cost per ulcer-day averted was $20.
An important limitation of the current evaluation is that it is a model with clinical results from one 20-week trial extended to a full year combined with associated costs based on a Delphi technique. However, extending the data from 20 weeks to 1 year was necessary because savings resulting from early healing could not be realized within 20 weeks. Also, health resource use data were not available from clinical studies. However, "real world" best clinical care rather than protocol-driven care may be more relevant in clinical practice. The use of the Delphi technique with health professionals actively involved with continuing medical education in this area and representing the range of disciplines that provide such care is expected to represent this care.
Over a 1-year period, the addition of up to 20 weeks of becaplermin to best clinical care resulted in an important clinical benefit of 26 fewer ulcer-days per patient, along with an increase in total treatment cost. From the societal perspective recommended by most methodological guidelines,15-17 the incremental cost-effectiveness ratio is $6 per ulcer-day averted.
Best clinical practices plus a trial of becaplermin gel implemented into a community setting for use in nonhealing neuropathic diabetic foot ulcers of the type in the clinical trial could offer a significant opportunity for improved patient outcomes. This study provides the information required by decision-makers wishing to include cost-effectiveness data in treatment decisions. Decision-makers in other healthcare systems can use this modeled approach as a basis to investigate the cost-effectiveness of becaplermin in their setting.
The authors wish to acknowledge the clinical expertise contributed by Expert Panel members.
Panel members included: Ms. Patricia Barton, Toronto, Ontario; Ms. Karen Campbell, London, Ontario; Ms. Kim Clarke, Toronto, Ontario; Ms. Sunita Coelho, Toronto, Ontario; Ms. Maureen Cooling, Toronto, Ontario; Ms. Patricia Coutts, Mississauga, Ontario; Mr. Ted Hyder, Toronto, Ontario; Dr. Anne Kenshole, Toronto, Ontario; Dr. David Lawee, Toronto, Ontario; Dr. David McNeely, Toronto, Ontario; Ms. Betty McTague, Georgetown, Ontario; Dr. Krystyna Ostrowska, Mississauga, Ontario; Dr. Kim Papp, Waterloo, Ontario; and Dr. Gary Sibbald, Toronto, Ontario.
These nurses, chiropodists, and physicians provided valuable input regarding treatment practice patterns in the Canadian setting.
The authors thank Janssen-Ortho Inc. for an unrestricted grant to fund this study.