A cost-utility analysis on two treatment strategies [systemic therapy with propranolol oral solution versus systemic therapy with corticosteroids] was conducted to assess the costs and clinical benefits of propranolol in comparison to corticosteroids in the treatment of proliferating IH in Italy. The cost-utility results were reported in terms of incremental cost per QALY gained and expressed by Incremental Cost-Utility Ratio (ICUR).
The results of economic evaluation were expressed in terms of resource consumption and in terms of clinical effectiveness measured by QALYs gained. Therefore the primary outcome was the incremental total costs per QALY that expresses the additional costs implied by adoption of the innovative treatment with propranolol oral solution to gain an additional life year in perfect health. Then, the primary outcome of analysis was expressed in quantitative terms by incremental cost-utility ratio, exploring the cost-utility of propranolol oral solution in the treatment of proliferating IH requiring systemic treatment.
The Italian National Health Service perspective was adopted to evaluate the resource consumption with a life-time (30 years) time-horizon, by applying a 3.00% discount rate to both costs and health benefits as indicated in Italian guidelines for Economic Evaluation of Health Programs .
Probabilistic sensitivity analysis was performed to assess the uncertainty around the ICUR.
The probabilistic sensitivity analysis was performed by varying the model input, modelling the transition probability as beta-distributions and the costs as gamma-distributions for 1,000 iterations.
A mixed decision tree—Markov model was developed in order to describe the pathway of infants with proliferating IH requiring systemic treatment. The model, shown in Fig. 1, is divided in 3 macro-phases representing the evolution of IH and the current practice in management of the affected infants: “Active treatment phase”, “Stabilization and involution phase” and “Post-involution phase”. The Active treatment phase comprises three health states: “Success”, defined as a complete or nearly complete resolution of lesions; “No Success”, defined as lack of complete or nearly complete resolution of lesions; “Drop-out”, defined as the discontinuation of trial treatment. The other two phases comprise two health states: “Success” and “No success”.
The first phase represents the proliferation period of hemangioma, in which it is necessary to initiate the active therapy (propranolol or corticosteroids). In this phase the transition from “No Success” to “Success” health state depends on the clinical efficacy of the pharmaceutical therapy (Table 1).
Following the active treatment period, the second phase of the model focuses on spontaneous involution of IH and lasts from the age of 1 year until the age of 6 year, as demonstrated in a retrospective analysis . During the involution phase the model utilizes the transition from “No Success” to “Success” health state which can occur through spontaneous involution of the hemangioma or through active intervention of sequelaes (e.g. surgical resection, laser therapy, or a combination of both). Worsening in the involution phase (transition from “Success” to “No Success”) is not to be expected, due to clinical experience.
In the “Post-involution phase” the residual lesions do not resolve spontaneously anymore because the involution process is complete. From this time onward, resolution can only occur through active intervention, and it is further assumed that any active treatment will lead to a complete success.
The model was developed with Microsoft Excel.
Clinical data source
The model was populated with clinical efficacy data which define the transition probability from one health state to another. The clinical efficacy data related to propranolol treatment were obtained from the pivotal head-to-head trial versus placebo published in 2015 in the NEJM . The clinical effectiveness data related to corticosteroids treatment were derived from systematic literature review, more specifically from two meta-analyses of observational studies conducted by Izadpanah et al. and by Bennet et al. [1, 3, 15, 20].
The main clinical outcomes taken into account in the model were success rate (i.e. the complete or nearly complete resolution of IH), drop-out rate (i.e. the discontinuation of trial treatment) and rebound rate (i.e. the reintroduction of systemic hemangioma treatment after therapy with propranolol or corticosteroids). Furthermore, the model incorporates the ulceration (i.e. complications due to proliferation of hemangioma) and adverse events (i.e. complications related to pharmaceutical therapy) related to the different therapeutic approaches. Table 1 indicates all main clinical outcomes that define the transition probability between health states.
Moreover, the model assumes that during the “Stabilization and involution phase” in 50% of patients, for whom pharmacological treatment was not completely effective, hemangioma spontaneously involutes [21, 22]. Whereas in 20% of patients an active intervention, like laser treatment, surgery or a combination of both, is necessary to remove residual hemangioma . Table 2 shows treatment modality distribution for residual lesions in “Stabilization and involution phase” and in “Post-involution phase”. In the “Post-involution phase” the model assumes that residual lesions were presented in 69% of patients .
Since no utility value was estimated for the condition of IH (corresponding to the health state “No Success”) or is available in literature, the utility estimate for atopic dermatitis was selected as a proxy for IH utilities. The used utility data were extrapolated from a work done by Monti at al. that estimated the utility in an Italian sample of children (aged between 1–12 years) affected by atopic dermatitis using the Infants Dermatitis Quality of Life Index (IDQoL) and the Children’s Dermatology Life Quality Index (CDLQI).Footnote 1 In accordance with results of this study, the level of utility associated with a patient affected by IH, using atopic dermatitis as proxy, was set equal to 0.76 (adjusted on scale from 0 to 1) during the proliferative phase and modelled according to the evolution of the disease .
Moreover, in the estimation of the quality of life, the decrease of utility due to the use of corticosteroids was taken into account. Indeed, the administration of these drugs has a greater impact on utility perceived by patients and their families than the administration of propranolol, due to the higher incidence of toxicity and side effects of corticosteroids . The decrease of utility was quantified equal to 3%, in accordance with a study conducted in UK that assessed the Utility in patients with atopic dermatitis treated with corticosteroids compared to a less toxic treatment .
Resource use and costs
The evaluation of resource consumption was conducted on the basis of the clinical pathway of infants, modelled in the decision model and based on interviews done with key opinion leaders experienced in the treatment of IH. The interviews were focused on dosage and length of pharmaceutical therapy, treatment setting (outpatient, day-hospital or inpatient care) and the frequency and type of monitoring visits in follow-up plan.
The economic evaluation considered only direct medical costs associated with drug acquisition, hospital admissions and outpatient visits. The direct non-medical costs such as travelling; waiting periods and indirect costs were not included in the analysis according to the Italian National Health Service perspective adopted.
The estimation of the pharmaceutical costs was performed on base of ex-factory price of the respective drugs, as this represents the maximum cost for the public structures. The cost of propranolol was fixed at €180.50 (3.75 mg/mL, oral solution for paediatric use, bottle of 120 mL. The cost of corticosteroids was obtained from “Compendio Farmaceutico Telematico-Farmadati 2013” equivalent to €1.04 (5.00 mg, tablets). While the dosage and the duration of therapy were determined according to the data collected in the interviews with expert opinion.
The resource consumption for hospitalization was estimated following the DRG-based reimbursement system, using version 24 of DRG-Grouper and the national tariffs established by Ministerial Decree of 18th October, 2012 . The hospital admission for the first administration of the drug, necessary for both evaluated clinical pathways, was coded by DRG 284 “Minor skin disorders without complications” to which a tariff equal to €153.00 in day hospital setting is associated . The first administration of the therapy was considered in day hospital setting also for corticosteroids according to the opinion of Clinical Experts of three Reference National Centres of this pathology, based in northern, central and southern Italy, considering the distinctive features of patient under one year of age and the well-known adverse event related to the corticosteroids treatment. The hospitalization for the surgical resection of residual lesions was coded by DRG 120 “Other acts on cardiovascular system” to which a tariff equal to €1,898.00 in day hospital setting is associated . While the estimate of the cost related to laser therapy to remove residual lesions was performed according to DRG 270 “Other acts on skin” and to the number of sessions necessary to remove the lesions (requiring 3 sessions on average). The total cost associated with laser therapy resulted in €3,297.00 .
The resources associated with outpatient visits for the follow-up of the pharmaceutical treatment were estimated on basis of National Outpatient Tariffs established by Ministerial Decree of 18th October, 2012 . In particular, the follow-up plan for the patients treated with propranolol involves a dermatological visit scheduled monthly, to which a tariff equal to €20.66 is associated. While for the patients treated with corticosteroids the follow-up plan is more intensive, consisting of a dermatological and a general outpatient visit conducted monthly, to which a tariff equal to €20.66 for both visits is associated.
The model also included the representation of ulceration and adverse events resulting from pharmaceutical therapies (the ulceration is not generally related to the therapy but it is a potential complication of IH most frequently during the 3–4th month, potentially requiring laser and weekly outpatient visits to prevent infection). The costs associated with treatment of ulceration were estimated with reference to the current clinical practice, that involves two sessions of laser treatment performed in day-hospital setting and coded by DRG 270 “Other skin procedure without complications”, to which a tariff of €1,099.00 is associated. In succession, four dermatological outpatient visits are performed to monitor evolution of the ulceration. On the base of this information, the total cost of managing ulceration was estimated at €2,280.64.
The costs associated with adverse events related to propranolol treatment were quantified assuming that a hospital admission in day hospital setting is necessary to treat the complication. This hospitalization was coded by DRG 284 “Minor skin disorders without complications,” to which a tariff equal to €153.00 is associated. For the management of complications resulting from the use of corticosteroids the costs were estimated by the average of costs related to the most frequent adverse events, like: hypertension caused by corticosteroids, interruption of growth, glaucoma, obesity, ‘moon face’ . This estimate was conducted taking into consideration additional costs for pharmaceutical treatments, hospital admissions and outpatient visits, and amounts to €799.49. Table 3 reports details about all cost items considered in the model associated with the two treatment scenarios.
A probabilistic sensitivity analysis was conducted to assess the uncertainty around the ICUR and the probability to be cost-effective at a given cost per QALYs threshold.
In the probabilistic sensitivity analysis the transition probability was modelled as beta-distributions, while the costs were modelled as gamma-distributions, in accordance with method of moments estimation .