Social costs of icatibant self-administration vs. health professional-administration in the treatment of hereditary angioedema in Spain

Background Icatibant is the only subcutaneous treatment for acute Type I and Type II hereditary angioedema with C1-esterase inhibitor deficiency (HAE-C1-INH) licensed for self-administration in Europe. Aim To compare the economic impact of two icatibant administration strategies: health professional-administration only (strategy 1) versus including the patient self-administration option (strategy 2). Methods Economic evaluation model based on the building of a decision tree. Both strategies are assumed to have equivalent effectiveness. The payer (Spanish National Health System) and the social perspectives were considered. All relevant cost-generating factors were taken into account. The time horizon was one year. Sources of information included scientific evidence, official data and experts’ opinion. A deterministic sensitivity analysis was carried out to quantify the underlying uncertainty in the model. Results From the social perspective, which considers both direct (health care costs) and indirect costs (productivity losses), strategy 2 would result into average savings of €121.30 per acute attack compared to strategy 1. For Spain, this would achieve in an annual savings of €551,371. The reduction in direct costs accounts for 74% of the savings and lower indirect costs account for the remaining 26%. Savings per acute attack may range from €79.50 to €169.80; accordingly, the annual savings in Spain may vary between €90,319 and €2,315,360. Conclusion Costs related to the management of acute HAE attacks with C1 inhibitor deficiency may be substantially reduced through interventions targeting home treatment by training patients to self-administer icatibant.


Background
Hereditary angioedema with C1-esterase inhibitor deficiency (HAE-C1-INH) is an autosomal dominant hereditary disease caused by deficiency or dysfunction of the protein C1-esterase inhibitor (C1-INH) [1,2]. The decrease in C1-INH activity may increase plasma concentration of bradykinin, the key mediator of HAE-C1-INH symptoms. Patients with this condition experience recurrent episodes of oedema in subcutaneous tissue or submucosa. Areas affected by this swelling include upper respiratory tract, face, limbs, genitals, and digestive tract [1,2]. Laryngopharyngeal oedema may be life-threatening due to upper respiratory tract obstruction [3,4].
Icatibant acetate (Firazyr W , Shire HGT), a selective bradykinin B2 receptor competitive antagonist, is among the therapies available for acute HAE-C1-INH attacks [5,6]. Recommended dosage to treat HAE-C1-INH episodes is a 30 mg subcutaneous injection, preferably in the abdomen [7]. A recent clinical study, named "Evaluation of the Safety of Self-Administration with Icatibant (EASSI)", evaluated the safety, tolerability, convenience, and effectiveness of Firazyr W self-administration [8]. Based on the study's favourable results, the European Medicines Agency (EMA) has recently approved Firazyr W selfadministration for appropriately trained patients in the drug's self-administration technique [9]. Icatibant is the only subcutaneous treatment for acute Type I and Type II HAE-C1-INH licensed for self-administration in Europe.
Patient's self-administration of icatibant may reduce healthcare costs and may improve the drug's effectiveness due to the accessibility of the treatment. Additionally, by reducing the need of health centres or emergency services visits, the patient would miss fewer hours of work, study, or leisure time.
To our knowledge, there is a lack of evaluation studies comparing icatibant administration methods. This study was designed to fill that gap and compare the economic costs of two strategies to manage acute HAE-C1-INH attacks in Spain. The first strategy assumes that only healthcare professionals administer the drug (health professional-only administration) and the second strategy contemplates that, in addition to visiting a health professional, patients have the option of self-administering the drug (self-administration).

Design
This economic evaluation study was developed by building a deterministic decision tree model with sensitivity analysis. The theoretical model compares the costs of health professional-only administration (strategy 1) vs. the costs of self-administration option (strategy 2). Since the model assumes that both strategies are equally effective [8,10,11], this is a cost-minimization study. Analyses were performed from two different perspectives: The payer (Spanish National Health System) perspective which considers direct costs only; and the social perspective which considers both direct and indirect costs. Indirect costs are defined as those incurred from labour productivity losses (lost working hours). The time horizon is one year.

Model building
The model was built in two phases: 1) Decision tree structure; and 2) Values assignation (probabilities and possible values for cost variables). For building the decision tree, the sequence of events was established, and the cost-generating variables and the categories for each variable for each of the two strategies were defined (Table 1). Variables were defined as follows: Attack severity: mild (discomfort that does not disrupt regular daily activities); moderate (discomfort that reduces or impacts regular daily activities); and severe (symptoms preventing work or daily activities) [1]. Episode duration is the time between the onset of symptoms and their complete resolution. Length of stay is the time patients spend in a healthcare facility. Labour force participation categories are: Employed, Unpaid household work and Other (unemployed, students, retired or early retirees due to permanent disability, recipients of pensions other than retirement or early retirement, volunteers in social work, charity organizations, and other [12]). Since icatibant is not approved for younger than 18 years, the age category "under 18" was not considered. The final product was a decision tree structure applicable to HAE episode management for the two treatment strategies. Figure 1 shows a simplified model of the decision tree. During the second phase, the model content was completed assigning probabilities to each of the variables' categories, and allocating values to the model's cost variables.

Probability assignment
Probabilities were assigned based on the best available scientific evidence. In some cases, data were not available or were available for other populations and, thus, the experts (TC, MG) deemed them inappropriate for modelling Spanish events. In these cases, the assignment of probabilities was informed by experts' opinions.
Official data sources were used for the age variable (Spain's population as of April 1st, 2011 [13]) and for the labour participation variable (Labour Force Survey, First Quarter, 2011 [12]).
Experts' opinions informed the parameterization of the remaining variables by assigning estimates of average, maximum, and minimum values for each variable. The experts parameterized the variables anonymously in two separate rounds with no interaction during the first round. The mean of these estimates became the synthesis estimators and during the second round the final values were allocated by consensus.

Cost allocation
The payer's perspective accounts only for direct medical costs (treatment, hospital emergency room visit, primary care emergency room visit, HAE specialist visit, hospital admission, ICU admission, and tracheotomy) and nonmedical costs (transportation). Because the healthcare system in Spain is governed by the autonomous communities (AC) (political geographic divisions), there are not national prices. Thus, a resource unit cost based on the official AC lists of resource unit costs [14][15][16][17][18][19][20][21][22][23][24][25][26][27][28][29][30][31] was estimated. This cost was calculated as the average price for healthcare services rendered by the AC Departments of Health. Treatment cost was the pharmaceutical laboratory sale's price (LSP) plus the value added tax (VAT).
Transportation costs were estimated using data from the Centro de Estudios y Experimentación de Obras Públicas (CEDEX) (Center for Public Works Studies and Experimentation) [32]. CEDEX provides estimates of the cost per kilometre of private transportation taking into account the investment, maintenance, vehicle repairs, tires change, insurance policy, motor vehicle tax, gasoline, parking, fines, and tolls.   The social perspective accounts for both direct and indirect costs. Indirect costs are defined as patient's and caretaker's working hours lost per episode plus working hours lost in case of patient's death. Leisure time lost was not included in the calculation. For employed workers, labour cost was estimated using the human capital method, based on the cost of the working hour reported in the Quarterly Survey of Labour Costs during the Fourth Quarter of 2010 [33]. For unpaid work (household work) the substitution cost method was used.
The cost of lost labour due to the patient's death was estimated multiplying the years the patient would no longer contribute to the national wealth (contributing years lost) times the per capita gross domestic product (GDP). The per capita GDP was calculated dividing the national 2009 GDP (last available datum) [34] by the Spanish population count as of July 1st, 2009 [13]. To calculate contributing years lost, it was assumed that, in average, patients under 65 years-old die at 41.5 years of age. For those 65 and over, death occurs half way between 65 and life expectancy at 65. The amount of contributing years lost is the difference between life expectancy [35] and age at death.
The product of this phase was the final model: a decision tree loaded with the probability values and the costs associated with each path.

Sensitivity analysis
A sensitivity analysis was performed to evaluate the underlying uncertainty dependent on the variability in the experts' estimators and the resources unit costs. Three scenarios were built under this analysis: base case scenario; most favourable (for strategy 2 vs. strategy 1); and least favourable (for strategy 2 vs. strategy 1). Unit costs for resources were allowed a potential variability of ± 5%.
Base case scenario employs the average unit costs and the average experts' estimators. The most favourable scenario uses the maximum unit costs and maximum experts' estimators, except for the following variables in strategy 2: Facility in which episode is treated, probability of death given a severe episode, probability of tracheotomy given a severe episode, and probability of admission given a severe episode for which the minimum estimators were entered. For the least favourable scenario the minimum unit costs and minimum experts' estimators were considered, except for the variables mentioned above in strategy 2 for which the maximum estimators were entered. Table 1 shows the values assigned to the variables in the model. Table 2 shows unit costs of the resources included in the model. In the base case scenario, HAE-C1-INH prevalence is 2 per 100,000 persons and each patient suffers an average of 6 acute attacks per year. Consequently, it is estimated that in Spain there would be 758 HAE-C1-INH patients experiencing a total of 4,545 acute attacks per year. Using the social perspective, the average cost of managing one of these episodes with strategy 1 (health professional-administration) would be €1,315.14 versus €1,193.84 with strategy 2 (self-administration option). This translates into an average savings of €121.30 (9.2% cost reduction) per episode with strategy 2, representing an annual saving of €551,371 nationwide. A reduction in direct costs would account for 74% of the savings and a decrease in indirect costs would make up the remaining 26% of the money saved. With the payer perspective, the average savings per episode would be €89.8 (7%), which  would result into an annual saving of €408,157 nationwide. The decrease in healthcare services use (emergency services and visits to specialists) accounts for the bulk of the savings ( Table 3).

Results
The number of episodes and the average treatment cost (with the social perspective), according to patient's age and severity of the attack is showed in Table 4. The more severe the episode, the greater the savings are. Savings are also greater in patients under 65 than in those 65 and over, due to the greater reduction in indirect costs.

Sensitivity analysis
In the most favourable scenario for strategy 2, HAE-C1-INH prevalence is 4 per 100,000 persons and each patient experiences an average of 9 acute attacks per year. This would add up to 1,515 patients suffering a total of 13,636 acute attacks per year. Under the social perspective, the average cost of managing an episode would be €1,664.21 with strategy 1 and €1,494.41 with strategy 2. Therefore, strategy 2 (self-administration option) would save an average of €169.80 (10.2%) per episode (Table 5) which would translate into an annual saving of €2,315,360 nationwide. The reduction in direct costs would account for 70% of the savings and a decrease in indirect costs would explain the remaining 30% (Table 6). With the payer perspective, the average savings would be €119.21 (7.5%) which represent an annual saving of €1,625,616 nationwide.
In the least favourable scenario for strategy 2, HAE-C1-INH prevalence is 1 per 100,000 persons and each patient suffers an average of 3 acute attacks per year.  Strategy 1= Health professional-administration only; Strategy 2= Self-administration also available. *Includes visits to hospital emergency rooms, primary care centres emergency rooms, and to HAE specialist offices.
That is, there would be 379 patients experiencing a total of 1,136 acute attacks per year. Under the social perspective, the average cost of managing an episode would be €1,011.89 with strategy 1 and €932.41 with strategy 2. Thus, strategy 2 yields an average savings of €79.48 per episode (7.9%) ( Table 5). This leads to an annual saving of €90,319 nationwide. Direct costs reduction would account for 80% of the savings and a decrease in indirect costs would explain the remaining 20% (Table 6). Using the payer perspective, the average saving per episode would be €63.62 (6.5%) which would represent an annual saving of €72,289 nationwide. Table 7 shows the number of episodes and the average treatment cost (with the social perspective) according to patient's age and severity of the attack, for the most and least favourable scenarios.
In summary, the cost comparison between strategy 2 and 1, shows that with the social perspective, savings per episode would be of €121.3 that could range from €79.5 to €169.8. This equals to save the 9.2% of the costs,   20. This equals to a savings of 7% which could range from 6.5% to 7.5%. Annual cost reduction in Spain would be €408,157 ranging between €72,289 and €1,625,616.

Discussion
According to this study's findings, the possibility of patients self-administering icatibant to control acute HAE-C1-INH attacks brings about substantial reduction in both direct and indirect costs, resulting in savings for the National Health System and for the society as a whole. To our knowledge and based on our literature search, this is the first study to date that has evaluated the economic costs involved in self-administration.
One of the strengths of the current study is that it comprises all relevant variables impacting the cost of managing acute HAE-C1-INH attacks, including the indirect costs related to patients' and caretakers' loss of labour hours. Further, these indirect costs also account for the years of life lost due to the unfortunate premature deaths which, though uncommon, do occur in this patient population.
The main study limitation is the scant scientific evidence available on the study variables. In the absence of scientific evidence, the next best alternative to fill this information gap is the experts' opinion. Although the panel consisted of only two experts, they are both renowned opinion leaders on this disease with ample experience in the management of HAE-C1-INH in Spain. Another possible limitation of the study is that the costs of training patients to self-administer the treatment were not taken into account. However, these costs would be irrelevant, since, in practice, the patients receive the training in the regular visits to the specialist. Finally, the model assumes neither underuse nor overuse of icatibant in the two compared strategies. However, it would be possible that in some settings, overuse or underuse may occur.
The alternative of self-administration of icatibant opens the possibility of early treatment of acute attacks at the first symptoms which may decrease attack severity [36,37]. For instance, potentially severe attacks may turn into mild attacks thanks to early treatment which is likely to lead to even greater savings than those estimated in this study.
In addition, the unpredictability of the timing, frequency, and severity of acute HAE-C1-INH attacks generate a substantial amount of stress in patients. Having the means to control an acute attack quickly and effectively may reduce that stress. This piece of mind compounds another unquantifiable benefit such as the patient's quality of life improvement derived from the reduction of previous restrictions in daily activities as well as work and school absenteeism [37,38].

Conclusions
Both the clinical aspects of HAE-C1-INH and, now, the economic aspects revealed in this study, strongly suggest the need for targeted interventions. These interventions would address home availability of the treatment specific to acute attacks as proposed by national and international consensus [38][39][40][41] and HAE-C1-INH patients' training on the proper technique of icatibant selfadministration.  Strategy 1= Health professional-administration only; Strategy 2= Self-administration also available. Sensitivity Analysis.