DISCUSSION

To our knowledge, this is the first study that evaluates the cost-effectiveness of HDM SCIT+ICS for patients with moderate persistent AA using medication step down and reduction of exacerbations as measures of clinical effectiveness. This analysis suggests that treatment with SCIT+ICS is a non-dominant but cost-effective therapy over ICS alone in pediatric and adult patients with AA with or without AR. The probabilistic sensitivity analysis confirmed the robustness of our model. Despite our observation of a reduction in costs per cycle after SCIT discontinuation, the magnitude of costs accumulated through the first three years led to a higher total cost associated to SCIT at the end of the time horizon. We consider that these results are driven by the substantial low cost of the IC defined in the case scenario (i.e., BDP) and thus, the resulting cost reduction did not compensate the initial additional costs associated to the SCIT. However, the resulting ICERs fell below the willingness to pay threshold per QALY of one to three GDP per capita of Colombia, making SCIT cost-effective in all scenarios defined.
Although our study was conducted using some parameters from the Colombian context, we consider that our results provide relevant inputs for the decision-making process in different contexts, especially in those where a significant pressure on health budgets exists. We developed a novel Markov model based on a guideline accepted in the clinical practice worldwide. Furthermore, Markov models are suitable for modelling chronic diseases like asthma that are characterized by varying symptomatic episodes of different severity over time (27). This aspect constitutes one of the major strengths of our study, as this may allow clinicians to obtain evidence regarding the economic implications of the SCIT as add-on therapy to the commonly used pharmacological treatments in the current practice and may base their decisions considering not only a clinical dimension (38).
A relevant aspect in our study was the inclusion of parameters from studies conducted in real-world settings. The efficacy of SCIT with HDM extracts in the reduction of symptoms and medications has been reported in previous experimental studies but its effectiveness in real-world settings is scarce (11,21,22). A previous observational study by Jutel et al. in Germany reported a 10.8% reduction in prescription of AA medications and a 59.7% reduction of AR medications among pediatric patients who received SCIT (39). Although the parameter for measuring the impact of SCIT in our study was the proportion of patients achieving reduction/discontinuation of medication usage, our results coincide in the fact of a positive performance of SCIT in reducing the most important factor that determines the cost of illness.
In a similar way, a previous population-based retrospective cohort study by Schmitt et al. evaluated the protective effect of AIT in asthma progression using the GINA treatment steps as a subrogate of disease severity in a real-world setting. Authors suggest that exposure to AIT is associated with a decreased risk of asthma progression from GINA Step 1 to Step 3 (HR 0.87 95% CI 0.80‐0.95) and from GINA Step 3 to GINA Step 4 (HR 0.66 95% CI 0.60‐0.74) (40). Although authors adopted a different definition of clinical effectiveness than that used in our study, their results highlight the protective effect of the SCIT in a real-world setting. In addition, our studies coincide with the use of a GINA-based conceptual framework for the simplification of the course of asthma.
We were able to obtain the probability of medication-step down and discontinuation of medications in patients with moderate persistent asthma that received SCIT from a real-world study in Colombia by Sánchez et al (23). In addition, we estimated the baseline probability of an asthma exacerbation using data form a population-based study by Dennis et al. that included a sample of 5,978 individuals in six cities in Colombia, and related the effect of the SCIT in this parameter using an observational study from El-Qutob et al. (28,29).
Our study addresses limitations previously identified economic evaluations of AIT. As stated by Asaria et al and Ehteshami-Afshar et al., the evaluation of possible differences in the cost-effectiveness of AIT across subgroups of patients remains one of the broader gaps in the literature. Population-based treatment decisions may potentially led to a loss of efficiency, as an intervention that is found to be cost-effective in a general population of patients may not be equally cost-effective among subgroups (or vice versa) (10,12). Our results indicate that the SCIT with either ICS or ICS+LABA would reach the highest cost-effectiveness in patients with AA+AR. We consider that these results were mainly driven due to the higher baseline probability of an asthma exacerbation compared to patients with AA only (0.465 vs 0.331) (28).
According to the latest guidelines on AIT for HDM mite-driven allergic asthma by the European Academy of Allergy and Clinical Immunology (EAACI), the reduction in asthma exacerbations and medications are considered relevant co-primary outcomes in the assessment of the efficacy of AIT (19). A previous study by Bruggenjurgen et al., evaluated the cost-effectiveness of HDM SCIT+ICS in the German setting through Markov models under the societal perspective (41). The study evaluated the strategies across different populations, but no attempt to reflect the clinical efficacy of SCIT either through a reduction of asthma medications or exacerbations was made. This aspect makes difficult to compare our estimated QALYs gains with those reported by the authors. In addition, relevant inputs in the model were retrieved only through consultation of experts and no detailed description of methods was conducted. Thus, it was defined as a low quality study by a previous systematic review (10).
A previous study by Reinhold et al. evaluated the economic implications of HDM SCIT+ICS in children with AA by a retrospective analysis of a clinical trial (65 patients). After three years, SCIT+ICS was found to be more expensive compared to ICS alone (7). Although authors considered medication step-down, they were unable to account for the effect of the SCIT in asthma exacerbations, neither were able to translate the clinical efficacy of the strategies through QALYs. Their estimations may thus underestimate the effect of the intervention in costs and QALYs.
This study has limitations and our results should be interpreted with caution. Firstly, utility estimations attributed to health states in the model were obtained from a survey conducted in Hungary, and they reflect the preferences for specific health states in that population (35). Utility parameters were associated with the greater uncertainty in our model as evidenced in the deterministic sensitivity analysis and this would have influenced the estimated QALYs gains. Nevertheless, the parameters used were retrieved from a population of patients across different GINA-defined disease categories using the EuroQol-5D instrument, one of the recommended health utility measures for the expression of clinical effectiveness in utility measures (42).
Secondly, given the limitations of information regarding the effectiveness of the alternatives under evaluation in the reduction of exacerbations and medication dosage in Colombia, we relied on important assumptions regarding the effectiveness of SCIT and ICS. In the base case scenario we assumed that the BDP would be the main IC administered in this study, as it is the only IC covered by the Colombian health care system and has the highest market share (24). We consider that this assumption had an important influence in the estimated costs of the controller treatment in our model. As BDP has a considerably lower cost compared to other controller treatments, we potentially underestimated the costs of ICS in the base case scenario. However, we assessed this limitation in a complementary analysis using ICS+LABA as the controller treatment, where the SCIT also resulted cost-effective. Moreover, we could not take into account potential limitations and confounding variables in all the observational studies used as the main source of effectiveness parameters of the evaluated strategies. Nevertheless, we included the effectiveness parameters in the deterministic sensitivity analysis and results indicate that even under both conservative and optimistic values, the SCIT+ICS resulted cost-effective. These limitations could be minimized if there were more high-quality studies that reported the frequency of the selected measures of effectiveness for both treatment schemes in Colombia.