2.1 Modeling Approach2.1.1 Model Type and Purpose

A budget impact analysis (BIA) [18] was conducted to estimate and predict the economic effects of increasing BIC/FTC/TAF in the Italian setting. The model, implemented through MS Excel, was developed to understand if increasing BIC/FTC/TAF in both treatment-naïve and treatment-experienced PWH could potentially generate savings and improve the hospital system capacity.

2.1.2 Comparators Considered and Model Scenarios

The model included all currently recommended HIV medications used in Italy. As such, two different scenarios were considered. The current situation of treatment patterns (Scenario A) was compared with a potential change in the rate of consumption of the most administered ART regimens, assuming an increased use of BIC/FTC/TAF (Scenario B).

2.1.3 Analysis Perspective and Time Horizon

The BIA assumed the Italian NHS perspective and represented the evolution of HIV-related healthcare expenditure up to 3 years (36 months). This time horizon was selected as suggested by international guidelines on budget impact [18].

2.1.4 Population

The BIA included the overall treatment-naïve or treatment-experienced Italian HIV populations taking ART.

2.1.5 Organizational Impact Assessment

Besides the BIA, an organizational impact assessment was conducted to determine the impact on the use of healthcare resources, assessing the release of organizational hospital assets, only focusing on the management of drug-related adverse events.

The organizational impact was assessed by focusing on two key metrics: (1) average minutes for outpatient access for adverse events resolution (this metric measures the time required for outpatient visits to manage and resolve adverse events associated with different ART regimens) and (2) average length of stay for hospitalization due to adverse events (this metric measures the duration of inpatient stays required to manage adverse events associated with ART regimens). Information was stratified by ART regimen and by treatment-naïve or treatment-experienced status, to better understand the impact of different ART regimens on hospital resource utilization. Based on the above metrics, the analysis considered the following: (i) an evaluation related to the release of outpatient slots (in terms of average minutes for outpatient visits for the resolution of any adverse events) and (ii) an evaluation related to the release of hospital days (in terms of average length of stay for any resolution of adverse events).

2.2 Model Inputs2.2.1 Model Population

For the definition of the HIV target population, the most recent epidemiological information concerning HIV prevalence and incidence rates was considered [19, 20] and applied to the overall Italian resident population in 2022 [21]. Evidence reported that the HIV prevalent individuals are equal to 130,000, of whom 12% are not diagnosed, with a consequent amount of 114,400 individuals with a proper HIV diagnosis [19]. In addition, the HIV incidence rate applied is equal to 0.002% [20].

The model assumed that, in any year, 88% of HIV patients are currently being treated with HIV medications [22] and three groups of HIV population were included: (1) the prevalent treatment-naïve population (HIV patients who have never experienced any therapeutic switch from the time of diagnosis), representing 24% of the overall prevalent individuals assuming ART [23]; (2) the prevalent treatment-experienced population (HIV patient who have experienced at least one therapeutic switch over the years), representing 76% of the overall prevalent individuals assuming ART [23]; and (3) the incident treatment-naïve population, in terms of new HIV diagnosis, based on the above-mentioned incidence rate. As such, the incident naïve are added to the population each year, whose treatment uptake is always equal to 88% [22].

2.2.2 Model Interventions, Included Treatments and Assumptions

The BIA compares the current therapeutic options based on current market shares for the three populations (Scenario A) with a scenario assuming a greater utilization of BIC/FTC/TAF (Scenario B).

Scenario B was informed based on expert declarations regarding potential modifications in the current market share of antiretroviral therapies, derived from a consensus among leading clinicians, pharmacists and health economists (according to a Delphi method approach [24, 25]), reflecting their insights into evolving clinical practices and anticipated regulatory support for BIC/FTC/TAF, thus being representative of the Italian clinical practice.

Specifically, experts anticipated an 11% increase in the prevalent-naïve population (from 24% to 35% BIC/FTC/TAF administration rate), a 10% increase in the incident-naïve population (from 50% to 60% BIC/FTC/TAF administration rate) and a 15% increase in the treatment-experienced population (from 25% to 40%), as shown in Table 1.

Table 1 Treatment regimens: Scenario A versus Scenario B, derived from the most recent guidelines [7, 26] and validated by Italian clinical practice

Scenario B assumes that the BIC/FTC/TAF market share is taken proportionally from the other therapies. In the scenarios, the model utilizes the guideline-recommended treatments [7, 26].

2.2.3 Treatment Efficacy and Safety Data and Assumptions

The efficacy and safety parameters considered in the model (achievement of virological control as measured by HIV RNA < 50 copies/mL by population and rate of occurrence of drug-related adverse events) for each ART regimen considered in the model were taken from randomized clinical trials (RCTs) [9,10,11,12, 27,28,29,30,31,32,33,34,35,36,37,38] (Supplementary Table 1, see electronic supplementary material [ESM]). While it is understood that RCTs often do not reflect real-world effectiveness of medications, they still represent the gold standard of evidence for decision making [39] and therefore observational studies were not included as a basis of the evidence for this model.

With reference to the data on the rate of occurrence of drug-related adverse events, a stratification was performed among the incidence rates occurring in treatment-experienced or in treatment-naïve PWH, where data was available. Data was also stratified by time horizon. In most cases, the literature provided the incidence data in the 48 or 96 weeks from exposure to antiretroviral treatment (in some cases, data is also available at 144 weeks). For consistency, however, the BIA assumed an incidence of adverse events over 3 years (144 weeks), equal to that of 2 years (96 weeks) for all the therapeutic regimens considered. However, most adverse events occur within the first 48 weeks after exposure to HIV treatment.

2.2.4 Model Costs and Healthcare Resources

The cost inputs of PWH were collected by means of an activity-based costing approach [40]. The model considered the direct HIV-related medical costs for the Italian NHS, also including ART costs and other direct NHS-borne expenditures related to HIV or HIV comorbidities, such as medications, diagnostics and surgical procedures, outpatient activities and inpatient admissions. These costs also include those concerning the management of ART-related adverse events, in terms of laboratory exams, diagnostic and surgical procedures, medical examinations and hospital admissions required for their resolution.

The quantity and the typology of any medical/surgical procedures performed derived from real-life data collected during the years 2019–2021 within a cohort of treatment-experienced, virologically suppressed PWH in three different centers in the Lombardy Region, thus representing the base cost of the economic assessment. It should be noted that the three centers, even if located in a specific region, could be considered as representative of the pathways and approaches generally used for the management of HIV patients and related treatments, with a broader national perspective. Protocols and procedures used are superimposable in the different regional contexts and well aligned with the national and international guidelines.

As such, anonymous administrative hospital flows, derived from the management control offices, were collected and included three different databases (inpatient hospitalization flow, outpatient flow and FILE F drug flow), thus referring to all clinical procedures performed for the management of HIV and comorbidities.

The economic assessment of these clinical pathways also included the costs related to hospitalizations, emergency room visits and the administration of drugs for managing any HIV patients’ comorbidities.

Data on unit costs were derived from the Italian national outpatient and inpatient reimbursement tariffs valid for 2023 [41, 42]. ART costs were derived from a regional public tender [26].

The economic evaluation of PWH was stratified based on their clinical history (treatment-naïve or treatment-experienced) and on the achievement of virological suppression (HIV RNA < 50 copies/mL).

Based on the economic data related to the management of treatment-experienced PWH under virological control (assuming the above real-life data collection) to derive the costs for the other patients’ categories, deviations from this base cost were applied as reported in the most recent literature on the topic [43], thus assuming differences also among treatment regimens, as reported in Table 2.

Table 2 Deviations used for the economic evaluation of clinical pathways

It should be noted that the economic assessment of the clinical pathway also included the costs related to hospitalizations, accesses to the emergency room (ER) department and the administration of drugs for the management of any HIV patient comorbidities (unrelated to the main condition).

2.2.5 Organizational Inputs

The average minutes of outpatient appointments and the average length of stay were derived from actual Italian clinical practice, informed from the declarations of infectious disease clinicians through a Delphi method, in terms of minutes spent for each visit and days of hospitalizations required to manage each drug-related adverse event.

Results have been expressed in terms of hours saved, as well as human resource costs saved, based on the Italian average public salaries (valorized in accordance with the Italian National Labor Contracts per professional class, considered as labor costs), related to the healthcare professionals involved.

This approach was useful to allocate a cost for each activity in terms of the work factor, thus calculating the cost per minute or hour for each healthcare professional involved in the treatment and care of the drug-related adverse events, considering both the outpatient visits and the entire number of days spent in hospitals by the HIV individuals developing any adverse events.

2.3 Sensitivity Analyses

To ensure the robustness of the BIA results, in addition to the previously mentioned analysis of the scenarios concerning the modification of the uptake of the HIV treatments (from 88% to 95% of the target population), two sensitivity analyses were developed, modifying the overall economic assessment of the HIV patients’ clinical pathway.

The first sensitivity analysis considered the exclusion from the medical management cost of the PWH of all the healthcare activities and procedures performed for the management of any concomitant disease experienced by patients and unrelated to their HIV disease.

The second sensitivity analysis considered a modification of the deviations reported in Table 2, informed by infectious disease clinicians through a Delphi method, which is widely used to derive consensus in the absence of comprehensive empirical data, thus ensuring that the sensitivity analysis parameters are grounded in clinical reality and expert judgment, providing a robust basis for the economic model [44]. The analysis assumed (i) a 32% decrease in the costs related to a treatment-naïve patient in virological control with respect to a treatment-experienced patient in virological control; (ii) a 7% increase in the costs related to a treatment-experienced patient not in virological control with respect to a treatment-experienced patient in virological control and (iii) a 25% decrease in the costs related to a treatment-naïve patient not in virological control with respect to a treatment-experienced patient in virological control.

Finally, Bayesian statistical methods were used [45, 46] to assess the robustness of the economic assessment of an HIV subject’s management costs, excluding ART.

Gamma distributions were developed for the cost parameters to incorporate uncertainty, thus considering the average value and the standard deviations related to HIV individuals’ management costs. The shape and scale parameters of these gamma distributions were informed by empirical data from the hospitals involved. Specifically, the parameters were set to reflect observed variations in cost data, ensuring a realistic representation of potential cost outcomes.

The probability of having any cost average value of the Gamma distribution of BIC lower than the other main prevalent treatment regimen was detected and included both treatment-naïve and treatment-experienced subjects. Based on this assumption, the sensitivity analysis compared BIC/FTC/TAF with 3TC/ABC/DTG, FTC/TAF+DTG and DTG/3TC in the treatment-naive population, as well as BIC/FTC/TAF with DTG/3TC, FTC/TDF/RPV and RPV/DTG in the treatment-experienced population, since the treatment alternatives were delivered to 95% and 90% of treatment-naive and treatment-experienced HIV subjects, respectively, based on the percentage reported in Table 1, thus ensuring these sensitivity analyses remained grounded in plausible real-word outcomes.