In this study, two TTO surveys were developed to estimate utilities associated with haemophilia treatment (details in the Appendix). Based on the recommendations of several HTA agencies, the study population for both surveys was recruited from the general population [17,18,19, 22]. Respondents were recruited through existing email panels and were awarded points equivalent to EUR/CAD/USD 1–2 for their participation.

Survey 1 (S1) was designed to estimate utilities for people with haemophilia, and survey 2 (S2) was designed to estimate utilities for caregivers of children with haemophilia, i.e. the impact on caregivers’ HRQoL when being responsible for treating a child with haemophilia.

The inclusion criteria were consent to participate and at least 18 years of age. Since haemophilia mainly exists among men, only male respondents were included in S1. In S2, only respondents with at least one child under the age of 15 years were included. This followed recommendations by Powell et al. (2021) [23] and increased relatability for respondents when answering questions.

No sensitive information was collected or revealed in the surveys, and all answers were made anonymous. The surveys were conducted according to the codes of conduct of the Market Research Society and followed the applicable ESOMAR guidelines. Ethical review board approval was not required, since the study was not a clinical trial, it did not include patients, and it did not gather biological or human samples or identifiable personal information.

Based on findings from previous focus group interviews in the UK and the US including 22 people with haemophilia and 16 caregivers (not published) as well as insights from medical experts, descriptions of haemophilia and haemophilia treatment were included in the surveys.

At the beginning of each survey, respondents were introduced to haemophilia; however, to minimise bias in answers, the disease name was never mentioned. Instead, respondents were asked to imagine either having an inborn bleeding disorder (S1) or having a 3-year-old child with an inborn bleeding disorder (S2) (Fig. 1).

Description of haemophilia in the two surveys

Following the disease introduction, three aspects of haemophilia treatment were presented in each survey: 1) frequency of treatment, 2) treatment device and 3) injection site reactions (Table 1). In order to make the results as realistic as possible, the descriptions of the treatment aspects were based on product information from existing treatments and a definition of injection site reactions from a previous phase 3 clinical trial [24]. Additionally, the final descriptions were validated by medical experts. The description of treatment aspects is included in the Appendix.

To elicit utilities associated with the three treatment aspects, nine health states describing different hypothetical haemophilia treatment options were designed (Table 2). In S1, health states were described as permanent conditions that respondents should imagine living with for their remaining lifespan. In S2, health states were described as conditions lasting for 11 years in which respondents should imagine treating their child from the age of 3 until the age of 14. Since previous findings in the literature indicate that the perspective introduced to respondents in vignettes-based research has an impact on respondents’ evaluations – especially if the perspective includes imagination of a child [23, 25,26,27] – the age interval of the child presented in S2 was chosen based on earlier research [28]. Since different treatment aspects usually are less impactful than the efficacy, safety or symptom profile associated with a specific treatment, the impact of these factors was minimised by describing all health states as treatments that keep the disease under control. Examples of how health states were presented for respondents are provided in Table A1 and Table A2.

The online surveys were programmed using a commercial survey software package (SurveyXact).

To make the questions as realistic as possible, time horizons presented for respondents varied depending on their remaining life expectancy (details in the Appendix) [29, 30]. Thus, in both surveys, respondents were asked to trade some of their own remaining life years to avoid living in an impaired health state in which they were treating themselves (S1) or their child (S2).

A number of features were built into the surveys to ensure that respondents understood and accepted the premise of the method. First, a warm-up TTO question was implemented as the first TTO question. This followed recommendations in the literature and helped familiarise respondents with the concept [31]. These results were not included in further analyses. Second, respondents’ trading behaviour was screened carefully if they were willing to trade the largest possible amount of life or if they were not willing to trade at all. In these cases, respondents were excluded from the analyses if they reported reasons for the behaviour that compromised the TTO method (details in the Appendix). Third, each survey was divided into three blocks containing three to five of the nine investigated health states (Table 2). To avoid fatigue, respondents were randomised to one block and received only questions about the health states included in that block. Finally, to ensure that results were not affected by the order of the questions, that was randomised as well.

A pilot study was conducted in the UK to test the survey functionality. Since it did not show unexpected or unusual results, the main data collection was initiated in the UK, Canada and the US without further changes and was conducted in August and September 2023.

To estimate utilities associated with haemophilia treatment, average utilities for each of the nine health states were calculated. Thereafter, utility differences per year associated with the three investigated treatment aspects were calculated as the difference between the utilities of two health states (details in the Appendix). Only differences between utilities derived from the same block were estimated.

To enhance result reliability and reduce susceptibility to extreme data points, the analysis excluded the most extreme 5% of values (2.5% from each end). In a sensitivity analysis, all values were included. Nonparametric bootstrapping with 10,000 iterations was employed to simulate standard errors and confidence intervals (CIs) and to assess the difference between the parameters. All statistical analyses were conducted using SAS version 9.4 statistical software.

All results are presented as utility gains or disutilities per year associated with the investigated aspects of haemophilia treatment. Results assessed to be the most relevant for future economic evaluations are presented in this article, with additional results and results of sensitivity analysis presented in the Appendix (Tables A3 – A6).