The data used in this study were obtained via a web-based cross-sectional survey conducted in China between May and June 2022. The research team collaborated with a patient association to recruit individuals with SMA (Meier Advocacy & Support Centre for SMA). The inclusion criteria were as follows: (1) aged 16 years or older at the time of the study, (2) having no cognitive problems, and (3) being able to provide informed consent. Information regarding the study was sent to all eligible participants via the patient organization’s internal social network. Thereafter, all interested members were invited to join an online chat group, and a link to the study introduction and questionnaire was shared with the group. Participants could participate in the formal survey by clicking on the link provided. All participants were required to complete the EQ-5D-5L and PROMIS-29 themselves. Besides, other information about their demographics, socioeconomic status, and health status was also collected.

The research team has extensive experience in conducting online surveys with patients who have rare diseases. The data collection procedure we used is similar to what we have published in papers collaborating with other rare disease patient associations in China [24, 26, 27]. In this study, the research team closely collaborated with the patient association during data collection to ensure data quality. Detailed instructions were provided for each section of the questionnaire to ensure that respondents understood the questions. Two research assistants managed the surveying group and were available for assistance from 9 am to 10 pm, responding within 30 min. The chief manager of the patient association emphasized the importance of data quality in the surveying group throughout the data collection period. Two research assistants from the research team were responsible for checking the data quality of each response independently. They reported any completed questionnaires that took an unreasonable amount of time ( < = 15 or > = 30 min) or showed abnormal response patterns (e.g., selecting answer “1” for all questions) to the principal investigator. After a double check by the principal investigator, suspected responses were reported to the chief manager of the patient association, who then contacted the respondents and required them to re-complete the questionnaire. To prevent survey fatigue, all questionnaires had a pause function, allowing respondents to complete them at different time points. Additionally, all questions were mandatory to avoid missing data.

The study protocol and informed consent were approved by the Institutional Review Board Ethics Committee of the XXX University with Ref no.: XXXX (details are provided in the title page). Written informed consent was obtained from all participants.

EQ-5D-5L consists of two sections. The first is a health state classification system, which comprises five dimensions (mobility, self-care, usual activities, pain/discomfort, and anxiety/depression), with a five-response level option ranging from “no problem” to “extreme problems.” All health states described by the classification system can be summarized as utility scores ranging from 0 (death) to 1 (full health); a negative score indicates a health state worse than death. In this study, the EQ-5D-5L utility score was estimated based on Chinese preference weights. The second section is the Visual Analog Scale (EQ-VAS). It ranges from 0 (worst imaginable health) to 100 (best imaginable health), which represent a person’s global assessment of health.

The PROMIS-29, version 2.0, comprises 29 items under seven core domains, including physical function, fatigue, pain, anxiety, depression, sleep disturbance, and the ability to participate in social roles and activities. Each item of the PROMIS-29 is rated on a five-point Likert scale ranging from “never” to “very much,” except for the pain intensity domain, which is a VAS with a score ranging between 0 and 10. The domain score, expressed as either the T-score or the theta score, can be calculated using a web-based calculator (healthmeasures.net/score-and-interpret/calculate-scores). Higher scores indicate more symptoms and impairment in the depression, anxiety, pain interference, fatigue, and sleep disturbance subscales, whereas lower scores indicate weaker physical and social functioning.

PROPr in this study was calculated based on the six domain scores (depression, fatigue, pain, physical function, sleep disturbance, and ability to participate in social roles and activities) of PROMIS-29 and another domain, cognition. As cognition is not a part of the PROMIS-29, it was calculated based on the scores of the six PROMIS-29 domains. The formula used to calculate the cognition score was introduced by Dewitt et al. [28].The theta scores of all seven PROPr domains were then fed into the online PROPr utility function to obtain a PROPr utility score [29].

The Pediatric Quality of Life Inventory (PedsQL) 4.0 Generic Core Scales is an HRQoL measure that has demonstrated good reliability and construct validity in various rare disease-specific populations [30,31,32]. It is typically used among children and adolescents aged 2–18 years but has now been extended for use with adults [33]. The benefit of using the PedsQL in patients with rare diseases is that it can provide consistent results to capture and compare the changes in HRQoL in a life-long process, given that nearly all rare diseases are congenital.

The 22-item SMA independence scale (SMAIS) upper-limb module is a newly developed instrument that measures the level of assistance required to perform daily activities [34]. A higher score indicates a better ability to perform daily activities in individuals with SMA.

A descriptive analysis was used to describe the patients’ background characteristics and health status. The EQ-5D-5L and PROMIS-29 profiles, including mean, standard deviation, median, score range, ceiling (percentage of highest possible scores), and floor effects (percentage of lowest possible scores), were presented. The ceiling (floor) effects were moderate (10–15%), minor (5–10%), and negligible (< 5%) [35].

Construct validity, including convergent and divergent validity, was assessed using hypothesis testing. We hypothesized correlations (1) between domains of EQ-5D-5L and PROMIS-29 (e.g., EQ-5D “Mobility” and PROMIS-29 “Physical function”); (2) between EQ-5D-5L dimensions/PROMIS-29 subscales and PedsQL core questionnaire (e.g., EQ-5D “Mobility” and PedsQL “Physical functioning” and PROMIS-29 “Physical function” and PedsQL “Physical functioning”); and (3) between EQ-5D-5L utility score/PROPr and SMAIS score. Spearman’s correlation coefficient (ρ) was used to assess the strength of the hypotheses (weak, ρ ≤ 0.35; moderate, 0.36 ≤ ρ ≤ 0.5; or strong, ρ > 0.5) [36]. Additionally, agreement between the EQ-5D-5L utility score and PROPr was assessed based on the intraclass correlation coefficient (ICC, > 0.7, satisfactory) and Bland-Altman (B-A) plot [37]. A bootstrap method (resamples = 1,000) was used to calculate the robust 95% confidence interval (95% C.I.) of the coefficient.

Known-group validity was examined using the Wilcoxon signed-rank test or Kruskal-Wallis test based on the individual’s reported symptoms and types of SMA [38]. We hypothesized that individuals with clinical symptoms would be more likely to report a lower utility score. We also calculated the effect size using the formula \(r=z/\sqrt\) (dividing the z value by the square root of N), to examine the relative efficiency of the measures in differentiating individuals whose conditions differed: r < 0.3 indicates a small effect, between 0.30 and 0.5 indicates a moderate effect, and r ≥ 0.5 indicates a large effect [39].

A multivariate linear regression model adjusted for demographics (sex and age) and health status (duration and SMA types) was used to further evaluate the discriminative ability of the EQ-5D-5L and PROMIS-29 in predicting the change in overall health status (EQ-VAS) at both the utility score and dimension levels. The R software (R Foundation, Austria) was used to perform all analyses, and the significance level was set at p ≤ 0.05.