A cross-sectional design was used, including patients with RLS recruited from the Swedish RLS Association, a nationwide patient organization with 1500 members. Inclusion criteria for filling out the cross-sectionally administered postal survey were as follows: (1) being 18 years or older of age, (2) diagnosed and treated for RLS, (3) able to speak and understand Swedish, and (4) provide a written informed consent.

Information about the aim of the study was sent to the Swedish RLS Association Board, which allowed it to be shared with their listed members. To participate, eligible members had to return a written informed consent form and the completed survey in a prestamped envelope. Information provided by the participants also included their age, gender, employment, economic situation, as well as years since their RLS diagnosis, self-reported comorbidities, and treatment.

Back-and-forth translated Swedish versions of eHEALS, Restless Legs Syndrome-6 Scale (RLS-6), Pittsburgh Sleep Quality Index (PSQI), Epworth Sleepiness Scale (ESS), Patient Health Questionnaire-9 (PHQ-9), and CollaboRATE were used to collect data.

The study received ethics approval from the Swedish Ethical Review Authority (Dnr 2022-01515-01). All procedures and administration of data were conducted in line with the General Data Protection Regulation and ethical principles outlined in the Helsinki Declaration [65]. Informed written consent was obtained from all study participants regarding data collection and the analysis of the data. The survey was submitted anonymously by participants, and no compensation was provided.

A total of 788 patients with a mean age of 70.8 (SD 11.3; range 28‐94) years responded (ie, response rate of 52%). Among them, 64.7% (n=510) were women, 73.8% (n=582) were married or living together, and 43.5% (n=343) had attained a university education. Additionally, 72.9% (n=575) were retired. The mean time since being diagnosed with RLS was 16.4 (SD 10.5) years. Of comorbidities, hypertension (n=281, 35.6%) and cardiovascular disease (n=143, 18.1%) were the most common. Iron deficiency was reported by 9.8% (n=78), and 3.6% (n=29) reported severe depressive symptoms. Dopamine agonists were the most common drugs used by 79.3% (n=625) of the patients. Satisfaction with prescribed RLS treatment was reported by one-fifth. Regarding RLS symptom severity, 44% (n=347) and 37% (n=292) reported severe symptoms during night- and daytime, respectively, based on the RLS-6 questionnaire. A total of 43% (n=339) experienced excessive daytime sleepiness. A median eHEALS score of 28 (IQR 22‐33) was reported. Table 1 presents patient demographics and clinical characteristics.

The internal consistency, as measured by Cronbach α and McDonald ω and composite reliability, was above 0.95. Moreover, the item-total correlations (corrected for overlap) were all above 0.84 (Table 2).

Pearson correlation analyses showed that eHEALS scores were positively associated with shared decision-making as measured by the CollaboRATE total score (r=0.159; P<.001), indicating evidence of convergent validity. Weak but statistically significant negative correlations were found between eHEALS and depressive symptoms (PHQ-9; r=–0.09; P=.01) as well as daytime sleepiness (ESS; r=–0.081; P=.03), supporting aspects of discriminant validity. No significant correlations were observed between eHEALS and sleep quality (PSQI; r=–0.018; P=.63) or overall RLS symptom severity (r=–0.064; P=.08).

The factor structure of the eHEALS, examined by the CFA, is provided in Table 2. The unidimensional structure of the eHEALS showed a very good model fit with χ²₂₀=52.5; P<.001; CFI=0.997; TLI=0.996; SRMR=0.049, except for RMSEA=0.047 (90% CI 0.031‐0.062). All factor loadings were significant and ranged from 0.856 (item 7) to 0.920 (item 3).

The unidimensional structure of eHEALS was then further examined to determine whether it could be interpreted similarly in different age and gender subgroups of the patients. As Table 3 shows, all items were perceived similarly by gender subgroups (a nonsignificant χ² difference, ∆CFI<0.01, ∆RMSEA<0.03, and ∆SRMR<0.01). Although the χ² difference test indicated a significant difference between the configural and metric models for age subgroups, the changes in CFI, RMSEA, and SRMR were all below the established thresholds for significance (ie, ∆CFI<0.01, ∆RMSEA<0.03, and ∆SRMR<0.01).

Item fit statistics from the Rasch model are presented in Table 2. All infit and outfit MNSQ values for the items were within the acceptable range: 0.75‐1.21 for infit MNSQ and 0.71‐1.26 for outfit MNSQ. Item 4 was reported to be the easiest item to interpret, while item 8 was perceived as the most difficult one (Table 4).

The results of the DIF analyses are presented in Multimedia Appendix 1. No substantial DIF was found (ie, contrast >0.5) across age, gender, medication use, shared decision-making condition (CollaboRATE), depressive symptoms (PHQ-9), or sleep quality (PSQI). However, item 8 showed a potential DIF across sleep quality, indicating that those with sleep problems reported higher difficulty on item 8 compared to those without sleep problems (DIF=0.51).

Comparisons of patients’ characteristics between low and high health literacy groups are shown in Table 5. The results indicated that participants with higher literacy were significantly younger (mean age 68.1, SD 11.6 vs 73.1, SD 10.6 years; P<.001), more likely to be women (n=259, 70.2% vs n=225, 58.4%; P<.001), and more often had higher education levels (n=203, 56.4% vs n=126, 33.5%; P<.001). No significant differences were observed in civil status or comorbidity burden. Although RLS symptom severity, sleep quality (PSQI), and daytime sleepiness (ESS) were similar between groups, those with higher literacy reported significantly greater involvement in shared decision-making (CollaboRATE score: 6.89 vs 6.13; P=.001). A nonsignificant trend toward lower depressive symptoms (PHQ-9) was observed in the high literacy group (P=.05).

This is the first study that has investigated the psychometric properties of the eHEALS among patients with RLS. Our findings showed a unidimensional structure in both the CFA and the Rasch model with high fit. The unidimensionality was not affected by age or gender, as all items were perceived similarly by younger and older patients as well as by women and men. Moreover, no substantial DIF was found across age, gender, medication use, shared decision-making condition, depressive symptoms, or sleep quality. Internal consistency also proved to be good. These findings support the use of a total score and that eHEALS can be an adequate tool to evaluate competence in health information–seeking behavior in patients of different ages with various clinical and treatment conditions related to RLS.

To begin with, a single factor solution has been found for eHEALS in several studies using general populations; among others, a recent Swedish study [43]. Diviani et al [36] found a single dimension for the Italian version in a community sample. Similarly, Mialhe et al [37], who translated the instrument into Portuguese, found excellent internal consistency for 1 dimension. However, studies focusing on different long-term conditions have reported various results. For example, Lin et al [38], who used classical test theory and Rasch analysis in a population with cardiovascular disease from Iran, found a single-factor structure. On the other hand, Richtering et al [66], who used patients with moderate to severe cardiovascular disease, found a good overall model fit, ordered response thresholds, reasonable targeting, and good internal construct validity, but that eHEALS measured 2 constructs of eHealth literacy (ie, using eHealth and understanding eHealth). Bäuerle et al [41], who evaluated the German version in patients with coronary heart disease and congestive heart failure, confirmed the 2-factor structure, construct, and criterion validity, as well as measurement invariance at the scalar level for age, gender, and educational level. Finally, Brørs et al [40], who investigated the psychometric properties in Norwegian patients after a percutaneous coronary intervention for ischemic heart disease, also found a multidimensional construct. When comparing our construct validity to the earlier-mentioned studies that have validated eHEALS in various conditions, one might have in mind that aspects of importance for the ability to seek, find, understand, and appraise health information from electronic sources may differ. These differences may be based on the presence of sociodemographic factors, such as age, gender, and education, as well as pathophysiological and symptomatologic effects related to the actual condition. Recently, the Swedish Internet Foundation reported that 95% of the adult Swedish population uses the internet, of which over 80% use eHealth services for digital health care visits or specific tasks such as prescription of medicines [2]. This may indicate that our validation carried out in a Swedish context may have slightly different prerequisites. Furthermore, when comparing our data collected during the COVID-19 pandemic to other validation studies [36-40], all conducted before the pandemic, an increased use of eHealth services in Sweden was seen during that period. Due to the social restrictions, digital alternatives then largely replaced health care delivered in a traditional way. Interestingly, even considering these aspects, the construct validity of the eHEALS appears stable, as our finding of a unidimensional factor structure supports adding the scores of individual items to calculate a total score.

Second, as patients with RLS in general are older people [16], often experience comorbidities, and report poor sleep and mood disturbances [8], as well as decreased QoL [15], it is vital to investigate DIF. Importantly, our results showed no substantial DIF for any of the items across age, gender, medication use, shared decision-making condition, depressive symptoms, and sleep quality, while item 8 showed potential DIF across sleep quality. However, the value of 0.51 was marginally above the threshold, which gives a small probability that poor sleep quality will have a decisive importance for how the current question is answered. Even if the chi-square difference test indicated a significant difference between the configural and metric models for age subgroups, the model fit indices changes were all below the established thresholds for significance. Therefore, our findings indicate the psychometric properties of eHEALS to be acceptable, which implies it to be a useful tool for researchers and clinicians to measure digital skills, informational needs, or self-care behaviors among patients with RLS. This has not been done in an RLS context, but digital skills that are deemed as important in general populations involve active information seeking, information use or sharing, and 2-way interactive communication [67]. Personal and socioeconomic factors, cultural factors, attitudes toward the internet, as well as health status have also proven to be of importance for eHealth literacy. Moreover, improved health literacy has been associated with increased health interest, promotion of health behaviors, and increased use of shared decision-making [68]. Positive relationships have also been found between eHealth literacy and various health care processes [67]. When we compared patient characteristics between low and high health literacy groups, we found that participants with higher literacy were significantly younger, more likely to be women, and more often had higher education levels. However, no significant differences were observed in civil status or comorbidity burden. Although RLS symptom severity, sleep quality, and daytime sleepiness were similar between groups, those with higher literacy reported significantly greater involvement in shared decision-making. To foster digital behaviors for care in patients with RLS without knowledge of their eHealth literacy (ie, to seek, find, understand, and appraise health information) might be difficult, especially since those with more pronounced RLS symptoms might experience significant sleep disturbances (ie, more light sleep, less deep sleep, and longer periods of wakefulness) [17], causing decreased cognitive ability [8,13]. However, using digital technology in an optimal way can be a key to developing health skills [69], which in turn can facilitate the mastery of RLS symptoms.

Fitzpatrick [69] stresses that digital tools such as eHEALS can facilitate patient education and self-care and provide empowerment possibilities. However, there are both facilitators and barriers for the implementation of digital tools for older people, which has been proven in other long-term conditions. Factors that have been found to act as both facilitators and barriers involve demographic, social, and socioeconomic factors, as well as health-related, dispositional, and technology-related factors [70]. On the other hand, facilitators often concern active engagement of the end users in the design and implementation of an eHealth program, support for overcoming concerns, privacy, and enhancing self-efficacy in the use of technology, and integration of the actual program across health services to accommodate the multimorbidity [71]. Implementing digital technology into the available RLS care can, as shown in other patient groups [68,69], probably lead to a transformation of health care delivery. Specifically, this may improve treatment options and communication among providers and patients [69], which in turn may give older patients with RLS improved involvement in their self-care as well as in RLS-related clinical decision-making [25]. However, studies on general populations have found that barriers often relate to a lack of self-efficacy, knowledge, support, functionality, and information provision about the benefits of eHealth [71]. Challenges and limitations associated with digital health literacy often include issues related to access, reliability, and privacy [69]. Therefore, the earlier-described aspects need to be explored in an RLS context using various designs and instrumentation.

Shifting the focus toward RLS care delivered through the internet, which patients today describe as a need [13], involves equipping them with digital skills to explore information to optimize their understanding of their disease and its pharmacological treatments and encouraging them to take an active role in managing their condition. However, obstacles when accessing digital health solutions concern technology literacy issues, affordability, the time burden to participate, and a perceived risk of losing in-person contact [72,73]. Studies have determined that involvement of user perspectives about what makes the best digital solution for those living with a chronic condition varies, but there is a strong conviction that tools providing feelings of reassurance increase the ability to manage their condition [72]. Several studies, not performed on patients with RLS, highlight the need for co-designing digital health interventions [72,73], as this is also particularly beneficial for providing more equitable access [72]. Guidelines for RLS treatment provide information regarding pharmacological treatment [19]. Nonpharmacological self-care interventions for RLS exist and could be assessed [74]. However, according to a meta-analysis by Harris et al [20], these are seldom used and need more evidence. Patients with RLS face several barriers to fulfill basic human needs [13] and may be perceived to be more sensitive to treatment and therefore have a greater need for easily accessible and relevant medical information than other patients [75]. A recent qualitative study [76] found that accessible information through the internet could increase motivation to perform RLS-related self-care actions. Knowledge describing self-care treatment may therefore need to be made available via digital channels to clarify self-care as a potential complement to medical treatment. Tailored patient-centered digital health care interventions, informed by the eHEALS, should be designed to promote digital health literacy at the individual and organizational level [74] to provide patients with RLS with user-friendly eHealth solutions. In this way, digital health can thus both empower and motivate various parts of RLS treatment.

It is important to consider some methodological aspects. Even if the sample is relatively large, the predominance of female and older retired patients may have influenced response patterns of the survey and the eHEALS as well. However, RLS is more common among women and older people [16], so our sample can be assumed to reflect the age and gender aspects of a clinical sample. The data collection was conducted via the nationwide Swedish RLS patient organization using a cross-sectional design. This, unfortunately, created a major limitation, as it limited the ability to perform test-retest analysis and explore changes in relation to different treatment interventions. Moreover, all assessments for DIF were self-reported, which might create recall bias. Future RLS studies should use a prospective design with repeated measurements to enable test-retest. They should also assess reliability and competence in health information–seeking behaviors (ie, predictive validity) in relation to self-care activities among patients with RLS.

This study showed promising psychometric properties for the eHEALS among patients with RLS. The instrument operated equivalently for men and women of different ages with various clinical and treatment conditions related to RLS. Accordingly, health care professionals can use eHEALS as a psychometrically sound tool to explore the digital health literacy level among patients with RLS.