The cross-sectional study was conducted from August 2022 to July 2023 through a nationwide, web-based survey and a community survey. Eligible criteria for the study were Thai citizens aged 45 years and older on the date of the survey who could read and communicate in the Thai language and had no underlying conditions or diseases that limited their ability to complete the survey or use mHealth apps (eg, dementia, active psychological problems, or severe visual problems). The web-based survey was disseminated through an assortment of social media platforms, including the department websites, Facebook, Line, Twitter (rebranded as X in 2023), and Instagram. The information on community survey setting and recruitment is described in the section below. For the survey data collection, the respondents to both the web-based and community surveys used the Research Electronic Data Capture (REDCap; Vanderbilt University) survey platform to self-complete the questionnaires. REDCap [21,22] is a secure, web-based software platform designed to support data capture for research studies, providing (1) an intuitive interface for validated data capture, (2) audit trails for tracking data manipulation and export procedures, (3) automated export procedures for seamless data downloads to common statistical packages, and (4) procedures for data integration and interoperability with external sources. All study data were collected and managed using REDCap tools hosted at the Faculty of Medicine, Chiang Mai University. All respondents provided their informed consent, which was included in the screening questionnaire and study information sheet, before participating in this survey. The study excluded incomplete respondents or participants who spent less than 2 minutes or more than 60 minutes on the survey. This study was reported in accordance with COSMIN (Consensus-Based Standards for the Selection of Health Status Measurement Instruments)

reporting guidelines for studies on measurement properties of patient-reported outcome measures [29].

The community survey was distributed by the investigator team, consisting of medical students and health care personnel at primary care units from 10 subdistricts in Chiang Mai province. To identify eligible participants in the target area, officers from the subdistrict primary care units reviewed periodic health survey data for community-dwelling adults aged 45 years and older. Subsequently, patients’ information was verified with the health-promoting hospital databases to exclude individuals with health conditions that impeded survey participation or mHealth use as described above. The subdistrict primary health care teams invited eligible individuals to participate in the study through individual contact by community health care volunteers, community radio announcements, and posters displayed at primary care units.

The ethical consideration of the human subject research was approved by the Institutional Review Board of the Faculty of Medicine, Chiang Mai University (COM-2565-09079). All respondents provided their informed consent, as outlined in the screening questionnaire and study information page, before participating in this survey. For the web-based survey, respondents remained anonymous, and no identification data were recorded. In the case of the community survey, identification data of eligible participants were used solely for recruitment purposes within each target area and were not recorded in either the survey form or the study database. Participants received 100 Thai Baht (US $3) as compensation for answering the questionnaires.

The original, English, 38-item STAM is a 10-point Likert scale consisting of 10 subscales and 38 items that capture the acceptance of general technology use for the older adult population. The total ranges from 38 to 380 points, with a higher score indicating greater acceptance of technology. The validity and reliability of 38-item STAM have been established on a satisfactory scale in 1012 older adults aged 55 years and older in Hong Kong [24]. The construct validity of the STAM was also evaluated with the confirmatory factor analysis (CFA) and revealed a satisfactory model fit with the proposed structure (comparative fit index [CFI]=0.938, root mean square error of approximation [RMSEA]=0.054, and standardized root mean square residual [SRMR]=0.075). The reliability of each subscale with Cronbach α coefficients ranged from 0.67 to 0.95.

Translation and adaptation of the Thai mHealth STAM was performed in accordance with the second edition of the International Test Commission (ITC) Guidelines for Translating and Adapting Tests [30]. In accordance with the ITC precondition guidelines, permission from the holder of the intellectual property rights relating to the 38-item STAM was obtained before performing any translation and adaptation of the STAM. The forward and backward translation with an expert reconciliation design was performed as recommended by the ITC test development guidelines. Before beginning the forward translation process, we decided to include a new subscale, perceived barriers, in the Thai STAM version due to the findings from the previous scoping review [31] on adopting mobile apps for health-related interventions among older adults. It revealed that barriers to adopting mHealth apps among older adults were the most common topics identified in the included studies. Insufficient technological skills, perceived lack of capability and time, concerns regarding personal data privacy, and trust in mHealth providers were the four items comprising the perceived barriers subscale. Following the translation protocol, the original, English, 38-item STAM was adapted to specify mHealth apps in all items and then forward translated into Thai by a professional translator to ensure accuracy for the target audience. The expert panel, which included a digital health expert (family physician and epidemiologist), 2 gerontology physicians, and a public health expert in community medicine, reviewed the forward translation of the Thai STAM questionnaire to ensure readability and transcultural adaptation. The backward translation was done by another professional translator into English. Then, the expert panel reconciled the backward translation version with the original STAM version. The investigator’s team resolved any discrepancies by reaching a final consensus through discussions with the expert panel. To ensure the face and content validity of the proposed questionnaire, a literature review, an expert review, and public interviews were incorporated into the adaptation of the Thai mHealth STAM. In total, 15 older adults participated in this phase to complete the pilot 40-item Thai STAM. Participants were subsequently interviewed to assess the following: overall questionnaire readability, clarity of instructions and items/response options, comprehension of the questionnaire, and other feedback on each item. Then, the pilot 40-item Thai STAM was reworded and revised as recommended on input from both participants and expert interviews. Finally, the pilot 40-item Thai mHealth STAM was given to a group of 40 older adults to verify its reliability and scale usability.

The sample size was estimated based on three parameters, which are as follows: (1) a stable structure for an exploratory factor analysis (EFA) based on the rule of thumb, which is 10 cases per question; (2) expected CFI for a CFA based on the structural equation modeling; and (3) expected Cronbach α for the internal consistency of the questionnaire. For the first parameter, according to the rule of thumb, at least 440 respondents, accounting for 10% of the dropout rate, were required for an EFA. To achieve the expected CFI of 0.95 for a CFA, at least 459 respondents, accounting for 10% of the dropout rate, were required based on an average factor loading of 0.60 and an average factor correlation of .30 to ensure a .05 α (type I) error and power of 90% [32]. For testing overall reliability, at least 146 total respondents were required based on expected Cronbach α=0.80 (SD .05), a confidence level of 95%, and a dropout rate of 10% [33]. All sample size estimation was performed by the web-based sample size calculator [34]. Finally, the minimal required sample size for this study was 920, which was divided into 460 each for the EFA and CFA, respectively.

After reviewing the forward translation, the panel of experts decided to remove 2 items from the gerontechnology self-efficacy subscale, as they were redundant with the facilitating condition (FC) subscale (FC1 and FC2). Independent back-translation provided an additional check of the semantic equivalence of the translation. A total of 4 items, including PU2, PEOU2, P4, and P8, were modified based on the backward translation. For face and content validity, we conducted interviews with 15 older adults similar to the target population. Based on participants’ feedback, 4 items (FC1, FC2, C4, and P2) were slightly modified for clarity. In addition, 2 gerontology experts suggested rephrasing 2 items (A1 and A2) regarding attitude to aging and life satisfaction due to the sensitive wording. Finally, the 40-item Thai mHealth STAM in the pilot group of 40 older adults indicated acceptable internal consistency (Cronbach α =0.91). The details of the full 40 items (10 dimensions) of the Thai mHealth STAM are presented in Table S1 in Multimedia Appendix 1.

From the total of 1100 participants, the mean age was 62.3 (SD 8.8) years. The majority of participants were female (776/1100, 70.5%). Among the 1100 participants, 360 (32.7%) were adults aged 45-59 years, and 740 (67.3%) were older adults aged 60 years and older. Statistically significant differences in the characteristics between adults and older adults were observed in marital status (P=.003), education levels (P<.001), income (P<.001), underlying diseases (P<.001), and technology experience (P<.001). The characteristics of the participants of the study population are presented in Table 1. The derived data were randomly divided in a 1:1 ratio into 2 datasets in preparation for the EFA and CFA. The characteristics of the participants involved in the EFA and CFA are described in Table S2 in Multimedia Appendix 1.

According to the item analysis, we excluded 6 items from the physical function subscale (P2, P3, P4, P5, P7, and P8) due to a floor effect or ceiling effect of >80% (Table S3 in Multimedia Appendix 1). An EFA was conducted using PCA with 34 remaining items. The Bartlett test of sphericity obtained P<.001, indicating that the correlation matrix was not random [37]. The KMO statistic was 0.875, well above the minimum standard for conducting factor analysis [36]. Therefore, we determined that the input data were appropriate for EFA. Subsequently, the rotation of principal components was performed using both orthogonal rotation (varimax) and oblique rotation (promax) in an attempt to achieve a simple structure. Given the fact that an oblique rotation is generally recommended by measurement specialists to facilitate the emergence of factor intercorrelations [48-50], almost all social sciences measurements exhibit some degree of correlation [51]. In addition, the correlation matrix for the factors with oblique (promax) rotation indicated that the highest correlation was 0.445 (Table S3 in Multimedia Appendix 1); we thereby determined that the factors were correlated, and hence, oblique rotation was an appropriate approach. The results of parallel analysis (Table S4 in Multimedia Appendix 1) and PCA with or without oblique (promax) rotation all recommended the retention of 7 factors. According to the previous criteria, 2-item factors were identified, including factor 4 (PBR1 and PBR2) and factor 6 (S1 and S2). The internal consistency of seven factors demonstrated Cronbach α of 0.884 with 95% CI (0.875-0.894), which met acceptable thresholds. Within the context and theoretical framework of the STAM [24] and the UTAUT [11,21,26,52], social factors significantly influence behavioral intentions to use technology, particularly in the use of mHealth. Perceived barriers also play a role in determining intentions to use mHealth, as demonstrated by the aforementioned scoping review [31]. The inclusion of factor 4 and factor 6, which represented perceived barriers and social relationships, was considered appropriate. Based on the priori criteria and consensus of the panel experts, the EFA identified 22 candidate items (ATT1, ATT2, PU1, PU2, PU3, PEOU1, PEOU2, PB1, PB2, ANX1, ANX2, FC2, FC4, FC5, H1, H2, H5, C2, C3, C4, S1, and S2) with factor loadings greater than 0.4 that encompassed the 7 factors. The final EFA result is presented in Table 2.

From the EFA, the 22 items of the 7-factor Thai mHealth STAM explained 91.45% of the variance. The unidimensionality of each factor (subscale) and the overall models were assessed by analyzing modification indices in the CFA. Of the 7 factors from the EFA, the CFA of each factor (subscale) showed that only 5 factors, consisting of cognitive ability (C2, C3, C4), perceived barriers (PB1 and PB2), facilitating conditions (FC2, FC4, and FC5), self-reported health conditions (H1, H2, and H5), and social relationships (S1 and S2), showed satisfactory information criteria indices of the CFA, as presented Table 3. Factor 1, which included items from the attitude toward using (ATT1 and ATT2), perceived usefulness (PU1, PU2, and PU3), and perceived ease of use (PEOU1 and PEOU2), did not meet the CFA criteria due to over-factoring issues (Table S5 in Multimedia Appendix 1). Factor 2, combining perceived barriers (PB3 and PB4) and gerontechnology anxiety (ANX1 and ANX2), was unfit according to CFA criteria, with a low CFI (0.799), low TLI (0.698), and high RMSEA (0.306, 90% CI 0.293-0.319). Attempts to combine subscales also did not meet CFA criteria (Table S5 in Multimedia Appendix 1). However, when items were separated as in the original STAM, including attitude toward using (ATT1 and ATT2), perceived usefulness (PU1, PU2, and PU3), perceived ease of use (PEOU1 and PEOU2), and gerontechnology anxiety (ANX1 and ANX2), these separated factors showed a good fit with CFA criteria (Table S5 in Multimedia Appendix 1). Out of the 9 factors from the single latent factor analysis, 5 were 2-item factors. These were kept in the final CFA model because 3 factors (attitude toward using, perceived ease of use, and gerontechnology anxiety) were originally designed as 2-item factors, similar to the original STAM. The perceived barriers and social relationships were also retained because of their contextual relevance, as described above. Finally, the CFA confirmed 9-dimensional sets of 22 items with satisfactory fit indices, as shown in Table 3. The details of the CFAs of evaluated and reevaluated models are described in Table S4 in Multimedia Appendix 1.

A nonparametric IRT analysis also affirmed the unidimensionality, local independence, and monotonicity of the 22-item model with 8 factors (Table S6 in Multimedia Appendix 1). For the scalability, all 22 items of the Thai mHealth STAM had H^s coefficients over 0.4, which indicates medium to strong scalability properties (Table S6 in Multimedia Appendix 1). The correlation among the final 22-item Thai mHealth STAM subscales ranged from 0.040 to 0.685 (Table S7 in Multimedia Appendix 1). The final 22-item Thai mHealth STAM questions, along with the English version and modeling indices, are described in Table 4.

Each item is scored on a 10-point Likert scale from 1 (very unsatisfied or strongly disagree) to 10 (very satisfied or strongly agree), with reverse scaling for perceived barriers and gerontechnology anxiety.

Considering the absence of a reference standard, it is theoretically reasonable that more participants with higher STAM scores will result in greater acceptance and adoption of technology. The discriminative indices, including sensitivity, specificity, and AUROC, were used to determine the cutoff scores for the proposed questionnaire, considering the intention to use mHealth from the external question. The 6 proposed sets of the final 22-item Thai mHealth STAM bands were classified into low, moderate, and high acceptance, as presented in Table 5. The set D of the possible banding was preferred as the optimal 22-item Thai mHealth STAM cutoff score based on the highest sensitivity of 89% (95% CI 86.1%-91.5%) and AUROC of 72.4% (95% CI 70%-74.8%). This finding also confirmed the discrimination performance of the 22-item Thai mHealth STAM in identifying persons with and without the intention to use mHealth. For set D, low, moderate, and high scores are defined as ≤151, 152-180, and ≥181, respectively. In addition, we conducted a subgroup analysis based on age groups: pre-older adults (aged 45-59 years) and older adults (aged 60 years and older). The result revealed that the set D banding had robust discriminant validity in older adults (AUROC 73%, 95% CI 70%-76%), but the discriminant validity decreased in the pre-older adult group (AUROC 67.7%, 95% CI 63.3%-71.9%). The discriminant validity of the 22-item Thai mHealth STAM by the subpopulation cohorts is shown in Table S8 in Multimedia Appendix 1.

Out of 1100 overall participants, the final 22-item Thai mHealth STAM demonstrated an excellent internal consistency in both the Cronbach α (0.88, 95% CI 0.87-0.89) and the McDonald ω coefficients (0.85, 95% CI 0.83-0.87), as shown in Table 6. By subpopulation, the Cronbach α and the McDonald ω coefficients were 0.88 (95% CI 0.86-0.90) and 0.84 (95% CI 0.81- 0.89) for adults aged 45-59 years and 0.88 (95% CI 0.86- 0.89) and 0.83 (95% CI 0.81-0.86) for older adults. All 22 items revealed the corrected item-total correlations ranging from 0.26 to 0.71, achieving a level of acceptance between 0.20 and 0.80 (Table 6).

The study aimed to adapt and validate the STAM questionnaire for assessing mHealth technology acceptance among pre-older and older populations regarding the use of health support. The results confirmed the scale’s factor structure, supported an 8-factor model with 22 items, and showed good discriminant validity in predicting mHealth intention. The optimal version was a 22-item Thai mHealth STAM using the scoring cutoff (≥152). Subgroup analysis indicated no significant difference in discriminant validity between pre-older and older adults. The scale demonstrated strong internal consistency and stability, with reliability confirmed by Cronbach α and McDonald ω coefficients. This adapted 22-item version is more relevant for assessing mHealth intention among older adults and is suitable for public surveys and routine practice, which take less than 15 minutes to complete.

Our findings are consistent with the previous study conducted by the owner of the original STAM [53], which was subsequently developed into a brief form to save administration time and reduce the burden on respondents. The 14-item brief version of the STAM questionnaire consisted of a 4-factor structure: attitudinal beliefs, control beliefs, gerontechnology anxiety, and health. These findings are consistent with ours, reflecting the original STAM model constructs and the age-related health characteristics of older adults. We observed a decrease in discriminant validity within the pre-older adult group, indicating a need for additional factors to explain their behavioral intentions. For example, older adults with different genders, education levels, income, marital status, and ethnicity may have different intentions and purposes to use mHealth for their health [54].

On the strength side, this is the first Thai version of the STAM questionnaire suitable for evaluating technology acceptance in Thai older adults. The 22-item Thai STAM version demonstrates structural balance, reliability, and validity in assessing technology acceptance among older individuals. The evaluation process is time-efficient. In addition, this tool can be used with both pre-older adults and older adults to prepare them for engaging with technology in their future lives. Furthermore, the evaluation takes into account the influence of Thai cultural norms on the adoption and acceptance of mHealth.

However, there are some limitations to consider. Although the psychometric properties of the 22-item Thai mHealth STAM are satisfied through transcultural adaptation in terms of validity and reliability in both the pre-older and older populations, this scale can be applied for use in a broad. However, our study participants may not be representative of the overall Thai pre-older and older populations, as almost all of the participants lived in the northern part of Thailand, particularly in Chiang Mai province. In order to address this concern, future studies, including those based on different regions of Thailand and other specific populations (eg, teenagers, vulnerable groups, minorities, and specific groups of patients) that could potentially derive advantages from mHealth usage, are recommended to expand the generalizability and usability of this scale. Finally, the 22-item Thai mHealth STAM was evaluated based on the board’s definition of mHealth. It is possible that the proposed questionnaire may not be compatible with all of the existing mHealth technologies due to the diverse range of mHealth technologies in health care. The patient’s choice may vary depending on several factors, such as health care providers, types of services, or the specific application. Hence, we suggest using this questionnaire to assess their acceptance and intention to use it in conjunction with the designated mHealth technology.

The 22-item Thai mHealth STAM offers a practical assessment of patients’ acceptability—a crucial factor often overlooked, as evidenced by a recent systematic review of technology acceptability in health care, which revealed that only 10% (142/1219) of the reviewed studies examined patient acceptance [55]. This publicly available questionnaire has the potential to support health care professionals, policy makers, and developers in making informed decisions [56,57], particularly regarding the adoption and acceptance of mHealth within Thai cultural norms. This questionnaire can be incorporated into the research and development (R&D) processes of mHealth and used as a questionnaire to define the target population based on levels of acceptability, as well as ascertain the factors that encourage or hinder the adoption of their mHealth technologies [58,59]. This information is important for informing stakeholders and developers in advance of the mHealth R&D and implementation stages, which necessitate user data for resource allocation and planning in consideration of user requirements and experiences [60,61].

The increasing number of older people, along with their growing adoption of technology, indicates that mHealth technologies might offer a new approach to enhancing the health of older adults with lower health care expenses. Although there are many advantages to using mHealth apps, it is important to consider their acceptance and intention to use them for health-related objectives. We proposed the 22-item Thai mHealth STAM as the questionnaire to evaluate the levels of acceptability and intention to use mHealth in the Thai community of pre-older and older adults. The 22-item Thai mHealth STAM has demonstrated satisfactory psychometric properties in terms of validity and reliability. As a result, it is now feasible to use this questionnaire in a public survey to support stakeholders in making informed decisions. Nevertheless, to improve generalizability and long-term use, further study is needed to investigate the various demographic groups with the specific mHealth interventions.