To predict the usefulness and accuracy of health information obtained online for consumers' HPB, this study is grounded in the health-promotion model (HPM) [17] and the technology acceptance model (TAM) [18].

The TAM is well known in technology-related research. However, despite the increased recognition of the importance of OHI during the COVID-19 pandemic, its significance in terms of health promotion or health behavior has not been fully addressed. The HPM provides useful ideas in terms of prevention and promotion, and studies have included the perceived benefits and self-efficacy variables. However, the HPM has received little validation among eHealth consumers. Using both models, the study predicts the health-promotion behavior of eHealth consumers.

Prior health related behaviors included in HPM are considered important to form habits so that previous behaviors are coded and memorized and so that they can naturally participate in future behaviors without much effort [19]. Prior health related behaviors naturally induce behaviors with healthy lifestyle habits that have been maintained so far, and they directly affect perceived benefits and self-efficacy. In addition, they affect intentions and HPBs directly or indirectly [17,20].

As a result of identifying HPBs and related factors, we have found that prior health related behaviors of Iranian women [19] and female worker [21] were related to perceived benefits, self-efficacy, behavioral intentions, and HPBs. Yang and Kim [22] and Lee [23] confirmed the relationship between prior health related behaviors and perceived benefit as a motivator, and Lee [23] and Shin et al. [24] confirmed perceived efficacy as a motivator.

Perceived ease of use and perceived usefulness, pivotal components in the TAM, have been extensively utilized as factors to elucidate behavioral intention and actual behavior in relation to technology utilization [25,26]. Perceived ease of use is defined as the degree to which a consumer believes that using OHI would be free from effort, and perceived usefulness is defined as the degree to which a consumer believes that using OHI would enhance his or her performance. The TAM is widely accepted because, in contemporary society, consumers tend to depend on technology to seek useful health information themselves rather than depending on passive methods (TV, newspaper etc.) [18,27].

On the other hand, perceived ease of use and perceived usefulness as an antecedent not only lead to perceived benefit, which facilitates health behavior, but also affect HPBs through the medium of self-efficacy [17,28]. Similarly, behavioral intention has been proven to have effects on perceived benefit [24,29] and self-efficacy [30,31].

Perceived benefit has been commonly employed to denote individual motivations. In this study, we also focused on personal motivation, which is more likely to change within the individual dimension. Consumers anticipated benefits from engaging in health behaviors. This is because consumers tend to maintain health behavior longer when they realize that health information benefits them [32]. Perceived self-efficacy of HPB is a mediating factor. Because HPB is the final goal, we focused on the self-efficacy of health behavior in the study rather than internet or technology efficacy. This is a construct of the HPM [17] and has been proven to be the most important predictor of lifestyle changes in meta-analyses [33,34].

Finally, with regard to behavior, behavioral intention refers to a specific plan of action, including when, where, and how, for an organized health behavior [17]. In addition to the HPM, several theories including the Theory of Planned Behavior, the Theory of Reasoned Action, and the TAM have demonstrated that behavioral intention is a significant predictor of health behavior [25,35].

Based on the studies mentioned above, Figure 1 outlines the identified predictors. Prior health related behavior has an effect on perceived benefit, perceived self-efficacy, behavioral intention, and HPB. Perceived ease of use and usefulness impact perceived benefit and perceived self-efficacy, which in turn affect behavioral intentions and HPB. Additionally, there is a direct relationship between behavioral intentions and HPB.

The objective of this study is to investigate the effect of information utilization on the intentions and behaviors associated with health promotion among eHealth consumers who use OHI , within the frameworks of HPM and TAM.

This study is a longitudinal path analysis and tested the factors affecting the behavioral intention and HPB of eHealth consumers based on the HPM and the TAM.

The subjects of this study were OHI consumers aged 18 years and older who had sought OHI at least once a month. Participants (n = 360) were recruited using the convenience sampling method. The inclusion criteria were selected using the following sampling: 1) male or female adults who understood the purpose of this study and agreed to participate voluntarily in it; 2) people who had at any time sought OHI (through a webpage or application on a mobile phone or PC) regarding health promotion on topics such as health responsibility, physical activity, nutrition, spiritual growth, self-actualization, stress management, and interpersonal influences within the previous month.

Upon receiving approval from the institutional review board at the researcher's institution, the collected data guided the research which adhered to the principles outlined in the Declaration of Helsinki and followed ethical guidelines. Before obtaining participants' consent and administering the online questionnaires, the researchers provided an explanation of the study's objectives. The sample size was set at 360 individuals, as it is desirable to extract a sample of around 200 to 400 individuals for Structural Equation Modeling using Maximum Likelihood Estimation [36]. Longitudinal data were collected from the five main cities in South Korea—Seoul, Daejeon, Daegu, Busan, and Gwang-ju. The baseline data collection took place from June 10 to 20, 2015, and follow-up data were collected from September 10 to 22, 2015. Three months after the initial survey, respondents were surveyed again to assess various aspects including their online mobile information searches, primary health information services and devices used, inconveniences experienced, and the impact on their HPBs.

General characteristics of the participants and each variable were analyzed using descriptive statistics. The descriptive statistics with normal distribution were expressed by mean ± standard deviation (M ± SD). Skewness and kurtosis were analyzed to ensure normal distribution of variables. Instrument reliability was tested using Cronbach’s alpha. To assess construct validity, convergent validity was tested by exploratory factor analysis and confirmatory factor analysis using principal component analysis. Multicollinearity among variables was analyzed using Pearson’s product moment correlation coefficient. Twelve insincere responses were removed from the 372 collected three months after the initial survey, resulting in a total of 360 responses used for analysis. Missing data were handled by expectation-maximization in SPSS 22.0 (IBM Corporation, Armonk, NY, USA). Model verification was done using maximum likelihood with AMOS 25.0 (IBM Corporation, Armonk, NY, USA). To evaluate model fit, we applied chi-square; normed chi-square; absolute fit indices, which included GFI (goodness-of-fit index), AGFI (adjusted GFI), and RMSEA (root mean square error of approximation); increment fit indices, which included NFI (normed fit index) and CFI (comparative fit index); and parsimony of fit indices, which included PNFI (parsimony NFI). It was considered statistically significant when a two-tailed probability value was less than 0.05.

This study was approved by the Institutional Review Board of the College of Nursing at Seoul National University in Korea (IRB No. E1603/002-004). Detailed informed consent procedures were followed to ensure that all participants fully comprehended the study's purpose, benefits, research questions, and participation process. Participation in the study posed no harm or risk to the participants. Additionally, they were assured that all collected data would be treated confidentially, and they retained the right to withdraw their data at any point. The data were anonymized to prevent the identification of individuals from the survey data, thereby ensuring both the anonymity and confidentiality of the participants. Furthermore, participants were informed that their personal information would be immediately discarded after the study's conclusion, while data devoid of personal identifiers would be securely stored on a locked computer in the laboratory for a period of 5 years.

The general characteristics of the 360 respondents were analyzed, including age, sex, occupation, level of education, household income/month in KRW 10,000, cohabitation, and number of chronic diseases (Table 1).

Consumers mainly use a PC (44.7%, n = 161) or mobile device (55.3%, n = 199) one or two times a week (27.8%, n = 100) with two or three apps (50.3%, n = 181). The categories of health information most sought were related to health responsibility (39.7%, n = 143) and physical activity (32.8%, n = 118). The purposes of using services were seeking health information (69.4%, n = 250) and searching for materials (10.0%, n = 36). Inconvenient factors in using health information were low credibility (50.4 %, n = 181) and too much information (24.8%, n = 89; Table 1).

As endogenous variables, the average score for prior health related behavior was 18.07 (SD = 3.72), that for perceived ease of use was 11.68 (SD = 2.02), and that for perceived usefulness was 13.78 (SD = 2.81), within a score range from 4 to 20 for each item. The average score for perceived benefit was 15.96 (SD = 5.96) within a score range from 5 to 25, that for self-efficacy was 10.91 (SD = 1.77) with a range from 3 to 15, that for behavioral intention was 21.61 (SD = 3.47 ) with a range from 6 to 30, and that for HPB was 65.07 (SD = 11.18 ) with a range from 20 to 100 (Table 2).

As a result of testing univariate normality, the absolute value of skewness of all variables was from -.64 to .06, less than 2, and the absolute value of kurtosis was from -.09 to .34, less than 7, satisfying the assumption of univariate normality. Therefore, the data set was adequately modeled by a normal distribution because univariate normality is a necessary condition for multivariate normality. The parameters that fit the information were estimated utilizing the maximum likelihood method. Before model verification, correlations among variables and multicollinearity were examined. The absolute values of the correlation coefficients in this study were from .09 to .51, whereas a correlation of above .7 is generally considered to be a very strong correlation. Thus, there was no multicollinearity problem.

As a result of testing the significance of hypothetical paths and the goodness-of-fit of the hypothetical path model, both were found to be satisfactory. Nine out of 13 OHI consumers’ HPB paths were found to be significant. The fit indices were computed as follows: χ² = 7.92, GFI = .99, AGFI = .94, Normed χ² =2.64, RMSEA = .068, NFI = .99, CFI = .99. The fit indices of the hypothesis model satisfied all fit criteria except PNFI.

Nine out of 13 paths of the model were significant. The squared multiple correlations (SMCs), which indicate the explanatory power of behavioral intention and HPB as consequence variables, were 10.4% and 15.3%, respectively. The SMC of perceived self-efficacy, a factor affecting the consequence variables, was 15.2% (Table 3).

Perceived ease of use and usefulness directly and positively influenced perceived benefit, and prior health related behavior and Perceived ease of use directly and positively influenced perceived self-efficacy. Prior health related behavior, perceived benefit, and self-efficacy directly influenced behavioral intention. Prior health related behavior and behavioral intention directly influenced HPB, whereas perceived ease of use, perceived usefulness, perceived benefit, and perceived self-efficacy indirectly influenced HPB (Table 3).