JMIR Publications

ABSTRACT

Background:

The proliferation of wearable devices and advances in data analytics are accelerating personalized digital healthcare. Patient-Generated Health Data (PGHD), created directly by individuals, plays a key role in this transformation. However, willingness to share such data remains limited due to privacy concerns, perceived risks, and uncertainty over data use. Prior studies have examined factors influencing health data sharing but often focused on specific patient populations and lacked a comprehensive analysis of psychological and social determinants in the general public.

Objective:

This study examined factors influencing individuals’ willingness to share health data by applying Privacy Calculus and Communication Privacy Management (CPM) theories, focusing on digital literacy and interpersonal trust.

Methods:

A cross-sectional analysis used data from the 2023 Intelligent Information Society User Panel Survey, a nationally representative sample of 4,518 Korean Internet users aged 15–69. The outcome variable was willingness to share health data. Predictors included perceived risk and benefit, digital literacy (use, understanding, engagement), interpersonal trust, and control variables. Digital literacy was modeled as a latent construct in structural equation modeling (SEM). Wald tests assessed differential effects of digital literacy subcomponents, and mediation effects were estimated.

Results:

Of the 4,518 respondents (weighted N = 38.4 million), 55.8% were female, and the largest age group was ≤20. The average willingness to share health data was 2.73 on a 5-point scale. SEM revealed that perceived risk negatively affected willingness (β = –0.045, P = .049), while perceived benefit (β = 0.046, P = .024), interpersonal trust (β = 0.073, P < .001), and moral motivation (β = 0.309, P < .001) had positive effects. Digital literacy showed no direct effect (β = –0.001, P = .945) but had an indirect effect via perceived benefit (β = 0.035, P = .038). Interpersonal trust reduced perceived risk (β = 0.045, P < .001). Wald tests (χ² = 54.496, df = 2, P < .001) confirmed significant differences in digital literacy subcomponents’ indirect effects via perceived risk, with “understanding” showing the strongest impact.

Conclusions:

Willingness to share health data is shaped by perceived risk, benefit, moral motivation, digital literacy, and interpersonal trust. Understanding and trust emerged as key indirect drivers. These findings support trust-based governance frameworks, context-aware consent processes, and targeted digital literacy education. Future studies should prioritize health-specific literacy tools and explore contextual influences on data-sharing behaviors.