Peer health literacy level and health literacy among school-aged children and adolescents: a multilevel moderated mediation model of internet health information acquisition

Abstract

Background

Health literacy is a critical determinant of health outcomes among school-aged children and adolescents. Although peer environments are known to shape adolescent health-related behaviors, their association with health literacy—particularly through internet health information acquisition—remains underexplored in multilevel settings.

Objective

This study examined the association between peer health literacy context and health literacy among Chinese school-aged children and adolescents, testing the mediating role of internet health information acquisition and the moderating effects of contextual factors (e.g., perceived teacher support, policy environment, economic status) within a multilevel moderated mediation framework.

Methods

We analyzed cross-sectional data from 3307 participants in Shaanxi Province, China, collected in 2021. Multilevel mediation and moderation analyses used the Monte Carlo method with 5000 simulations.

Results

Peer health literacy context was negatively associated with health literacy (β = – 17.513, P < 0.001). Internet health information acquisition showed a significant negative indirect pathway at the group level (β = – 0.052, 95% CI: –0.081, –0.027). Left-behind status, economic indicators, and policy variables significantly moderated the mediating pathway.

Conclusions

A higher peer health literacy context was associated with lower individual health literacy, and this pattern was partially explained by internet health information acquisition. Contextual factors at the school, policy, and economic levels play significant moderating roles in this pathway. These findings suggest that adolescent health literacy should be understood not only as an individual competence but also as a process shaped by peer context and broader ecological environments.

Introduction

Background

Health literacy serves as a comprehensive indicator of socioeconomic development and public health status and is a key metric for assessing a nation's basic public service provision and population health outcomes [1]. The school-age and adolescent years represent critical developmental stages and pivotal periods for addressing health, social, and educational challenges, as well as cultivating essential competencies and healthy behaviors. The capabilities and habits formed during this phase profoundly impact both individual well-being and the health of subsequent generations [2]. School-aged children and adolescents face numerous health challenges, including obesity [3], myopia [4], and mental health issues [5]. Enhancing health literacy is one of the most fundamental, economical, and effective measures for improving public health standards.

The proliferation of digital technologies has fundamentally transformed how individuals, particularly adolescents, access and engage with health information [6]. The internet has become a primary source of health knowledge for this demographic, making internet health information acquisition a critical component of modern health literacy [7]. For digital-native generations, seeking health-related information online is a normative and often preferred behavior [8, 9]. The ability to effectively seek and appraise online health information thus represents a critical component of adolescent health literacy.

To understand the social determinants of this digital health behavior, it is essential to adopt a multi-level perspective. Bronfenbrenner's Ecological Systems Theory [10, 11] provides a valuable framework, positing that individual development is embedded within interconnected environmental systems, from the immediate microsystem (e.g., peer groups) to the broader mesosystem (e.g., school environment) and exo/macrosystems (e.g., policy and economic contexts). This theoretical lens justifies examining how factors at different ecological levels jointly shape adolescents' health literacy.

Among the multiple factors embedded in this ecological framework, peer context warrants particular attention at this stage for three reasons. First, during adolescence, peers constitute one of the most developmentally salient and proximal social environments, often exerting more immediate influence on self-evaluation, motivation, and everyday behavior than more distal structural factors [10–12]. Second, in the digital era, many health-related behaviors—such as searching for, sharing, interpreting, and discussing online health information—are closely embedded in peer interaction [13–16]. This makes peer context especially relevant for understanding internet health information acquisition, which is a central behavioral pathway in the present study. Third, compared with broader school-, family-, or policy-level influences, the naturally occurring peer health literacy context in everyday classrooms remains relatively underexamined, particularly in terms of how it may shape adolescents’ health literacy through digital information practices [17–22]. For these reasons, peer context is not treated as the only important factor, but as the most proximal and theoretically strategic entry point for explaining adolescent health literacy within a broader multilevel system.

Within this multi-level framework, the peer group represents the most proximal and developmentally salient microsystem influence for adolescents.. Existing research has extensively demonstrated the effectiveness of peer interventions in enhancing health literacy among children and adolescents, particularly in areas such as mental health [17–20]. However, these studies predominantly focus on evaluating organized "peer education" or "peer support" programs [21, 22]. There remains a lack of systematic exploration into how spontaneously formed, more pervasive “peer health literacy environments” —the average health literacy level of one's classmates—subtly shape health beliefs and behaviors through everyday social comparison processes. Furthermore, existing research predominantly operates at a single level (e.g., schools) and has not systematically examined the mediating pathways (such as internet health information acquisition) or the moderating effects of multi-level contexts (e.g., school environment, policy factors) on this influence process.

Therefore, to address these gaps, this study adopts ecological systems theory as the overarching framework and uses a multilevel model to analyze large-scale questionnaire data collected in 2021. Within this framework (Figure 1), peer context represents the microsystem most proximal to adolescents, while school-, policy-, and regional-level conditions represent broader ecological layers that may shape how peer influence operates. In this study, we focus on a peer health literacy context, operationalized as the average health literacy level of other students in the same class, and examine whether it is associated with individual health literacy. We further investigate whether internet health information acquisition serves as a behavioral pathway linking peer context to health literacy, and whether this pathway varies across contextual conditions such as perceived teacher support, left-behind status, policy support, and regional economic conditions. By doing so, this study seeks to provide a more integrated account of adolescent health literacy formation in the digital era.

Fig. 1.

The Theoretical Framework for Peers on health literacy

Theoretical foundation

Ecological system theory was proposed by Bronfenbrenner [10, 11]. This theory emphasizes that individual development is embedded within interconnected environmental systems. These systems can be categorized into microsystems, mesosystems, exosystems, and macrosystems, forming the core components of the ecological system model [23]. In this study, the microsystem refers to the individual's immediate environment, specifically peer groups. The mesosystem refers to the school atmosphere. The exosystem and macrosystem denote broader societal environments, including school health policies and regional economic levels. While these do not directly influence individuals, they exert effects through the underlying systems. This theory provides the overarching logic for incorporating multi-level moderating variables from the classroom to the regional level in this study, clarifying the necessity of examining factors such as peers, schools, policies, and economics.

Research hypotheses

Relationship between peer health literacy level and individual health literacy

Peer environments are among the most immediate social contexts in adolescent development and may shape health-related cognition and behavior through comparison, reinforcement, and access to informal support. Existing studies on peer education have generally emphasized the beneficial role of structured peer-led interventions in improving adolescents’ health knowledge, help-seeking intentions, and health behaviors [17–19, 24]. However, these organized interventions differ conceptually from the naturally occurring peer context examined in the present study. In everyday classroom settings, adolescents are continuously exposed not only to peers’ behaviors but also to the overall competence level of their peer group.

When the average health literacy level of peers is high, this context may serve as a salient comparison standard. According to social comparison theory [12], adolescents often evaluate their own abilities relative to similar others, especially peers. Under such conditions, a high-performing peer context may not always produce encouragement. For some students, especially those with relatively limited health knowledge or low confidence, upward comparison may instead trigger a contrast effect, making the gap between themselves and their peers appear difficult to bridge [25, 26]. This may weaken self-efficacy, reduce motivation to engage with health-related content, and ultimately be associated with lower health literacy. Similar contrast effects have been documented in educational and psychological research, such as the big-fish-little-pond effect [2] and peer comparison studies [27] showing that stronger peer contexts can lower individuals’ self-evaluations or performance expectations.

Based on this reasoning, we propose the following hypothesis:

• Hypothesis 1: peer health literacy level is significantly negatively associated with individual health literacy among school-aged children and adolescents

Mediating Role of internet health information acquisition

Internet health information acquisition is not solely an individual behavior; it is also shaped by how adolescents interact with peers in everyday digital and social environments. Adolescents frequently search for and appraise health information online [8, 9], and peer communication may influence whether such information is perceived as relevant, trustworthy, and worth further exploration [14]. In peer settings, online information can gain salience when it becomes part of ongoing conversations, shared concerns, or collective interpretations among adolescents. Because internet health information acquisition provides a key behavioral pathway linking digital engagement to health literacy [28, 29], it is theoretically plausible that peer health literacy context may be associated with individual health literacy through adolescents’ internet health information acquisition. Thus, peer’s context may be associated with health literacy through adolescents’ internet health information acquisition. Based on this, we propose:

• Hypothesis 2: internet health information acquisition mediates the association between peer health literacy context and health literacy.

Multilevel Moderated mediation

Perceived teacher support may shape how adolescents interpret and respond to peer differences in everyday school life. When students perceive teachers as supportive, they may feel more secure in asking questions, seeking clarification, and engaging with health-related information instead of avoiding it. In this sense, teacher support may buffer the negative consequences of upward peer comparison by reducing perceived threat and strengthening students’ confidence in handling health information [12]. At the same time, adolescence is a developmental period in which peer influence may be more salient than teacher influence in many domains [30]. Therefore, perceived teacher support may condition, but not necessarily override, the mediated pathway linking peer health literacy context to health literacy through internet health information acquisition. Based on this reasoning, we propose:

• Hypothesis 3: Perceived teacher support moderates the mediating role of internet health information acquisition in the association between peer health literacy context and health literacy.

Broader policy and economic conditions may condition how adolescents respond to peer health literacy context. According to ecological systems theory, peer context operates at the microsystem level, whereas policy and economic environments belong to broader exo- and macro-level contexts that shape the resource conditions under which peer comparison occurs [10, 11, 23]. Previous studies have shown that regional economic conditions and health-related policy environments are associated with variation in adolescent health literacy [31–34]. In addition, access to online health information is influenced by the availability of digital resources, information environments, and the broader conditions that support health information seeking [35–37]. In resource-rich and health-promoting contexts, adolescents may be more likely to view the gap between themselves and higher-performing peers as manageable and actionable rather than discouraging. Under such conditions, peer comparison may be less likely to trigger avoidance and more likely to promote engagement with online health information. Accordingly, we propose:

• Hypothesis 4: Policy and economic environments moderate the mediating role of internet health information acquisition in the association between peer health literacy context and health literacy.

Methods

Data source

Data were derived from a survey conducted by our research team between July and November 2021. A stratified random sampling method was employed. One city was selected from each of the northern, central, and southern regions of Shaanxi Province. From each city, two primary schools, two junior high schools, and two senior high schools were randomly selected. Approximately 200 students from different grades in each school were randomly selected for the questionnaire survey. The questionnaire used in this research consists of published scales and questionnaires [38–40]. Specific references, please refer to the variable measurement section. Considering questionnaire readability and the practical circumstances of students in graduation years, participants excluded students in grades 1–2 and those in graduation grades.

The survey was anonymous, administered on-site by investigators using paper questionnaires. A total of 3600 student questionnaires were distributed and collected upon completion. After excluding invalid questionnaires, 3307 valid student questionnaires remained. Invalid questionnaires were identified based on: (1) Questionnaires with more than 2/3 missing answers; (2) Questionnaires with obvious logical errors, such as multiple answers for single-choice questions; (3) Damaged questionnaires resulting in significant loss of valid content.

Variable measurement

Dependent variable

The dependent variable was health literacy. Based on the characteristics of school-aged children and adolescents and drawing from The Chinese National Residents Health Literacy Monitoring Survey Questionnaire [38], health literacy was measured using 67 items across three dimensions: health knowledge, healthy lifestyle & behaviors, and basic skills. Items were dichotomously scored (correct = 1, incorrect = 0). The total score ranged from 0 to 67, with higher scores indicating higher health literacy.

Independent variable

The independent variable was peer health literacy context. It was operationalized as the average health literacy score of all other students in the same class, excluding the focal student’s own score. A higher value indicates that the focal student is situated in a classroom context with a higher average level of peer health literacy. Although this measure does not directly capture perceived normative pressure or shared expectations, it reflects the health literacy context created by classmates and serves as a class-level peer reference environment in this study.

Mediating variable

The mediating variable was internet health information acquisition, measured by 5 items from the eHEALS [39]: (1) I know how to find helpful health resources on the Internet; (2) I know how to use the Internet to answer my health questions; (3) I know what health resources are available on the Internet; (4) I know where to find helpful health resources on the Interne t; (5) I know how to use the health information I find on the Internet to help myself. Responses were on a 5-point Likert scale (1 = Strongly Disagree to 5 = Strongly Agree). Cronbach's α was 0.865.

Moderating variables

The moderating variables were contextual factors at different ecological levels, including left behind, perceived teacher support, and economic and policy factors.

Left-behind students. Determined based on whether both parents are migrant workers. If both parents are migrant workers, the student is classified as left-behind and assigned a value of 1. If only one parent is a migrant worker, or if neither parent is a migrant worker, the student is not considered left-behind and assigned a value of 0.

Perceived teacher support was measured using the Teacher Support dimension of the Perceived School Climate Scale. Developed by Jia et al. in 2009 for Chinese students [40], this scale comprises 25 items, with the Teacher Support Dimension consisting of 7 items. It utilizes a 4-point rating scale. Responses were coded as follows: “1” indicated never; “2” indicated occasionally; ‘3’ indicated often; and “4” indicated always. In this study, the Cronbach's alpha for the Teacher Support subscale was 0.881.

Economic factors include variables such as the Top 100 Counties Ranking, per capita disposable income, and regional GDP levels. Top 100 Counties: Based on the National Bureau of Statistics of China's assessment, counties selected for the Top 100 Counties Ranking are assigned a value of 1, while others receive a value of 0. Per capita disposable income is identified based on the region's per capita disposable income during the survey period (2021). Ordinal (1 = Lowest, 2 = Middle, 3 = Highest) based on 2021 regional data. This variable is treated as an ordered categorical variable. Regional GDP level is identified based on the GDP level of the region during the data collection period (2021). Ordinal (1 = Lowest, 2 = Middle, 3 = Highest) based on 2021 GDP data. This variable is an ordered categorical variable.

Policy factors included health-promotion counties/districts, health school designation, and regional health literacy levels. These indicators were used to reflect whether the local institutional environment had formal health-promotion arrangements in place during the survey period. In the Chinese policy context, Health-Promoting Counties/Districts refer to counties or districts that have received official designation for implementing coordinated health-promotion initiatives, which typically involve public health promotion, school health education, and supportive institutional arrangements. Regions with this designation were coded as 1; others were coded as 0.

Healthy schools were identified based on official national and provincial designation lists. In general, these designations indicate that a school has implemented health-promotion activities, such as health education, campus health management, and supportive school health environments. Schools included in the relevant lists were coded as 1; others were coded as 0. To distinguish variation in policy intensity, national and provincial healthy school designations were entered separately in the moderated mediation models.

Regional health literacy levels were based on the prevalence of adequate health literacy among school-aged children and adolescents in the surveyed regions. Because the number of regions in this study was limited, this variable was coded as an ordinal contextual indicator rather than treated as a continuous regional-level variable: central Shaanxi (29.80%) = 3, southern Shaanxi (21.72%) = 2, and northern Shaanxi (17.76%) = 1. This coding was used to represent relative differences in regional health literacy context rather than precise interval-level variation.

Statistical analysis

Statistical analysis proceeded in five stages. First, descriptive statistics and Common Method Bias (CMB) assessment were conducted. To mitigate CMB, anonymous reporting and some reverse-scored items were used. Harman's single-factor test was applied post-hoc; the first factor explained 25.44% of variance (< 40%), indicating no severe CMB [41]. Second, a null model (intercept-only model) was used to calculate Intra-class Correlation Coefficients (ICC) to justify multilevel modeling [42, 43]. When ICC is less than 0.059, it indicates low intraclass correlation; 0.059 < ICC < 0.138 indicates moderate intraclass correlation; ICC > 0.138 indicates high intraclass correlation. ICC for health literacy and its dimensions ranged from 0.183 to 0.416, indicating high clustering supporting the use of multilevel analysis. Third, multilevel regression analysis identified factors associated with health literacy. Fourth, multilevel mediation analysis tested the mediating role of internet health information acquisition in the peer health literacy and individual health literacy association. Finally, multilevel moderated mediation analysis examined whether contextual factors moderated the mediating pathway. All multilevel mediation and moderation analyses used the Mlmed macro, developed by Rockwood in 2017 [44], with 5,000 simulations to generate multilevel mediation moderation Monte Carlo confidence intervals. Effects were considered statistically significant if the 95% bias-corrected confidence interval (95% CI) did not include zero [45, 46]. Analyses were performed using SPSS (version 26.0; IBM Corp), with a two-sided significance level of p < 0.05.

For multilevel regression models that included covariates with missing data (e.g., parental migration status, parental education), listwise deletion was applied. The effective sample size for Table 2 is 1,181. For mediation and moderated mediation models that excluded these covariates, the full sample (N = 3,307) was used.

Table 2.

Analysis of factors associated with health literacy among school-aged children and adolescents using multilevel regression

Factor	Health Literacy	Health Knowledge	Healthy Lifestyle & Behaviors	Basic Skills
	β (P-value)	β (P-value)	β (P-value)	β (P-value)
Intercept	878.469 (< 0.001)	370.619 (< 0.001)	388.623 (< 0.001)	213.118 (< 0.001)
Gender (Ref: Female)
Male	0.226 (0.194)	0.338 (0.007)	−0.104 (0.025)	0.030 (0.278)
Age	0.026 (0.647)	0.037 (0.371)	0.007 (0.638)	−0.006 (0.514)
Ethnicity (Ref: Other)
Han	0.522 (0.741)	−0.163 (0.885)	0.253 (0.549)	0.091 (0.716)
Korean	−2.588 (0.343)	−1.159 (0.552)	−1.428 (0.050)	−0.450 (0.297)
Manchu	1.761 (0.518)	1.174 (0.547)	0.283 (0.698)	0.056 (0.897)
Hui	0.260 (0.895)	0.341 (0.810)	−0.745 (0.160)	−0.072 (0.817)
Residency (Ref: Rural)
Urban	1.206 (< 0.001)	0.712 (< 0.001)	0.264 (< 0.001)	0.140 (< 0.001)
Only Child (Ref: No)	0.502 (0.324)	0.147 (0.686)	0.168 (0.216)	0.062 (0.438)
Region (Ref: Southern Shaanxi)
Northern Shaanxi	−52.583 (0.052)	−24.171 (0.063)	−18.322 (0.150)	−13.813 (0.022)
Central Shaanxi	10.480 (0.674)	1.143 (0.929)	11.135 (0.345)	1.542 (0.780)
Parental Migration (Ref: Neither)
Father only	0.031 (0.944)	−0.047 (0.882)	−0.016 (0.892)	0.071 (0.312)
Mother only	0.297 (0.771)	0.311 (0.670)	−0.190 (0.486)	0.178 (0.269)
Both parents	0.385 (0.491)	0.158 (0.692)	0.132 (0.377)	0.059 (0.506)
Left-behind Child (Ref: No)	−1.500 (0.001)	−0.944 (0.004)	−0.302 (0.014)	−0.152 (0.037)
Weekly Gaming (hours)	−0.003 (0.975)	0.000 (0.998)	−0.003 (0.888)	0.008 (0.559)
Mother's Education	0.581 (0.069)	0.515 (0.024)	0.022 (0.795)	0.058 (0.250)
Father's Education	0.293 (0.372)	0.184 (0.433)	0.095 (0.281)	0.005 (0.922)
Educational Stage	−4.160 (0.748)	0.183 (0.977)	−11.127 (0.072)	5.103 (0.079)
Peer Health Literacy	−17.513 (< 0.001)	−12.835 (< 0.001)	−27.409 (< 0.001)	−29.662 (< 0.001)
R2	53.44%	41.28%	77.12%	79.89%

Due to listwise deletion for missing values on covariates (parental migration status, mother's education, father's education), the effective sample size for this analysis is N = 1,181. All other tables (Tables 3, 4, 5, 6 and 7) using the full sample (N = 3,307) are clearly indicated

Results

Descriptive statistics

Because some background variables contained missing responses, percentages for selected variables in Table 1 were calculated based on the number of non-missing observations for that variable rather than the full sample size. The prevalence of adequate health literacy among school-aged children and adolescents was 23.37%. Specifically, it was 33.79% for health knowledge, 10.85% for healthy lifestyle & behaviors, and 54.26% for basic skills. Table 1 presents the basic characteristics of the sample (N = 3307). Herman's single-factor test indicated no severe CMB, with the first factor accounting for 25.44% of the variance. The ICCs for health literacy and its three dimensions ranged from 0.183 to 0.416, all exceeding the 0.138 threshold for high correlation, justifying multilevel modeling.

Table 1.

Basic characteristics of the sample (N = 3307)

Variable	Category	N (%)
Gender	Male	1682 (50.86%)
	Female	1625 (49.14%)
Age (years)	≤ 6	5 (0.15%)
	7–10	945 (28.58%)
	11–14	1270 (38.40%)
	15–18	1085 (32.81%)
	> 18	2 (0.06%)
Educational Stage	Primary School	1350 (40.82%)
	Junior High School	977 (29.55%)
	Senior High School	980 (29.63%)
Residential Area	Urban	589 (25.69%)
	Rural	1704 (74.31%)
Ethnicity	Han	3275 (99.18%)
	Korean	3 (0.09%)
	Manchu	4 (0.12%)
	Hui	17 (0.51%)
	Other	3 (0.09%)
Weekly Internet Time	≤ 3 h	941 (79.59%)
	> 3 h	240 (20.41%)
Parental Migration Status	Father only migrated	317 (26.84%)
	Mother only migrated	35 (2.96%)
	Both parents migrated	144 (12.19%)
	Neither parent migrated	685 (58.00%)
Mother's Education	Primary school or below	327 (27.69%)
	Junior high school	372 (31.50%)
	Senior high school/Technical school	212 (17.95%)
	Associate degree	59 (5.00%)
	Bachelor's degree	98 (8.30%)
	Master's degree	3 (0.25%)
	Doctoral degree	7 (0.59%)
	Unknown	103 (8.72%)
Father's Education	Primary school or below	151 (12.79%)
	Junior high school	465 (39.37%)
	Senior high school/Technical school	277 (23.45%)
	Associate degree	74 (6.27%)
	Bachelor's degree	134 (11.35%)
	Master's degree	5 (0.42%)
	Doctoral degree	6 (0.51%)
	Unknown	129 (10.92%)

Percentages were calculated using non-missing responses for each variable; therefore, totals may not sum to 100% of the full sample

Multilevel regression

Table 2 presents the multilevel regression results. Specifically, a one-unit increase in peer health literacy context was associated with a 17.513-unit decrease in overall health literacy (β = −17.513, P < 0.001), a 12.835-unit decrease in health knowledge (β = −12.835, P < 0.001), a 27.409-unit decrease in healthy lifestyle and behaviors (β = −27.409, P < 0.001), and a 29.662-unit decrease in basic skills (β = −29.662, P < 0.001). These findings indicate that differences in classroom peer health literacy context corresponded to differences in adolescents’ health literacy outcomes in this cross-sectional sample. Several covariates were also significantly associated with health literacy. Compared with rural students, urban students showed 1.206 units higher overall health literacy (β = 1.206, P < 0.001), 0.712 units higher health knowledge (β = 0.712, P < 0.001), 0.264 units higher healthy lifestyle and behaviors (β = 0.264, P < 0.001), and 0.140 units higher basic skills (β = 0.140, P < 0.001). In contrast, left-behind children showed 1.500 units lower overall health literacy (β = −1.500, P = 0.001), 0.944 units lower health knowledge (β = −0.944, P = 0.004), 0.302 units lower healthy lifestyle and behaviors (β = −0.302, P = 0.014), and 0.152 units lower basic skills (β = −0.1520, P = 0.037).

Multilevel mediation analysis

The mediation analysis revealed distinct patterns across health literacy dimensions (Tables 3, 4, 5 and 6). At the group level, a one-unit increase in peer health literacy context was associated with an estimated 0.052-unit decrease in overall health literacy through internet health information acquisition (95% CI: −0.081, −0.027). For health knowledge, the corresponding indirect association was an estimated 0.043-unit decrease (95% CI: −0.075, −0.017). For healthy lifestyle and behaviors, the estimated indirect association was a 0.075-unit decrease (95% CI: −0.156, −0.001), although this pattern was weaker and close to the threshold of statistical significance. No meaningful indirect association was observed for basic skills (β = 0.000, 95% CI: −0.001, 0.001). Across all four outcomes, individual-level indirect associations were non-significant, suggesting that the mediating pathway was primarily observable at the classroom/contextual level rather than at the individual-within-class level. Overall, these indirect associations were modest in magnitude, but they indicate that differences in peer health literacy context were linked to small yet measurable differences in adolescents’ health literacy, particularly in overall health literacy and health knowledge, through internet health information acquisition.

Table 3.

Results of multilevel mediation analysis in health literacy

Path	β	SE	P-value	95% CI(MCLL, MCUL)
Group Level
Direct effect
Peer → mediator	−0.55	0.080	< 0.001	−0.706, −0.394
Mediator → health literacy	0.094	0.021	< 0.001	0.052, 0.135
Peer → health literacy	−2.449	0.062	< 0.001	−2.570, −2.328
Indirect effect
Peer → mediator → health literacy	−0.052	0.014	< 0.001	−0.081, −0.027
Individual Level
Direct effect
Peer → mediator	0.038	0.007	< 0.001	0.024, 0.053
Mediator → health literacy	0.065	0.068	0.346	−0.072, 0.201
Peer → health literacy	0.143	0.003	< 0.001	0.137, 0.149
Indirect effect
Peer → mediator → health literacy	0.003	0.003	0.358	−0.003, 0.008

Abbreviations: SE Standard Error, CI Confidence Interval, MCLL/MCUL Monte Carlo Lower/Upper Limit

Table 4.

Results of multilevel mediation analysis in health knowledge

Path	β	SE	P-value	95% CI (MCLL, MCUL)
Group Level
Direct effect
Peer → mediator	−0.503	0.103	< 0.001	−0.705, −0.302
Mediator → health knowledge	0.086	0.024	< 0.001	0.039, 0.132
Peer → health knowledge	−2.714	0.088	< 0.001	−2.887, −2.541
Indirect effect
Peer → mediator → health knowledge	−0.043	0.015	0.004	−0.075, −0.017
Individual Level
Direct effect
Peer → mediator	0.059	0.012	< 0.001	0.036, 0.083
Mediator → health knowledge	−0.009	0.084	0.916	−0.176, 0.158
Peer → health knowledge	0.237	0.006	< 0.001	0.224, 0.249
Indirect effect
Peer → mediator → health knowledge	−0.001	0.005	0.918	−0.010, 0.009

Abbreviations: SE Standard Error, CI Confidence Interval, MCLL/MCUL Monte Carlo Lower/Upper Limit

Table 5.

Results of multilevel mediation analysis in healthy lifestyle & behaviors

Path	β	SE	P-value	95% CI (MCLL, MCUL)
Group Level
Direct effect
Peer → mediator	−2.172	0.273	< 0.001	−2.707, −1.638
Mediator → healthy lifestyle & behaviors	0.034	0.017	0.049	0.000, 0.069
Peer → healthy lifestyle & behaviors	−12.764	0.175	< 0.001	−13.107, −12.422
Indirect effect
Peer → mediator → healthy lifestyle & behaviors	−0.075	0.039	0.058	−0.156, −0.001
Individual Level
Direct effect
Peer → mediator	0.119	0.023	< 0.001	0.074, 0.165
Mediator → healthy lifestyle & behaviors	−0.098	0.072	0.178	−0.241, 0.046
Peer → healthy lifestyle & behaviors	0.418	0.011	< 0.001	0.397, 0.439
Indirect effect
Peer → mediator → healthy lifestyle & behaviors	−0.012	0.009	0.194	−0.030, 0.005

Abbreviations: SE Standard Error, CI Confidence Interval; MCLL/MCUL, Monte Carlo Lower/Upper Limit

Table 6.

Results of multilevel mediation analysis in basic skills

Path	β	SE	P-value	95% CI (MCLL, MCUL)
Group Level
Direct effect
Peer → mediator	−2.061	0.465	< 0.001	−2.972, −1.15
Mediator → basic skills	0.000	0.000	0.952	−0.001, 0.001
Peer → basic skills	0.813	0.005	< 0.001	0.803, 0.822
Indirect effect
Peer → mediator → basic skills	0.000	0.001	0.953	−0.001, 0.001
Individual Level
Direct effect
Peer → mediator	0.247	0.055	< 0.001	0.137, 0.356
Mediator → basic skills	−0.102	0.060	0.097	−0.223, 0.019
Peer → basic skills	0.825	0.019	< 0.001	0.788, 0.862
Indirect effect
Peer → mediator → basic skills	−0.025	0.016	0.122	−0.059, 0.004

Abbreviations: SE Standard Error, CI Confidence Interval, MCLL/MCUL Monte Carlo Lower/Upper Limit

Multilevel moderated mediation analysis

Contextual factors showed significant conditional variation in the indirect association linking peer health literacy context to health literacy through internet health information acquisition, particularly at the individual level (Table 7). For overall health literacy, the negative indirect association was more pronounced among left-behind children (β = −0.094, 95% CI: −0.169, −0.038), indicating that the cross-sectional indirect association through internet health information acquisition was stronger under left-behind conditions. A similar pattern was observed for health knowledge, for which left-behind status was associated with a stronger negative indirect association (β = −0.129, 95% CI: −0.243, −0.044).

Table 7.

Results of multilevel mediation and moderation analysis

Outcome	Moderator	Individual Level Indirect Effect			Group Level Indirect Effect
		β	MCLL	MCUL	β	MCLL	MCUL
Health literacy	Left-behind	−0.094	−0.169	−0.038	0.000	−0.001	0.002
	Perceived Teacher Support	0.000	−0.015	0.016	0.001	−0.001	0.003
	Top 100 County	−0.091	−0.154	−0.041	0.001	−0.002	0.006
	Disposable Income per Capita	0.075	0.039	0.118	0.001	−0.001	0.003
	Regional GDP Level	0.004	−0.025	0.033	−0.001	−0.003	0.001
	Health-Promoting County	0.091	0.040	0.152	−0.001	−0.006	0.002
	National Health School	−0.081	−0.156	−0.020	−0.001	−0.007	0.003
	Provincial Health School	0.088	0.042	0.140	−0.001	−0.006	0.002
	Regional Health Literacy Level	−0.004	−0.033	0.023	0.001	−0.001	0.003
Health knowledge	Left-behind	−0.129	−0.243	−0.044	0.000	−0.003	0.003
	Perceived Teacher Support	0.003	−0.021	0.035	0.000	−0.005	0.003
	Top 100 County	−0.126	−0.224	−0.049	0.000	−0.006	0.005
	Disposable Income per Capita	0.097	0.042	0.163	0.000	−0.004	0.004
	Regional GDP Level	0.007	−0.035	0.050	0.000	−0.003	0.003
	Health-Promoting County	0.126	0.047	0.220	0.000	−0.006	0.006
	National Health School	−0.131	−0.251	−0.037	0.000	−0.007	0.007
	Provincial Health School	0.124	0.050	0.208	0.000	−0.006	0.007
	Regional Health Literacy Level	−0.007	−0.050	0.032	0.000	−0.003	0.003
Healthy lifestyle & behaviors	Left-behind	−0.032	−0.109	0.016	−0.001	−0.010	0.005
	Perceived Teacher Support	0.009	−0.028	0.065	0.000	−0.005	0.005
	Top 100 County	−0.037	−0.108	0.006	−0.010	−0.032	0.004
	Disposable Income per Capita	0.051	0.000	0.114	−0.003	−0.014	0.005
	Regional GDP Level	−0.010	−0.047	0.017	0.005	−0.002	0.016
	Health-Promoting County	0.037	−0.006	0.104	0.010	−0.004	0.032
	National Health School	−0.041	−0.130	0.017	0.009	−0.007	0.036
	Provincial Health School	0.022	−0.015	0.077	0.012	−0.005	0.039
	Regional Health Literacy Level	0.010	−0.017	0.045	−0.005	−0.016	0.002
Basic skills	Left-behind	−0.067	−0.203	0.022	−0.003	−0.024	0.014
	Perceived Teacher Support	−0.011	−0.093	0.049	−0.009	−0.031	0.005
	Top 100 County	−0.074	−0.196	0.004	−0.022	−0.068	0.008
	Disposable Income per Capita	0.033	−0.014	0.101	−0.011	−0.042	0.010
	Regional GDP Level	0.025	−0.024	0.090	0.011	−0.004	0.034
	Health-Promoting County	0.074	−0.004	0.190	0.022	−0.008	0.068
	National Health School	0.061	−0.044	0.204	0.029	−0.005	0.085
	Provincial Health School	0.084	0.001	0.207	0.026	−0.006	0.078
	Regional Health Literacy Level	−0.025	−0.088	0.021	−0.011	−0.034	0.005

Several contextual indicators were associated with a weaker negative indirect association. For overall health literacy, higher disposable income per capita corresponded to a less negative indirect association (β = 0.075, 95% CI: 0.039, 0.118), and the same pattern was observed for health knowledge (β = 0.097, 95% CI: 0.042, 0.163). Similarly, health-promoting county status was associated with a weaker negative indirect association for overall health literacy (β = 0.091, 95% CI: 0.040, 0.152) and health knowledge (β = 0.126, 95% CI: 0.047, 0.220). Provincial healthy school designation also corresponded to a weaker negative indirect association for overall health literacy (β = 0.088, 95% CI: 0.042, 0.140) and health knowledge (β = 0.124, 95% CI: 0.050, 0.208).

By contrast, national healthy school status corresponded to a more negative indirect association for overall health literacy (β = −0.081, 95% CI: −0.156, −0.020) and health knowledge (β = −0.131, 95% CI: −0.251, −0.037). Top 100 county status was also associated with a stronger negative indirect association for overall health literacy (β = −0.091, 95% CI: −0.154, −0.041) and health knowledge (β = −0.126, 95% CI: −0.224, −0.049).

For healthy lifestyle and behaviors and for basic skills, the moderated mediation patterns were more limited. Among these outcomes, only disposable income per capita showed a marginally weaker negative indirect association for healthy lifestyle and behaviors (β = 0.051, 95% CI: 0.000, 0.114), whereas provincial healthy school designation showed a small positive conditional indirect association for basic skills (β = 0.084, 95% CI: 0.001, 0.207). No significant moderated mediation was observed for perceived teacher support, regional GDP level, or regional health literacy level across the four outcomes. Taken together, these findings suggest that the indirect association between peer health literacy context and adolescent health literacy varied depending on whether adolescents were in more vulnerable or more resource-supportive contexts, although all such patterns should be interpreted as conditional associations observed in cross-sectional data.

Summary of hypothesis testing

Overall, Hypothesis 1 was supported, as peer health literacy context showed a significant negative association with adolescents’ health literacy. Hypothesis 2 was partially supported: internet health information acquisition showed a significant indirect pathway for overall health literacy and health knowledge, limited evidence for healthy lifestyle and behaviors, and no significant mediation for basic skills. Hypothesis 3 was not supported, as perceived teacher support did not significantly moderate the mediated pathway. Hypothesis 4 was partially supported, as several policy- and economy-related contextual variables, as well as left-behind status, significantly moderated the indirect pathway.

Discussion

Main findings

Using a multilevel moderated mediation framework, this study examined how peer health literacy context was associated with adolescents’ health literacy and whether internet health information acquisition served as a linking pathway under different ecological conditions. Three main findings emerged. First, a higher peer health literacy context was associated with lower individual health literacy. Second, internet health information acquisition partially explained this pattern, particularly for overall health literacy and health knowledge. Third, this pathway varied across contexts: left-behind status intensified the negative indirect pathway, whereas several economic and policy-related contextual factors weakened it. By contrast, perceived teacher support did not show a significant moderating effect.

Multilevel regression

The results supported Hypothesis 1, confirming a significant negative association between peer health literacy level and individual health literacy. At first glance, this finding appears to contradict previous studies that have generally concluded that peer-based programs improve health literacy [17–19, 24]. However, this apparent contradiction can be reconciled by recognizing the fundamental difference between organized peer interventions (which are designed to support and educate) and the spontaneous influence of the peer health literacy environment (which serves as a reference standard for social comparison).

From the perspective of Social Comparison Theory [12], the peer health literacy level establishes a clear standard for upward comparison. For students with lower individual health literacy, this high standard may not motivate improvement but instead trigger a contrast effect [47]. Individuals assess their relative position within the peer group [48]. Perceiving a large, potentially unbridgeable gap may lead students to adopt avoidance strategies, steering clear of health information and behaviors that highlight their shortcomings, thereby resulting in lower health literacy [49].

This counterintuitive finding aligns with a well-established phenomenon in international educational psychology known as the "Big-Fish-Little-Pond Effect" [2, 49]. It describes how students in academically selective schools (high-ability peer environments) tend to have lower academic self-concepts compared to equally able students in non-selective schools. Similarly, research in the Singapore context has demonstrated that passively observing high-performing peers triggers social comparison and leads to negative self-evaluation [50]. These parallel findings from Western and Asian educational settings suggest that the negative contrast effect we observed is not culturally unique but represents a general social-psychological process that operates across contexts.

Multilevel mediation

The core findings of this study stem from the multilevel mediation analysis. Hypothesis 2 was partially supported, revealing a significant within-level mediating pathway. Unlike conventional mediation analyses, multilevel mediation parsing of within- and between-group paths helps elucidate the complex mediating role of internet health information acquisition. More specifically, the indirect pathway was evident primarily for overall health literacy and health knowledge, whereas the evidence was weaker for healthy lifestyle and behaviors and absent for basic skills. This pattern suggests that internet health information acquisition may be more closely linked to cognitive and knowledge-based aspects of health literacy than to all dimensions equally.

The peer health literacy environment demonstrated a negative within-level association with internet health information acquisition, indicating that being in a class with a higher average health literacy suppresses an individual's ability to acquire online health information. From a social comparison perspective, comparing themselves to peers with higher health literacy may not motivate learning but instead foster frustration and diminished self-efficacy due to the perceived large gap, thereby reducing their confidence and motivation to seek health information online independently.

Concurrently, this psychological mechanism may be reinforced by observable peer behaviors. Prior research indicates that in groups with higher collective health or e-health literacy, individuals often rely on trusted peers as filters and interpreters of online information, rather than engaging in unfiltered personal searches [51, 52]. These peers act as primary information conduits, reducing the necessity for others to navigate the vast and often unvetted digital landscape on their own [53, 54]. Thus, a high peer health literacy environment may not only undermine the individual's motivation to seek information online (as per social comparison) but also reduce the practical need to do so, as reliable information is socially sourced and vetted within the peer network.

Second, internet health information acquisition was positively associated with health literacy. The digital age has not only transformed how health information is disseminated but also opened new avenues for enhancing public health literacy. Accessing health information online provides a practical arena and reinforcement opportunities for the skill dimension of health literacy. For instance, some scholars found that the act of conducting more frequent online health information searches itself is a significant predictor of higher levels of health literacy [55]. This indicates that practice enhances confidence and competence in evaluating and utilizing information. Hansen et al. documented in their early observational study of adolescents how they employed trial-and-error learning strategies during searches—a process that constitutes vital competency development [56]. When addressing specific health needs, online information access directly empowers individuals as a means to enhance health literacy. For populations with relatively weak health knowledge, the internet can rapidly bridge information gaps. For instance, parents from low-income families or those with children having special health needs can better understand their children's conditions and care options by searching online, directly enhancing their concrete health management capabilities [57]. More significantly, accessing online health information can stimulate the development of critical health literacy. When individuals encounter diverse or even conflicting online information, they naturally activate comparison and evaluation mechanisms. Research indicates that adolescents experience emotions like anxiety or relief after searching online and express a desire to discuss findings with health professionals [58]. This reflects the activation of their health information awareness and the need to identify credible authorities—a crucial step in developing critical literacy. This improvement does not occur automatically, its effectiveness is influenced by the quality of information resources, individuals' initial skills, and the presence of professional guidance. Research indicates that the public strongly desires guidance from healthcare professionals to identify reliable online information [59, 60]. This suggests that transforming aimless online browsing into conscious, resource-supported information acquisition is key to maximizing its benefits for literacy enhancement.

Finally, the mediating pathway of internet health information acquisition was negative. Although internet health information acquisition is beneficial for health literacy, the contrast link generated by a high peer health literacy context suggests an inhibitory association with internet health information acquisition. Ultimately, this pathway results in a negative indirect relationship. This indicates that to leverage the role of internet health information acquisition, the individual's external environment must be recognized. Consequently, we further analyzed how different levels of the ecological system moderate this mediating pathway.

Although the observed coefficients were modest in magnitude, small effects should not be interpreted as trivial in population-based educational and public health research. Health literacy is shaped by multiple interacting influences, and peer context is only one component within a broader ecological system. Even relatively small contextual effects may accumulate over time or become meaningful when considered across large student populations. Therefore, the practical significance of these findings lies less in large individual-level shifts and more in identifying peer context as a measurable contextual factor that may contribute incrementally to adolescents’ health literacy development.

Multilevel moderated mediation

Building on the findings from the multilevel mediation analysis, this study further detected that the mediated pathway is moderated by external contexts. Based on the multilevel regression results and the tenets of ecosystem theory, we specified these external contexts as left-behind status, perceived teacher support, and economic and policy factors. We found that left-behind status, economic, and policy factors moderate this pathway at the within level, confirming Hypothesis 4.

Our results indicate that left-behind status exacerbates the negative association in this pathway. Peer groups play a significant role in the development of left-behind children, exerting complex influences spanning social, psychological, and educational dimensions [61]. Firstly, due to the physical distance from their migrant parents, left-behind children lack direct parental protection and may consequently become targets of peer bullying [62, 63]. Peers may engage in verbal aggression or physical bullying, severely impacting the mental health of left-behind children [63]. Such experiences may foster resistance to peer health literacy among left-behind children, where their status further intensifies the contrast effect.

Economic and policy factors weakened this negative pathway. This finding can be understood through an integrated lens of social comparison and ecosystem theories. The role of the external environment essentially operates by altering the psychological context and behavioral consequences of social comparison for the individual, potentially transforming a negative "contrast effect" into a positive "assimilation effect." According to ecosystem theory, individual development is nested within multi-layered environments from the micro- to the macro-system. In this study, classroom peer health literacy constitute the microsystem most directly influencing the individual, while regional economic levels and health policies belong to the exo- and macro-systems. These higher-level systems, while not interacting directly with the individual, set the stage and provide resources for activities within the microsystem, shaping the nature and outcomes of the social comparison process occurring within it.

Specifically, when a region enjoys a higher level of economic development, it implies that individuals have more universal and equitable access to resources supporting health development, such as convenient internet access and smart terminal devices. In regions with higher levels of economic development, residents typically enjoy more widespread internet access and more advanced smart devices [8], which lowers the barriers to accessing health information [9]. This makes it more likely for individuals to believe that their own efforts can narrow the gap with their peers, thereby converting a contrast effect into an assimilation effect.

Research indicates that robust digital infrastructure facilitates easier access to health information and online support for individuals [64], particularly for populations with special health needs [65]. This resource accessibility not only mitigates health inequalities stemming from information asymmetry [66] but also enhances individuals' confidence and capacity to improve health through self-management [67]. Consequently, it increases the likelihood of transforming social comparison pressures into positive health behavior changes [68].

Policies such as Health-Promoting Counties/Districts and Schools reshape the meaning of school health norms from the external environment. When health promotion becomes an explicit institutional arrangement, health literacy is no longer merely a matter of individual competency competition but evolves into a form of population-wide literacy supported and encouraged by top-level design. This institutional environment can effectively buffer the lateral comparison pressure among peers, creating a more supportive health culture for all students. This aligns with leading international research findings that whole-school health promotion interventions effectively improve student health behaviors and the school social environment [69–71]. For example, Bond et al. demonstrated how multi-level school interventions influence emotional well-being and health risk behaviors, confirming that early social and school connections predict subsequent positive outcomes [70, 71]. Bonell et al. demonstrated through a cluster randomized controlled trial in UK secondary schools that the “Learning Together” intervention effectively reduced bullying and aggressive behaviors [69]. The core of health-promoting regions or schools lies in their systematic approach, particularly their commitment to fostering a safe, supportive socio-emotional environment. Such environments reduce anxiety and stress from peer comparisons, transforming individual competition into shared, supportive group norms. Consequently, in high-performing health-promoting schools, accessing online information ceases to be a passive coping mechanism for stress and instead becomes an active process of constructing health literacy within a supportive environment.

An unexpected finding was that national healthy school status appeared to strengthen, rather than weaken, the association under some models. One possible explanation is that policy designation reflects institutional commitment at the school level, but does not necessarily translate into uniformly perceived support for every student. In addition, in schools where health-related norms and expectations are more visible, students may become more aware of differences in health literacy among peers, thereby intensifying social comparison processes rather than buffering them. This interpretation also suggests that formal policy status may be too coarse an indicator, and future research should incorporate measures of implementation quality and students’ actual exposure to school-based health promotion practices.

In summary, for digital health interventions, the construction of the macro-environment and the guidance of micro-psychological mechanisms must proceed simultaneously, empowering individuals by building supportive ecosystems.

It is noteworthy that Hypothesis 3 was not supported; perceived teacher support did not exhibit a significant moderating effect on the mediation pathway. This finding suggests that teacher support may operate through different mechanisms or may be less salient in moderating peer-driven social comparison processes compared to other school-level factors. Research indicates that during adolescence, peer influence may surpass that of teachers, highlighting adolescents' heightened sensitivity to peer health literacy [71]. Furthermore, more structural and institutional aspects of the school environment—such as health-promoting school policies, the availability of health resources, and a school-wide culture of health—might be more potent in moderating social comparison processes than the interpersonal dimension of teacher support. Future research should employ comprehensive school climate measures that capture multiple dimensions (e.g., peer relationships, safety, institutional environment) to more fully test the moderating role of the mesosystem.

Study contributions

This study makes three main contributions. First, it extends research on adolescent health literacy by shifting attention from organized peer interventions to the naturally occurring peer health literacy context in everyday classrooms. In doing so, it shows that peer influence is not uniformly beneficial and may, under some conditions, be associated with lower individual health literacy. Second, it highlights internet health information acquisition as an important behavioral pathway linking peer context to health literacy, thereby connecting peer environments with digital health behavior in the formation of adolescent health literacy. Third, by incorporating left-behind status and broader policy and economic conditions into a multilevel framework, this study demonstrates that peer-related associations are embedded in wider ecological environments and may vary across contextual conditions.

Limitations and future research

This study has the following limitations. First, this study is based on cross-sectional data, and this design does not permit causal inference. Although the analyses identified statistically significant associations among peer health literacy context, internet health information acquisition, and health literacy, the temporal ordering of these variables cannot be established in the present dataset. As a result, the observed direct, indirect, and moderated associations should not be interpreted as evidence of causal pathways. Reverse associations are also plausible. For instance, individuals with low health literacy might inherently avoid health information seeking, affecting their perceived peer environment. Future research should employ longitudinal panel designs or intervention experiments to observe dynamic changes in variables, thereby more rigorously revealing causal directions and long-term impact mechanisms. Second, some constructs were measured in a relatively limited way. In particular, peer health literacy context was operationalized as the average health literacy level of classmates rather than as directly perceived normative pressure, and perceived teacher support was used as a narrower indicator rather than a full measure of school climate. Future research could combine perceived norms, peer network data, and broader school climate measures to provide a more comprehensive account of adolescent health literacy formation. Third, while the sample has certain regional balance within Shaanxi Province (covering Northern, Central, and Southern Shaanxi), China's vast territory exhibits significant disparities in economy, culture, and educational resource allocation across regions. Therefore, caution is needed when generalizing the results of this study to other parts of China. Future research could validate the universality of the conclusions through cross-regional collaboration, employing stratified random sampling to obtain nationally representative samples. To enhance the feasibility and operability of this recommendation, we suggest a longitudinal design with at least three measurement timepoints (e.g., baseline, 6 months, 12 months) to establish temporal ordering and examine the stability of the mediation pathways. For sample expansion, we recommend a multi-stage stratified random sampling strategy that includes representative provinces from China's eastern, central, and western regions, ensuring national representativeness and enabling cross-regional comparisons. Such an approach would help determine whether the observed negative peer effects are consistent across different socioeconomic and cultural contexts within China. Furthermore, although the main mediation analyses used the full sample (N = 3,307), the multilevel regression model in Table 2 was based on a reduced sample (N = 1,181) due to missing data on parental migration and education covariates. Auxiliary comparisons indicated no systematic differences on core study variables between the reduced and excluded samples, but the generalizability of covariate-adjusted estimates should be interpreted with caution.

Implications for practice and policy

Based on our findings, we propose several concrete, actionable recommendations for practitioners and policymakers:

• For Schools: Rather than simply highlighting high-performing students as health role models—which may inadvertently reinforce social comparison pressures—schools should consider implementing collaborative digital health literacy curricula. For example, group-based activities where students work together to search for, evaluate, and share health information can transform potential "peer competition" into "peer cooperation." This approach leverages the benefits of peer interaction while mitigating the negative contrast effects identified in this study.

• For Policymakers: Our findings underscore the protective role of macro-level interventions. The significant moderating effects of "Health-Promoting Counties" and "Healthy Schools" designations suggest that these policies do more than provide resources—they create a supportive socio-emotional environment that buffers negative peer comparisons. Therefore, expanding and strengthening these programs, particularly in underserved areas, represents a promising strategy for promoting health equity among adolescents. For vulnerable groups such as left-behind children, targeted support—including school-based mentoring programs and digital literacy training—may help counteract the exacerbated negative peer effects we observed.

• For Health Educators: The mediating role of internet health information acquisition highlights the importance of teaching not just what health information is reliable, but how to effectively navigate the digital landscape. Structured programs that build self-efficacy in online health information seeking may empower students to transform social comparison pressures into opportunities for self-improvement.

Conclusions

This study demonstrates that peer health literacy is negatively associated with individual health literacy among Chinese adolescents, and this association is partially statistically mediated by internet health information acquisition.. Contextual factors at the policy and economic levels were found to moderate this pathway. While causal conclusions cannot be drawn from the cross-sectional design, these findings suggest that interventions fostering supportive peer environments, enhancing digital health literacy skills, and addressing structural barriers—particularly for vulnerable groups such as left-behind children—may hold promise for promoting adolescent health literacy in the digital age.

Authors’ contributions

T.X designed the analysis plan, analyzed the data and results, and wrote the final version of the manuscript. YQ.X and Y.M were involved in revising and checking the final version of the manuscript. YQ.X and L.Z were involved in formulating the primary framework of the study. All authors have read and approved the published version of the manuscript.

Funding

This study was supported by the National Natural Science Foundation of China (72474162), Major Program of National Fund of Philosophy and Social Science of China (17ZDA079).

Data availability

The datasets generated and analyzed during the current study are not publicly available due to protecting participant privacy as containing sensitive personal information but are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the Biomedical Ethics Committee, School of Medicine, Xi ‘an Jiaotong University (No. 2021–1525). Informed consent was obtained from all individual participants included in the study. After detailing our study, informed consent was obtained from all participants or their legal guardians for those below 16 years old.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.