Introduction Background

With the proliferation of the Internet, the public’s online health information seeking behavior is increasing.1,2 An increasing number of patients begin to utilize online platforms to obtain health information and bring this information into the diagnostic and treatment process for discussion with physicians.3 Public online health information seeking behavior is evolving into a pre-positioned component of modern doctor-patient communication. This paradigm shift in behavior is deconstructing the traditional doctor-patient knowledge power structure, with medical authority gradually transitioning from unidirectional transmission to bidirectional negotiation, and patients progressively transforming into participants in communication. However, the proliferation of online health information has brought about paradoxical effects.

On the positive side, patients acquiring online health information can increase medical KR, obtain others’ treatment experiences, enhance the sense of empowerment, promote more equitable communication, elevate confidence and motivation in coping with uncertainty, and improve doctor-patient communication effectiveness.2,4–6 Research indicated that patients who searched for online health information prior to medical encounters often felt more capable and confident in communicating with physicians during the consultation.7–9 However, the uneven quality and rapid proliferation of online health information can affect the doctor-patient relationship.3 Meanwhile, due to the existence of professional barriers in medicine, patients who receive false health information on the Internet may find it difficult to distinguish truth from falsehood, potentially leading them to question physicians and subsequently affecting the doctor-patient relationship.10 The organizational-level construction of health-literate medical institutions to accommodate populations with low health literacy and disseminate accurate scientific health knowledge is receiving increasing attention.11,12

Furthermore, individuals may also be affected by the Dunning-Kruger Effect after acquiring online health information.13 This effect indicates that individuals with lower cognitive ability or professional knowledge level often overestimate their own capabilities, particularly tending to generate “overconfidence” when confronted with complex or highly specialized information.14–16 Existing research has found that this effect is particularly pronounced among individuals with insufficient health literacy.16 Although some existing studies have provided valuable insights in the field of online health information, research exploring the Dunning-Kruger effect in this domain remains relatively scarce. As legal guardians of adolescents and children, parents’ health information identification and communication capabilities directly influence their children’s consultation efficiency and the quality of the doctor-patient relationship.

To conduct targeted research with precision and clarify the associative mechanisms among variables, this study provided standardized definitions of core research variables. In this study, knowledge reserve (KR) refers to the level of parental mastery of myopia control knowledge, which serves as the foundation for forming proactivity and expressive confidence in doctor-patient communication. Participation in decision-making (PD) is defined as the enhancement of parental cognition regarding their roles in medical decision-making, as well as their willingness and confidence in participating in the medical process. Perceived a community with shared future for doctor-patient (PCSF) is defined in this study as the degree of parental identification with the cooperative relationship of “consensus and shared responsibility” between doctors and patients regarding health objectives. This study defines communication preparedness (CP) as the degree of parental psychological preparedness in terms of information acquisition, expressive capability, and interaction expectations prior to medical encounters. This study enhanced the medical cognitive baseline and information critical capability of parents of children with myopia through standardized health knowledge exposure. On this basis, this study further explored the pathway differences in parental CP and their associative mechanisms across different KR levels.

Theoretical Framework Digital Health Tools

The Internet has become a nexus connecting professional medical knowledge and patient knowledge.17 Online health information prompts numerous patients to transform from passive recipients of medical services into subjects capable of actively participating in the medical process.18 Furthermore, online health information has altered the subjectively perceived hierarchy between physicians and patients.19 Patients with adequate health literacy can read, comprehend, and take action to process healthcare information.20 In China, significant disparities in digital health literacy exist across different populations, revealing the urgent need for precision health communication and personalized intervention.21,22 With the rapid proliferation of digital technology, digital health interventions have been increasingly employed to enhance public comprehension, critical judgment, and CP regarding medical information. Intelligent health question-answering systems,23 mobile health applications,24 and patient education platforms25 contribute to reducing patient misinterpretation of medical information and enhancing their understanding of professional terminology and treatment recommendations. Furthermore, digital health tools have been demonstrated to strengthen patient acceptance of physician recommendations and promote more active engagement in diagnostic and therapeutic communication in scenarios such as chronic disease management26 and vaccination.27

Based on individual differences in KR, health literacy also exhibits characteristics of hierarchical progression. Research indicated that health literacy capability was typically classified into three levels: functional literacy (the first level), interactive literacy (the second level), and critical literacy (the third level).28 The latter two levels of health literacy enable individuals to transcend the role of mere information recipients and become participants in medical decision-making equipped with information transformation capability and critical thinking. Meanwhile, self-determination theory (SDT) encompasses four interrelated sub-theories: cognitive evaluation, organismic integration, psychological needs, and causality orientations. Among these, organismic integration theory posits that individual self-integration is influenced by environmental factors across a continuum from low to high self-determination.29–31 The degree to which different individuals’ motivations shift from external control to internal identification under environmental influence varies. Therefore, this study further analyzed the differences in PD, PCSF, and CP across groups with different KR under standardized health knowledge exposure. In summary, this study proposed the following hypothesis:

H1: Parents with high KR may exhibit higher levels of PD, PCSF, and CP compared to parents with low KR.

Dunning-Kruger Effect

The Dunning-Kruger Effect is a cognitive bias phenomenon describing how individuals tend to overestimate their own KR or skill levels when they possess insufficient capability in a particular domain. Its underlying mechanism originates from metacognitive deficit, whereby individuals with low capability are unable to accurately assess the boundaries of their own knowledge.14–16 As KR levels increase, individuals’ self-assessment becomes progressively more accurate. Previous studies employed the Dunning-Kruger Effect to explain the relationship between overconfident populations and opposition to vaccination policies, and found that such overconfidence was most pronounced among individuals with insufficient KR and those who strongly identified with misinformation.32 This effect may be equally applicable in the domain of health behavior decision-making. When KR is low, individuals who acquire only fragmented medical information may be influenced by the Dunning-Kruger Effect, generating unrealistic communication confidence. This study positioned this effect as a moderating mechanism in the “KR → CP” pathway. In summary, this study proposed the following hypothesis:

H2: Among parents with low KR, KR may be negatively correlated with CP; among parents with high KR, KR may be positively correlated with CP.

Self-Determination Theory

SDT, proposed by Gagné, Deci,33 aims to elucidate the motivational mechanisms underlying human self-determined behavior. This theory posits that the motivational process of human self-determined behavior is driven by three basic psychological needs: competence, autonomy, and relatedness.34 Within the framework of SDT, patients are regarded as behavioral subjects possessing self-determination capability.35 “Competence” refers to the medical knowledge and judgment capability required for individuals to perform tasks in relevant domains. Furthermore, autonomy represents genuine willingness to participate based on foundational knowledge and judgment capability, reflecting individuals’ intrinsic motivation and self-decision-making capability. This is fundamentally distinct from the “false confidence” resulting from the Dunning-Kruger Effect: the former is driven by thorough comprehension and internal motivation, whereas the latter originates from cognitive misjudgment. Existing empirical evidence also indicated that individuals with higher health literacy levels tended to establish bidirectional communication relationships and promote informed decision-making.36 Third, relatedness refers to individuals’ psychological needs for belonging and interpersonal connection.37 In this study, as legal guardians during adolescents’ diagnostic and treatment processes, parents’ understanding and perception of the doctor-patient relationship may be directly associated with their degree of CP. Therefore, parents who are well-informed and possess certain medical cognition are more inclined toward PD and establishing favorable relationships with physicians, and are more likely to demonstrate positive communication attitudes and behaviors. In summary, this study proposed the following hypothesis:

H3: PD and PCSF may mediate the relationship between KR and CP.

Methods Research Design and Materials

This study adopted a cross-sectional quasi-experimental design, simulating real-world scenarios through standardized digital knowledge exposure. This study did not establish a concurrent control group or baseline measurement, focusing primarily on assessing immediate effects following single knowledge exposure and their associated mechanisms, without directly isolating the independent contributions of digital content from participants’ pre-existing knowledge, attitudes, and behavioral tendencies.

The research team previously developed a set of science popularization videos on myopia control encompassing the entire process of pre-consultation, consultation, and post-consultation, based on physician recommendations and parental actual needs. All videos adhered to unified filming standards and were published on the BiliBili website (https://www.bilibili.com/). All parents were required to complete the designated video learning prior to questionnaire completion to fulfill the standardized knowledge exposure procedure.

Viewing Behavior and Quality Control

This study utilized objectively recorded page dwell time from the platform backend as the criterion for video viewing behavior, rather than relying on participants’ subjective self-assessment. To accurately identify valid viewing behavior, this study set the threshold for valid dwell time marginally above the actual video playback duration, combined with a mandatory page retention mechanism that restricted participants from exiting or switching pages prior to video completion. This design ensured that all participants were fully exposed to the experimental stimulus, effectively circumventing potential biases such as fast-forwarding, idling, or mid-session interruption that may occur under free-browsing modes, thereby standardizing and homogenizing viewing behavior at the procedural level.

The questionnaire comprised two sections. The first section consisted of self-designed items covering the child’s age, optical correction methods (such as orthokeratology, defocus spectacles, and frame spectacles), and pharmacological treatment modalities (such as atropine eye drops and pirenzepine eye drops), as well as the respondent’s educational attainment. The second section comprised scales. All scale items were preceded by the prefix “Through learning from the myopia science popularization videos,”. Given that this study employed single-timepoint observation without longitudinal tracking design, KR served as the core predictor variable, scored in open-ended question format with answers derived from the aforementioned science popularization videos. The questionnaire also included an introductory statement and informed consent, with a commitment to confidential handling of all collected information. Additionally, a common-sense question was specifically embedded in the questionnaire to assess participants’ response attentiveness and diligence.

Participants and Sampling

This study conducted surveys in representative major cities across five regions of China - eastern, southern, western, northern, and central (Shanghai, Chengdu, Guangzhou, Beijing, and Wuhan) - via the Wenjuanxing platform (https://www.wjx.cn). Constrained by research conditions, this study employed convenience sampling to recruit participants without random sampling. The external validity of the research findings carried certain limitations. A total of 529 questionnaires were collected in this survey, of which 524 were valid, yielding an effective response rate of 99.05%. All participants voluntarily participated in the survey under the premise of fully informed consent. If participants declined the informed consent, their response process was automatically terminated by the system.

Quality Control

This study utilized the VIP function of the Wenjuanxing platform to establish stringent rules for screening and excluding invalid questionnaires. Specific rules were as follows: selecting “No” for informed consent, questionnaire IP address mismatching the planned survey city, questionnaire completion time below 233 seconds (encompassing short video viewing, informed consent clause reading, basic information completion, and scale response time), incorrect response to the common-sense question (emergency ambulance telephone number), and selection of identical answers across all scale items. Beyond the aforementioned automatic screening rules, this study also employed manual verification by comparing respondents’ personal identification codes (composed of name initials and last four digits of mobile phone number), IP addresses, cities of residence, gender, age, and educational attainment to determine whether duplicate questionnaire submissions by the same individual existed. Furthermore, to control common method bias, this study adopted an anonymous survey approach and avoided employing questions that might elicit neutral responses.

Measurement Instruments

Except for myopia knowledge evaluation which employed open-ended questions, all other measurement instruments utilized scales. All scales used in this study were informed by prior research.38–40 In this study, the PD scale was used to measure participants’ cognitive enhancement regarding autonomy in medical decision-making and changes in participation confidence. The PCSF scale was used to assess participants’ cognition of doctor-patient cooperative awareness and identification with shared responsibility. The CP scale was used to evaluate changes in participants’ preparedness when communicating with physicians, inquiring, and expressing their own health needs. Definitions of the aforementioned core terms were presented in Table 1.

Table 1 Definitions of the Aforementioned Core Terms

Statistical Methods

This study conducted visual inspection of normality distribution for continuous variables through P-P plots and Q-Q plots. Data analysis was performed using SPSS 27.0, AMOS 24, and the lavaan package in RStudio.

This study split the sample according to data entry sequence, selecting the first 262 cases for confirmatory factor analysis and the remaining cases for exploratory factor analysis. Subsequently, this study employed confirmatory factor analysis to construct multi-factor competitive models and examine the discriminant validity of each variable. Meanwhile, this study utilized reliability analysis to assess the reliability and stability of the scales. Furthermore, this study conducted common method bias testing through Harman’s single-factor test.

In structural equation modeling and multilevel analysis research, determining critical thresholds through statistical testing or empirical criteria to accomplish continuous variable grouping represented a widely adopted analytical paradigm.41 In exploratory research, moderate grouping can enhance the interpretability of specific mechanism pathways.42 This study first employed the generalized additive model to examine the association pattern between KR and CP, identifying the existence of a nonlinear U-shaped effect and a distinct inflection point, and further determined the optimal critical segmentation value through threshold regression modeling. Subsequently, based on the data-driven objective threshold criterion, this study dichotomized the continuous KR variable into low and high KR groups for inclusion in subsequent structural equation modeling, thereby exploring the differentiated characteristics of variable pathway mechanisms under the Dunning-Kruger Effect.

This study constructed structural equation models separately for the high and low KR groups using AMOS 24.0. Furthermore, to control potential confounding factors and ensure between-group comparability, this study introduced children’s age, parental educational attainment, prior optical correction methods (orthokeratology/defocus spectacles/frame spectacles), and pharmacological treatment methods (atropine/pirenzepine, etc.) as exogenous covariates into the structural equation system via the lavaan package in R Studio. Additionally, this study comprehensively evaluated model adaptation effects across different KR groupings by integrating overall model fit indices.

Results Validity Analysis Structural Validity Analysis

This study conducted exploratory factor analysis using the maximum variance orthogonal rotation method. The KMO coefficients for all scales exceeded 0.6, satisfying the applicability conditions for factor analysis. With eigenvalues greater than 1 as the extraction criterion, all scales yielded only a single common factor, with cumulative variance explanation rates ranging from 78.148% to 79.280%. Therefore, the factor structures of all scales were acceptable (Table 2). Confirmatory factor analysis results indicated that PD and PCSF were saturated models with zero degrees of freedom, precluding the output of model fit parameters. The fit indices for CP were as follows: SRMR = 0.031, CFI = 0.974, TLI = 0.921, all surpassing the critical reference standards.

Table 2 Exploratory Factor Analysis

Convergent Validity Analysis

This study conducted convergent validity analysis using AMOS 24.0. Results indicated that the standardized factor loadings for all scales exceeded 0.4, AVE values exceeded 0.5, and CR values exceeded 0.6. Therefore, all scales demonstrated acceptable convergent validity (Table 3).

Table 3 Convergent Validity Analysis

Discriminant Validity Analysis

Table 4 showed that the square roots of AVE for PD, PCSF, and CP were 0.840, 0.833, and 0.832, respectively, all exceeding the maximum absolute value of inter-factor correlation coefficients. Therefore, all scales demonstrated acceptable discriminant validity (Table 4).

Table 4 Discriminant Validity Analysis

Table 5 Comparison of Fit Indices Across Different Factor Merger Models

As the measurement model iteratively changed from a single-factor structure toward a full-factor merged structure, the χ2/df values progressively increased, the TLI exhibited a gradual decline, and the SRMR showed a stepwise upward trend. Based on the comprehensive characteristics of changes in all fit indices, the baseline model without factor merging demonstrated the optimal fit (Table 5).

Reliability Analysis

The Cronbach’s α coefficients for all scales ranged from 0.861 to 0.901, all exceeding 0.7. After deleting any single item, the variation range of Cronbach’s α remained within acceptable limits. The corrected item-total correlations (CITC) for all scales exceeded 0.4. In summary, all scales demonstrated acceptable reliability (Table 6).

Table 6 Reliability Analysis

Common Method Bias

This study employed Harman’s single-factor test to determine whether common method bias existed in the data.43 Results indicated that the first principal component explained only 47.724%, below the recommended threshold of 50%.44 This suggested that severe common method bias was not present in the data.

Nonlinear Effect of KR and Group Delineation

A significant U-shaped nonlinear association existed between KR and CP, indicating the presence of a distinct effect turning point. To further objectively locate the position of this inflection point, this study executed threshold regression modeling, ultimately determining 8 points as the critical threshold for KR. Based on this data-driven objective result, this study dichotomized the research sample, defining KR scores of 0–8 (inclusive) as the low KR group and scores greater than 8 as the high KR group, thereby providing an objective grouping basis for subsequent structural equation modeling comparative analysis. This grouping scheme not only validated the necessity of KR as a key driving mechanism in the theoretical model, but also provided empirical evidence for subsequently examining model adaptability across populations with different health literacy levels.

Demographic and Clinical Characteristics of Parents in Low and High KR Groups

In the low KR group, participants’ children were predominantly aged 0–5 years (42.5%) and 6–14 years (41.7%). The educational attainment of these participants was mainly concentrated at the associate degree level (42.5%) and bachelor’s degree level (32.3%). The majority of their children had not utilized optical correction methods (70.9%) or pharmacological treatment methods (81.9%). In the high KR group, participants’ children were predominantly aged 6–14 years (43.3%) and 15 years and above (35.8%). The educational attainment of these participants was similarly concentrated at the associate degree level (38.5%) and bachelor’s degree level (45.3%). The majority of their children had utilized optical correction methods (59.4%) and pharmacological treatment methods (59.9%) (Table 7).

Table 7 Demographic and Clinical Characteristics of Parents in Low and High KR Groups

Correlation Analysis

Table 8 showed that in the low KR group, the mean KR was 5.64 (SD = 2.18); the means of PD, PCSF, and CP were 3.35 (SD = 1.00), 3.59 (SD = 0.89), and 3.50 (SD = 0.98), respectively. In the high KR group, the mean KR was 12.20 (SD = 1.80); the means of PD, PCSF, and CP were 3.79 (SD = 0.89), 4.00 (SD = 0.90), and 4.03 (SD = 0.88), respectively, all exceeding those in the low KR group.

Through Q-Q plots and P-P plots, combined with absolute kurtosis values below 10 and absolute skewness values below 3 for all data, the distribution of all data approximated normality (Table 8). Therefore, this study adopted Pearson correlation analysis.

Table 8 Skewness and Kurtosis

In the low KR group, KR was negatively correlated with PD, but without statistical significance (r = −0.068, P > 0.05). KR was significantly negatively correlated with CP (r = −0.246, P < 0.01); both PD (r = 0.411, P < 0.001) and PCSF (r = 0.516, P < 0.001) were significantly positively correlated with CP. Furthermore, PCSF was significantly positively correlated with PD (r = 0.328, P < 0.001).

In the high KR group, KR was positively correlated with CP, but without statistical significance (r = 0.097, P > 0.05). KR was significantly positively correlated with both PD (r = 0.189, P < 0.001) and PCSF (r = 0.158, P < 0.01). Furthermore, PD was significantly positively correlated with both PCSF (r = 0.354, P < 0.001) and CP (r = 0.340, P < 0.001). Additionally, PCSF and CP were significantly positively correlated (r = 0.378, P < 0.001) (Table 9).

Table 9 Correlation Analysis

Between-Group Comparison

Through independent samples t-tests, this study found that significant differences existed in PD, PCSF, and CP between the low and high KR groups, with the high KR group exhibiting higher levels of PD (t = −4.402, P < 0.001), PCSF (t = −4.508, P < 0.001), and CP (t = −5.436, P < 0.001) compared to the low KR group (Table 10). H1 was supported.

Table 10 Mean Comparison of Variables Between High and Low KR Groups

Model Construction

Variance inflation factors (VIF) below 5 and tolerance above 0.2 indicated that variables do not suffer from severe multicollinearity.45,46Table 11 showed that VIF for all variables were under 1.2, with tolerance values all exceeding 0.20, indicating that no severe multicollinearity existed among independent variables.

Table 11 Collinearity Diagnostics

This study constructed the initial structural equation model using AMOS 24.0. Evaluating measurement through the dual-index combination of SRMR with TLI or SRMR with CFI based on Maximum Likelihood can reduce the risk of model fit misjudgment.47 The degrees of freedom in this study’s model were relatively small (df = 39). However, RMSEA is prone to systematic overestimation in models with small degrees of freedom.48 Based on the aforementioned methodological foundations, this study employed SRMR, TLI, and CFI as core fit evaluation indices, supplemented by other fit indices for comprehensive determination of model adaptation. In the initial model, the model fit for the high KR group was superior to that of the low KR group. Specifically, in the high KR group model, except for χ2/df (3.961) and RMSEA (0.086) which marginally exceeded critical standards, the remaining indices including SRMR, GFI, TLI, and NFI all satisfied model fit criteria, indicating acceptable model fit for this group (Table 12).

Table 12 Model Fit Results

The path coefficient test results for the two-group models were presented in Table 13. In the low KR group, the path coefficients from KR to PD and from KR to PCSF did not reach significance (both P > 0.05); only the path coefficient from KR to CP was significant (β = −0.297, P < 0.001). The path coefficients from PD to PCSF (β = 0.382, P < 0.001) and from PD to CP (β = 0.217, P = 0.019) were both significant, and the path coefficient from PCSF to CP was likewise significant (β = 0.498, P < 0.001). In the high KR group, the path coefficients from KR to CP and from KR to PCSF showed no statistical significance (both P > 0.05); the path coefficient from KR to PD was significant (β = 0.206, P < 0.001), the path coefficient from PD to PCSF was significant (β = 0.346, P < 0.001), and the path coefficients from PD to CP (β = 0.261, P < 0.001) and from PCSF to CP (β = 0.337, P < 0.001) were both significant. H2 was supported in the low KR group but not in the high KR group.

Table 13 Path Coefficient Estimates for Each Group

This study further employed a multi-group model comparison. As shown in Table 14, the driving effect of KR on PD exhibited significant between-group heterogeneity. Specifically, in the high KR group, the path coefficient from KR to PD was 0.106 (P < 0.001); whereas in the low KR group, this path showed no statistical significance (B = −0.018, P = 0.651), with the difference in path coefficients between the two groups being significant (Z = −2.584, P < 0.001).

Table 14 Group Comparison of Path Coefficient Differences

The influence of KR on CP exhibited typical nonlinear characteristics. In the low KR group, the path coefficient from KR to CP was −0.111 (P < 0.001); whereas in the high KR group, this path showed no statistical significance (B = −0.006, P = 0.801), with significant between-group difference (Z = −2.806, P < 0.001). This result corroborated the moderating mechanism of the Dunning-Kruger Effect: at the low KR stage, information intake may trigger communication anxiety due to cognitive overload; whereas at the high KR stage, mere knowledge increment may no longer directly translate into communication behavior, but instead exert indirect effects through enhancing PD and PCSF.

The mediating transmission mechanism of PD and PCSF demonstrated strong consistency between the two groups. Specifically, the path coefficient from PD to PCSF was 0.314 in the high KR group and 0.337 in the low KR group; the path coefficients from PD to CP were 0.237 and 0.197, respectively; the path coefficients from PCSF to CP were 0.337 and 0.511, respectively. Statistical tests indicated that the coefficient values of these three paths did not exhibit significant differences between the two groups (all P > 0.05). This result suggested that regardless of parental KR levels, the psychological mechanism through which intrinsic motivation (PD) and relational cognition (PCSF) translate into actual performance (CP) was universally applicable.

To exclude confounding variable interference and verify result robustness, this study incorporated demographic and diagnosis-treatment related variables as covariates for controlled model testing. Table 15 showed that after controlling for covariates, the between-group path difference from KR to CP remained significant, while the previously significant between-group path difference from KR to PD became marginally significant (P = 0.075), with no obvious changes in the remaining between-group comparison results. Given that the core focus of this study was the relationship between KR and CP, the between-group path change corresponding to PD need not be prioritized. In summary, the model path differences between low and high KR groups primarily originated from the hierarchical differentiation of parental myopia control KR levels, with limited interference from confounding variables such as demographic characteristics and treatment experiences on between-group effects, indicating acceptable robustness of the research findings.

Table 15 Group Comparison of Path Coefficient Differences with Control Variables

Model Modification

After deleting non-significant paths in both groups, the model fit for both the low KR group and the high KR group improved significantly (Table 16).

Table 16 Model Fit Results of the Modified Model

Table 17 showed that in the low KR group, the standardized path coefficient from KR to CP was −0.288 (P < 0.001); the standardized path coefficients from PD to PCSF and from PD to CP were 0.377 (P < 0.001) and 0.215 (P = 0.019), respectively; the standardized path coefficient from PCSF to CP was 0.493 (P < 0.001) (Figure 1 ).

Table 17 Path Coefficient Estimates for Each Group of the Modified Model

Figure 1 Path diagram of the structural equation model for different KR groups.

In the high KR group, the standardized path coefficient from KR to PD was 0.210 (P < 0.001); the standardized path coefficients from PD to PCSF and from PD to CP were 0.367 (P < 0.001) and 0.259 (P < 0.001), respectively; the standardized path coefficient from PCSF to CP was 0.336 (P < 0.001) (Figure 1 ).

Simple Mediation Effect Test

Table 18 showed that in both the low and high KR groups, the mediating effect of PCSF between PD and CP accounted for 46.38% and 32.20% of the total effect, respectively; the mediating role of PCSF may be more pronounced in the low KR group. In the low KR group, the effect of PD on CP was more strongly dependent on the transmission pathway of PCSF; whereas in the high KR group, the pathway structure from PD to CP was more diversified, attenuating the mediating role of PCSF. Combined with the aforementioned between-group heterogeneity in path coefficients, H3 was supported in the high KR group but not in the low KR group.

Table 18 Mediation Effect Test and Decomposition

Discussion

This study achieved standardized health knowledge exposure through science popularization videos on whole-cycle health management for myopia control, and verified the nonlinear association between KR and CP, while revealing the moderating mechanism of the Dunning-Kruger Effect within the SDT framework. The concept of three levels of health literacy provided an explanatory basis for interpreting this moderating mechanism.

The high KR parental group exhibited significantly higher levels of PD, PCSF, and CP compared to the low KR parental group. Through the provision of standardized health knowledge exposure, the between-group differences in PD, PCSF, and CP were associated with disparities in health information literacy levels following knowledge exposure. According to the dimensions of health information literacy measurement tools, health information literacy is classified into literacy awareness, knowledge reserve, application capability, and integration capability.49 According to SDT, individuals with high KR may also possess stronger knowledge application and integration capabilities. Therefore, higher health information literacy is positively correlated with individuals’ superior health information identification, comprehension, and application levels, and is associated with medical decision-making capability and doctor-patient communication quality. The concept of empowerment is closely linked to health information literacy in this study. Within the context of this study, empowerment encompasses not only the enhancement of knowledge and capability, but also includes individuals’ elevated self-efficacy and agency in actual medical contexts.50 The high KR parental group consequently tends to participate more effectively in decision-making and establish more positive interactions with physicians. Therefore, this study suggests establishing a myopia control knowledge base certified by professional medical institutions and readily accessible for parental consultation.

In the low KR group, KR demonstrated a significant negative correlation with CP, consistent with the core mechanism of the Dunning-Kruger Effect. Specifically, individuals with low capability suffer from insufficient metacognitive ability, rendering them unable to accurately assess the boundaries of their own knowledge, often manifesting as overconfidence. Research has indicated that individuals affected by the Dunning-Kruger Effect fail to recognize their insufficient health literacy and perform poorly in misinformation identification. Following initial exposure to medical knowledge, these individuals may generate an “illusory superiority” accompanied by a tendency to overestimate their own capabilities. As KR gradually increases, individuals’ recognition of medical complexity and uncertainty tends to strengthen, and their self-assessment adjusts accordingly, exhibiting a phenomenon of “the more one learns, the less confident one becomes.” Therefore, this study suggests emphasizing the importance of sustained myopia control knowledge learning alongside the promotion of science popularization short videos, encouraging and attracting parents to continuously update their knowledge.

This study also found that in the low KR group, KR was not significantly correlated with PD or PCSF. According to the three levels of health literacy,28 individuals with low KR may only attain functional health literacy (the lowest level). This condition appears insufficient to effectively support their participation in medical decision-making or informed decision-making, and may additionally lead to doctor-patient relationship alienation and impede favorable interactions. When KR surpasses the critical threshold, the transition of individual self-assessment from “illusory superiority” toward “objective cognition” becomes more pronounced. Second, in the low KR group, knowledge has not yet constituted adequate cognitive support, and is therefore insufficient to generate significant correlations with their PD or relationship-building behaviors. In the high KR group, KR was not significantly correlated with CP or PCSF, but was significantly positively correlated with PD. This may reflect that after knowledge saturation, CP no longer relies on external KR accumulation, but is instead driven more by intrinsic motivation (such as sense of self-determination and responsibility identification). Some individuals with high KR may exhibit “information saturation” or “information fatigue” states, leading them to no longer actively seek further communication preparedness. Furthermore, this may also be associated with shifts in information processing patterns among individuals with high KR. The high KR group may signify possession of higher-level health literacy (such as interactive literacy and critical literacy). At this stage, individuals’ KR transforms into substantive decision-making efficacy, no longer excessively relying on knowledge as a direct driving force for CP, but instead functioning through other more intrinsic motivational mechanisms. Building upon this foundation, this study additionally obtained a novel finding: providing online health information for individuals with high KR may be more valuable for medical decision-making.

Results also indicated that PD was significantly positively correlated with PCSF and CP across groups with different KR levels. When patients possess greater rights to participate and make choices in medical participation in decision-making, they are more likely to perceive physicians as equal partners in jointly addressing health challenges, and to profoundly appreciate the important value of sharing health responsibility with physicians.39 Meanwhile, higher degrees of participation in decision-making may be associated with higher levels of participation in decision-making satisfaction and confidence.51 Furthermore, this study also found that PCSF was significantly positively correlated with CP. When patients perceive a community with shared future for doctor-patient, they recognize that they share common health objectives with medical personnel.39 Therefore, regardless of KR levels, PD was significantly positively correlated with PCSF and CP. This study suggests establishing a family-hospital collaborative participation in decision-making platform to facilitate physicians and parents in jointly formulating and adjusting children’s myopia control plans.

More importantly, this study found that PCSF exerted a partial mediating effect between PD and CP. In both the low and high KR groups, 46.38% and 32.20% of the effect of PD on CP, respectively, were mediated through PCSF. This mediating effect indicated that PCSF was not only associated with parental CP prior to medical encounters, but also partially explained how PD influenced this perception. Research has shown that shared doctor-patient participation in decision-making signifies shared power and joint assumption of responsibility.52,53 When individuals possess strong willingness and confidence to participate in the medical decision-making process, they not only feel more closely connected with physicians, but may also become more confident in future interactions with physicians. Furthermore, under the concept of a community with shared future for doctor-patient, patients are no longer merely passive recipients of medical services, but have transformed into collaborators actively participating in personal health management.39 The mediating pathway of PCSF suggests that when individuals perceive the possibility of PD, value identification with the community with shared future for doctor-patient is more readily formed and associated with CP behaviors. In the low KR group, individuals have not yet acquired sufficient myopia control knowledge or self-decision-making capability. Their “competence” drive is relatively weaker, and CP relies more on the sense of belonging and trust within the doctor-patient relationship. Research has shown that favorable doctor-patient relationships and trust levels are positively correlated with patient communication engagement.54 Within this context, PCSF served as an external activation mechanism for relational motivation, with its mediating effect being more strongly triggered, thereby explaining its higher proportion of 46.38%. Conversely, in the high KR group, parents already possessed relatively sufficient KR and medical decision-making capability; CP behaviors tended to be driven more by “competence” and “autonomy,” and although PCSF remained important, its relative contribution proportion decreased. Therefore, this study recommends incorporating empirical cases of “parents as collaborators in health management” into science popularization short videos. Through concrete presentation of how parents optimize children’s visual outcomes through participation in decision-making in doctor-patient interactions, their “community” identity identification can be effectively strengthened, and their CP enhanced.

This study also unexpectedly discovered that in the high KR group, the association between KR and CP may be realized through enhanced PD and PCSF. Based on SDT, sufficient KR is positively correlated with parental comprehension and sense of mastery over myopia control content, which contributes to satisfying the need for “competence,” thereby potentially enhancing their confidence when facing complex decisions. Meanwhile, abundant knowledge may endow parents with stronger capability for choice and foundation for judgment, contributing to satisfying the need for “autonomy” and facilitating their transition from passive acceptance to active participation. Furthermore, the accumulation of knowledge may help parents more comprehensively comprehend the collaborative relationship with the medical team, thereby enhancing “relatedness” and elevating trust and willingness to cooperate. Therefore, individuals with high KR levels are more likely to achieve the “internalization” of motivation, that is, to internalize externally provided information and platform functions into their own value identification and behavioral motivation, subsequently exhibiting more proactive CP and participation along the “knowledge-motivation-behavior” pathway. Thus, KR in the high KR group may not merely represent information accumulation, but rather serve as an important supportive resource for satisfying parents’ basic psychological needs and forming intrinsic motivation for participation in treatment decision-making, contributing to the promotion of their CP and participation. The aforementioned results are consistent with the Dunning-Kruger effect in health education within the SDT framework. This also confirms the importance of higher-level interactive health literacy and critical health literacy in enhancing parental participation in decision-making and decision selection capability among individuals with high KR. Truly achieving capability advancement may require the satisfaction of dual prerequisites: the acquired information must possess professionalism and reliability, and individuals must possess sufficient health literacy to effectively integrate information.

In summary, within the SDT framework, a nonlinear association exists between KR and CP, with its trajectory of change conforming to the typical characteristics of the Dunning-Kruger effect. The low KR group may fall into a “knowledge-confidence” paradox due to metacognitive limitations, whereas individuals with high KR may achieve decision-making empowerment through hierarchical advancement of health literacy. The advancement of health literacy from functional to interactive and critical levels may contribute to resolving dilemmas in communication and decision-making. Furthermore, this study also confirmed through the theoretical model that health information literacy was equally important as health literacy.

Significance Practical Significance

The nonlinear relationship between KR and CP revealed in this study indicated that myopia control health education should abandon homogeneous education and implement stratified precision intervention. For parents with low KR, the focus should be on correcting metacognitive biases, conducting popularized science education and interactive learning through a “knowledge empowerment and cognitive correction” strategy, dismantling their cognitive misconceptions of overconfidence, and enhancing myopia control cognition and treatment cooperation. For parents with high KR, immersive interactive education can be leveraged to conduct decision-making training, activating intrinsic participation motivation and strengthening their participation in decision-making capability in medical contexts. Simultaneously, the mediating mechanism of PCSF confirmed in this study provided practical support for optimizing doctor-patient interaction. By establishing a shared participation in decision-making platform between doctors and patients and promoting parental deep participation in diagnostic and treatment decision-making, medical trust and CP efficacy can be effectively enhanced. Furthermore, authoritative online health information demonstrates significant decision-making empowerment effects for parents with high literacy, suggesting that relevant institutions should establish a standardized authoritative information repository for myopia control, refine information screening and push mechanisms, circumvent false information interference, and assist parents in surpassing the critical threshold of health literacy to achieve scientific family-based myopia control management.

Theoretical Significance

Based on SDT combined with the Dunning-Kruger effect, this study constructed an integrated theoretical model of “knowledge reserve-health literacy hierarchy-communication preparedness,” breaking through the traditional linear research perspective of knowledge-behavior. Taking KR as the core antecedent, this study elucidated the evolutionary mechanism of dynamic advancement in parental health literacy from functional to interactive and then to critical levels, confirming that knowledge accumulation can effectively enhance medical communication preparedness through satisfying individuals’ basic psychological needs, and corroborating the dynamic attributes of health literacy encompassing both cognitive drive and motivational behavior. Meanwhile, by identifying the critical threshold of KR, this study revealed the qualitative transformation law of hierarchical advancement in health literacy, compensating for the deficiency of static analysis in traditional cognition-motivation-behavior models. Furthermore, this study validated the key mediating role of PCSF, expanding doctor-patient relationship research from the emotional trust dimension to the behavioral dimension of collaborative decision-making. Doctor-patient community identification can activate intrinsic medical motivation through identification regulation mechanisms, promoting parental transition from passive trust to proactive communication preparedness. This study effectively enriched the theoretical system of dynamic health literacy development and doctor-patient interaction, providing a novel integrated analytical paradigm for medical decision-making and health communication research.

Limitations

Although this study made contributions in theoretical integration and mechanism exploration, several limitations remained. This study employed an online voluntary recruitment sampling method. Participants were predominantly parent groups with higher attention to children’s myopia control and greater proficiency in digital tool usage, exhibiting self-selection bias. Second, this study only selected parents from five representative major cities in China as research subjects, without covering county-level, township, rural, and educationally disadvantaged regions. Therefore, the research findings were only applicable to urban parent groups in China, with significant regional and cultural context specificity. Caution was warranted when extrapolating to other regions or countries. Third, this study was cross-sectional in design, capable only of observing immediate effects following standardized knowledge exposure, lacking longitudinal follow-up tracking data, and unable to reveal long-term dynamic evolutionary patterns and causal temporal relationships among variables. Meanwhile, this study employed a single-group cross-sectional survey design following standardized knowledge exposure, without establishing a control group or baseline data, making it difficult to completely isolate interference from confounding factors such as individuals’ pre-existing cognition and medical experience. It could only verify associative characteristics among variables, unable to rigorously derive causal relationships. Furthermore, this study relied on subjective self-report scales for data collection, with limited items in the myopia knowledge assessment, and did not incorporate objective indicators of real clinical doctor-patient communication behavior, potentially introducing certain measurement bias. Subsequent research could optimize assessment tools and combine objective behavioral data with longitudinal research designs to further deepen the argumentation.

Conclusion

This study confirmed a significant nonlinear association between KR and CP, aligning with the core characteristics of the Dunning-Kruger effect. The low KR group may be prone to cognitive biases, whereas the high KR group may achieve knowledge empowerment and rational decision-making. PCSF exerted a partial mediating effect between PD and CP, with this mediating effect being more pronounced in the low KR group. Based on SDT, KR formed differentiated mechanisms of doctor-patient CP through satisfying individuals’ basic psychological needs. This study implied the medical interaction patterns of parents at different health literacy levels, providing theoretical support for precision science popularization education on adolescent myopia and collaborative doctor-patient control management.