A quasi-experimental design study on myopia preventive intervention based on the MMHEM health promotion model

Abstract

Background

Childhood myopia, driven by modifiable environmental factors like insufficient outdoor exposure, remains a global challenge. Existing interventions often lack frameworks bridging theory to practice. This study evaluated the Multisensory Multilevel Health Education Model (MMHEM)—a novel model integrating cultural, artistic, and scientific elements—to design a school-based myopia intervention in Yunnan’s minority areas.

Methods

We conducted a quasi-experimental study based at the school, employing the difference comparison analysis method. A total of 136 students from two middle schools were assigned by convenience sampling to either an MMHEM-based intervention group (n = 83) or a control group (n = 53). The experimental group consisted of first-grade students from an ethnic middle school, while the control group was composed of first-grade students from another middle school in the same city with similar economic conditions and student sources. Refractive error was measured using non-cycloplegic autorefraction. Outcome measures included myopia prevalence and incidence, spherical equivalent (SE), and axial length (AL), which were tracked over a three-year period (from 2020 to 2023).

Results

The DID analysis revealed a significant intervention effect: the myopic shift (SE) was reduced by 0.18 D (95% CI: 0.05 to 0.31, P = 0.007), and axial elongation (AL) was slowed by 0.08 mm (95% CI: 0.02 to 0.14, P = 0.012) in the intervention group relative to controls. Consistent with this, the cumulative incidence of myopia was significantly lower in the intervention group (29.4%) than in the control group (56.6%; P < 0.01). A similar protective effect was observed for moderate/high myopia incidence (21.3% to 42.6%; P < 0.05). Furthermore, outdoor activity scores improved significantly in the intervention group (from 10.31 to 11.23, P < 0.05), although no significant changes were noted in screen time or close-range eye habits.

Conclusion

A school-based comprehensive intervention, guided by the MMHEM and integrating strategies across the art, culture, and science fields, was associated with a lower incidence of myopia and slower axial elongation over two years compared to a control school. These research results indicate that the intervention measures with cultural customization attributes formulated based on the MMHEM model in this study have initially achieved the set goals, providing methodological references for behavioral intervention in ethnic minority areas. In the future, random controlled trials (RCTs) can be incorporated to further verify the effectiveness of the intervention measures and ensure their wider applicability.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12889-026-26695-9.

Introduction

Myopia represents a significant global public health concern among children and adolescents. Alarmingly, in recent decades, myopia has been presenting at younger ages, leading to a heightened overall risk of developing high myopia later in life [1]. However, effective and comprehensive interventions for this age group remain insufficiently explored [2]. Our baseline survey conducted in Lijiang City showed that the incidence rate among adolescents reached 67.9%. Analysis of baseline data, collected through pre-intervention surveys and examinations, revealed that while genetics play a role, environmental and lifestyle factors—particularly insufficient outdoor activity and prolonged screen time—were key contributors to myopia prevalence.

The Multisensory Multilevel Health Education Model (MMHEM) is an evidence-based conceptual framework that employs a multisensory and culturally tailored approach to guide the development of health education interventions [3]. While foundational theories such as the Social-Ecological Model (SEM) often address the ‘why’ behind behavior change, the MMHEM distinctively provides a practical blueprint for implementation. It achieves this by integrating actionable strategies across its three core fields—art, culture, and science—thereby providing a structured framework that specifies not only how to implement interventions but also what core components to deliver and when to sequence them for maximum impact. Originally developed to enhance stroke awareness, MMHEM’s applicability extends broadly. For instance, from an artistic perspective, parallels can be drawn between health communication and creative marketing strategies, encompassing various forms of advertising [4–6]. It demands achieving behavioral change through culturally-attuned guidance, measurable results, and sustainable mechanisms. In the art field, health programs need to consider an individual’s culture [7]. Campaigns in culturally diverse communities must prioritize reconciling targeted behavior change with dominant social norms, particularly when addressing collectivist value systems [8]. In fact, culturally discordant programs contribute to the failures of many health campaigns [9]. Finally, health education programs need to be guided by scientific evidence supporting their content, followed by a rigorous evaluation of their effectiveness prior to dissemination [10].

Current theoretical models of health education share a common limitations. SEM serves as a comprehensive health education framework within certain boundaries. While SEM explores the various levels influencing the intervention target behavior and addresses the ‘why/what,’ it does not account for the ‘how.’ In contrast, MMHEM provides specific, actionable strategies in public health to guide the design and implementation of health education, offering a novel perspective on addressing myopia through evidence-based interventions. Notably, the effectiveness of multi-component school-based interventions, particularly those promoting increased outdoor time, in reducing myopia incidence and progression is well-established, as demonstrated in randomized trials [6, 9]. While several school-based interventions have demonstrated efficacy in reducing myopia incidence—primarily through policies mandating increased outdoor time —no published studies have applied the MMHEM to myopia prevention. Existing approaches often focus on single behavioral levers and lack a comprehensive framework that integrates cultural, artistic, and scientific fields to address the ‘how’ of sustainable behavior change. This study therefore aimed to provide evidence by designing, implementing, and evaluating a culturally tailored, MMHEM-guided comprehensive myopia intervention among Naxi ethnic minority adolescents.

Lijiang City is home to a large population of Naxi ethnic group, which constitutes 58% of the city’s total population(and 68.5% of the total Naxi population in China). The Naxi ethnic group possesses rich cultural characteristics and a historical heritage, renowned for their musical and dance traditions, as well as their hardworking nature. The majority of students at the Ethnic Middle School exhibit high rates of participation, with basketball and traditional shuttlecock kicking being culturally embedded activities.

Given this, the present study creatively applied relevant health promotion intervention activities based on the three fields of MMHEM: art, culture, and science. Tailoring interventions to the cultural characteristics of the Naxi ethnic group, the study implemented activities such as basketball match, handwritten newspaper competitions, non-contact eye exercises, myopia prevention workshops, distribution of educational materials (e.g., brochures and customized folders), and classroom lighting optimization. Ultimately, this study aimed to evaluate the effectiveness of a school-based myopia prevention intervention guided by the MMHEM. We hypothesized that, compared to a matched control school, the MMHEM-based intervention would: (1) reduce the incidence of myopia, (2) slow the progression of spherical equivalent (SE) and axial length (AL), and (3) improve outdoor activity scores among adolescents.

Materials and methods

Study design

This myopia prevention study conducted in Lijiang, China, was a 3-year cohort study (2020–2023) designed to develop and evaluate a comprehensive intervention strategy guided by the MMHEM. The experimental plan is detailed in Appendix 1. In the initial phase of the study, based on baseline survey data from selected schools, we developed a comprehensive intervention strategy targeting identified risk factors for myopia and implemented corresponding activities at Ethnic Middle School. The impact of the intervention was subsequently assessed using a difference-in-differences model (DID) with a quasi-experimental design to analyze both within-group changes before and after the intervention, and between-group differences (intervention and control), as shown in Table 1. In this study, Xianghe Middle School located in the same city, was selected as the control group. Evaluation metrics include standardized prevalence, standardized incidence, cumulative standardized prevalence and incidence, as well as the standardized incidence and prevalence of moderate and high myopia, SE and AL, and scores for electronic screen use, close-range eye habits, and outdoor activities.

Table 1.

DID model

	Before intervention	After intervention	Difference value
Intervention group	Yt1	Yt2	△Yt= Yt1- Yt2
Control group	Yc1	Yc2	△Yc=Yc1- Yc2
Difference value			△△Y=△Yt-△Yc

Y_t1 and Y_t2 denote before- and after-intervention outcomes for the intervention group, respectively, while Y_c1 and Y_c2 represent corresponding measurements for the control group. The difference between before- and after -intervention changes (△Y_t and △Y_c) was calculated for both groups. The DID estimator (△△Y) isolates the intervention effect by subtracting the control group’s change from the intervention group’s change, accounting for temporal trends and confounding factors

In the process of calculating the prevalence and incidence, we analyses the adjusted prevalence/incidence rate by parents with myopia in order to eliminate the influence caused by the different internal composition factors of the two groups and to conduct a more accurate assessment of the intervention effect. The formulas are:

1

2

In the formulas, N represents the total size of the standard population, which was constructed by combining the two study groups, the term represents the number of students in the standard population, represents the baseline prevalence, X represents the expected number of participants, and represents the standardized prevalence. The standard population (N) was constructed based on the age and sex distribution of the middle school student population (aged 11–15 years) from the most recent national census data available for our study region.

The study was approved by the institutional review board of Kunming Medical University, School of Public Health (No.81960593). Parents were informed of the study prior to any examination, and all parents provided written informed consent for their children to participate (No.KMMU2019MEC01). This study followed the Transparent Reporting of Evaluations with Nonrandomized Designs (TREND) reporting guideline.

Intervention assessment

This study employed a quasi-experimental design, analyzing changes both before and after the intervention, as well as differences between the intervention and control groups. A scoring system was developed to assess various factors related to students’ eye health on and off campus, including visual environment, reading and writing posture, electronic screen use, close-range eye habits, outdoor activities, sleep, and dietary habits. The Close-Range Eye Habit Rating is a composite score (range: 0–28 points) that evaluates behaviors associated with prolonged near-work activities (e.g., reading, screen use), which are known to increase myopia risk. Higher scores indicate healthier habits and lower myopia risk. The Outdoor Activity Score, with a total range of 0–20 points, was derived from key components such as: daily outdoor time (< 1 h = 0 points; 1–2 h = 2 points; >2 h = 4 points) and type of activity (non-sport leisure, e.g., walking = 1 point; organized sports, e.g., basketball = 3 points). Higher scores reflect better adherence to recommended outdoor behavior. The detailed scoring methodology is provided in Appendix 2.

Intervention framework

The MMHEM is a conceptual framework designed to provide guidance for the development of health education interventions. This model represents an integrative approach to public health education operating at different levels of behavioral influence [3]. It comprises three major fields—art, culture, and science—and eight subfields. In the field of art, including aesthetic design and multisensory approaches; in the cultural field, including cultural customization, social norms, and digital information; and in the field of science, it includes cognitive strategies, evidence-based methods, and evaluation systems. Using this framework and the baseline identification of myopia risk factors, a comprehensive intervention was developed, as shown in Fig. 1.

Fig. 1.

Intervention framework based on MMHEM

Intervention activities

The intervention was designed to address modifiable myopia risk factors identified from baseline data, including suboptimal reading postures, inadequate classroom illumination, and limited outdoor exposure. Culturally adapted and multi-component strategies were implemented, with their delivery and fidelity systematically monitored through a structured process evaluation. Figure 2 presents an overview of the intervention framework and its core components.

Fig. 2.

Intervention content based on MMHEM guidance

• Classroom Environmental Modifications: Lighting Optimization: Classrooms were equipped with LED panels, and blackboards were retrofitted with anti-glare coatings to ensure uniform illuminance. This modification was quantitatively verified using portable light meters (UT383BT, Uni-Trend®). Pre-optimization measurements revealed an average desktop illuminance of 146 lux. Post-optimization, this was significantly increased to an average of 382 lux. Annual compliance checks confirmed that these improved lighting levels were maintained throughout the study period.

• Ergonomic Desk Adjustments: Adjustable desks and chairs were installed based on students' anthropometric measurements. Desks were positioned to enforce a 30–40 cm reading distance.

• Culturally Tailored Behavioral Activities: Guided by the MMHEM framework and leveraging the Naxi ethnic group's affinity for sports and art, the following activities were implemented to promote behavioral change:

  • Health Education Lectures: Three myopia prevention lectures were delivered annually at 4-month intervals. Each session lasted approximately 45 minutes. Educational materials (e.g., brochures, customized folders) were distributed biannually to all participants.
  • “Love Eye Cup” Basketball Tournament: This event was held once per semester, totaling four tournaments over the two-year intervention. The tournaments, spanning a week each, were estimated to provide a moderate-to-vigorous outdoor activity per participating student daily during the event weeks.
  • Non-Contact Eye Exercises: The non-contact eye exercises were conducted daily during the morning break (5 minutes per session). According to teacher supervision logs, adherence to these exercises exceeded 90% on regular school days.
  • Handwritten Newspaper Competition: "Myopia Prevention" themed handwritten newspaper competition was conducted to engage students creatively in learning about eye health. Students designed and created newspapers featuring relevant knowledge and prevention tips, and the entries were displayed on campus to reinforce key messages.
  • Implementation Context and Barriers: The most significant implementation barrier was the disruption caused by the COVID-19 pandemic in 2022, which led to temporary school closures and a shift to online learning for approximately three months. This suspension of on-site group activities and the associated increase in screen time are acknowledged as a contextual factor affecting the intervention delivery and outcomes.

Study population

Based on the research objectives, this research was conducted in Lijiang, where the Naxi ethnic group lives in concentration. The intervention group consisted of all third-year students (n = 83) from an Ethnic Middle School located in a minority region. This grade level was selected as it represents a critical period for myopia onset and development [11]. The control group was drawn from Xianghe Middle School (n = 53) within the same municipality.

The selection of the control school was based on its comparability to the intervention school in terms of age distribution,, socioeconomic status, regional education policies, and ethnic students composition. Both schools operated under identical academic schedules and had no pre-existing myopia prevention programs, thereby minimizing the risk of contamination and enhancing the validity of the difference-in-differences comparison.

A priori sample size calculation was not performed due to the pragmatic nature of this school-based intervention. However, a post-hoc power analysis, based on the observed cumulative change in spherical equivalent (ΔSE), indicated that with an effect size (Cohen’s d) of 0.88, a total sample size of 136, and a two-tailed alpha of 0.05, the study achieved a statistical power of 91%.

All students across both junior middle schools who met the inclusion criteria were invited to participate. Inclusion criteria were: (1) middle school students aged 11–15; (2) no plans to transfer schools within 3 years; (3) completion of baseline survey data; (4) absence of major physical illnesses. Exclusion criteria included: (1) intermittent participation by the school or parents; (2) amblyopia, hyperopia, or astigmatism exceeding 0.5 DC; (3) history of eye diseases, trauma, or surgery; (4) inability to participate in school collective activities.

Participant retention was systematically monitored throughout the three-year study. Of the 149 students enrolled at baseline, 13 (8.7%) were lost to follow-up, resulting in a final analytical sample of 136 students and a follow-up rate of 91.2%. The attrition distribution was 9 students from the intervention group and 4 from the control group.The primary reasons for attrition included school transfer and absence during examination days. The distribution of baseline characteristics between completers and those lost to follow-up was compared and found to be not significantly different (all P > 0.05), suggesting that The impact of loss to follow-up on the research results can be disregarded. Additionally, item-level missing data in the questionnaire responses were minimal (< 2% for all variables) and were considered missing completely at random.

The questionnaire was anonymously filled out, and students were explained that it is not related to exam scores, in order to minimize false statements in the questionnaire. Before completing the questionnaire, the investigators thoroughly explained the purpose and method of the study and the details of how to complete the questionnaire. The questionnaires used in this study were developed specifically for this study and have not been published elsewhere, as shown in Appendix 3.

To ensure validity and reliability, the Adolescent Myopia Behavior Scale underwent a rigorous validation process. First, content validity was established through literature reviews and consultations with experts in ophthalmology and public health to ensure comprehensive coverage of key myopia-related behaviors, such as electronic screen use, close-range eye habits, and outdoor activities. Second, a pilot survey involving 30 students was conducted to assess the clarity and cultural appropriateness of the questionnaire; ambiguous terms were revised based on feedback, including adjustments to localized vocabulary tailored to Naxi ethnic students. Internal consistency was evaluated using Cronbach’s alpha (α = 0.87), confirming high reliability. The self-administered questionnaire was written in Chinese and completed anonymously by students under researcher supervision to minimize bias while avoiding direct intervention in responses.

Data collection

Trained ophthalmologists and optometrists conducted eye examinations on subjects according to standard procedures. Distance visual acuity was measured using a standard logarithmic VAE chart (National Standard of People’s Republic of China, GB 11533 − 1989) mounted on an illuminated cabinet with a luminance of 80–320 cd/m2. Non-cyclopedic refractions were measured using an auto-refractor (NIDEK ARK-30). SE was calculated as sphere plus 1/2 cylinder. Myopia was defined as an SE ≤ -0.50 diopters (D) and uncorrected visual acuity < 5.0. Subjects with myopia were then divided into low (− 3.0 D < SE < − 0.5 D), moderate (− 6.0 D < SE ≤ − 3.0 D), or high (SE ≤ − 6.0D) myopia [12]. Trained specialist personnel were responsible for measuring the classroom floor area, blackboard area, blackboard and wall reflectance, class’s lighting coefficient, average illumination, and uniformity of desktop and blackboard light, as well as color temperature and wavelength. Detailed information regarding the research design is depicted in Fig. 3.

Fig. 3.

Flowchart of participant recruitment and selection process

Reliability and validity of the questionnaire

The self-administered Adolescent Myopia Behavior Scale demonstrated good psychometric properties. Content validity was confirmed through expert review and pilot testing. Exploratory factor analysis of baseline data supported a three-factor structure (Electronic Screen Use, Close-Range Eye Habits, and Outdoor Activity), which collectively explained 68.5% of the total variance. The scale showed high internal consistency, with a Cronbach’s α of 0.87 for the overall scale and ranging from 0.76 to 0.82 for the subscales.

Statistical analysis

Students from the selected schools who met the inclusion criteria were enrolled into the study. The cumulative change in spherical equivalent (ΔSE) served as the primary continuous outcome measure. A post-hoc power analysis was conducted based on the observed effect: with an effect size (Cohen’s d) of 0.88, a total sample size of 136, and a two-sided alpha of 0.05, the study achieved a statistical power of 91%.

Descriptive statistics, double sample test, and chi-square test were used to compare the baseline characteristics of the intervention group and the control group (including the prevalence of myopia, the incidence rate, moderate and high myopia incidence rate, the absolute value of above three rate and the difference of above three rate’s difference between intervention group and the control group .To address potential confounding, we employed two complementary strategies: the comparison of myopia rates was conducted using standardized rates, and the analyses of behavioral and ocular parameters were adjusted for key baseline covariates, including age, gender, parental myopia, parental education level, and socioeconomic status, where applicable. The primary analysis for assessing the intervention effect was conducted using DID approach. A generalized linear model was fitted to estimate the DID effect, quantified by the coefficient (β₃) of the interaction term between group (intervention vs. control) and time (before vs. after intervention). This model was adjusted for key baseline covariates—age, gender, and parental myopia—to improve estimate precision and control for potential confounding. We acknowledge that the DID approach relies on the parallel trends assumption, which we could not verify statistically due to having only one pre-intervention time point in our design. Furthermore, cluster-robust standard errors were employed to account for the clustering of observations at the school level. Beyond the primary DID analysis, changes in risky behaviors, SE, AL, and environmental factor scores were analyzed using analysis of variance and nonparametric rank-sum tests. The completeness and validity of the questionnaire were rigorously assessed. We checked the completeness and validity of the questionnaire. Statistical software SPSS 26.0 (SPSS Inc., version 26, Chicago, IL, USA) was used for analyses.

Results

Demographic characteristics and myopia related information

A total of 136 Naxi ethnic students participated in the study, as shown in Table 2. The number of students in Xianghe Middle School is 53, with an average age of (12.4 ± 0.5). The number of students in Ethnic Middle School is 83, with an average age of (12.5 ± 0.5). The difference of parents’ myopia between the two schools was statistically significant (P < 0.05). However, the influence of family lighting on different schools was not significant (P > 0.05).

Table 2.

Demography and myopia related information

Variable		The intervention group (n = 83)	The control group (n = 53)	χ²/t	P value
Age		12.54 ± 0.53	12.43 ± 0.54	1.161	0.248
Gender	Male	44(53.0)	31(58.5)	0.392	0.531
	female	39(47.0)	22(41.5)
Parents with myopia	myopia	10(12.0)	14(26.4)	4.594	0.032
	Non-myopia	73(88.0)	39(73.6)
Household lighting conditions	Good lighting and below	62.9% (78)	37.1% (46)	2.075	0.150
	Intense light	41.7% (5)	58.3% (7)
Parental education	Middle school or below	42(50.6)	21(39.6)	2.874	0.238
	High school/Technical	28(33.7)	22(41.5)
	College or above	13(15.7)	10(18.9)
Socioeconomic status	Low	35(42.2)	18(34.0)	1.326	0.515
	Middle	38(45.8)	28(52.8)
	High	10(12.0)	7(13.2)

in “whether the parents are myopic”, myopia includes (only the father, only the mother or both of them are myopic, which is judged to be myopic), and non-myopia means that both of them do not have myopia. The “home lighting” refers to the lighting conditions in students’ homes, including the internal lighting intensity, light source type (such as natural light or artificial light), and daily lighting environment.Parental education was defined as the highest education level attained by either parent. Socioeconomic status (SES) was assessed based on parental occupation, household income, and living conditions, and was categorized into three levels (low, middle, high)

Demographic characteristics and myopia prevalence of two middle schools in Lijiang City

We further analyzed the data according to whether the parents were myopic. Through three surveys, we compared the standardized myopia prevalence in two schools. The baseline standardized myopia prevalence rate of Ethnic Middle School was 48.5%, and that of Xianghe Middle School was 72.8%. The difference between the two schools was statistically significant (χ² = 16.778, P < 0.01). In the first follow-up survey, the prevalence of standardized myopia in Ethnic Middle School was 52.9%, and that in Xianghe Middle School was 79.4%. In the second follow-up, the prevalence of standardized myopia was 63.2% in Ethnic Middle School and 88.2% in Xianghe Middle School. The difference between the two schools was statistically significant (P < 0.01). As shown in Table 3.

Table 3.

Comparison of the standardized prevalence of myopia in different middle schools in Lijiang City

	Parents are not myopic			Parents are myopic			Total	Crude Prevalence % (95% CI)	Standardized prevalence%(95%CI)
	Standard population	Baseline prevalence	Expected number of participants	Standard population	Baseline prevalence	Expected number of participants
Baseline
The intervention group	112	47.9%	54	24	50.0%	12	66	47.0% (36.2, 58.1)	48.5%(37.5, 59.6)**
The control group		71.8%	80		78.6%	19	99	71.7% (58.5, 82.0)	72.8%(63.2, 80.8)
The first follow-up
The intervention group	112	53.4%	60	24	50.0%	12	72	53.0% (42.0, 63.8)	52.9%(41.8, 63.8)**
The control group		76.9%	86		92.9%	22	108	79.2% (66.5, 88.0)	79.4%(70.1, 86.4)
The second follow-up
The intervention group	112	64.4%	72	24	60.0%	14	86	62.7% (51.6, 72.6)	63.2%(52.2, 73.0)**
The control group		87.2%	98		92.9%	22	120	88.7% (77.4, 94.7)	88.2%(80.5, 93.2)

** represents  P＜0.01

A comparative analysis of myopia changes in two schools according to the absolute value of prevalence and incidence

In order to explore the increasing prevalence of myopia among students, we further compared the increasing rate of myopia investigated year by year. The results showed that the increase rate of myopia in Ethnic Middle School reached 4.4% and 10.3% in the first and second follow-up. The myopia increasing rate of Xianghe Middle School reached 6.6% and 8.8% in the first and second follow-up. Compared with the myopia rate in the first year, the cumulative myopia increase rate in the third year reached 14.7% and 15.4%, respectively. The results showed that although there was no statistically significant difference in the increased value of myopia prevalence after standardization in different middle schools (P > 0.05), the incidence rate of Xianghe Middle School was still higher than that of Ethnic Middle School, as shown in Fig. 4.

Fig. 4.

Comparison of the absolute value of the standardized prevalence of myopia in different middle schools in Lijiang City

We investigated the standardized incidence of myopia in students. The results showed that in the two follow-up visits, the standardized incidence rates of Ethnic Middle School and Xianghe Middle School showed a significant upward trend, and the incidence rate of Xianghe Middle School was higher than that of Ethnic Middle School. Compared with the first year, the standardized cumulative incidence of myopia in the third year reached 29.4% and 56.6%, respectively(χ²= 20.533, P < 0.01), which is statistically significant. In addition, the absolute value of cumulative incidence rate is 20.6% in Ethnic Middle School, which is lower than that of 32.1% in Xianghe Middle School (χ²=17.623, P < 0.01), as shown in Fig. 5.

Fig. 5.

Comparison of standardized incidence of myopia in different middle schools. Notes: *represents P＜0.05；** represents P＜0.01

Changes in moderate and high myopia among students in different middle schools

In the calculation of the prevalence of moderate and high myopia, we used the standardized prevalence for analysis. The results showed that in the two follow-up visits, the standardized prevalence rates of Ethnic Middle School and Xianghe Middle School showed a significant upward trend, and the prevalence rate of Xianghe Middle School was higher than that of Ethnic Middle School, with statistical significance, as shown in Fig. 6. In addition, we compared the progression of moderate and high myopia. The results showed that in the first and second follow-up, the number of students with moderate and high myopia in Ethnic Middle School and Xianghe Middle School showed an increasing trend. However, in the first follow-up, there was no statistical difference. Compared to the first year, the cumulative incidence rate of individuals with moderate and high myopia in the third year at Xianghe Middle School exceeded that of Ethnic Middle School, with an increase of 19.9% and 42.6%, respectively (χ²=16.445, P < 0.01), indicating statistically significant differences (Fig. 7). In addition, we compared the standardized incidence of moderate and high myopia. According to the results, in the first and second follow-up visits, the incidence of both Ethnic Middle School and Xianghe Middle School showed an increasing trend, and Xianghe Middle School was higher than Ethnic Middle School. According to the cumulative standardized incidence results, the incidence of Xianghe Middle School was higher than Ethnic Middle School, reaching 42.6% and 21.3% respectively. The results were statistically significant, as shown in Fig. 8.

Fig. 6.

Comparison of moderate and high myopia prevalence rate in standardization (%)

Fig. 7.

Comparison of the absolute value of moderate and high myopia prevalence in standardization. Notes: *represents P＜0.05；** represents P＜0.01

Fig. 8.

Comparison of incidence of moderate and high myopia in standardization. Notes: *represents P＜0.05；** represents P＜0.01

Progress of eye ocular refractive parameters in different middle schools

These findings were corroborated by the longitudinal data on SE and AL. At baseline, the median SE was comparable between the intervention and control groups (-1.00 vs. -1.06 D, P = 0.122). However, the control group was associated with a more pronounced myopic shift at both the first follow-up (-1.38 vs. -2.06 D, P = 0.008) and the second follow-up (-1.88 vs. -3.00 D, P = 0.011) (Table 4). Analysis of the inter-period changes (Table 5) further demonstrated that the intervention group had a significantly smaller SE progression than the control group during the first year (ΔSE: -0.38 vs. -0.75 D, P = 0.002) and cumulatively over the entire study period (ΔSE: -1.13 vs. -1.44 D, P = 0.001). A similar protective pattern was observed for axial elongation, with the intervention group showing consistently smaller increases in AL at each follow-up and cumulatively (P < 0.05, Tables 6 and 7).

Table 4.

Comparison of SE progression in two middle schools over three years [IQR (P25, P75)]

Group	SE
	Baseline	First follow-up	Second follow-up
The intervention group	-1.00 (-2.00, -0.25)	-1.38 (-2.50, -0.38)	-1.88 (-3.53, -0.88)
The control group	-1.06 (-2.50, -0.63)	-2.06 (-3.25, -1.03)	-3.00 (-4.00, -1.81)
Z	-1.545	-2.671	-2.545
P value	0.122	0.008**	0.011*

*represents P<0.05;** represents P<0.01

Table 5.

The progress of SE in different middle schools [IQR (P25, P75)]

	The intervention group	The control group	F	P value
1st ΔSE	-0.38(-0.88,-0.13)	-0.75(-1.13,-0.50)	-3.063	0.002**
2st ΔSE	-0.63(-1.13,-0.13)	-0.75(-1.00,-0.50)	-1.458	0.145
Accumulation ΔSE	-1.13(-1.53,-0.47)	-1.44(-2.00.-1.00)	-3.182	0.001**

** represents P<0.01

Table 6.

Comparison of AL in two middle schools over the three years (± s)

Group	AL			F	P value
	Baseline	First follow-up	Second follow-up
The intervention group	23.74 ± 0.91	23.95 ± 0.97	24.23 ± 1.04	89.634	< 0.01
The control group	24.25 ± 1.04	24.53 ± 1.05	24.81 ± 1.17	78.769	< 0.01
t	-3.048	-3.277	-2.998
P value	0.003	0.001	0.003

Table 7.

The progress of AL in different middle schools [M (P25, P75)]

	The intervention group	The control group	Z	P value
1st Δ AL	0.19(0.08,0.32)	0.28(0.18,0.39)	-3.617	< 0.01
2st Δ AL	0.25(0.11,0.39)	0.33(0.20,0.46)	-2.082	0.037
Accumulated ΔAL	0.45(0.23,0.69)	0.63(0.46,0.83)	-3.018	0.003

Data are presented as median [M (P25, P75)], where P25 represents the 25th percentile and P75 the 75th percentile

Intervention effects on SE and AL from adjusted DID models

To robustly quantify the net intervention effect while accounting for baseline differences and the clustered nature of the data, we performed an adjusted DID analysis. The model estimated the intervention effect (β3) on the primary ocular parameters, adjusting for age, gender, parental myopia, and baseline values, with standard errors clustered at the school level. As presented in Table 8, the intervention was associated with a statistically significant beneficial effect on both spherical equivalent and axial elongation. The positive DID coefficient for SE (β3 = + 0.18 D, 95% CI: +0.05 to + 0.31, P = 0.007) indicates a reduction in the myopic shift in the intervention group relative to the controls. Concurrently, the negative coefficient for AL (β3 = -0.08 mm, 95% CI: -0.14 to -0.02, P = 0.012) confirms a slower rate of axial elongation among intervention students. The robustness of the primary DID estimate was evaluated through a sensitivity analysis employing multiple imputations for missing data. Results derived from 10 imputed datasets were highly consistent with the complete-case analysis, yielding nearly identical effect sizes and confidence intervals for both SE (imputed β3 = + 0.17 D, 95% CI: +0.04 to + 0.30) and AL (imputed β3 = -0.07 mm, 95% CI: -0.13 to -0.01). This concordance confirms that the reported intervention effects are robust to different assumptions regarding missing data.

Table 8.

Adjusted DID analysis of the intervention effect on SE and AL

Outcome variable	DID coefficient (β3)	95% confidence Interval	P value
SE	+ 0.18 D	+ 0.05,+0.31	0.007
AL	-0.08 mm	-0.14, -0.02	0.012

The DID coefficient (β3) represents the net intervention effect derived from the interaction term (Group × Time) in a generalized linear model. A positive β3 for SE indicates a reduction in myopic shift in the intervention group compared to the control group. A negative β3 for AL indicates less elongation of the axial length in the intervention group

Comparison of risk factor exposure before and after myopia intervention in ethnic middle school

This study also compared the environmental factor scores of Ethnic Middle School before and after intervention. The results show that the use of electronic screens, close-range eye habits, and outdoor activity scores in Ethnic Middle School has statistical significance in the intervention, as shown in Table 9.

Table 9.

Comparison of the absolute increases in standardized prevalence of moderate and high myopia (± s) [M (P25, P75)]

Environmental factor	Total score	Before intervention	After intervention	t	P value
Electronic screen usage rating	20	12.73 ± 3.51	11.19 ± 3.64	2.953	0.004
Close-range eye habit rating	28	18.59 ± 3.93	17.32 ± 3.52	2.504	0.014
Outdoor activity rating	19	10.31 ± 3.47	11.23 ± 2.98	-2.513	0.014

Data are presented as median [M (P25, P75)], where P25 represents the 25th percentile and P75 the 75th percentile

Analysis on the difference of myopia between two middle schools

Because the prevalence of the two schools varied during the baseline survey period, the difference was significant. Intervention impact evaluated by analyzing the data and using the DID analysis method for calculation, the results obtained were shown in Table 10. Compared with Xianghe Middle School, the incidence of myopia in Ethnic Middle School decreased by 27.2%. The incidence of middle and high myopia was 21.3% lower than that of Xianghe Middle School, and the progress of eye axis was slowed down. The incidence of moderate and high myopia in Ethnic Middle School was slower than that in Xianghe Middle School, which decreased by 21.3%. It showed the progress of eye axis in Ethnic Middle School was slower than Xianghe Middle School. Since the prevalence and incidence of myopia in Xianghe Middle School are higher than those in Ethnic Middle School, negative values can indicate that the metrics of Ethnic Middle School after intervention are better than those of Xianghe Middle School.

Table 10.

Analysis on the difference of myopia between two middle schools

ΔΔ	1st	2st	Cumulative quantity
Prevalence of myopia	-2.2	1.5	-0.7
Prevalence of moderate and high myopia	-5.2	-8.8	-14.0
Incidence of myopia	-15.7	-12.5	-27.2
Incidence of moderate and high myopia	-9.6	-7.4	-21.3
Axial length	-0.09	-0.08	-0.18

ΔΔ = Δ Ethnic Middle School - ΔXianghe Middle School

Discussion

MMHEM, as a comprehensive health education framework mainly consisting of social ecological models and entertainment education theories, integrates the ‘Why/What,’ ‘Who/Where,’ and ‘How’ to conduct effective health education [3]. MMHEM further deconstructs how its main fields—art, culture, and science — operate in a way that synergizes with people’s social and physical environments and influences their personal health behaviors. Our baseline survey and other studies agree that many factors affect myopia in adolescents, such as the duration of electronic software use, the intensity of indoor learning in illumination, safe eye distance, duration and outdoor activity [13–15].

Since the influencing factors obtained from the baseline survey can guide us to design innovative intervention contents, the implementation of targeted behavioral intervention based on MMHEM theory is more theoretically based [3].therefore, we considered effective multisensory strategies, particularly in the field of art, supported by the best existing research evidence in science. Additionally, we tailored interventions to connect with specific cultural groups. The main risk factors for myopia were identified based on baseline surveys in two schools, and interventions were implemented at Ethnic Middle School to address these risk factors. The Naxi people possess distinctive national characteristics, which are not commonly found among other ethnic groups in the Yunnan region. They are inherently hardworking and kind, with a strong interest in labor and outdoor sports. As such, this study incorporates numerous artistic and cultural elements related to Naxi culture. Examples of these measures are hosting a basketball match named ‘Love Eye Cup’ organizing a hand-copy newspaper competition, implementing non-contact eye exercises, and conducting extensive health education. At the same time, adjustments are being made to classroom teaching facilities on campus, including the optimization of lighting facilities and improvement of desk facilities. Multiple intervention measures are focused on the fields of art, culture, and science. These measures include promoting popular science materials and organizing campus eye care lectures.

This study conducted a total of three follow-up surveys over three years in two middle schools in Lijiang City. Although the myopia detection rate among students in both middle schools increased over time, the annual incidence of myopia in Ethnic Middle School was lower than that in Xianghe Middle School. Throughout the follow-up of the survey, we found through our analysis of cumulative incidence, the incidence of myopia in Ethnic Middle School was also much lower than that in Xianghe Middle School. In addition, through the DID method, we found that the progression of myopia in Ethnic Middle Schools is slower than that in Xianghe Middle Schools. To further investigate the growth of myopia among students, we conducted comparisons of the myopia prevalence rate in two schools over the past three years. The results showed that, compared to Xianghe Middle School, the myopia prevalence rate of Ethnic Middle School was lower in the first follow-up but slightly higher in the second follow-up. This difference may be attributed to the disruptive impact of the COVID-19 pandemic in 2022. During this period, school closures and the shift to online learning led to a significant increase in students’ screen time for academic and recreational purposes, which is a known risk factor for myopia onset and progression. Interestingly, while we observed a significant improvement in outdoor activity time—a protective factor—this alone may not have been sufficient to counteract the negative effects of prolonged near-work and increased digital screen exposure mandated by the lockdowns and remote learning policies. In this study, we conducted interventions to Ethnic Middle School, and although the myopia prevalence rate in Ethnic Middle School was slightly higher than that in Xianghe Middle School during the second follow-up, the intervention guided by MMHEM was associated with beneficial outcomes overall in Ethnic Middle School.

Several studies have found that school-based outdoor activity interventions can reduce myopia changes in children with myopia [16–18]. We conducted further investigations on students who have already developed myopia. Our research result showed that the number of students with moderate and high myopia and the prevalence rate of moderate and high myopia in Ethnic Middle School were lower than those in Xianghe Middle School. Additionally, we conducted comparisons of the incidence rate of medium to high myopia over the past three years and found that the incidence rate of medium to high myopia in Ethnic Middle School has always been lower than that in Xianghe Middle School. Therefore, intervention measures based on MMHEM may contribute to the prevention and control of myopia progression of myopia towards severity. Compared to interventions solely based on outdoor activities, comprehensive interventions mediated by the MMHEM theory will have a more successful impact on the myopia situation of children with myopia.

In this study, we also measured the refractive error and axial length of the eyeballs of two groups and compared their respective growth rates. The research results indicate that the SE and AL values of these two schools have improved to varying degrees within three years. Compared with Ethnic Middle School, students in Xianghe Middle School have higher severity and average AL values for equivalent spherical mirrors. Additionally, from the difference in vision between the two schools, it can be seen that the degree of eye ocular refractive parameters change in the eyes of Xianghe Middle School is also more severe than that of Ethnic Middle School. However, in terms of the difference between SE and AL, there was no significant difference in the second follow-up survey. From the cumulative growth and trend of SE and AL, it can be seen that the visual acuity of Ethnic Middle School is progressing slowly compared to Xianghe Middle School. The results of this study suggest that, compared with Xianghe Middle School, the MMHEM intervention was associated with an attenuation in the development of ocular refractive parameters and axial changes among ethnic minority middle school students, which is consistent with a beneficial effect on controlling the occurrence and progression of myopia.

The observed attenuation of myopia progression through this behavioral and environmental intervention warrants consideration alongside established clinical modalities. Interventions such as orthokeratology, dual-focus lenses, and low-dose atropine act primarily by manipulating retinal defocus to slow axial elongation after myopia onset [19]. In contrast, the MMHEM model constitutes an upstream public health strategy that targets modifiable risk factors, notably insufficient outdoor time, to prevent or delay the onset and progression of myopia. Thus, this model is not a direct substitute for clinical tools but a complementary, population-based approach. By fostering sustainable behavioral change and eye health literacy within a culturally resonant framework, it may reduce the overall burden and severity of myopia at a community level, potentially diminishing the future need for intensive clinical interventions.

In this study, we also developed a questionnaire related to behavioral factors that affect myopia and assigned scores for different behavioral factors. By comparing the scores of behavioral factors before and after the intervention in Ethnic Middle School, we found that although the influence of electronic screen use and the habit of using eyes in close proximity after the intervention was not changed. These paradoxical increases likely due to the pandemic-induced modality shifts, as mandatory online learning during COVID-19 lockdowns exacerbated digital eye strain patterns [20]. Despite the impact of COVID-19, this did not affect the final outcome of the intervention. We believe that, although electronic screen use and close-range eye use have not been well improved, at least no more serious changes have occurred through this behavioral intervention. In accordance with a prospective, cluster-randomized study conducted by He et al. [16], wherein outdoor activities were exclusively controlled, a noteworthy reduction in incidence was observed. The disparities in the incidence rates of SE and AL were determined as -0.84 and 0.55, respectively. In comparison to the outcomes of the present study (-0.63, 0.25), He et al. [16] reported greater myopia severity among its participants. This suggests that the intervention measures guided by MMHEM in this study have yielded anticipated outcomes in students with myopia.

A key strength of this study is its robust quasi-experimental design, which integrates innovative conceptual, methodological, and practical elements. First, we developed and implemented an innovative intervention guided by the MMHEM model. Second, the intervention effect was rigorously evaluated using DID approach. This analytical strategy was further strengthened by adjusting for key baseline confounders, including parental myopia, and employing cluster-robust standard errors to account for school-based clustering, thereby providing a more unbiased estimate of the causal effect. Third, the two-year longitudinal follow-up within a cohort design moves beyond short-term assessment, offering valuable insights into the sustained impact of the intervention. Finally, conducting the study within a school-based natural population, as opposed to a clinical setting, enhances the ecological validity and generalizability of our findings to real-world public health practice.

This study has two primary limitations. Firstly, due to the small sample size in this study, our future research will expand the sample size to further validate the reliability and generalizability of the findings. We will also adopt a randomized controlled trial design to better control for potential confounding factors, thereby providing further validation of the intervention’s effect. Secondly, given that our study was a 3-year investigation, future research should consider adopting longer-term cohort studies to further validate the findings. Lastly, non-cycloplegic autorefraction may have led to an overestimation of myopia prevalence and incidence in both study groups due to accommodative tone. However, since the primary analysis was based on the DID approach, which compares the changes over time between the groups, systematic measurement error have been effectively controlled for between the two groups. Therefore, while the absolute rates might be overestimated, the assessment of the relative intervention effect is presumed to be unbiased.

Conclusion

A school-based comprehension intervention, guided by the MMHEM and integrating strategies across art, culture, and science fields, was associated with a lower incidence of myopia and slower axial elongation over two years compared to a control school. These findings suggest that MMHEM can serve as a methodological framework for future myopia research, offering a structured approach for exploring effective intervention strategies. Future studies should incorporate RCTs to further validate the intervention’s effectiveness, strengthen its generalizability, and address long-term impact.

Abbreviations

MMHEM

Multisensory Multilevel Health Education Model

SEM

Socio-ecological model

SE

Spherical equivalent

AL

Axial length

Authors’ contributions

Ying Huang and Zehui Yan have written the main manuscript. Litao Chang, Weimin Gao and Ying Huang have analyzed the data. Dafeng Huang has logically proof read the database. Jinjiao Zhang, Zixue Ma, Xixi Li, Xiao Luo, Yayi Dai and Chenxi Su have participated field survey. All authors reviewed the manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (Grant NO. 81960593).

Data availability

The data supporting this study are primary data collected by our research team. Ownership of the data resides with the Department of Child and Adolescent Health and Maternal &amp; Child Health Care, School of Public Health, Kunming Medical University. Researchers interested in accessing the data may submit a reasonable request to the corresponding author.

Declarations

Ethics approval and consent to participate

Ethics approval was obtained from the Kunming Medical University ethical review board and an informed consent form has been signed.

Consent for publication

The study has obtained written informed consent for personal details from the parents or legal guardians of all participants.

Competing interests

The authors declare no competing interests.