High myopia (spherical equivalent, >−6.00 D) leads to the irreversible retinal atrophy and will cause a lot of complications such as macular degeneration, retinal detachment, glaucoma, and premature cataracts (1–3). The risk of suffering from a retinal detachment is 20 times more in a person with high myopia than an emmetropic one (4). Recently, the high prevalence of myopia worldwide among children and teenagers is a serious threat of public health especially in Asia (5–7). In China, the myopia incidence rate in primary school is approximately 40%, in middle and high school students is even higher and could be even severer in near future (8–10). The increasing number of myopias in teenagers whose myopia progress even faster attracts more attention to develop novel strategies to slow down the progress of myopia (11).

To slow down the progress of myopia, it is the prerequisite to understand the risk factors and mechanisms of myopia. The previous studies suggested that the genetic factors play substantial roles than the environmental ones (12, 13). Thus, the children whose parents are myopia trend to be easier to suffer from myopia than those children who have only one or no myopic parent. However, there are solid evidences also clearly showed that the environmental factors could not be ignored for developing myopia especially at current age when it requires school-aged children to take longer time reading, sustainable homework and spend fewer hours for outdoor activities (14–16).

According to animal studies, it was demonstrated that hyperopic defocus in retina causes refractive development and eye axial excessive growth, which promotes myopia progression. Conversely, myopic defocus in retina could retard the eye axial growth, which slows myopia progression (17–20). On the basis of these observations, a novel lens, namely, multifocal lens (MFL) was developed with peripheral focus technology that provides a central zone containing the distance correction and periphery zone having a myopic defocus by adding an extra positive power, resulting in myopic defocus in retina to slow down the eye axial elongation. Currently, there are two main types of MFLs designs including concentric rings/bifocals (BF) and progressive power/peripheral add lens (PAL), which principally provide both near- and distance-vision spectrum (21). With BF, it has two zones of myopia correction for all gaze positions and two neighboring concentric treatment zones with plus power addition to simultaneously deliver peripheral myopic defocus (22, 23). However, PAL simultaneously produces constantly peripheral myopization defocus that increases gradually from the central optic axis toward the periphery (21, 23).

Although it is in theory that MFLs could slow down the myopia progression, the outcomes from clinic practice are controversy. Thus, the several earlier meta-analyses also could not achieve consistent conclusion due to multiple reasons. Li et al. (24) collected nine randomized controlled trials (RCTs) from 1989–2010 to compare effects of MLCs with single vision lenses (SVLs) in children. The data suggested that MLCs with powers ranging from + 1.50 to + 2.00 D were associated with a statistically significantly decrease in myopia progression in school-aged children compared with SVLs. In 2017, the other group (24) evaluated the possible difference between BFs and PALs. They found that both BFs and PALs are clinically effective to control myopia in school-aged children. However, it seems that BFs seem to have greater effect than PALs. Recently, Kaphle et al. (25) performed a meta-analysis to compare the absolute progression rates over the successive 6-months or 1-year periods to gauge how long the efficacy of the intervention lasts. They found that the treatment effect of MFLs is maximum in the first 6- and 12-months intervals and is reduced in subsequent intervals. A latest study (26) also suggested a similar tendency when comparing the relative increases rather than absolute increases in measures of myopia progression for PALs and SVLs that the relative efficacy of PALs tends to be weaker after the first 12 months. Overall, these studies indicate that MFL is effective to slow down the progress of myopia in children, but how long the treatment effects could be sustained and whether treatment interval affects the effects remain to be intensively reviewed.

In this study, we conducted a systematic meta-analysis focusing on currently available evidences from 15 high quality RCTs involving 1,840 children aged 6–18 years to assess the effects of MFLs vs. SVLs on slowing myopic progress. We extended our analyzed treatment period up to 36 months and subdivide into 6- and 12-months (annual) visit intervals. To more precisely evaluate the treatment effects, both spherical equivalent refraction (SER) and axial length (AL) are analyzed.

Currently, the most prevalent treatments to control the myopia progress involves in pharmacological agents such as atropine ophthalmologic drops, orthokeratology (OK), and MFL. Atropine ophthalmologic drops could reduce accommodation and increase pupillary diameter, resulting in well-controlled myopia progress (56). However, because of its side-effects such as photophobia, poor near visual acuity, increased pupillary diameter, and headache, many parents terminate the treatment with atropine for their children. In addition, the mechanism why atropine could control myopic progress and whether long-term application of antimuscarinic agents on ocular tissue is harmful or not remain to be determined (38). Orthokeratology is suitable for the low-to-moderate myopia, the children can wear it overnight to remodel the corneal epithelial into a flatter and less powerful refractive surface, achieving a transient emmetropia (57, 58). According to the clinical trials, the axial elongation with OK group could be significantly inhibited and the peripheral refractive status of the cornea could be less hyperopic defocus when comparing with single vision group (59). However, the children who take OK treatments are easy to suffer from microbial keratitis, contact lens irritation, dry eye, and corneal epithelial iron deposition (60, 61).

In this study, we summarized currently available evidences of controlling myopia progression in school-aged children aged 6–18 years from 15 RCTs to dissect the potential roles of MFLs treatment. The effects of MFLs to slow down the myopic progress could be sustained to 36 months but with slightly decreased effects from the second visit. While the treatment effects in both 6-months and annual-visit trials are positive, the annual visit is more profound comparing to 6-months visit group (Figure 7) although no statistic difference was detected which is likely due to the variability and small samples.

It is still unknown why a decreased effect happened after first visit in both 6-months visit and annual visit trials not only in our analysis but also in the previous reports (25, 26). Our analysis showed that in annual visit, differences of mean SER change is 0.29 D for the first year, 0.17 D for the 12–24 months and 0.18 D for the 24–36 months; of mean AL change is −0.12 mm for the first year, −0.07 mm for the 12–24 months and −0.17 mm for the 24–36 months. In 6-months visit, the difference of mean SER change is 0.14 D for the first visit and around 0.05–0.07 D for the following visit intervals; of mean AL change is −0.07 mm for the first visit and −0.04 to −0.02 mm for the following visits. Kaphle et al. proposed accommodation adaptation may play a role to explain this phenomenon (25). It was reported that the lag of accommodation gradually increases while a person wearing MFLs (62), the same addition power may not be as effective as it did earlier. Therefore, to maintain the effective treatment for MFLs as they were in the beginning of the trial, the power of the addition should be increased gradually. In addition, they also supposed that the age is a critical factor since myopia usually stabilizes when a child reaches a certain age which means the rate of myopia progression decreases over time in the SVLs control group, and hence the treatment effect of the myopia intervention reduces as it is determined via comparison with the progression of the SVLs group. This argument is however not very reliable since the average age of children involved in these clinic trails is around 9.5–12 years old at least in current study (Tables 1, 2), which is far from the stage for the myopia stabilization although a slower and more stable rates of change of myopia after onset was observed (63, 64).

In addition, we clearly present an obvious difference in distinct treatment intervals. Comparing to 6-months trails, the annual visit with MFLs show more profound effects [annual vs. 6-months visit: SER: 0.29 D (95% CI 0.21 ∼ 0.37) vs. 0.19 D (95% CI 0.11 ∼ 0.28) at the first year, 0.46 D (95% CI 0.32 ∼ 0.60) vs. 0.31 D (95% CI 0.18 ∼ 0.44) at the second year; AL: −0.12 (95% CI −0.14 ∼−0.11) vs. −0.07 (95% CI −0.10 ∼−0.04) at the first year, −0.19 (95% CI −0.22 ∼ -0.16) vs. −0.10 (95% CI −0.13 ∼−0.06) at the second year]. One possibility could be that frequent lens power adjustment (less than 1 year) affects the accommodation adaptation response which eventually blocks the treatment effects. Thus, the detailed mechanisms on why frequent lens power adjustment is not conducive to myopia control in this case should be further explored.

There are several limitations in our study. First, only the RCTs were included in the analysis. We excluded at least one clinic research from Paune et al. (36) which is a prospective, longitudinal, non-randomized study (36). In addition, a few studies, for example the trials from Fujikado et al. (39) and Berntsen et al. (41) met the inclusion criteria, but were not included because of unavailability of data at the required time points.

In addition, much fewer (only 4 in 15 RCTs) studies finished 3-year observation. One trial from Lam et al. (46) was a part of the data from Lam (35) by year of 2021. However, the data at 36 months from the trail of Lam et al. (35) was discarded because it lacks of control data since the control group of this trial was switched from SVLs to MFLs at this time point. Besides, we could not exclude the possibility that the third-year interval was terminated due to either good or no obvious treatment effects at the first and/or especially the second year, which could affect the treatment effects in the meta-analysis. In this analysis, we could reach the data of 3-year treatments with annual visit but not 6-months visit. Therefore, the difference between annual and 6-months visit for a long-term treatment effect is not determined.

Third, our analysis may also have publication bias since the studies we selected were peer-reviewed. Currently, the studies with positive treatment effects are much easy to be published comparing to the studies with negative or no obvious treatment effects. Besides, some clinic trails may be terminated because no significant treatment effects are observed at early stage. This likely would overestimate the treatment effect of intervention.

Lastly, heterogeneity is a common problem for meta-analysis and it is difficult to deal with especially when the enrolled studies is selected in certain cases. In current study, we noticed a high heterogeneity which promotes us to do subgroup analysis. By analyzing the potential factors such as the county/region, added power and visit intervals (6-months and annual-visit), we found that subgroup with the visit intervals shows smaller and tolerable heterogeneity. Actually, this factor is exactly the one we would like to analyze because it seems distinct visit intervals did affect the treatment effects from literatures and in clinic practice. However, the part of data still shows moderate heterogeneity after subgroup. Therefore, a random-effect model for analysis was selected as other previous studies conducted. In addition, it is also possible to eliminate/reduce the heterogeneity by excluding certain studies. Thus, we found that by excluding the studies from Lam et al. (46) and Cheng et al. (55) could significantly reduce the heterogeneity. However, excluding specific study is not applicable when no decent reasons arising. Nevertheless, we found that the mean differences when including or excluding these two studies are only tiny changed, which did not alter any of the interpretations. Furthermore, a meta-regression could be also performed to find the potential factors that arise the heterogeneity. However, the meta-regression would not eventually dissolve the problem since we have failed to reduce the heterogeneity when subgroup with country/region and added power and we are mainly focusing on the treatment effects from distinct visit intervals in current study.

In conclusion, our analysis showed that the treatment effect of MFLs is positive in either 6-months or annual visit although annual visit shows more profound effects. In addition, it also highlights that while a slightly decreased effect was observed from the second visit in both 6-months and annual visit, a longer treatment likely acquires a better effect. Thus, the data encourages the MFLs users to maintain the long-term treatment (at least for 36 months) with annual visit. Due to the presence of heterogeneity in this analysis, a standardized large scale multi-center clinical trial should be conducted to provide an explicit direction.

The original contributions presented in this study are included in the article/supplementary material, further inquiries can be directed to the corresponding authors.

MC and YW designed and supervised the project. MC performed literature review, data analysis, figure preparation, and prepared the manuscript with contributions from all authors. LX and CH prepared the tables. HL, FC, and HW reviewed the literatures and datasets. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.