Interventions for myopia control in children: a living systematic review and network meta-analysis

1Myopia, or short‐ or near‐sightedness, is a common refractive anomaly of the eye that occurs when parallel rays of light are brought to a focus in front of the retina with accommodation at rest, causing distant objects to appear blurred and near objects to remain clear (Morgan 2012). Myopia most often results from the eyeball being too long (i.e. there is excessive axial elongation), but can also occur when the image‐forming structures of the eye are too strong (Flitcroft 2019).

2The prevalence of myopia shows significant age, ethnic and regional variation (Rudnicka 2016). Currently, 30% to 50% of adults in the USA and Europe have myopia (Dolgin 2015). Myopia is already reaching 'epidemic' proportions in children and young adults in urban areas of East and South East Asia, with over 80% of children being myopic by the time they complete their high school education (Dolgin 2015). If current trends continue, it is estimated that by 2050 there will be approximately 5 billion (5000 million) people with myopia (i.e. about 50% of the world's population), with around 10% having high myopia (when defined as a spherical equivalent of −5.00 dioptres (D) or worse) (Holden 2016).

3The aetiology of myopia involves a complex interaction between environmental and genetic factors. Although genetic inheritance is a well‐established predisposing factor for myopia, genetic factors cannot explain the rapidly rising prevalence of the condition (Williams 2019). A Mendelian randomisation study, using the UK Biobank cohort, provided strong evidence for the cumulative effect of additional years in education on myopia development (Mountjoy 2018). Mendelian randomisation is a statistical approach that uses genetics to provide information about the relationship between an exposure and outcome. This study estimated that for each additional year in education, myopic spherical equivalent increased by −0.27 D. Evidence from a number of observational studies further supports the causal association between environmental and social factors and myopia development (Morgan 2018).

4Epidemiological studies have shown that myopia is an established risk factor for a number of ocular pathologies, including cataract, glaucoma and retinal detachment (Flitcroft 2012). Although myopia‐related complications can occur irrespective of age and degree of myopia (Dhakal 2018), the excessive axial elongation associated with higher degrees of myopia causes biomechanical stretching of the outer coat of the eye, increasing the risk of sight‐threatening pathologies such as posterior staphyloma and myopic maculopathy (Saw 2005; Verkicharla 2015). A meta‐analysis of population studies reporting blindness and visual impairment due to myopic maculopathy (Fricke 2018), estimated that in 2015, approximately 10 million people had visual impairment due to myopic macular degeneration, of whom three million were blind. Although the sight‐threatening pathologies associated with myopia usually occur later in life, the underlying myopia develops during childhood and therefore interventions to reduce the progression of myopia have the potential to reduce future visual impairment.

5Most cases of myopia develop during childhood and the prevalence of myopia begins to increase noticeably after the age of six years (McCullough 2016). Progression rates vary significantly, with rates in Asian children being approximately 0.20 D per year faster than their age‐matched European counterparts (Donovan 2012). Since myopia tends to stabilise in late adolescence, interventions to slow myopia progression need to be delivered in childhood.

6Interventions to slow progression of myopia can be grouped into three broad categories: optical, pharmacological and environmental (Wildsoet 2019). Optical interventions include a variety of spectacle and contact lens designs. Spectacles are the least invasive and most accessible method for potentially slowing myopia progression. Spectacle options include refractive under‐correction, bifocal and progressive addition lenses and, more recently, specialised 'myopia control' designs. Soft multifocal and approved myopia control contact lenses are increasingly being used for myopia management in children (Efron 2020). Centre‐distance soft multifocal lens designs incorporate a central zone that contains the distance refractive correction, with peripheral regions of the lens having relatively increased positive power (myopic defocus). This is achieved by either a gradual increase in power towards the periphery or using concentric peripheral zones of alternating myopic defocus and distance correction. Orthokeratology involves the use of specialised rigid contact lenses that are worn during sleep to change the topography of the cornea to reduce myopic refractive error and also manipulate peripheral retinal defocus. Safety remains a concern because of the greater risk of sight‐threatening microbial keratitis with overnight wear compared with daily contact lens wear modalities (Dart 2008).

7The most commonly used topical pharmacological intervention for myopia control is atropine, a non‐selective muscarinic antagonist, which has been widely used in clinical trials in concentrations ranging from 0.01% to 1.0%. Although higher atropine concentrations have been shown to be effective in retarding myopia progression in children, the higher incidence of side effects with higher doses, including cycloplegia (inhibition of accommodation) and pupil dilation (which causes blur for near vision and photophobia) limits its use. Furthermore, a rebound effect (involving more rapid myopia progression) after discontinuation of therapy is more pronounced with higher concentrations of atropine (Chia 2014). More recent studies have evaluated the efficacy of lower concentrations to reduce side effects and lessen the likelihood of rebound. The results of these studies have led to a renewed interest in the clinical application of low‐dose atropine (i.e. 0.01% to 0.05%) for myopia control (Wu 2019). Other pharmacological agents that have been evaluated for myopia control include topical tropicamide, cyclopentolate and pirenzipine (a selective M1 muscarinic antagonist) and the oral adenosine antagonist, 7‐methylxanthine.

8Evidence that more time spent on near work activities is associated with higher odds of developing myopia (Huang 2015), and the observation that increased time spent outdoors is protective against myopia, after adjusting for near work, parental myopia and ethnicity (Rose 2008), have raised the possibility that environmental or behavioural interventions could be effective for myopia control. Trials of school‐based programmes that promote outdoor activities, conducted in East Asia, have reported a lower incidence of myopia onset but have limited impact on progression following onset of myopia (Dhakal 2022).

9Animal studies have shown that optically‐induced changes to the effective refractive status of the eye can regulate eye growth and influence refractive development (Troilo 2019). Specifically, the observation that imposed relative myopic defocus (image focused in front of the retina) can slow axial elongation has been the impetus for the development of novel multifocal spectacles and contact lenses that provide clear central vision, whilst at the same time presenting myopic defocus over a large proportion of the visual field. The critical area ratio required for these simultaneous competing defocus signals to dominate eye growth is currently unclear. However, the relative treatment effects reported for different optical treatment regimens suggest that there appears to be an eccentricity‐dependent decrease in the efficacy of myopic defocus beyond the near periphery (Smith 2014; Smith 2020).

10Orthokeratology involves corneal reshaping lenses that are worn overnight to flatten the central cornea and reduce its dioptric power. The geometry of these lenses also creates a corneal profile that produces relative myopic defocus.

11The precise mechanism by which anti‐muscarinic agents reduce myopic progression is not fully understood. A non‐accommodative mechanism is thought to be the most likely, and alternative targets have been proposed, including eye growth regulatory pathways that arise in the retina and are relayed to the sclera via the retinal pigment epithelium and choroid (McBrien 2013; Upadhyay 2020).

12The protective effect of increased time outdoors on myopia development is thought to be related to the higher light intensity of sunlight and possibly its spectral composition (French 2013). Light levels have been shown to influence refractive development in animal models (Smith 2012). Higher light intensities stimulate retinal dopamine production, which is thought to inhibit axial elongation (Feldkaemper 2013).

13As a result of its increasing global prevalence and association with sight‐threatening pathologies, myopia is emerging as a major public health concern. Myopia is predicted to affect almost half of the world’s population by 2050, and the pathologic consequences of high myopia increase the risk of irreversible visual impairment and blindness. There has been considerable interest in the development of strategies to delay the onset of myopia and slow its progression. Myopia control interventions are increasingly being used in routine clinical practice (Efron 2020; Wolffsohn 2016). Evidence from randomised controlled trials (RCTs) indicates that the progression of myopia can be slowed by different interventions, although treatment efficacy is highly variable.

14There is a broad consensus that the primary endpoints for judging efficacy in clinical trials of myopia control interventions should include change in axial length, in addition to change in refractive error (Brennan 2020; Walline 2018; Wolffsohn 2019). Myopia development and progression usually occur due to abnormal axial elongation. Therefore, axial length may be a better predictor of future progression and consequent risk of posterior pole complications (Brennan 2020). In terms of a minimal clinically important difference of the key efficacy outcomes in myopia control studies, an expert panel concluded that a mean difference between intervention groups of 0.25 D per year would be regarded as clinically significant (i.e. 0.75 D over the course of a three‐year study) (Walline 2018). This would correspond to a difference in axial length of approximately 0.3 mm.

15An updated Cochrane systematic review, published in January 2020 (Walline 2020), evaluated the efficacy of a number of interventions, including spectacles, contact lenses and pharmaceutical agents, for slowing the progression of myopia in children. Walline 2020 concluded that topical anti‐muscarinic medication was effective in slowing myopia progression. Multifocal lenses, either spectacles or contact lenses, also conferred a small benefit. Although the update was published in 2020, the review only included evidence published up to the end of 2018. In this rapidly moving field, the results of additional important trials have subsequently been reported.

16Eye care professionals often find it difficult to assimilate potentially conflicting evidence to inform their clinical decision‐making (Douglass 2020). It is therefore important that practitioners can access high‐quality and up‐to‐date evidence to inform practice. Moreover, parents of myopic children also need reliable information to help them to understand and interpret research findings. Given the large number of different interventions available for myopia control and the large number of completed and ongoing RCTs on this topic, there is an urgent need to evaluate the comparative effectiveness of different interventions. A network meta‐analysis (NMA) offers an advantage over a standard pairwise meta‐analysis in that it provides both direct comparisons of individual trials and indirect comparisons not directly evaluated in trials across a network of studies, thus generating the comparativeness of all interventions in a coherent manner. A NMA can also provide relative rankings of interventions to inform clinical decision‐making.

17There are significant resource implications associated with myopia for both individuals and healthcare systems. This includes both corrected and uncorrected myopic refractive error. Lim 2009 estimated the mean direct costs of managing myopia in school‐aged children in Singapore. These costs included optometrist visits, spectacles, contact lenses and travel costs. The mean cost was estimated as USD 148 (median SGD 83.33) per year in 2006. In addition, Zheng 2013 estimated the lifetime costs for a person with myopia over an 80‐year lifespan to be USD 17,020 in 2011. There are also associated costs and quality‐of‐life impacts associated with uncorrected refractive error. Tahhan 2013 found a significant reduction in health state utility (a preference‐based quality‐of‐life measure) associated with uncorrected refractive error. Fricke 2012 estimated that the direct costs of correcting all cases of uncorrected refractive error globally would be approximately USD 28 billion (USD 28,000 million; price year not stated). Given these cost estimates, understanding the current evidence base for myopia control is key for both individuals and healthcare decision‐makers.

18We plan to maintain this review as a living systematic review. This will involve searching the literature every six months and incorporating new evidence as it becomes available. This approach is appropriate for this review since it addresses an important clinical topic and there is currently significant uncertainty as to the most effective intervention. It is therefore important that consumers and healthcare providers have access to the most up‐to‐date evidence to make informed decisions. The review authors are aware of several relevant ongoing trials that will be important to incorporate in a timely manner.

19To assess the comparative efficacy of optical, pharmacological and environmental interventions for slowing myopia progression in children using network meta‐analysis (NMA). To generate a relative ranking of myopia control interventions according to their efficacy. To produce a brief economic commentary, summarising the economic evaluations assessing interventions for myopia control in children. To maintain the currency of the evidence using a living systematic review approach.

20We included randomised controlled trials (RCTs) of optical, pharmacological and environmental interventions used alone or in combination for slowing the progression of myopia in children.

21This review considered studies that included children 18 years old and younger. We excluded studies in which the majority of participants were older than 18 years at the start of the study. We also excluded studies that included participants with spherical equivalent myopia less than −0.50 D at baseline. The spherical equivalent is calculated by the sum of the spherical power plus half the cylindrical power of the refractive error.

22We included studies that compared interventions of interest and reported having measured the relevant outcomes, irrespective of whether data for the outcomes were available.

23We included studies that compared any of the interventions listed below with a control group, or with each other. For the purposes of the analysis, we defined a control group as a placebo intervention or single vision spectacles or contact lenses.

24Undercorrection of myopia with single vision spectacle lensesMultifocal (bifocal or progressive addition) spectacle lenses, peripheral defocus spectacle lensesMultifocal soft contact lenses (MFSCL; concentric ring or progressive designs), rigid gas‐permeable contact lenses or corneal reshaping (orthokeratology) contact lensesAtropine (stratified according to dosing regime as high (≥ 0.5%), moderate (0.1% to < 0.5%) and low (< 0.1%)Other pharmaceutical agents (e.g. pirenzepine, 7‐methylxanthine)Environmental interventions (e.g. time spent outdoors, modifications to the performance of near work)

25Undercorrection of myopia with single vision spectacle lenses

26Multifocal (bifocal or progressive addition) spectacle lenses, peripheral defocus spectacle lenses

27Multifocal soft contact lenses (MFSCL; concentric ring or progressive designs), rigid gas‐permeable contact lenses or corneal reshaping (orthokeratology) contact lenses

28Atropine (stratified according to dosing regime as high (≥ 0.5%), moderate (0.1% to < 0.5%) and low (< 0.1%)

29Other pharmaceutical agents (e.g. pirenzepine, 7‐methylxanthine)

30Environmental interventions (e.g. time spent outdoors, modifications to the performance of near work)

31Progression of myopia was assessed by:

32mean change in refractive error (spherical equivalent in D) from baseline for each year of follow‐up and measured by any method (e.g. objective or subjective refraction); andmean change in axial length for each year of follow‐up in millimetres (mm) from baseline for each year of follow‐up and measured by any method (e.g. ultrasound or optical biometry).

33mean change in refractive error (spherical equivalent in D) from baseline for each year of follow‐up and measured by any method (e.g. objective or subjective refraction); and

34mean change in axial length for each year of follow‐up in millimetres (mm) from baseline for each year of follow‐up and measured by any method (e.g. ultrasound or optical biometry).

35Rebound was evaluated when children in the treatment group were switched to the control treatment and then followed for a minimum period of one year.

36We described adverse events relating to the interventions as reported in the included studies, irrespective of severity. These included but were not limited to blurred vision, photophobia, hypersensitivity reactions, corneal infiltrative events and infections. In studies that graded clinical signs using standard anterior eye grading scales from normal to severe, we recorded the number of clinically significant signs (grade 3 or 4) that would usually require a clinical action.

37Where data were available we documented withdrawals due to adverse events and number of 'serious' events.

38We documented vision‐related or health‐related quality of life when reported, measured by any validated questionnaire (e.g. National Eye Institute (NEI) Visual Function Questionnaire 25 (NEI VFQ‐25), or EuroQol questionnaire, EQ‐5D).

39Studies evaluated adherence with the prescribed treatment regimen using a variety of compliance measures, including daily wearing time with contact lenses and spectacle interventions as reported by parents or children, or both, or the proportion of participants in pharmacological studies following the required dosing regime.

40We have reported outcomes at one year, two years and as available for the duration of the study. We imposed no restrictions based on the length of follow‐up.

41We present evidence regarding relevant economic evaluations, as a brief economic commentary.

42The Cochrane Eyes and Vision Information Specialist searched the electronic databases below for RCTs and controlled clinical trials. There were no restrictions to language or date of publication. Given the similarity in the PICO and corresponding search strategies between the current review and a previous Cochrane Review on interventions for myopia control in children (Walline 2020), and the likelihood that studies included in Walline 2020 would meet the inclusion criteria for this review, we ran the search for the current review in parallel with the search strategy used by Walline 2020 up to the search date for the earlier review (26 February 2019) and removed duplicates. We combined the search results with all records identified up to 4 February 2022.

43We did not perform the generic search described in Electronic searches for adverse events, however we added a filter to the search strategy to identify systematic reviews of adverse events associated with myopia control interventions. We compared the findings of these reviews to the adverse events reported in the studies included in the current review.

44In addition to these searches we carried out a MEDLINE and Embase search using economic search filters to specifically identify economic studies.

45We have developed this review as a living systematic review, and we will re‐run the searches on a six‐monthly basis.

46Cochrane Central Register of Controlled Trials (CENTRAL; which contains the Cochrane Eyes and Vision Trials Register; 2022, Issue 2) in the Cochrane Library (Appendix 1)MEDLINE Ovid (1946 to 26 February 2022; Appendix 2)MEDLINE Ovid ‐ economic search (1946 to 26 February 2022; Appendix 3)MEDLINE Ovid ‐ adverse events (1946 to 26 February 2022; Appendix 4)Embase Ovid (1980 to 26 February 2022; Appendix 5)Embase Ovid ‐ economic search (1980 to 4 February 2022; Appendix 6)Embase Ovid ‐ adverse events (1980 to 26 February 2022; Appendix 7)ISRCTN registry (www.isrctn.com/editAdvancedSearch) (Appendix 8)US National Institutes of Health Ongoing Trials Register ClinicalTrials.gov (www.clinicaltrials.gov; Appendix 9)World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP; www.who.int/ictrp; Appendix 10)

47Cochrane Central Register of Controlled Trials (CENTRAL; which contains the Cochrane Eyes and Vision Trials Register; 2022, Issue 2) in the Cochrane Library (Appendix 1)

48MEDLINE Ovid (1946 to 26 February 2022; Appendix 2)

49MEDLINE Ovid ‐ economic search (1946 to 26 February 2022; Appendix 3)

50MEDLINE Ovid ‐ adverse events (1946 to 26 February 2022; Appendix 4)

51Embase Ovid (1980 to 26 February 2022; Appendix 5)

52Embase Ovid ‐ economic search (1980 to 4 February 2022; Appendix 6)

53Embase Ovid ‐ adverse events (1980 to 26 February 2022; Appendix 7)

54ISRCTN registry (www.isrctn.com/editAdvancedSearch) (Appendix 8)

55US National Institutes of Health Ongoing Trials Register ClinicalTrials.gov (www.clinicaltrials.gov; Appendix 9)

56World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP; www.who.int/ictrp; Appendix 10)

57We searched the reference lists of identified study reports to identify additional studies. We also contacted the principal investigators of included studies for details of other potentially relevant studies not identified by the electronic searches, and of recently completed or ongoing studies.

58The Information Specialist at Cochrane Eyes and Vision downloaded all titles and abstracts retrieved from the electronic searches to EndNote (Endnote X9 2013) and removed duplicates before uploading to Covidence. Two review authors (from JGL, RS, BH, RD, PV) independently reviewed the titles and abstracts of the search results based on the eligibility criteria stated above. We categorised Abstracts for inclusion as 'Yes', 'Maybe' or 'No'. We obtained the full text of articles for the studies categorised as 'Maybe' and 'Yes', and reassessed them for final eligibility. After examining the full text, we labelled studies as 'include' or 'exclude'. Studies selected as 'exclude' by both authors were excluded from the review. We documented the reasons for exclusion. We resolved any screening discrepancies through discussion and, if necessary, through consultation with a third review author. One review author (AK) screened the economic search results.

59We plan to screen any new citations retrieved by the six‐monthly searches immediately.

60For eligible studies, two review authors independently extracted the data. We contacted the authors of the original reports to obtain further details if the data reported were unclear or incomplete. We exported the collected data into Review Manager Web (RevMan Web) (RevMan Web 2022). We extracted the following study characteristics.

61Methods: study design, number and location of study centre(s), date of study and total durationParticipants: inclusion and exclusion criteria, number randomised, number lost to follow‐up or withdrawn, number analysed, mean age and standard deviation (SD), age range, genderInterventions: description of intervention and comparatorOutcomes: primary and secondary outcomes specified and collected, and time points reported. Unit of analysisNotes: funding for study and conflicts of interest of study authors

62Methods: study design, number and location of study centre(s), date of study and total duration

63Participants: inclusion and exclusion criteria, number randomised, number lost to follow‐up or withdrawn, number analysed, mean age and standard deviation (SD), age range, gender

64Interventions: description of intervention and comparator

65Outcomes: primary and secondary outcomes specified and collected, and time points reported. Unit of analysis

66Notes: funding for study and conflicts of interest of study authors

67Pairs of review authors (from JGL, BH, RS, RD, PV, SM, DL) independently assessed the risk of bias in the included studies for all outcomes using the revised Cochrane risk of bias tool for randomised trials (RoB 2) 22 August 2019 version, described in Chapter 8 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2022a). RoB 2 covers five domains of bias:

68bias arising from the randomisation process;bias due to deviations from intended interventions;bias due to missing outcome data;bias in measurement of the outcome; andbias in selection of the reported result.

69bias arising from the randomisation process;

70bias due to deviations from intended interventions;

71bias due to missing outcome data;

72bias in measurement of the outcome; and

73bias in selection of the reported result.

74These domain‐level judgements provide the basis for an overall risk of bias judgement for the specific outcome being assessed. The response options for an overall risk of bias judgement in RoB 2 are the same as for individual domains (i.e. 'low risk of bias'; 'some concerns'; 'high risk of bias'). The following criteria were adopted:

75Low risk of bias: low risk of bias for all domains;Some concerns: 'some concerns' in at least one domain, but not at high risk of bias for any domain;High risk of bias: high risk of bias in at least one domain or the study is judged to have some concerns for multiple domains in a way that substantially lowers confidence in the result.

76Low risk of bias: low risk of bias for all domains;

77Some concerns: 'some concerns' in at least one domain, but not at high risk of bias for any domain;

78High risk of bias: high risk of bias in at least one domain or the study is judged to have some concerns for multiple domains in a way that substantially lowers confidence in the result.

79To implement RoB 2 assessments we used the Excel tool available at https://www.riskofbias.info/welcome/rob-2-0-tool/current-version-of-rob-2.

80We did not include cluster‐randomised trials. In the case of cross‐over trials, we only used data from the first phase prior to the cross over and therefore used the version of the tool for parallel trials. Should cluster‐randomised and cross‐over trials be included in future updates of the review, we will use the versions of RoB 2 with additional considerations for these designs.

81For all outcomes we assessed the effect of assignment to intervention (the intention‐to‐treat effect).

82We conducted the review according to this published protocol and have reported any deviations from it in the Differences between protocol and review section of the review.

83We used mean differences (MDs) as the measure of treatment effect for the critical outcome 'progression of myopia', that is, difference in mean change in refractive error (SER) and axial length from baseline at each year of follow‐up.

84When studies randomised only one eye per participant, the unit of analysis was the individual eye (participant). When studies randomised both eyes from the same participant (either to the same or different interventions), we analysed data adjusted for clustering or paired‐eye design. In the NMA, we accounted for the correlation between the effect sizes derived from the same study.

85In multiple‐arm trials, to overcome a unit‐of‐analysis error for a study that could contribute multiple, correlated data, we combined groups to create a single pair‐wise comparison.

86If we identify cluster‐RCTs in future updates, we will include them in meta‐analyses directly, where the sample size has been adjusted for clustering. We will combine them with the results from individual studies if there is little heterogeneity between the study designs and the interaction between the effect of the intervention and the unit of randomisation is considered to be unlikely. If studies present outcomes at individual level (i.e. a unit of analysis error), we will use established methods to adjust for clustering by calculating an effective sample size by dividing the original sample size by the design effect. This can be calculated from the average cluster size and the intra‐class correlation coefficient (ICC). Where the ICC is unknown, we will use an estimation from similar trials (Higgins 2022b).

87We contacted study authors to verify key study characteristics and to obtain missing outcome data. If we did not receive a response within eight weeks, we analysed the studies based on available data. We used the RevMan calculator to calculate missing standard deviations using other data from the study (e.g. confidence intervals) based on methods outlined in Chapter 10 of the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2022).

88We assessed clinical and methodological heterogeneity for each pairwise meta‐analysis by comparing the characteristics of included studies and by visual inspection of forest plots. We assessed statistical heterogeneity quantitatively for pairwise comparisons using the values of the Chi2 test and the I2 statistic (Higgins 2003). We interpreted I2 statistic values according to Chapter 10 of the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2022), as follows:

890% to 40% may not be important;30% to 60% may represent moderate heterogeneity;50% to 90% may represent substantial heterogeneity;75% to 100% represents considerable heterogeneity.

900% to 40% may not be important;

9130% to 60% may represent moderate heterogeneity;

9250% to 90% may represent substantial heterogeneity;

9375% to 100% represents considerable heterogeneity.

94For the NMA, we assumed a common estimate for the heterogeneity variance across the different comparisons. The assessment of statistical heterogeneity was based on the magnitude of the heterogeneity variance parameter (Tau2) estimated from the NMA models.

95To evaluate the presence of inconsistency locally, we used the node splitting approach (Dias 2010), which assesses the agreement between direct and indirect evidence for each treatment comparison.

96To check the assumption of consistency across the entire network, we used the 'design by treatment' interaction model (White 2015). This method accounts for different sources of inconsistency that can occur when studies with different designs are incorporated into the network (e.g. two‐arm trials versus multi‐arm trials), as well as inconsistency between direct and indirect evidence.

97If there are sufficient studies in future updates, we plan to run network meta‐regression models to detect associations between study size and effect size.

98We initially carried out standard pairwise meta‐analyses to combine outcome data using random‐effects models in RevMan Web. For comparisons with three or fewer trials, we used a fixed‐effect model. We combined change from baseline data in meta‐analyses with mean outcome data using the generic inverse variance (unstandardised) MD method, as outlined in Chapter 10 of the Cochrane Handbook for Systematic Interventions (Deeks 2022). In the case of substantial clinical, methodological or statistical heterogeneity, we generally did not attempt to combine data from individual trials but reported study results separately, however, subtotals were included in some analyses when presenting subgroups with varying degrees of heterogeneity.

99For cross‐over trials we only extracted data from the first phase prior to cross over.

100We conducted a NMA using the network suite of programs available in STATA (http://www.stata.com) for myopia progression, as defined by difference in change in SER and axial length at 12 and 24 months, using random‐effects multivariate models (Chaimani 2013; Chaimani 2015; White 2015). An important concept in NMA is 'transitivity', which implies that the distribution of effect modifiers is similar across all sources of direct evidence. The statistical manifestation of transitivity is consistency, which refers to the statistical agreement between the direct and indirect sources of evidence. We checked for consistency in the network both locally (node‐splitting approach) and globally (design by treatment model).

101We assumed a common heterogeneity across all comparisons in the network. We used te surface under the cumulative ranking curve (SUCRA) to rank the interventions for all available outcomes. SUCRA values range from 0% to 100%. The higher the SUCRA value (i.e. the closer to 100%), the greater the probability of an intervention ranking best. (Chaimani 2015; Salanti 2012).

102In the primary NMA, we considered MFSCL, rigid gas‐permeable lenses and orthokeratology lenses as separate nodes. For spectacle lens interventions, there were separate nodes for undercorrected single vision spectacle lenses, multifocal spectacle lenses and peripheral plus spectacle lenses. We considered each pharmacological intervention as a separate node regardless of the dose. We did not anticipate a strong dose‐response effect except for atropine. We grouped atropine according to dosing regime as high (≥ 0.5%), moderate (0.1 % to < 0.5%) and low (< 0.1%). We grouped all control arms (single vision spectacle lenses, single vision contact lenses, placebo eyedrops or no treatment) into a single node.

103When we were unable to perform a meta‐analysis, we undertook a narrative synthesis following guidance in Chapter 12 of the Cochrane Handbook for Systematic Reviews of Interventions (McKenzie 2022b). Specifically, we presented the effect estimates in structured tables and provided a descriptive summary of the range and distribution of the observed effects. In particular, we noted the direction of effects and whether these were consistent in the individual studies.

104Following the search outlined in the Search methods for identification of studies, we developed a brief economic commentary to summarise the availability and principal findings of the full economic evaluations assessing interventions for myopia control in children as outlined in Chapter 20 of the Cochrane Handbook for Systematic Reviews of Interventions (Aluko 2022). This brief economic commentary was planned to encompass full economic evaluations (i.e. cost‐effectiveness analyses, cost‐utility analyses and cost‐benefit analyses) conducted as part of a single empirical study, such as a RCT, a model based on a single such study or a model based on several such studies.

105Whenever we identify new evidence in future updates (i.e. new studies, data, or other information) that is relevant to the review, we will extract the data and assess risk of bias, as appropriate. We will wait until the accumulating evidence changes one or more of the following components of the review before incorporating it and re‐publishing the review.

106The findings of one or more outcomes (e.g. clinically important change in size or direction of effect)Credibility (e.g. change in the overall confidence in the effect estimates for critical outcomes)

107The findings of one or more outcomes (e.g. clinically important change in size or direction of effect)

108Credibility (e.g. change in the overall confidence in the effect estimates for critical outcomes)

109We will not use formal sequential meta‐analysis approaches for updated meta‐analyses.

110We will review the scope and methods of this review annually in light of potential changes in the topic area or in evidence available for inclusion in the review. Each year, we will consider the necessity for the review to be a living systematic review by assessing ongoing relevance of the question to decision‐makers and by determining whether uncertainty is ongoing in the evidence and whether further relevant research is likely.

111We performed predefined subgroup analyses for types of intervention modalities (i.e. spectacle and contact lens designs, and dose of particular pharmaceutical interventions (e.g. low‐, moderate‐ and high‐dose atropine)). There were insufficient data to carry out other proposed subgroup analyses.

112We planned a sensitivity analysis on the exclusion of studies that we judged to be at high risk of bias or to raise some concerns in at least one domain of RoB 2. However, since we judged almost all the included studies at high risk of bias or with some concerns we did not seek to conduct a sensitivity analysis.

113We planned to follow methods presented in Yepes‐Nunez 2019 to prepare summary of findings tables for the NMA, however because the network was not well‐connected, we primarily based our comparisons on direct evidence from classical pairwise meta‐analyses, except for moderate‐dose atropine. We prepared summary of findings tables for progression of myopia at one and two years, with separate tables for change in spherical equivalent and change in axial length.

114Instead of the planned CINeMA framework for evaluating confidence in the domains (Nikolakopoulou 2020; Salanti 2014) we summarised four levels of confidence for each relative treatment effect, corresponding to the usual GRADE approach: very low, low, moderate, or high (Schünemann 2022). In fact, because most evidence was direct versus control in NMAs, we used NMA estimates only when direct evidence was not available.

115We considered that all studies that met the inclusion criteria for Walline 2020 would potentially meet the inclusion criteria for the current review.

116The searches performed by Walline 2020 to 26 February 2020 identified 41 studies with 74 ongoing studies and 25 studies awaiting classification. Updated electronic searches for the current review identified a further 1473 potentially eligible studies after removal of duplicates. We independently screened these studies for inclusion. We discarded 1290 citations and examined the full texts of the remaining 183 records. In total, we included 64 studies (reported in 225 records) and two studies published as conference abstracts are awaiting classification (for a full description see Characteristics of included studies and Characteristics of studies awaiting classification).

117The economic search was carried out on 4 February 2022 and yielded 80 studies that were screened by AK. No studies met the inclusion criteria.

118A search for systematic reviews of adverse events was carried out on 8 July 2022 and yielded 79 studies. These were screened, and we discuss relevant reviews in Agreements and disagreements with other studies or reviews.

119For a summary of the screening process, see the study flow diagram (Figure 1; Liberati 2009).

120Sixty‐one studies used a parallel‐group design and three studies used a cross‐over design (Anstice 2011; Fujikado 2014; Hasebe 2008). The median sample size was 150 (range 24 to 660). Most participants were recruited from academic clinic settings, hospitals and in a few cases from private optometry or ophthalmology practices. The studies took place in China or other Asian countries (39 studies, 60.9%), North America (13 studies, 20.3%), Europe (7 studies, 10.9%), Australasia (2 studies, 3.1%), Israel (1 study, 1.6%) and Ghana (1 study, 1.6%); one multicentre study recruited participants in both Europe and Asia (1 study, 1.6%).

121Fifty‐seven studies (89%) compared one or more myopia control interventions against a placebo intervention (generally single vision spectacles or contact lenses for optical interventions, and placebo or no treatment for pharmacological interventions). Four studies included a combined intervention group compared with control (Han 2019; MIT Study 2001; Schwartz 1981), and eight studies compared single or combined interventions with each other (ATOM 2 Study 2012; Cui 2021; Guo 2021; Kinoshita 2020; Shih 1999; Swarbrick 2015; Tan 2020; Zhao 2021).

122Twenty‐two (34.4%) of the studies were of 12‐month duration, five studies(7.8%) had a duration of 18 to 20 months, 25 (39.1%) studies were 24 months, 11 (17.2%) up to 36 months and only one reported data over 36 months (Zhu 2021).

123Seven studies were conducted before the year 2000 (Fulk 1996; Houston Study 1987; Jensen 1991; Pärssinen 1989; Schwartz 1981; Shih 1999; Yen 1989). Of the 49 studies that declared a source of funding, 19 (38.8%) were funded by the optical or pharmaceutical industry.

124The review included 64 studies that randomised a total of 11,617 children, aged between 4 and 18 years, with a pooled mean age of 10.35 (range 7.6 to 14.0) years and 48% of participants were male. In the 58 studies that documented the level of myopia for inclusion, all but five studies recruited low to moderate myopes of −6.00 D or less; the other five studies included participants with higher levels of myopia up to −8.75 D (Charm 2013; Garcia‐del Valle 2021; Lyu 2020; Shih 1999; Zhu 2021). Most studies adopted an upper astigmatism limit of 1.00 D or 1.50 D. Three studies specifically recruited myopes with both myopia and near esophoria (Fulk 1996; Fulk 2002; STAMP Study 2012). One study selectively recruited participants with anisomyopia with an interocular difference of 1.00 D or greater (Zhang 2021). Eight studies restricted recruitment to those demonstrating a minimum myopic progression rate of at least 0.50 D in the year prior to enrolment (ATOM 2 Study 2012; Anstice 2011; Cheng 2010; CONTROL Study 2016; Lu 2015; Swarbrick 2015; LAMP Study 2019; Zhu 2021). Participants were sufficiently similar to satisfy the transitivity assumption for the NMA, that is, that there were no systematic differences between the available comparisons other than the treatments being compared.

125Undercorrection versus fully corrected single vision spectacle lenses (SVLs) (3 studies; Adler 2006; Chung 2002; Koomson 2016). These studies, conducted in Israel, China and Ghana, compared the effect of under correcting myopia by either 0.50 D or 0.75 D versus fully corrected SVL. The follow‐up periods were 18 months for Adler 2006 and 24 months for Chung 2002 and Koomson 2016.Multifocal spectacle lenses (MFSLs) versus single vision spectacle lenses (SVLs) (13 studies; Cheng 2010; COMET Study 2003: COMET2 Study 2011; Edwards 2002; Fulk 1996; Fulk 2002; Hasebe 2008; Houston Study 1987; Jensen 1991; MIT Study 2001; Pärssinen 1989; STAMP Study 2012; Yang 2009). These studies were conducted in North America (7 studies), Asia (4 studies) and Europe (2 studies). All studies enroled children aged 8 to 15 years. MFSLs were either bifocal (6 studies) or progressive addition lenses (7 studies) with near additions between +1.00 D and +2.00 D. The study durations were between 18 and 36 months. Eight studies had two arms and five studies had three arms. Hasebe 2008 compared bifocals with two add powers (+1.00 D and +2.00 D) to SVLs. Jensen 1991 randomised children to one of three groups, bifocals, SVLs or timolol maleate eye drops, and Pärssinen 1989 compared a group wearing bifocals (+1.75 D add) to a group wearing SVLs for distance vision only and a reference group wearing SVLs continuously. Peripheral plus spectacle lenses (PPSL) versus single vision spectacle lenses (SVLs) (6 studies; Bao 2021; Hasebe 2014; Han 2018; Lam 2020: Lu 2015: Sankaridurg 2010). Novel spectacle lens designs have been developed that aim to reduce peripheral hyperopic defocus. These lenses, designated PPSLs, were compared to SVLs in Chinese and Japanese myopic children aged 6 to 16 years. Study durations were 1 to 2 years. Sankaridurg 2010 tested three lens designs (designated types I, II and III) that provided different relative peripheral power against SVLs in children aged 6 to 16 years. Hasebe 2014 compared two positively aspherised progressive addition lens designs, with +1.00 D or +2.00 D near add powers and a relative plus power in the upper portion of the lens, to SVLs. Lu 2015 randomised children to receive either PPSLs with up to a +2.50 D near addition or SVLs. Han 2018 conducted a three‐arm study in which children were randomised to PPSLs, SVLs, or orthokeratology lenses. Lam 2020 adapted a design that had previously been used in contact lenses (DISC Study 2011), to develop a spectacle lens with a clear central zone for distance correction and an annular peripheral zone consisting of a multiple array of segments approximately 1 mm in diameter, providing +3.50 D of myopic defocus. The lens, which is termed the ‘Defocus Incorporated Multiple Segments' (DIMS) lens, was tested in a two‐year study involving Chinese children aged 9 to 13 years, who were randomised to wear either DIMS lenses or SVLs. Finally, Bao 2021 tested a lens design based on the same principal that consisted of concentric rings of aspheric lenslets to provide myopic defocus. Children aged 8 to 13 years were randomised in a three‐arm study to receive either a lens with highly aspherical lenslets, a lens with slightly aspherical lenslets, or SVLs. The study reported interim results on myopia progression at one year.

126Undercorrection versus fully corrected single vision spectacle lenses (SVLs) (3 studies; Adler 2006; Chung 2002; Koomson 2016). These studies, conducted in Israel, China and Ghana, compared the effect of under correcting myopia by either 0.50 D or 0.75 D versus fully corrected SVL. The follow‐up periods were 18 months for Adler 2006 and 24 months for Chung 2002 and Koomson 2016.

127Multifocal spectacle lenses (MFSLs) versus single vision spectacle lenses (SVLs) (13 studies; Cheng 2010; COMET Study 2003: COMET2 Study 2011; Edwards 2002; Fulk 1996; Fulk 2002; Hasebe 2008; Houston Study 1987; Jensen 1991; MIT Study 2001; Pärssinen 1989; STAMP Study 2012; Yang 2009). These studies were conducted in North America (7 studies), Asia (4 studies) and Europe (2 studies). All studies enroled children aged 8 to 15 years. MFSLs were either bifocal (6 studies) or progressive addition lenses (7 studies) with near additions between +1.00 D and +2.00 D. The study durations were between 18 and 36 months. Eight studies had two arms and five studies had three arms. Hasebe 2008 compared bifocals with two add powers (+1.00 D and +2.00 D) to SVLs. Jensen 1991 randomised children to one of three groups, bifocals, SVLs or timolol maleate eye drops, and Pärssinen 1989 compared a group wearing bifocals (+1.75 D add) to a group wearing SVLs for distance vision only and a reference group wearing SVLs continuously.

128Peripheral plus spectacle lenses (PPSL) versus single vision spectacle lenses (SVLs) (6 studies; Bao 2021; Hasebe 2014; Han 2018; Lam 2020: Lu 2015: Sankaridurg 2010). Novel spectacle lens designs have been developed that aim to reduce peripheral hyperopic defocus. These lenses, designated PPSLs, were compared to SVLs in Chinese and Japanese myopic children aged 6 to 16 years. Study durations were 1 to 2 years. Sankaridurg 2010 tested three lens designs (designated types I, II and III) that provided different relative peripheral power against SVLs in children aged 6 to 16 years. Hasebe 2014 compared two positively aspherised progressive addition lens designs, with +1.00 D or +2.00 D near add powers and a relative plus power in the upper portion of the lens, to SVLs. Lu 2015 randomised children to receive either PPSLs with up to a +2.50 D near addition or SVLs. Han 2018 conducted a three‐arm study in which children were randomised to PPSLs, SVLs, or orthokeratology lenses. Lam 2020 adapted a design that had previously been used in contact lenses (DISC Study 2011), to develop a spectacle lens with a clear central zone for distance correction and an annular peripheral zone consisting of a multiple array of segments approximately 1 mm in diameter, providing +3.50 D of myopic defocus. The lens, which is termed the ‘Defocus Incorporated Multiple Segments' (DIMS) lens, was tested in a two‐year study involving Chinese children aged 9 to 13 years, who were randomised to wear either DIMS lenses or SVLs. Finally, Bao 2021 tested a lens design based on the same principal that consisted of concentric rings of aspheric lenslets to provide myopic defocus. Children aged 8 to 13 years were randomised in a three‐arm study to receive either a lens with highly aspherical lenslets, a lens with slightly aspherical lenslets, or SVLs. The study reported interim results on myopia progression at one year.

129Multifocal soft contact lenses (MFSCL) versus single vision soft contact lenses (SVSCLs) (9 studies; Anstice 2011; BLINK Study 2020; Chamberlain 2019; CONTROL Study 2016; DISC Study 2011; Fujikado 2014; Garcia‐del Valle 2021: Ruiz‐Pomeda 2018; Sankaridurg 2019). Nine studies investigated the efficacy of a variety of MFSCL designs compared to SVSCL. The MFSCLs incorporated a central zone to provide clear distance vision with relatively more positive peripheral lens power, which either increased gradually towards the periphery (progressive design) or presented as discrete peripheral annular zones (concentric ring design). Three studies followed participants for 12 months, four provided data to 20 to 24 months, and two had a duration of 36 months (BLINK Study 2020; Chamberlain 2019). Seven studies used a parallel‐group design, comparing MFSCLs with SVSCLs, and two studies used a cross‐over design (Anstice 2011; Fujikado 2014). Six studies adopted similar eligibility criteria and randomised children, aged 6 to 18 years with low to moderate myopia up to −6.00 D; Garcia‐del Valle 2021 included myopes to −8.75 D. Anstice 2011 and CONTROL Study 2016) only included children with documented myopia progression of −0.50 D or greater in the previous year, and the CONTROL Study 2016 additionally restricted inclusion to myopic children with near esophoria. Three studies used a similar centre distance dual focus concentric ring design with alternating distance correction zones and peripheral zones providing +2.00 D of defocus (Anstice 2011; Chamberlain 2019; Ruiz‐Pomeda 2018). These studies were conducted in New Zealand (Anstice 2011), Spain (Ruiz‐Pomeda 2018) and at sites in Europe, Asia and Canada (Chamberlain 2019). Garcia‐del Valle 2021 tested a MFSCL with a progressive design (+2.00 D addition) compared to SVSCL in Spanish schoolchildren age 7 to 15 years. Two studies, conducted in the USA (CONTROL Study 2016; BLINK Study 2020), used commercially available MFSCLs. The CONTROL Study 2016 evaluated children aged 8 to 18 years with progressive myopia, randomised to wear either a concentric bifocal soft contact lens or SVSCLs. The near add was selected based on the add power to neutralise the associated esophoria. The ‘Bifocal Lenses in Near‐sighted Kids' (BLINK) study (BLINK Study 2020), tested the efficacy of bifocal soft contact lenses with a central correcting zone for myopia and either a medium add (+1.50 D) or high add (+2.50 D) compared to SVSCLs. Three studies, conducted in China and Japan, used novel custom MFSCL designs (DISC Study 2011; Fujikado 2014; Sankaridurg 2019). The DISC Study 2011 tested the ‘Defocus Incorporated Soft Contact (DISC) lens’, a custom‐made bifocal soft contact lens of concentric ring design with a +2.50 D addition alternating with the normal distance correction. The DISC lens was compared to SVSCL in Chinese school children aged 8 to13 years, who were followed for two years. Fujikado 2014 used a cross‐over study design, in which Japanese children aged 6 to 16 years were randomised to wear a progressive MFSCL with a peripheral power of +0.50 D or SVSCLs in both eyes for 1 year and then were switched to the other type of lens for the second year. Sankaridurg 2019 randomised Chinese children aged 8 to 13 years to one of five groups: two groups wore MFSCLs that imposed peripheral myopic defocus of +1.50 D or +2.50 D with a stepped, relative positive power centrally of up to +1.00 D; and two groups wore extended depth of focus soft lens designs to optimise focus in front of and on the retina and degrade focus behind the retina. The control lens was a SVSCL.Spherical aberration soft contact lenses versus single vision soft contact lenses (SVSCLs) (1 study; Cheng 2016). This study randomised children aged 8 to 11 years to receive soft contact lenses with or without positive spherical aberration. Although the study was conducted in the USA, it enroled mostly Asian children (91%). The study was planned for two years, but was stopped early and reported only one‐year data.Rigid gas‐permeable (RGP) contact lenses versus single vision soft contact lenses (SVSCLs) or single vision spectacle lenses (SVLs) (2 studies; CLAMP Study 2004; Katz 2003). Two studies investigated the impact of RGP lenses on myopia progression compared to SVLs. Katz 2003 randomised Singaporean children aged 6 to 12 years to SVLs or RGP lenses. Myopia progression was evaluated at 1 and 2 years. The Contact Lens and Myopia Progression (CLAMP) Study (CLAMP Study 2004), was conducted in the USA and randomised children to RGP or soft single vision contact lenses. Annual myopia progression was reported based on change in SER and axial length, for the three‐year duration of the study.Orthokeratology lenses versus single vision spectacle lenses (SVLs) or contact lenses (9 studies; Bian 2020; Charm 2013; Han 2018; Jakobsen 2022; Lyu 2020; Ren 2017; ROMIO Study 2012; Tang 2021; Zhang 2021). Eight parallel‐group studies compared overnight orthokeratology contact lenses or SVLs, and in one study SVSCLs (Tang 2021). Participants were followed for 1 to 2 years. Seven studies enroled children with low to moderate degrees of myopia (up to −6.00 D), and two studies selectively recruited children with myopia 5.00 D or greater (Charm 2013; Lyu 2020). Zhang 2021 included participants with anisomyopia with a difference in myopia between eyes of 1.00 D or greater. Eight of the nine studies were conducted in China and one in Denmark (Jakobsen 2022). Axial length was the primary outcome in all studies. The 'Retardation of Myopia in Orthokeratology' (ROMIO) Study (ROMIO Study 2012), randomised 102 Chinese children aged 6 to 12 years to overnight orthokeratology lenses or SVLs, who were followed for two years. Charm 2013 randomised 52 highly myopic children (aged 8 to 11 years), with a SER of at least −5.75 D to partial reduction overnight orthokeratology lenses and daily SVLs for residual myopia, or a control group who were fully corrected with SVLs. Axial length was measured at six‐monthly intervals for two years. Lyu 2020 similarly investigated partial reduction orthokeratology lenses in participants with myopia up to −8.75 D. They randomised 102 children aged 8 to12 years into three groups: (1) orthokeratology lenses with a target reduction of 6.00 D; (2) orthokeratology lenses with a 4.00 D target reduction; or (3) SVLs. Axial length was measured at baseline and at 12 months. Jakobsen 2022 randomised 60 Danish children aged 6 to 12 years to orthokeratology lenses or SVLs, and followed them for 18 months. Four studies compared orthokeratology lenses to SVLs in Chinese children aged 8 to 15 years with myopia (Bian 2020; Han 2018; Ren 2017; Tang 2021). Ren 2017 also included a group that was treated with 0.01% atropine, and Han 2018 included a group wearing PPSLs.

130Multifocal soft contact lenses (MFSCL) versus single vision soft contact lenses (SVSCLs) (9 studies; Anstice 2011; BLINK Study 2020; Chamberlain 2019; CONTROL Study 2016; DISC Study 2011; Fujikado 2014; Garcia‐del Valle 2021: Ruiz‐Pomeda 2018; Sankaridurg 2019). Nine studies investigated the efficacy of a variety of MFSCL designs compared to SVSCL. The MFSCLs incorporated a central zone to provide clear distance vision with relatively more positive peripheral lens power, which either increased gradually towards the periphery (progressive design) or presented as discrete peripheral annular zones (concentric ring design). Three studies followed participants for 12 months, four provided data to 20 to 24 months, and two had a duration of 36 months (BLINK Study 2020; Chamberlain 2019). Seven studies used a parallel‐group design, comparing MFSCLs with SVSCLs, and two studies used a cross‐over design (Anstice 2011; Fujikado 2014). Six studies adopted similar eligibility criteria and randomised children, aged 6 to 18 years with low to moderate myopia up to −6.00 D; Garcia‐del Valle 2021 included myopes to −8.75 D. Anstice 2011 and CONTROL Study 2016) only included children with documented myopia progression of −0.50 D or greater in the previous year, and the CONTROL Study 2016 additionally restricted inclusion to myopic children with near esophoria. Three studies used a similar centre distance dual focus concentric ring design with alternating distance correction zones and peripheral zones providing +2.00 D of defocus (Anstice 2011; Chamberlain 2019; Ruiz‐Pomeda 2018). These studies were conducted in New Zealand (Anstice 2011), Spain (Ruiz‐Pomeda 2018) and at sites in Europe, Asia and Canada (Chamberlain 2019). Garcia‐del Valle 2021 tested a MFSCL with a progressive design (+2.00 D addition) compared to SVSCL in Spanish schoolchildren age 7 to 15 years. Two studies, conducted in the USA (CONTROL Study 2016; BLINK Study 2020), used commercially available MFSCLs. The CONTROL Study 2016 evaluated children aged 8 to 18 years with progressive myopia, randomised to wear either a concentric bifocal soft contact lens or SVSCLs. The near add was selected based on the add power to neutralise the associated esophoria. The ‘Bifocal Lenses in Near‐sighted Kids' (BLINK) study (BLINK Study 2020), tested the efficacy of bifocal soft contact lenses with a central correcting zone for myopia and either a medium add (+1.50 D) or high add (+2.50 D) compared to SVSCLs. Three studies, conducted in China and Japan, used novel custom MFSCL designs (DISC Study 2011; Fujikado 2014; Sankaridurg 2019). The DISC Study 2011 tested the ‘Defocus Incorporated Soft Contact (DISC) lens’, a custom‐made bifocal soft contact lens of concentric ring design with a +2.50 D addition alternating with the normal distance correction. The DISC lens was compared to SVSCL in Chinese school children aged 8 to13 years, who were followed for two years. Fujikado 2014 used a cross‐over study design, in which Japanese children aged 6 to 16 years were randomised to wear a progressive MFSCL with a peripheral power of +0.50 D or SVSCLs in both eyes for 1 year and then were switched to the other type of lens for the second year. Sankaridurg 2019 randomised Chinese children aged 8 to 13 years to one of five groups: two groups wore MFSCLs that imposed peripheral myopic defocus of +1.50 D or +2.50 D with a stepped, relative positive power centrally of up to +1.00 D; and two groups wore extended depth of focus soft lens designs to optimise focus in front of and on the retina and degrade focus behind the retina. The control lens was a SVSCL.

131Spherical aberration soft contact lenses versus single vision soft contact lenses (SVSCLs) (1 study; Cheng 2016). This study randomised children aged 8 to 11 years to receive soft contact lenses with or without positive spherical aberration. Although the study was conducted in the USA, it enroled mostly Asian children (91%). The study was planned for two years, but was stopped early and reported only one‐year data.

132Rigid gas‐permeable (RGP) contact lenses versus single vision soft contact lenses (SVSCLs) or single vision spectacle lenses (SVLs) (2 studies; CLAMP Study 2004; Katz 2003). Two studies investigated the impact of RGP lenses on myopia progression compared to SVLs. Katz 2003 randomised Singaporean children aged 6 to 12 years to SVLs or RGP lenses. Myopia progression was evaluated at 1 and 2 years. The Contact Lens and Myopia Progression (CLAMP) Study (CLAMP Study 2004), was conducted in the USA and randomised children to RGP or soft single vision contact lenses. Annual myopia progression was reported based on change in SER and axial length, for the three‐year duration of the study.

133Orthokeratology lenses versus single vision spectacle lenses (SVLs) or contact lenses (9 studies; Bian 2020; Charm 2013; Han 2018; Jakobsen 2022; Lyu 2020; Ren 2017; ROMIO Study 2012; Tang 2021; Zhang 2021). Eight parallel‐group studies compared overnight orthokeratology contact lenses or SVLs, and in one study SVSCLs (Tang 2021). Participants were followed for 1 to 2 years. Seven studies enroled children with low to moderate degrees of myopia (up to −6.00 D), and two studies selectively recruited children with myopia 5.00 D or greater (Charm 2013; Lyu 2020). Zhang 2021 included participants with anisomyopia with a difference in myopia between eyes of 1.00 D or greater. Eight of the nine studies were conducted in China and one in Denmark (Jakobsen 2022). Axial length was the primary outcome in all studies. The 'Retardation of Myopia in Orthokeratology' (ROMIO) Study (ROMIO Study 2012), randomised 102 Chinese children aged 6 to 12 years to overnight orthokeratology lenses or SVLs, who were followed for two years. Charm 2013 randomised 52 highly myopic children (aged 8 to 11 years), with a SER of at least −5.75 D to partial reduction overnight orthokeratology lenses and daily SVLs for residual myopia, or a control group who were fully corrected with SVLs. Axial length was measured at six‐monthly intervals for two years. Lyu 2020 similarly investigated partial reduction orthokeratology lenses in participants with myopia up to −8.75 D. They randomised 102 children aged 8 to12 years into three groups: (1) orthokeratology lenses with a target reduction of 6.00 D; (2) orthokeratology lenses with a 4.00 D target reduction; or (3) SVLs. Axial length was measured at baseline and at 12 months. Jakobsen 2022 randomised 60 Danish children aged 6 to 12 years to orthokeratology lenses or SVLs, and followed them for 18 months. Four studies compared orthokeratology lenses to SVLs in Chinese children aged 8 to 15 years with myopia (Bian 2020; Han 2018; Ren 2017; Tang 2021). Ren 2017 also included a group that was treated with 0.01% atropine, and Han 2018 included a group wearing PPSLs.

134Atropine eye drops versus placebo or untreated control (11 studies; ATOM Study 2006; Han 2019; Hieda 2021; LAMP Study 2019; Moriche‐Carretero 2021; Ren 2017; Wang 2017; Wei 2020; Yen 1989; Yi 2015; Zhu 2021). Twelve parallel‐group studies compared atropine to either placebo or an untreated control. These studies enroled children with low to high myopia (up to −8.00 D), aged from 4 to 15 years. Eligibility criteria in LAMP Study 2019 and Zhu 2021 additionally included a documented level of myopic progression in the past year. All studies were conducted in Asia, except for Moriche‐Carretero 2021, which was conducted in Spain. Participants were followed for periods ranging from one to four years. High‐dose atropine (≥ 0.5%): four studies compared 1% atropine to placebo or untreated control. The 'Atropine in the Treatment of Myopia' (ATOM) Study (ATOM Study 2006), was conducted in Singapore and involved 400 children aged 6 to 12 years, who were randomised to receive either 1% atropine or placebo to one eye and were followed for two years. Yi 2015 randomised 140 Chinese children, aged 7 to 12 years with low myopia (−0.50 to −2.00 D) to 1% atropine or placebo eyedrops nightly for 12 months. Zhu 2021 compared 1% atropine to placebo using a novel dosing regime in 660 Chinese children. The study was divided into three phases. In Phase 1, the treatment group received 1% atropine once per month for 24 months, this was reduced to once every two months for 12 months (Phase 2), followed by no drops for 12 months in Phase 3. The placebo group received the same dosing regime. Wang 2017 compared daily 0.5% atropine with placebo in 126 Chinese children with low myopia, who were followed for one year. Two, three‐armed studies included a 1% atropine arm. Han 2019 randomised 150 Chinese children aged 6 to 12 years in a 1:2:2 ratio to either an untreated control group, 1% atropine or a combination of 1% atropine with 0.5% raceanisodamine (a non‐selective muscarinic antagonist, used as an ingredient of traditional Chinese medicines). Yen 1989 included a group receiving 1% cyclopentolate.Low‐dose atropine (< 0.1%): five studies tested lower doses of atropine, ranging from 0.01% to 0.05%. The 'Low‐concentration Atropine for Myopia Progression' (LAMP) Study (LAMP Study 2019), randomised 438 Chinese children aged 4 to 12 years with myopia of at least −1.00 D to four groups (in a 1:1:1:1 ratio): low‐concentration atropine eye drops at 0.05%, 0.025%, or 0.01% concentration or placebo. Participants were followed for one year. Four studies tested the efficacy of 0.01% atropine eyedrops versus placebo or an untreated control in participants aged between 5 and 15 years with low or moderate myopia, who were followed for one or two years (Hieda 2021; Moriche‐Carretero 2021; Ren 2017; Wei 2020). Participants in Hieda 2021 included 171 Japanese children, Wei 2020 included 220 Chinese children and Moriche‐Carretero 2021 randomised 339 Spanish children. In Ren 2017, 150 Chinese children were randomised (1:1:1 ratio) to 0.01% atropine, orthokeratology or single vision spectacle lenses. Pirenzepine eye drops versus placebo (2 studies; PIR‐205 Study 2004; Tan 2005). These two studies compared 2% pirenzepine gel, a selective M1 muscarinic receptor antagonist, to placebo. PIR‐205 Study 2004 was a two‐year, multicentre study conducted in the USA that randomised 174 myopic children aged 8 to 12 years in a 2:1 ratio to twice‐daily pirenzepine gel or placebo. Tan 2005 was conducted in centres in Singapore, Thailand and China, and randomised 353 children aged 6 to 13 years to one of three arms: (1) 2% pirenzepine gel twice daily; (2) placebo once daily and 2% pirenzepine gel once daily; or (3) placebo twice daily.

135Atropine eye drops versus placebo or untreated control (11 studies; ATOM Study 2006; Han 2019; Hieda 2021; LAMP Study 2019; Moriche‐Carretero 2021; Ren 2017; Wang 2017; Wei 2020; Yen 1989; Yi 2015; Zhu 2021). Twelve parallel‐group studies compared atropine to either placebo or an untreated control. These studies enroled children with low to high myopia (up to −8.00 D), aged from 4 to 15 years. Eligibility criteria in LAMP Study 2019 and Zhu 2021 additionally included a documented level of myopic progression in the past year. All studies were conducted in Asia, except for Moriche‐Carretero 2021, which was conducted in Spain. Participants were followed for periods ranging from one to four years.

136High‐dose atropine (≥ 0.5%): four studies compared 1% atropine to placebo or untreated control. The 'Atropine in the Treatment of Myopia' (ATOM) Study (ATOM Study 2006), was conducted in Singapore and involved 400 children aged 6 to 12 years, who were randomised to receive either 1% atropine or placebo to one eye and were followed for two years. Yi 2015 randomised 140 Chinese children, aged 7 to 12 years with low myopia (−0.50 to −2.00 D) to 1% atropine or placebo eyedrops nightly for 12 months. Zhu 2021 compared 1% atropine to placebo using a novel dosing regime in 660 Chinese children. The study was divided into three phases. In Phase 1, the treatment group received 1% atropine once per month for 24 months, this was reduced to once every two months for 12 months (Phase 2), followed by no drops for 12 months in Phase 3. The placebo group received the same dosing regime. Wang 2017 compared daily 0.5% atropine with placebo in 126 Chinese children with low myopia, who were followed for one year. Two, three‐armed studies included a 1% atropine arm. Han 2019 randomised 150 Chinese children aged 6 to 12 years in a 1:2:2 ratio to either an untreated control group, 1% atropine or a combination of 1% atropine with 0.5% raceanisodamine (a non‐selective muscarinic antagonist, used as an ingredient of traditional Chinese medicines). Yen 1989 included a group receiving 1% cyclopentolate.

137Low‐dose atropine (< 0.1%): five studies tested lower doses of atropine, ranging from 0.01% to 0.05%. The 'Low‐concentration Atropine for Myopia Progression' (LAMP) Study (LAMP Study 2019), randomised 438 Chinese children aged 4 to 12 years with myopia of at least −1.00 D to four groups (in a 1:1:1:1 ratio): low‐concentration atropine eye drops at 0.05%, 0.025%, or 0.01% concentration or placebo. Participants were followed for one year. Four studies tested the efficacy of 0.01% atropine eyedrops versus placebo or an untreated control in participants aged between 5 and 15 years with low or moderate myopia, who were followed for one or two years (Hieda 2021; Moriche‐Carretero 2021; Ren 2017; Wei 2020). Participants in Hieda 2021 included 171 Japanese children, Wei 2020 included 220 Chinese children and Moriche‐Carretero 2021 randomised 339 Spanish children. In Ren 2017, 150 Chinese children were randomised (1:1:1 ratio) to 0.01% atropine, orthokeratology or single vision spectacle lenses.

138Pirenzepine eye drops versus placebo (2 studies; PIR‐205 Study 2004; Tan 2005). These two studies compared 2% pirenzepine gel, a selective M1 muscarinic receptor antagonist, to placebo. PIR‐205 Study 2004 was a two‐year, multicentre study conducted in the USA that randomised 174 myopic children aged 8 to 12 years in a 2:1 ratio to twice‐daily pirenzepine gel or placebo. Tan 2005 was conducted in centres in Singapore, Thailand and China, and randomised 353 children aged 6 to 13 years to one of three arms: (1) 2% pirenzepine gel twice daily; (2) placebo once daily and 2% pirenzepine gel once daily; or (3) placebo twice daily.

139Tropicamide and multifocal spectacles (MFSLs) (1 study; Schwartz 1981). This study was conducted in the USA and randomised 26 monozygous twin pairs aged 7 to 14 years to either a combination of MFSLs combined with 1% tropicamide or SVLs.Atropine and multifocal spectacles (MFSLs) versus placebo (2 studies; MIT Study 2001; Yen 1989). The 'Myopia Intervention Trial' (MIT) (MIT Study 2001), was conducted in Taiwan and evaluated SVLs, progressive addition lenses and progressive addition lenses combined with 0.5% atropine eyedrops. Yen 1989 randomly divided 247 Taiwanese children aged 6 to 14 years into three groups. Group 1 received 1% atropine and bifocal spectacles; group 2 received 1% cyclopentolate; and group 3 received saline eye drops. All groups were followed for 12 months.

140Tropicamide and multifocal spectacles (MFSLs) (1 study; Schwartz 1981). This study was conducted in the USA and randomised 26 monozygous twin pairs aged 7 to 14 years to either a combination of MFSLs combined with 1% tropicamide or SVLs.

141Atropine and multifocal spectacles (MFSLs) versus placebo (2 studies; MIT Study 2001; Yen 1989). The 'Myopia Intervention Trial' (MIT) (MIT Study 2001), was conducted in Taiwan and evaluated SVLs, progressive addition lenses and progressive addition lenses combined with 0.5% atropine eyedrops. Yen 1989 randomly divided 247 Taiwanese children aged 6 to 14 years into three groups. Group 1 received 1% atropine and bifocal spectacles; group 2 received 1% cyclopentolate; and group 3 received saline eye drops. All groups were followed for 12 months.

142Timolol eyedrops versus single vision spectacle lenses (SVLs) (1 study; Jensen 1991). One arm of Jensen 1991 investigated topical 0.25% timolol maleate, a non‐selective beta antagonist. Timolol eyedrops were given twice a day for two years and compared to MFSL or SVL control.Systemic 7‐methylxanthine versus placebo (1 study; Trier 2008). This study investigated the effectiveness of systemic 7‐methylxanthine, an adenosine receptor antagonist, in 83 Danish children aged 8 to 13 years. Participants were randomised to once daily 7‐methylxanthine or a placebo tablet.

143Timolol eyedrops versus single vision spectacle lenses (SVLs) (1 study; Jensen 1991). One arm of Jensen 1991 investigated topical 0.25% timolol maleate, a non‐selective beta antagonist. Timolol eyedrops were given twice a day for two years and compared to MFSL or SVL control.

144Systemic 7‐methylxanthine versus placebo (1 study; Trier 2008). This study investigated the effectiveness of systemic 7‐methylxanthine, an adenosine receptor antagonist, in 83 Danish children aged 8 to 13 years. Participants were randomised to once daily 7‐methylxanthine or a placebo tablet.

145Comparison of atropine doses (2 studies; ATOM 2 Study 2012; Cui 2021). The ATOM 2 Study 2012 was conducted in Singapore and compared the efficacy and safety of three doses of topical atropine: 0.5%, 0.1%, and 0.01% in 400 children of Chinese ethnicity. Cui 2021 evaluated the safety and efficacy of 0.02% and 0.01% atropine in 400 myopic Chinese children, who were randomly allocated to atropine 0.02% (138 children) or 0.01% (142 children). The study also included a non‐randomised control group wearing single vision spectacle lenses (120 children). All participants were followed for two years.Atropine and multifocal spectacle lenses (MFSLs) versus tropicamide (1 study; Shih 1999). This study evaluated low‐concentration atropine in Taiwanese children aged 6 to 13 years. It randomly allocated 200 children to one of three atropine groups (0.5%, 0.25% or 0.1%) or 1% tropicamide as a control. The 0.5% atropine group were advised to wear MFSLs and the 0.25% atropine group were advised to wear slightly undercorrected SVLs.Combined orthokeratology plus atropine versus orthokeratology alone (3 studies; Kinoshita 2020; Tan 2020; Zhao 2021). Kinoshita 2020 randomly allocated 80 Japanese children with low to moderate myopia, aged 8 to 12 years, to receive either a combined orthokeratology and 0.01% atropine, or orthokeratology monotherapy. Tan 2020 randomised 72 Chinese children aged 6 to 11 years to receive combined orthokeratology/0.01% atropine compared to monotherapy. Similarly, Zhao 2021 included as a separate parallel‐group comparison, combined 0.01% atropine/orthokeratology versus orthokeratology alone. Zhao 2021 randomised 40 children who had been wearing orthokeratology lenses for three months to orthokeratology and 0.01% atropine or orthokeratology only.Comparison of orthokeratology designs (1 study; Guo 2021). This study compared two designs of orthokeratology lenses with different back optic zone diameters. It randomly assigned 82 Chinese children aged 6 to 11 years to wear orthokeratology lenses with either a 6 mm or 5 mm back optic zone diameter and followed them for two years.Orthokeratology versus rigid gas‐permeable contact lenses (RGP) (1 study; Swarbrick 2015). This study conducted a randomised, contralateral‐eye cross‐over study over a one‐year period. Although the study was conducted in Australia, all 26 children were of East Asian ethnicity. Participants were fitted with an orthokeratology lens in one eye, chosen at random, and a conventional RGP lens worn daily in the contralateral eye. Children wore the lenses for six months. After a two‐week washout period, the lenses were reversed and lens wear was continued for a further six months.Orthokeratology versus atropine (2 studies; Ren 2017; Zhao 2021). Both studies compared 0.01% atropine to orthokeratology.

146Comparison of atropine doses (2 studies; ATOM 2 Study 2012; Cui 2021). The ATOM 2 Study 2012 was conducted in Singapore and compared the efficacy and safety of three doses of topical atropine: 0.5%, 0.1%, and 0.01% in 400 children of Chinese ethnicity. Cui 2021 evaluated the safety and efficacy of 0.02% and 0.01% atropine in 400 myopic Chinese children, who were randomly allocated to atropine 0.02% (138 children) or 0.01% (142 children). The study also included a non‐randomised control group wearing single vision spectacle lenses (120 children). All participants were followed for two years.

147Atropine and multifocal spectacle lenses (MFSLs) versus tropicamide (1 study; Shih 1999). This study evaluated low‐concentration atropine in Taiwanese children aged 6 to 13 years. It randomly allocated 200 children to one of three atropine groups (0.5%, 0.25% or 0.1%) or 1% tropicamide as a control. The 0.5% atropine group were advised to wear MFSLs and the 0.25% atropine group were advised to wear slightly undercorrected SVLs.

148Combined orthokeratology plus atropine versus orthokeratology alone (3 studies; Kinoshita 2020; Tan 2020; Zhao 2021). Kinoshita 2020 randomly allocated 80 Japanese children with low to moderate myopia, aged 8 to 12 years, to receive either a combined orthokeratology and 0.01% atropine, or orthokeratology monotherapy. Tan 2020 randomised 72 Chinese children aged 6 to 11 years to receive combined orthokeratology/0.01% atropine compared to monotherapy. Similarly, Zhao 2021 included as a separate parallel‐group comparison, combined 0.01% atropine/orthokeratology versus orthokeratology alone. Zhao 2021 randomised 40 children who had been wearing orthokeratology lenses for three months to orthokeratology and 0.01% atropine or orthokeratology only.

149Comparison of orthokeratology designs (1 study; Guo 2021). This study compared two designs of orthokeratology lenses with different back optic zone diameters. It randomly assigned 82 Chinese children aged 6 to 11 years to wear orthokeratology lenses with either a 6 mm or 5 mm back optic zone diameter and followed them for two years.

150Orthokeratology versus rigid gas‐permeable contact lenses (RGP) (1 study; Swarbrick 2015). This study conducted a randomised, contralateral‐eye cross‐over study over a one‐year period. Although the study was conducted in Australia, all 26 children were of East Asian ethnicity. Participants were fitted with an orthokeratology lens in one eye, chosen at random, and a conventional RGP lens worn daily in the contralateral eye. Children wore the lenses for six months. After a two‐week washout period, the lenses were reversed and lens wear was continued for a further six months.

151Orthokeratology versus atropine (2 studies; Ren 2017; Zhao 2021). Both studies compared 0.01% atropine to orthokeratology.

152We excluded studies that reported the impact of environmental interventions (e.g. elevated light levels in classrooms, increased outdoor time or regulated near working distances) mostly because the populations included participants both with and without myopia, or the primary outcome was incident myopia. One ongoing study, The 'Shanghai Time Outside to Reduce Myopia' (STORM) Study (NCT02980445), is a two‐year, school‐based, prospective, cluster‐randomised study that is investigating the effect of two 'doses' of increased outdoor time (40 and 80 minutes over normal time outdoors). Outcomes include the incidence of myopia in non‐myopic children, and the progression of myopia in myopic children.

153All the included studies evaluated progression of myopia, either by measuring the mean change in refractive error, defined as spherical equivalent refraction (SER), mean change in axial length or both. Nine studies reported SER only (Adler 2006; Han 2018; Hasebe 2008; Jensen 1991, Pärssinen 1989; Schwartz 1981; Shih 1999; Yang 2009; Yen 1989), six studies investigating the efficacy of orthokeratology lenses reported axial length only (Bian 2020; Jakobsen 2022; Kinoshita 2020; ROMIO Study 2012; Swarbrick 2015; Zhang 2021), and the remaining 49 studies provided data on both SER and axial length.

154Six studies investigated change in refractive error and axial length following cessation of treatment (commonly referred to as 'rebound'). In the STAMP Study 2012, children were randomly assigned to MFSL or SVLs for one year and all children wore SVLs in the second year. Cheng 2016 invited participants who had been randomised to soft contact lenses with positive spherical aberration or single vision soft lenses for one year to participate in a withdrawal phase where all children wore single vision contact lenses. To assess a potential rebound effect, Ruiz‐Pomeda 2018 invited children to participate in an additional year of follow‐up. Children were divided into three groups: a group in which children from the original study group continued wearing MFSCL; a group in which children discontinued MFSCL wear; and an SVL group, in which children from the original control group continued wearing SVLs. Three studies investigated the impact of terminating atropine treatment. In the ATOM Study 2006, children received 0.5%, 0.1% or 0.01% atropine for 12 months, after which treatment was terminated and the children were followed for a further 24 months. In the ATOM 2 Study 2012, children who received topical atropine 0.5%, 0.1% or 0.01% for 24 months entered Phase 2, the washout phase, where atropine was discontinued and SER and axial length assessed at 26, 32 and 36 months. In Zhu 2021, the frequency of 1% atropine eyedrop instillation was reduced from year 1 from once per month to once every two months in years 2 and 3, and withdrawn completely in year 4.

155Twenty‐six studies provided data on safety outcomes in terms of the occurrence of adverse events. These included five studies reporting on adverse events with spectacle lens interventions, 11 reporting on contact lens interventions (including orthokeratology) and 10 studies reporting on various pharmacological interventions.

156Only one study (LAMP Study 2019), which investigated the efficacy of low‐dose topical atropine, measured vision‐related quality of life. At the 12‐month follow‐up visit, the Chinese version of the 25‐Item National Eye Institute Visual Function Questionnaire was administered to all participants to determine the impact of different treatment groups on vision‐related quality of life.

157Twenty‐one studies provided data on treatment adherence. Ten studies reported on compliance with spectacle lens wear including undercorrection with SVL, bifocal or progressive addition lenses. Compliance was typically based on parent or child, or both, self‐reporting wearing time (hours per day) and overall compliance, expressed as a % of participants in each arm. Similarly, six studies using contact lens interventions, reported on wearing time and percentage compliance between the intervention and control lenses. Four studies of pharmacological interventions provided data on self‐reported compliance with the study medication.

158We identified 120 ongoing studies (see Characteristics of ongoing studies). These studies compare contact lenses or spectacle lenses (MFSCL, MFSL or orthokeratology) or pharmacological interventions to a control or other myopia control interventions. The majority of the ongoing studies investigate the efficacy of various doses of atropine used alone or in combination with other interventions.

159We classified two studies published as conference abstracts as awaiting classification (see Characteristics of studies awaiting classification; Wang 2005; Viswanath 2022).

160We excluded a total of 137 studies. We excluded the Cambridge Anti‐Myopia Study 2013, which had been included in Walline 2020. The main reasons for exclusion were that the study was not randomised, population not eligible, intervention not eligible or ineligible outcome (see Characteristics of excluded studies for further details).

161We assessed risk of bias using the RoB 2 (Higgins 2022a). A graphical representation of risk of bias for each comparison for the critical outcome 'progression of myopia' can be seen in Analysis 1.1; Analysis 1.2; Analysis 2.1; Analysis 2.2; Analysis 2.3; Analysis 3.1; Analysis 3.2; Analysis 4.1; Analysis 4.2; Analysis 5.1; Analysis 5.2; Analysis 6.1; Analysis 7.1; Analysis 7.3; Analysis 7.2; Analysis 7.4; Analysis 7.5; Analysis 8.1; Analysis 8.2; Analysis 9.1.

162The overall risk of bias across studies in each analysis reporting this outcome ranged from some concerns to high risk of bias depending on the particular intervention used.

163For the comparison 'undercorrection versus full correction spectacles', we assumed an overall high risk of bias, since two out of three studies had a high risk of bias in one domain, due to either an inappropriate method used to measure the outcome or no reasons given for missing data Table 188; Table 189.

164Of the 10 studies that reported on progression of myopia in the analysis 'multifocal spectacle lenses versus single vision spectacles', we judged one study to be at high risk of bias (Cheng 2010). We judged the remainder as 'some concerns', usually due to insufficient information concerning allocation concealment and lack of an a priori statistical analysis plan. We therefore assumed an overall bias of 'some concerns' (see Table 190; Table 191). For 'peripheral plus spectacle lenses versus single vision spectacles', three of the six studies reporting this outcome had some concerns, with three studies judged to be at high risk of bias (Han 2018 Hasebe 2014; Lu 2015); see Table 193; Table 194).

165Seven of the eight studies reporting progression of myopia in the comparison 'multifocal soft contact lenses versus single‐vision soft contact lenses' were judged as 'some concerns', primarily due to failure to describe the method of allocation concealment and no information on the predetermined analysis plan. We gave an overall judgement of 'some concerns' for this outcome (see Table 195; Table 196).

166Only two studies evaluated progression of myopia following RGP wear. We judged one study at high risk (Katz 2003). We therefore gave an overall judgement of 'high risk' for this outcome (see Table 199; Table 200).

167We judged four of the seven studies reporting change in axial length from baseline after wearing orthokeratology lenses as 'some concerns' (Analysis 1.2). We judged three studies at high risk of bias (Lyu 2020; Ren 2017; ROMIO Study 2012). We gave an overall judgement of 'some concerns' for this outcome (see Table 201).

168For progression of myopia in the comparison 'anti‐muscarinics versus control', we assumed an overall risk of bias of 'some concerns' for studies using different doses of atropine, as we judged the majority of the studies reporting the outcome as 'some concerns'. However, we judged both of the studies reporting on 2% pirenzepine to be at high risk of bias (PIR‐205 Study 2004; Tan 2005), which gave an overall judgement of 'high risk' for this outcome (see Table 202; Table 204; Table 203; Table 205).

169For the outcome 'change in refractive error and axial length following cessation of treatment', three of the four included studies had some concerns and one was at high risk of bias (Zhu 2021; see Analysis 2.3; Analysis 4.3; Analysis 4.4; Analysis 7.5; Analysis 7.6).

170Detailed risk of bias assessments are available at: osf.io/ms83h/

171See: Table 1; Table 2; Table 3; Table 4

172See summary of findings tables for overall treatment effects for any myopia control intervention on progression of myopia compared to placebo. The certainty of the evidence is also provided and if appropriate, the reasons for downgrading (Table 1; Table 2; Table 3; Table 4).

173We performed standard pairwise and network meta‐analyses for optical and pharmacological interventions compared to a control group (consisting of either standard SVLs or contact lenses or a placebo) for the critical outcome 'progression of myopia'. We also compared myopia control interventions to each other. In total, we included 52 studies, analysing 8152 participants, in either the standard or network meta‐analysis.

174Twelve studies did not contribute directly to the quantitative synthesis (Cheng 2016; Fulk 1996; Houston Study 1987; Han 2018; Han 2019; Hasebe 2008; Lu 2015; Schwartz 1981; Shih 1999; Swarbrick 2015; Wang 2017; Yen 1989). For these studies, we reported study‐specific results, as appropriate.

175Quantitative synthesis was not possible for the outcomes 'risk of adverse events', 'quality of life', or 'treatment adherence' due to insufficient available data. For these outcomes we provided a descriptive summary of the range and distribution of the observed effects, and summarised the study‐level findings in structured tables.

176Direct treatment estimates from pairwise meta‐analysis for optical and pharmacological interventions versus control/placebo are reported in Analysis 1.1; Analysis 1.2; Analysis 2.1; Analysis 2.2; Analysis 3.1; Analysis 3.2; Analysis 4.1; Analysis 4.2, Analysis 5.1 Analysis 6.1; Analysis 7.1; Analysis 7.3; Analysis 7.2; Analysis 7.4; Analysis 8.1; Analysis 8.2). We assessed progression of myopia as mean change in refractive error (defined as SER) from baseline for each year of follow‐up or change in axial length from baseline, or both. To facilitate interpretation of the forest plots, negative mean differences (MDs) for changes in refractive error represent faster progression of myopia in the intervention group compared to progression in the control group. Thus, point estimates to the left of the null on the forest plots favour the control group. For the measurement of changes in axial length, negative MDs for changes in axial length represent faster axial elongation in the control group compared to the intervention group, and therefore estimates to the left of the null favour the intervention group.

177Optical interventions: spectaclesSpectacle interventions designed to reduce accommodative demand and lag during near work by undercorrection, or the use of MFSLs have been common in practice for many years. Three studies, involving 292 participants, compared spectacles that were undercorrected by −0.50 D to −0.75 D to fully corrected SVLs (Analysis 1.1; Analysis 1.2). There was no evidence that undercorrection slowed myopic progression, based either on change in refractive error (MD at 1 year −0.15 D (95% CI −0.29 to 0.00); MD at 2 years 0.02 D (95% CI −0.05 to 0.09)) or change in axial length (MD at 1 year 0.05 mm (95% CI −0.01 to 0.11); MD at 2 years −0.01 mm (95% CI −0.06 to 0.03)).Thirteen studies compared bifocal or progressive addition lenses to SVLs. We included 10 studies with 1612 participants in a quantitative synthesis (Analysis 2.1; Analysis 2.2). Eight studies provided data up to two years and four studies followed participants for up to three years (Cheng 2010; COMET Study 2003; COMET2 Study 2011; Pärssinen 1989). There was a small reduction in myopia progression at both one‐ and two‐year follow‐up (change in refractive error at 1 year, MD 0.14 D, 95% CI 0.08 to 0.21; and at 2 years, MD 0.19 D, 95% CI 0.08 to 0.30). The three‐year results showed considerable heterogeneity (I2 = 86%), however after removing two studies judged to be at high risk of bias (Cheng 2010; Pärssinen 1989), heterogeneity was substantially reduced (I2 = 0%). The pooled three‐year MD after removing these studies was 0.21 D (95% CI 0.08 to 0.34). The three studies not included in the meta‐analysis reported inconsistent results (Fulk 1996; Hasebe 2008; Houston Study 1987). Hasebe 2008 reported significantly less myopia progression in children wearing progressive addition lenses over the first 18 months of a cross‐over study compared to children wearing SVLs, but no difference in the second 18‐month period. In another 18‐month study (Fulk 1996), children wearing bifocals progressed at a rate of −0.39 D per year compared to −0.57 D per year in the SVL group, however these differences were not significant (P = 0.26). The Houston Study 1987 found no significant difference between groups wearing bifocals (+1.00 D and +2.00 D add) and SVLs. We included four studies with 896 participants that reported change in axial length with progressive addition lenses in a quantitative analysis (Analysis 2.2). There was a small reduction in axial elongation in progressive addition lens wearers for each year of follow‐up (1‐year MD −0.06 mm, 95% CI −0.09 to −0.04; 2‐year MD −0.07 mm, 95% CI −0.12 to −0.03; 3‐year MD −0.12 mm, 95% CI −0.18 to −0.07).The rationale for prescribing peripheral plus spectacle lenses (PPSL) is to reduce hyperopic defocus in the peripheral retina. Six studies compared PPSL to SVLs (Bao 2021; Han 2018; Hasebe 2014; Lam 2020; Lu 2015; Sankaridurg 2010). Changes in refractive error from baseline showed considerable heterogeneity (I2 = 89% to 91%; Analysis 3.1). Mean differences at one year ranged from 0.02 D to 0.97 D. Only two studies followed children for two years (Hasebe 2014; Lam 2020). These studies showed contrasting results. Hasebe 2014 found no difference in myopia progression with positively aspherised progressive addition lenses (MD 0.12 D, 95% CI −0.06 to 0.31). In contrast, Lam 2020 found a significant reduction in progression using the Defocus Incorporated Multiple Segments (DIMS) spectacle lens (MD 0.55 D, 95% CI 0.38 to 0.72). Four studies provided data on changes in axial length from baseline, showing similarly high heterogeneity (Analysis 3.2). The combination of all three novel lenses tested by Sankaridurg 2010 were not significantly different from SVLs at one year (MD −0.02 mm, 95% CI −0.08 to 0.04), contrasting with designs used by Lam 2020 and Bao 2021, which showed less axial elongation. Only two studies provided two‐year data for axial length. Hasebe 2014 reported no significant difference (MD −0.07mm, 95% CI −0.20 to 0.07), whereas Lam 2020 showed that the reduced axial elongation demonstrated at one year, continued into the second year (−0.32 mm, 95% CI −0.39 to −0.25).Optical interventions: contact lensesStudies tested a variety of contact lens design, including multifocal soft contact lenses (MFSCL), positive spherical aberration contact lenses, rigid gas‐permeable (RGP) and orthokeratology lenses. Conceptually, MFSCL use a progressive or concentric ring design to create myopic defocus and reduce myopia progression. Nine studies, compared MFSCL to single vision soft contact lenses (SVSCL). We excluded Fujikado 2014 from the quantitative analysis since it used a lens design that was distinct from the other lenses in the comparison. Consequently, we included eight studies with a total of 1135 participants in a quantitative synthesis (Analysis 4.1; Analysis 4.2). Five studies provided data on change in refractive error and axial length up to two years and two studies followed children for up to three years (BLINK Study 2020; Chamberlain 2019). Over the three‐year reporting period, there was a progressive reduction in myopia progression with MFSCL compared to SVCL, although there was a suggestion that the change in refractive error reduced over time (1‐year MD 0.26 D, 95% CI 0.17 to 0.35; 2‐year MD 0.30 D, 95% CI 0.19 to 0.41; 3‐year MD 0.47 D, 95% CI 0.13 to 0.82). A significant reduction in axial elongation was seen across all three years (1‐year MD −0.11 mm, 95% CI −0.13 to −0.09; 2‐year MD −0.15 mm, 95% CI −0.19 to −0.12; 3‐year MD −0.22 mm, 95% CI −0.34 to −0.10).Cheng 2016 investigated the effect of soft contact lenses with positive spherical aberration. The mean reported change in refractive error in this study was 0.137 D (95% CI −0.007 to 0.281) amongst 52 children in the spherical aberration group compared with 57 children in the single vision contact lens group at one‐year follow‐up. In terms of axial elongation, children in the positive spherical aberration group showed 0.143 mm (95% CI −0.188 to −0.098) less elongation compared with the control at one year.Two studies investigated the use of rigid gas‐permeable contact lenses (RGPs) in slowing the progression of myopia compared to single vision contact lenses in one study (CLAMP Study 2004), and SVLs in the other (Katz 2003) (Analysis 5.1; Analysis 5.2). The CLAMP Study 2004 followed up participants for three years and Katz 2003 followed up participants for two years. We did not pool data on change in refractive error due to considerable heterogeneity (I2 > 90%). The two studies reported contrasting results, with the CLAMP Study 2004 showing significantly less myopia progression over three years with the RGPs compared to participants wearing single vision contact lenses (1‐year MD 0.40 D, 95% CI 0.19 to 0.61; 2‐year MD 0.54 D, 95% CI 0.27 to 0.81; 3‐year MD 0.63 D, 95% CI 0.30 to 0.96). By contrast, Katz 2003 observed no difference in myopia progression over two years (1‐year MD −0.02, 95% CI −0.14 to 0.10; 2‐year MD −0.05 D, 95% CI −0.25 to 0.15). Neither study was able to show any effect on axial elongation over three years (pooled estimate 1‐year MD 0.02 mm, 95% CI −0.05 to 0.10; 2‐year MD 0.03 mm, 95% CI −0.05 to 0.12; 3‐year MD 0.05 mm, 95% CI −0.12 to 0.22).Eight studies, involving 787 participants, compared orthokeratology lenses to SVLs or SVSCLs and provided data up to two years (Analysis 6.1). Overnight wear of orthokeratology lenses flattens the central cornea and temporarily reduces refractive error. Since it is not possible to assess the true progression of refractive error without ceasing lens wear for a period of time to allow the cornea to return to its pre‐treatment state, the included studies presented change in axial length as the primary efficacy outcome. A significant reduction in axial elongation was seen across both years (1‐year MD −0.19 mm, 95% CI −0.23 to −0.15; 2‐year MD −0.28 mm, 95% CI −0.38 to −0.19).Pharmacological interventions: antimuscarinicsAtropine: 11 studies compared topical atropine to control (placebo, no treatment or SVLs). These were grouped according to dosing regime into high dose (≥ 0.1%) or low dose (< 0.1%). The majority of studies testing atropine reported data at one year, with only four studies (ATOM Study 2006; Hieda 2021; Moriche‐Carretero 2021; Zhu 2021) reporting at two years. Several studies tested atropine as a co‐intervention or in head‐to‐head dose comparisons and these are described below.High‐dose atropine: three studies (1072 participants) compared high‐dose atropine (1%) to control (ATOM Study 2006; Yi 2015; Zhu 2021). At one year, effect sizes for change in refractive error ranged from MD 0.79 D to 1.1 7 D in favour of high‐dose atropine Analysis 7.1. A reduction in axial elongation was also seen with 1% atropine at one year (MD −0.31 to −0.35 mm; Analysis 7.2). Studies reporting at two years similarly showed that high‐dose atropine had greater efficacy. The ATOM Study 2006 showed a change in refractive error of MD 0.92 D (95% CI 0.75 to 1.09); and change in axial length of MD −0.40 mm (95% CI −0.48 to −0.32). Zhu 2021 showed a change in refractive error of MD 1.41 D (95% CI 1.30 to 1.52) and change in axial length of MD −0.54 mm (95% CI −0.57 to −0.51) (Analysis 7.3; Analysis 7.4). Two studies, not included in the meta‐analysis, also investigated 1% atropine versus placebo and reported significantly less myopia progression in children in the atropine group at the end of the follow‐up period, however the data were not presented in a form that could be included in the meta‐analysis (Han 2019; Wang 2017). We also excluded Yen 1989 and MIT Study 2001 since the atropine groups also wore MFSL.Low‐dose atropine: five studies (1143 participants) compared lower atropine doses (0.01% to 0.05%) to control (Hieda 2021; LAMP Study 2019; Moriche‐Carretero 2021; Ren 2017; Wei 2020). Results for these comparisons for each year of follow‐up showed considerable heterogeneity (I2 > 90%), however all effects were in the same direction, and we included subgroup summary effect estimates in the forest plots as the best estimate of the intervention effect. At one year, effect sizes for change in refractive error were in the range 0.08 D to 0.80 D, and change in axial length ranged from −0.04 mm to −0.35 mm in favour of low‐dose atropine (Analysis 7.1; Analysis 7.2). Studies reporting at two years similarly showed a greater efficacy for low‐dose atropine (Hieda 2021 change in refractive error MD 0.22 D, 95% CI 0.09 to 0.35; change in axial length −0.14 mm, 95% CI −0.20 to −0.08; Moriche‐Carretero 2021, change in refractive error MD 0.25 D, 95% CI 0.17 to 0.33; change in axial length −0.17 mm, 95% CI −0.22 to −0.12) in favour of atropine (Analysis 7.3; Analysis 7.4).Pirenzepine: two studies investigated 2% pirenzepine eyedrops (PIR‐205 Study 2004; Tan 2005; see Analysis 7.1; Analysis 7.2 Analysis 7.3). At one‐year follow‐up, average myopia progression was less for participants treated with pirenzepine compared to placebo (PIR‐205 Study 2004 MD 0.27 D, 95% CI 0.11 to 0.43; Tan 2005 MD 0.47 D, 95% CI 0.16 to 0.78). The difference in progression between groups continued at two years (PIR‐205 Study 2004 MD 0.41 D, 95% CI 0.13 to 0.69). Data for axial length were only available at one year. Tan 2005 found a slowing of axial elongation (MD −0.13 mm, 95% CI −0.14 to −0.12), whereas the PIR‐205 Study 2004 found no significant difference in axial length (MD −0.04 mm, 95% CI −0.15 to 0.07).Other pharmacological interventionsOne arm of Jensen 1991 investigated the non‐selective beta‐antagonist timolol maleate compared to a SVL control. The differences in myopia progression were not significant at one year (MD −0.05 D, 95% CI −0.21 to 0.11) or at two years (MD −0.04 D, 95% CI −0.30 to 0.22). This study did not measure axial length.Trier 2008 compared systemic 7‐methylxanthine, an oral adenosine receptor antagonist, versus a placebo tablet for one year. At one‐year follow‐up, the differences in myopia progression were not significant (change in refractive error MD 0.07 D, 95% CI −0.09 to 0.24; change in axial length MD −0.03 mm, 95% CI −0.10 to 0.03; Analysis 8.1; Analysis 8.2).

178Optical interventions: spectacles

179Spectacle interventions designed to reduce accommodative demand and lag during near work by undercorrection, or the use of MFSLs have been common in practice for many years. Three studies, involving 292 participants, compared spectacles that were undercorrected by −0.50 D to −0.75 D to fully corrected SVLs (Analysis 1.1; Analysis 1.2). There was no evidence that undercorrection slowed myopic progression, based either on change in refractive error (MD at 1 year −0.15 D (95% CI −0.29 to 0.00); MD at 2 years 0.02 D (95% CI −0.05 to 0.09)) or change in axial length (MD at 1 year 0.05 mm (95% CI −0.01 to 0.11); MD at 2 years −0.01 mm (95% CI −0.06 to 0.03)).

180Thirteen studies compared bifocal or progressive addition lenses to SVLs. We included 10 studies with 1612 participants in a quantitative synthesis (Analysis 2.1; Analysis 2.2). Eight studies provided data up to two years and four studies followed participants for up to three years (Cheng 2010; COMET Study 2003; COMET2 Study 2011; Pärssinen 1989). There was a small reduction in myopia progression at both one‐ and two‐year follow‐up (change in refractive error at 1 year, MD 0.14 D, 95% CI 0.08 to 0.21; and at 2 years, MD 0.19 D, 95% CI 0.08 to 0.30). The three‐year results showed considerable heterogeneity (I2 = 86%), however after removing two studies judged to be at high risk of bias (Cheng 2010; Pärssinen 1989), heterogeneity was substantially reduced (I2 = 0%). The pooled three‐year MD after removing these studies was 0.21 D (95% CI 0.08 to 0.34). The three studies not included in the meta‐analysis reported inconsistent results (Fulk 1996; Hasebe 2008; Houston Study 1987). Hasebe 2008 reported significantly less myopia progression in children wearing progressive addition lenses over the first 18 months of a cross‐over study compared to children wearing SVLs, but no difference in the second 18‐month period. In another 18‐month study (Fulk 1996), children wearing bifocals progressed at a rate of −0.39 D per year compared to −0.57 D per year in the SVL group, however these differences were not significant (P = 0.26). The Houston Study 1987 found no significant difference between groups wearing bifocals (+1.00 D and +2.00 D add) and SVLs. We included four studies with 896 participants that reported change in axial length with progressive addition lenses in a quantitative analysis (Analysis 2.2). There was a small reduction in axial elongation in progressive addition lens wearers for each year of follow‐up (1‐year MD −0.06 mm, 95% CI −0.09 to −0.04; 2‐year MD −0.07 mm, 95% CI −0.12 to −0.03; 3‐year MD −0.12 mm, 95% CI −0.18 to −0.07).

181The rationale for prescribing peripheral plus spectacle lenses (PPSL) is to reduce hyperopic defocus in the peripheral retina. Six studies compared PPSL to SVLs (Bao 2021; Han 2018; Hasebe 2014; Lam 2020; Lu 2015; Sankaridurg 2010). Changes in refractive error from baseline showed considerable heterogeneity (I2 = 89% to 91%; Analysis 3.1). Mean differences at one year ranged from 0.02 D to 0.97 D. Only two studies followed children for two years (Hasebe 2014; Lam 2020). These studies showed contrasting results. Hasebe 2014 found no difference in myopia progression with positively aspherised progressive addition lenses (MD 0.12 D, 95% CI −0.06 to 0.31). In contrast, Lam 2020 found a significant reduction in progression using the Defocus Incorporated Multiple Segments (DIMS) spectacle lens (MD 0.55 D, 95% CI 0.38 to 0.72). Four studies provided data on changes in axial length from baseline, showing similarly high heterogeneity (Analysis 3.2). The combination of all three novel lenses tested by Sankaridurg 2010 were not significantly different from SVLs at one year (MD −0.02 mm, 95% CI −0.08 to 0.04), contrasting with designs used by Lam 2020 and Bao 2021, which showed less axial elongation. Only two studies provided two‐year data for axial length. Hasebe 2014 reported no significant difference (MD −0.07mm, 95% CI −0.20 to 0.07), whereas Lam 2020 showed that the reduced axial elongation demonstrated at one year, continued into the second year (−0.32 mm, 95% CI −0.39 to −0.25).

182Optical interventions: contact lenses

183Studies tested a variety of contact lens design, including multifocal soft contact lenses (MFSCL), positive spherical aberration contact lenses, rigid gas‐permeable (RGP) and orthokeratology lenses. Conceptually, MFSCL use a progressive or concentric ring design to create myopic defocus and reduce myopia progression. Nine studies, compared MFSCL to single vision soft contact lenses (SVSCL). We excluded Fujikado 2014 from the quantitative analysis since it used a lens design that was distinct from the other lenses in the comparison. Consequently, we included eight studies with a total of 1135 participants in a quantitative synthesis (Analysis 4.1; Analysis 4.2). Five studies provided data on change in refractive error and axial length up to two years and two studies followed children for up to three years (BLINK Study 2020; Chamberlain 2019). Over the three‐year reporting period, there was a progressive reduction in myopia progression with MFSCL compared to SVCL, although there was a suggestion that the change in refractive error reduced over time (1‐year MD 0.26 D, 95% CI 0.17 to 0.35; 2‐year MD 0.30 D, 95% CI 0.19 to 0.41; 3‐year MD 0.47 D, 95% CI 0.13 to 0.82). A significant reduction in axial elongation was seen across all three years (1‐year MD −0.11 mm, 95% CI −0.13 to −0.09; 2‐year MD −0.15 mm, 95% CI −0.19 to −0.12; 3‐year MD −0.22 mm, 95% CI −0.34 to −0.10).

184Cheng 2016 investigated the effect of soft contact lenses with positive spherical aberration. The mean reported change in refractive error in this study was 0.137 D (95% CI −0.007 to 0.281) amongst 52 children in the spherical aberration group compared with 57 children in the single vision contact lens group at one‐year follow‐up. In terms of axial elongation, children in the positive spherical aberration group showed 0.143 mm (95% CI −0.188 to −0.098) less elongation compared with the control at one year.

185Two studies investigated the use of rigid gas‐permeable contact lenses (RGPs) in slowing the progression of myopia compared to single vision contact lenses in one study (CLAMP Study 2004), and SVLs in the other (Katz 2003) (Analysis 5.1; Analysis 5.2). The CLAMP Study 2004 followed up participants for three years and Katz 2003 followed up participants for two years. We did not pool data on change in refractive error due to considerable heterogeneity (I2 > 90%). The two studies reported contrasting results, with the CLAMP Study 2004 showing significantly less myopia progression over three years with the RGPs compared to participants wearing single vision contact lenses (1‐year MD 0.40 D, 95% CI 0.19 to 0.61; 2‐year MD 0.54 D, 95% CI 0.27 to 0.81; 3‐year MD 0.63 D, 95% CI 0.30 to 0.96). By contrast, Katz 2003 observed no difference in myopia progression over two years (1‐year MD −0.02, 95% CI −0.14 to 0.10; 2‐year MD −0.05 D, 95% CI −0.25 to 0.15). Neither study was able to show any effect on axial elongation over three years (pooled estimate 1‐year MD 0.02 mm, 95% CI −0.05 to 0.10; 2‐year MD 0.03 mm, 95% CI −0.05 to 0.12; 3‐year MD 0.05 mm, 95% CI −0.12 to 0.22).

186Eight studies, involving 787 participants, compared orthokeratology lenses to SVLs or SVSCLs and provided data up to two years (Analysis 6.1). Overnight wear of orthokeratology lenses flattens the central cornea and temporarily reduces refractive error. Since it is not possible to assess the true progression of refractive error without ceasing lens wear for a period of time to allow the cornea to return to its pre‐treatment state, the included studies presented change in axial length as the primary efficacy outcome. A significant reduction in axial elongation was seen across both years (1‐year MD −0.19 mm, 95% CI −0.23 to −0.15; 2‐year MD −0.28 mm, 95% CI −0.38 to −0.19).

187Pharmacological interventions: antimuscarinics

188Atropine: 11 studies compared topical atropine to control (placebo, no treatment or SVLs). These were grouped according to dosing regime into high dose (≥ 0.1%) or low dose (< 0.1%). The majority of studies testing atropine reported data at one year, with only four studies (ATOM Study 2006; Hieda 2021; Moriche‐Carretero 2021; Zhu 2021) reporting at two years.

189Several studies tested atropine as a co‐intervention or in head‐to‐head dose comparisons and these are described below.

190High‐dose atropine: three studies (1072 participants) compared high‐dose atropine (1%) to control (ATOM Study 2006; Yi 2015; Zhu 2021). At one year, effect sizes for change in refractive error ranged from MD 0.79 D to 1.1 7 D in favour of high‐dose atropine Analysis 7.1. A reduction in axial elongation was also seen with 1% atropine at one year (MD −0.31 to −0.35 mm; Analysis 7.2). Studies reporting at two years similarly showed that high‐dose atropine had greater efficacy. The ATOM Study 2006 showed a change in refractive error of MD 0.92 D (95% CI 0.75 to 1.09); and change in axial length of MD −0.40 mm (95% CI −0.48 to −0.32). Zhu 2021 showed a change in refractive error of MD 1.41 D (95% CI 1.30 to 1.52) and change in axial length of MD −0.54 mm (95% CI −0.57 to −0.51) (Analysis 7.3; Analysis 7.4). Two studies, not included in the meta‐analysis, also investigated 1% atropine versus placebo and reported significantly less myopia progression in children in the atropine group at the end of the follow‐up period, however the data were not presented in a form that could be included in the meta‐analysis (Han 2019; Wang 2017). We also excluded Yen 1989 and MIT Study 2001 since the atropine groups also wore MFSL.

191Low‐dose atropine: five studies (1143 participants) compared lower atropine doses (0.01% to 0.05%) to control (Hieda 2021; LAMP Study 2019; Moriche‐Carretero 2021; Ren 2017; Wei 2020). Results for these comparisons for each year of follow‐up showed considerable heterogeneity (I2 > 90%), however all effects were in the same direction, and we included subgroup summary effect estimates in the forest plots as the best estimate of the intervention effect. At one year, effect sizes for change in refractive error were in the range 0.08 D to 0.80 D, and change in axial length ranged from −0.04 mm to −0.35 mm in favour of low‐dose atropine (Analysis 7.1; Analysis 7.2). Studies reporting at two years similarly showed a greater efficacy for low‐dose atropine (Hieda 2021 change in refractive error MD 0.22 D, 95% CI 0.09 to 0.35; change in axial length −0.14 mm, 95% CI −0.20 to −0.08; Moriche‐Carretero 2021, change in refractive error MD 0.25 D, 95% CI 0.17 to 0.33; change in axial length −0.17 mm, 95% CI −0.22 to −0.12) in favour of atropine (Analysis 7.3; Analysis 7.4).

192Pirenzepine: two studies investigated 2% pirenzepine eyedrops (PIR‐205 Study 2004; Tan 2005; see Analysis 7.1; Analysis 7.2 Analysis 7.3). At one‐year follow‐up, average myopia progression was less for participants treated with pirenzepine compared to placebo (PIR‐205 Study 2004 MD 0.27 D, 95% CI 0.11 to 0.43; Tan 2005 MD 0.47 D, 95% CI 0.16 to 0.78). The difference in progression between groups continued at two years (PIR‐205 Study 2004 MD 0.41 D, 95% CI 0.13 to 0.69). Data for axial length were only available at one year. Tan 2005 found a slowing of axial elongation (MD −0.13 mm, 95% CI −0.14 to −0.12), whereas the PIR‐205 Study 2004 found no significant difference in axial length (MD −0.04 mm, 95% CI −0.15 to 0.07).

193Other pharmacological interventions

194One arm of Jensen 1991 investigated the non‐selective beta‐antagonist timolol maleate compared to a SVL control. The differences in myopia progression were not significant at one year (MD −0.05 D, 95% CI −0.21 to 0.11) or at two years (MD −0.04 D, 95% CI −0.30 to 0.22). This study did not measure axial length.

195Trier 2008 compared systemic 7‐methylxanthine, an oral adenosine receptor antagonist, versus a placebo tablet for one year. At one‐year follow‐up, the differences in myopia progression were not significant (change in refractive error MD 0.07 D, 95% CI −0.09 to 0.24; change in axial length MD −0.03 mm, 95% CI −0.10 to 0.03; Analysis 8.1; Analysis 8.2).

196We conducted NMAs for change in SER and axial length at 12 and 24 months. See Figure 2 for the network maps for each comparison, and Table 5 presenting the number of study arms (participants) for each intervention. Direct comparisons between interventions were limited to different doses of atropine, meaning that only indirect comparisons were possible. League‐tables presenting all indirect and mixed comparisons for SER and axial length at 12 and 24 months can be seen at https://osf.io/ms83h/. Figure 3 presents forest plots of NMA comparisons with control.

197Change in SER at 1‐yearThe NMA included 30 studies (4694 participants) with two connected closed loops comparing different doses of atropine or control. There was no overall inconsistency (p=0.185). The only two closed loops were partly overlapping and showed no inconsistency.The overall NMA between‐study SD was large (0.19 D), which made most NMA estimates versus control as, or more, imprecise than the corresponding direct evidence. For this reason, Table 1 presents direct and indirect evidence for all comparisons versus control, including the certainty of evidence assessment, except for moderate‐dose atropine versus control, for which only indirect evidence was available.Change in axial length at 1‐yearThe NMA included 31 studies (4864 participants) with two connected closed loops comparing different doses of atropine or control. There was no overall inconsistency (P = 0.236). The only two closed loops were partly overlapping and showed no inconsistency.The overall NMA between‐study SD was large (0.048 mm), which made most NMA estimates versus control as, or more, imprecise than the corresponding direct evidence. Table 3 presents direct and indirect evidence for all comparisons versus control, except for moderate‐dose atropine versus control, for which only indirect evidence was available.Change in SER and change in axial length at 2‐yearsThe NMAs of the change in SER and change in axial length at two years included 24 studies (4485 participants) and 21 studies (4010 participants), respectively. Table 2 and Table 4 present the evidence from direct and indirect comparisons, except for moderate‐dose atropine versus control, for which only indirect evidence was available.SUCRAs and mixed comparisons in NMAsAll indirect comparisons are presented for illustrative purposes as league‐tables at https://osf.io/ms83h/, where differences not including nil are marked in grey.

198Change in SER at 1‐year

199The NMA included 30 studies (4694 participants) with two connected closed loops comparing different doses of atropine or control. There was no overall inconsistency (p=0.185). The only two closed loops were partly overlapping and showed no inconsistency.

200The overall NMA between‐study SD was large (0.19 D), which made most NMA estimates versus control as, or more, imprecise than the corresponding direct evidence. For this reason, Table 1 presents direct and indirect evidence for all comparisons versus control, including the certainty of evidence assessment, except for moderate‐dose atropine versus control, for which only indirect evidence was available.

201Change in axial length at 1‐year

202The NMA included 31 studies (4864 participants) with two connected closed loops comparing different doses of atropine or control. There was no overall inconsistency (P = 0.236). The only two closed loops were partly overlapping and showed no inconsistency.

203The overall NMA between‐study SD was large (0.048 mm), which made most NMA estimates versus control as, or more, imprecise than the corresponding direct evidence. Table 3 presents direct and indirect evidence for all comparisons versus control, except for moderate‐dose atropine versus control, for which only indirect evidence was available.

204Change in SER and change in axial length at 2‐years

205The NMAs of the change in SER and change in axial length at two years included 24 studies (4485 participants) and 21 studies (4010 participants), respectively. Table 2 and Table 4 present the evidence from direct and indirect comparisons, except for moderate‐dose atropine versus control, for which only indirect evidence was available.

206SUCRAs and mixed comparisons in NMAs

207All indirect comparisons are presented for illustrative purposes as league‐tables at https://osf.io/ms83h/, where differences not including nil are marked in grey.

208Table 6 presents all SUCRA values for NMAs of change in SER and axial length at one and two years, where the three highest SUCRAs are highlighted in bold print. As can be seen, high‐dose atropine and moderate‐dose atropine were amongst the three best SUCRAs for all outcomes. Low‐dose atropine and PPSLs were third for SER at one and two years respectively, while orthokeratology was amongst the best three for axial length at one and two years.

209Three studies compared antimuscarinics combined with MFSLs to SVLs (Schwartz 1981; MIT Study 2001; Yen 1989). Schwartz 1981 randomised monozygous twin pairs to receive either bifocal spectacles combined with 1% tropicamide or SVLs. The paper did not present any numerical results, but the study authors stated that control twins showed more progression of myopia than their co‐twins who received tropicamide and bifocals, but the difference was not statistically significant.

210The Myopia Intervention Trial (MIT) (MIT Study 2001), evaluated progressive addition spectacle lenses combined with 0.5% atropine compared to placebo eyedrops plus SVLs. At the end of the 18‐month follow‐up period, participants in the atropine plus progressive addition spectacle lenses group showed significantly less myopia progression (MD 0.98 D, 95% CI 0.76 to 1.20) and significantly less axial elongation (MD −0.47 mm, 95% CI −0.47 to −0.27) than participants in the placebo plus SVL group.

211Yen 1989 randomly divided children into three groups. Group 1 received 1% atropine and bifocal spectacles; group 2 received 1% cyclopentolate, and group 3 received placebo eyedrops plus SVLs. At one year there was less myopia progression in the atropine plus bifocal group compared to the control group (MD 0.70 D, 95% CI 0.43 to 0.97).

212Three studies compared different doses of topical atropine to each other ( ATOM 2 Study 2012; Cui 2021; Shih 1999). The ATOM 2 Study 2012 compared the efficacy of 0.5%, 0.1%, and 0.01% atropine in children of Chinese ethnicity. The mean change in refractive error at two years was −0.30 D (95% CI −0.40 to −0.20); −0.38 D (95% CI −0.48 to 0.29) and −0.49 D (95% CI −0.64 to −0.35) in the atropine 0.5%, 0.1%, and 0.01% groups, respectively. Pairwise differences were statistically significant for the comparison between 0.01% and 0.5% atropine (P = 0.02). Changes in axial length were 0.27 mm (95% CI 0.21 to 0.33); 0.28 mm (95% CI 0.24 to 0.33) and 0.41 mm (95% CI 0.36 to 0.46) for the high, moderate and low doses.

213Cui 2021 evaluated the efficacy of 0.02% and 0.01% atropine in Chinese children. The mean changes in refractive error at two years were −0.80 D (95% CI −0.90 to −0.70) for the 0.02% concentration and −0.93 D (95% CI −1.04 to −0.82) for the 0.01% concentration. The corresponding changes in axial length were 0.62 mm (95% CI 0.56 to 0.68) and 0.72 mm (95% CI 0.66 to 0.78).

214Shih 1999 compared three doses of atropine (0.1%, 0.25% and 0.5%) versus 1% tropicamide control. Participants in the 0.5% group wore bifocal spectacles, those in the 0.25% group were provided with slightly undercorrected SVLs and the 0.1% group wore fully corrected SVLs. Myopia progression at the end of the two‐year follow‐up period was significantly slowed for each atropine group compared with tropicamide, with the 0.5% atropine dose showing the least progression compared with the tropicamide group, MD 1.95 D (95% CI 1.60 to 2.30) for 0.1% atropine; MD 1.98 D (95% CI 1.68 to 2.28) for 0.25% atropine; and MD 2.42 D (95% CI 2.16 to 2.68) for 0.5% atropine.

215Three studies evaluated the combination of low‐dose atropine (0.01%) combined with orthokeratology, compared to orthokeratology alone (Kinoshita 2020; Tan 2020; Zhao 2021). These studies were conducted in Japan and China and followed participants for up to two years. The primary outcome was change in axial length (Analysis 9.1). We found a reduction in axial elongation for the combination therapy group compared to monotherapy at each year of follow‐up (1‐year MD −0.13 mm, 95% CI −0.16 to −0.09; 2‐year MD −0.11 mm, 95% CI −0.21 to −0.01).

216Two studies included treatment arms that compared low‐dose atropine (0.01%) to orthokeratology at one year. There was no significant difference in axial length between treatments (Ren 2017 MD 0.03 mm, 95% CI −0.17 to 0.03; Zhao 2021 MD 0.05 mm, 95% CI−0.02 to 0.12).

217Guo 2021 compared two designs of orthokeratology lenses with 6 mm or 5 mm back optic zone diameters. The rationale was that a smaller back optic zone diameter would increase myopia control efficacy by inducing a steeper distribution of the relative corneal refractive power profile within the pupillary diameter and further increase higher order aberrations. Axial elongation was lower in the 5 mm group (MD 0.04 mm, 95% CI −0.005 to 0.08) than the 6 mm group (MD 0.17 mm, 95% CI 0.13 to 0.21).

218Swarbrick 2015 conducted a randomised, contralateral‐eye cross‐over study comparing a regular RGP lens in one eye and an orthokeratology lens in the other, conducted over a one‐year period. Lenses were worn for six months and then crossed over after a two‐week washout period for a further six months. This study did not report data eligible for analysis.

219Six studies investigated changes in refractive error and axial length following cessation of the myopia control intervention (ATOM Study 2006; ATOM 2 Study 2012; Cheng 2016; Ruiz‐Pomeda 2018; STAMP Study 2012; Zhu 2021). These studies compared the rate of myopia progression in the intervention group after switching to the control intervention to progression in the original control group.

220In the STAMP Study 2012, children wore MFSL or SVLs for one year and all children wore SVLs in the second year. At the end of year 2 there was no difference in myopia progression between groups (MD 0.00 D, 95% CI −0.17 to 0.17; Analysis 2.3). This study did not measure progression of axial length.

221In Ruiz‐Pomeda 2018, children who had worn MFSCL or SVSCL for two years were invited to participate in an additional one‐year follow‐up study to investigate rebound. One group discontinued MFSCL wear and switched to SVLs and progression was then compared to the original control group, who continued wearing SVLs. After one year there was no significant difference in progression of refractive error (MD 0.09 D, 95% CI −0.16 to 0.34) or axial length (MD 0.01, 95% CI −0.05 to 0.07; Analysis 4.4).

222Cheng 2016 invited participants who had worn either novel soft contact lenses with positive spherical aberration or conventional single vision soft lenses to participate in a 12‐month withdrawal study, where all children wore single vision lenses. The study authors reported that they found no evidence of a rebound effect at one year.

223Three studies investigated the impact of terminating atropine treatment. The ATOM Study 2006 discontinued atropine treatment after children had received 1% atropine for two years. Children were followed for a further year. At the end of this period, the study compared myopia progression to the placebo‐treated group. Progression of refractive error in the original 1% atropine‐treated group was significantly greater than the control group at the end of the second year (MD −0.76 D, 95% CI −0.90 to −0.62; Analysis 7.5). The study authors reported axial length data as change from baseline over the entire three‐year duration of the study and therefore these data were not suitable for evaluating rebound.

224Zhu 2021 used a novel dosing regime. Participants received 1% atropine eye drops once per month for 24 months, then every other month for 12 months followed by no drops for 12 months. Progression at the end of the one‐year withdrawal period was evaluated and compared to the placebo group. One year after terminating treatment the atropine group still showed a slowing in the progression of refractive error (MD 0.34 D, 95% CI 0.26 to 0.42; Analysis 7.5) and a reduction of axial elongation (MD −0.21 mm, 95% CI −0.23 to −0.19; Analysis 7.6) compared to the placebo group.

225The ATOM 2 Study 2012 was a dose comparison study, with participants randomised to receive 0.5%, 0.1% or 0.01% atropine for 24 months followed by a 12‐month withdrawal phase. During the washout period, myopic progression was greater in participants treated with 0.5% atropine (MD −0.87 D, 95% CI −0.96 to −0.78) compared to 0.1% (MD −0.68 D, 95% CI −0.76 to −0.61) and 0.01% atropine (MD −0.28 D, 95% CI −0.36 to −0.20; P < 0.001). Axial elongation was also greater in the 0.5% group (MD 0.35 mm, 95% CI 0.32 to 0.38) compared to the 0.1% (MD 0.33 mm, 95% CI 0.30 to 0.36) and 0.01% (MD 0.19 mm, 95% CI 0.16 to 0.22) groups.

226The risk of adverse events was generally poorly reported. We extracted data on the frequency of adverse events from 26 studies (see Table 7; Table 8; Table 9), comprising five studies reporting on the effects of spectacle interventions (Adler 2006; Bao 2021; COMET2 Study 2011; Hasebe 2008; Sankaridurg 2010), 11 on contact lens interventions, including orthokeratology (BLINK Study 2020; Chamberlain 2019; Cheng 2016; CLAMP Study 2004; Garcia‐del Valle 2021; Guo 2021; Jakobsen 2022; Kinoshita 2020; Lyu 2020; Ruiz‐Pomeda 2018; Tan 2020) and 10 using pharmacological interventions (ATOM 2 Study 2012; Cui 2021; Hieda 2021; LAMP Study 2019; PIR‐205 Study 2004; Shih 1999; Tan 2005; Wei 2020; Yen 1989; Zhu 2021). Multiple adverse events occurring in the same study participant, including events affecting both eyes, were counted as independent events.

227Studies used a variety of methods to record adverse events. Symptoms were usually elicited using questionnaires, telephone interviews or were self‐reported at follow‐up appointments by parents or children, or both. Objective clinical signs were usually based on clinical examination at each follow‐up visit, however in some studies, participants were advised to return to the clinic for unscheduled visits should adverse events arise.

228Data on adverse events were available from 446 participants wearing spectacle lens interventions (undercorrection, MFSLs and PPSLs) and 302 SVL‐wearing controls. Study duration was one to three years. MFSLs and PPLSs were usually well tolerated following a short adaptation period and the reported adverse events were generally mild. There were 55 events in the active arm and 41 in the control arm. Dizziness and blurred vision were the most commonly reported adverse events with similar rates in SVL controls (dizziness: active arm 13/446, controls 15/302; blurred vision: active arm 31/446, controls 18/302) see Table 7. Overall there were three withdrawals due to adverse events in studies using MFSLs.

229Eleven studies provided safety data on 1068 participants receiving contact lens interventions, which included various soft contact lens designs (including MFSCLs and SVSCLs), RGP and orthokeratology. Study duration was one to three years. The control arm in orthokeratology studies was typically SVLs. Two studies compared orthokeratology monotherapy to combined orthokeratology and low‐dose atropine. Safety outcomes were monitored by clinical examination of the anterior segment of the eye using the slit‐lamp biomicroscope at follow‐up appointments. Many studies graded clinical signs using standard grading scales, which used either artist‐rendered or photographic images to grade corneal and conjunctival signs on a 0 to 4 scale from normal to severe. Grade 3 and 4 are regarded as clinically significant and usually require a clinical action. For the most part, studies using MFSCLs and SVSCLSs reported adverse events separately for each arm, however the largest study (BLINK Study 2020), reported safety data for all arms combined (294 children). The most commonly reported adverse events in studies involving soft contact lenses were corneal infiltrative events (17/664 wearers), conjunctival papillary reaction (20/664), and corneal staining (12/664), see Table 8. The number of events were similar for test and control lenses. These events were generally not serious, with only one grade 3 event, and one participant in the BLINK Study 2020 was reported as a 'probable microbial keratitis'. There were four reported adverse effect‐related withdrawals in these studies (incidence: approx 0.6%).

230Adverse events in orthokeratology studies were more common: corneal infiltrates (7/254 wearers), corneal staining (36/254), with four cases of corneal staining graded 3 or higher. There were 12 withdrawals due to adverse events from these studies (incidence approx. 5%).

231Safety data were available for eight studies using various doses of atropine. The three most common adverse events were photophobia or glare, blurred vision (particularly for near vision) and hypersensitivity reactions see Table 9. Atropine studies used high (≥ 0.5%; 564 children), moderate (0.1% to < 0.5%; 251 children), and low (< 0.1%; 829 children) atropine doses, with study durations ranging from 12 to 48 months. Adverse events were generally dose dependent with a greater likelihood of adverse events with higher atropine doses (high‐dose 437 events in 564 children; moderate‐dose 150 events in 251 children; low‐dose 138 events in 829 children). There were higher numbers of withdrawals due to adverse events in studies using high‐dose atropine (7% over 1 year in ATOM Study 2006 and 21% over 2 years in Zhu 2021) compared to 2% or fewer in studies using lower atropine doses (Hieda 2021; Moriche‐Carretero 2021: Wei 2020).

232Evaluating the rates of photophobia and difficulties with near vision was confounded by the use of photochromic spectacle lenses or sunglasses to mitigate photophobia, and multifocal lenses for near vision problems in some studies, which may have reduced reporting of symptoms.

233PIR‐205 Study 2004 and Tan 2005 (259 children in the active treatment arms) documented ocular and systemic adverse events. The three systemic adverse events most frequently reported were headache, common cold, and flu syndrome in the PIR‐205 Study 2004, and increased cough, respiratory infection, and rhinitis in Tan 2005. The three ocular adverse events most frequently reported by both studies were symptoms of decreased accommodation, papillae/follicles, and medication residue on the eyelids or eyelashes. Forty‐three children in the active treatment arms withdrew due to adverse events.

234One study (LAMP Study 2019) reported on vision‐related quality of life using the Chinese version of the 25‐item National Eye Institute Visual Function Questionnaire (NEI VFQ‐25). This validated quality‐of‐life instrument assesses 11 subscales: general health, general vision, ocular pain, near vision, distance vision, social function, mental health, role limitations, dependency, colour vision and peripheral vision. LAMP Study 2019 evaluated quality of life in 438 participants receiving one of three doses of atropine (0.05%, 0.02% or 0.01%) or placebo. The study authors reported no difference between groups in vision‐related quality of life, with similar scores across all 11 domains.

235Quantitative data on adherence to myopia control treatment were available in 21 studies, comprising 10 studies investigating spectacle interventions (Bao 2021; COMET Study 2003; COMET2 Study 2011; Fulk 2002; Hasebe 2008; Koomson 2016; Lam 2020; Pärssinen 1989; STAMP Study 2012; Yang 2009), six studies evaluating contact lens interventions (Anstice 2011; BLINK Study 2020; Chamberlain 2019; DISC Study 2011; Fujikado 2014; Katz 2003), and five using pharmacological interventions (ATOM 2 Study 2012; Hieda 2021; LAMP Study 2019; PIR‐205 Study 2004; Trier 2008). Where adherence data were available at multiple time points we report the results at the longest time point (see Table 10; Table 11; Table 12).

236Studies usually assessed adherence to optical interventions through an estimate by parents or children, or both, of wearing time per day of spectacles or contact lenses and the number of days per week the optical appliances were worn (6 to 7 days per week was usually judged as being fully compliant). Studies used a variety of methods for data collection, including questionnaires or discussion of compliance at follow‐up appointments. These data were available for 1731 participants wearing a variety of spectacle lens interventions (undercorrection, multifocal, peripheral plus and single vision). The range of daily wearing times were between 13.1 and 15.5 hours per day. The percentage of participants who were judged to be compliant were similar between test and control lenses, although the COMET Study 2003 and Pärssinen 1989 reported a lower proportion of participants wearing MFSL than the SVL controls.

237Adherence data were available for 873 participants in contact lens studies, including MFSCL, SVSCL and RGP. Daily wearing times were between 6.3 and 13.7 hours per day, with no statistical differences between multifocal and single vision contact lenses. In Katz 2003, which investigated RGP lenses versus SVLs, the percentage compliance at 24 months in the RGP group was 31.5% compared to 98.4% in spectacle lens‐wearing controls. Adherence was not formally assessed in orthokeratology studies, but these studies were often associated with high dropout rates (over 50% in some studies).

238For most pharmacological interventions, adherence was monitored by self‐reported questionnaires. Only PIR‐205 Study 2004 used electronic monitoring. Compliance was defined as using the study medication 75% to 80% of the time. Percentage compliance in 919 participants taking low‐dose atropine ranged from 83.3% to 98.8%, with similar levels of compliance between active and placebo arms. Compliance in the intervention arm of PIR‐205 Study 2004 and Trier 2008 were 79% and 89% respectively, which were similar to participants taking the placebo.

239This review summarises evidence from 64 studies, involving a total of 11,617 participants with low to moderate myopia. Studies investigated 11 interventions to slow the progression of myopia in children. Participants were girls and boys aged between 4 and 18 years, with an average age of 10.4 years. Interventions were broadly categorised into optical, pharmacological and environmental modalities. Fifty‐seven studies compared one or more myopia control interventions relative to a control or placebo intervention. Four studies included a combined intervention arm compared to control, and seven studies compared single or combined interventions to each other. Over 60% of studies were conducted in China or other Asian countries. In terms of study duration, 34% of the studies had a 12‐month duration, 46% reported to 24 months, 17% up to 36 months and only one study measured outcomes over 36 months. We defined the critical outcome 'progression of myopia' as both change in refractive error (as SER from baseline) and the more clinically meaningful, change in axial length. We judged most of the studies to be at 'high' or 'some concern' for risk of bias. Because the network was not well‐connected, we based our comparisons on direct evidence from classical pairwise meta‐analyses, except for moderate‐dose atropine.

240In terms of SER and axial length at 12 and 24 months, all interventions, except for undercorrection with SVLs, RGP and the adenosine antagonist 7‐methylxanthine, were superior to placebo in reducing the change in SER and slowing axial elongation. The certainty of evidence ranged from very low to moderate, depending on the comparison (see Table 1; Table 2; Table 3; Table 4). Although statistically significant, many of the efficacy estimates were small and clinically insignificant. There was evidence of retardation in efficacy of myopia control treatment over time, with most of the reduction in progression occurring in the first year.

241Overall, high‐dose topical atropine (≥ 0.5 %) and orthokeratology were the most effective interventions in slowing axial elongation at two years of follow‐up, corresponding to a 0.3 mm to 0.5 mm slowing of axial elongation (moderate‐certainty evidence). MFSCLs were similar to low‐dose topical atropine (< 0.1%), with a reduction in axial elongation of 0.15 mm and 0.16 mm respectively (moderate‐certainty evidence).

242The most commonly studied combination therapy was orthokeratology plus low‐dose atropine. Compared to orthokeratology monotherapy, the combination was associated with a significant reduction in axial elongation.

243We did not identify any relevant studies reporting on the effect of environmental interventions on childhood myopia progression.

244Data on changes to SER and axial length following cessation of treatment ('rebound') were available in four studies. There was no evidence of rebound in two studies that investigated optical interventions (Ruiz‐Pomeda 2018; STAMP Study 2012), but there was inconsistent evidence on rebound for topical atropine. One study, which abruptly terminated 1% topical atropine, found that there was a significant rebound effect (ATOM Study 2006), whilst another, which reduced the frequency of atropine instillation over time, reported a maintained slowing of myopia progression (Zhu 2021).

245In terms of the risk of adverse events, based on limited evidence that was often poorly reported, spectacle interventions were well tolerated with minimal and mild adverse events, similar to controls. In contact lens studies, the incidence of adverse events for multifocal soft contact lenses was also similar to the single vision soft contact lens controls. Adverse events in these studies generally consisted of expected contact lens‐related adverse events that were generally non‐serious. However, the incidence and severity of corneal staining was higher in studies using orthokeratology. The most commonly reported adverse events with antimuscarinic agents were photophobia, blurred near vision and hypersensitivity reactions, which increased with increasing drug concentration.

246Treatment adherence was generally high with levels of adherence that were similar across study arms.

247Only one study provided information on the effect of myopia control treatment on vision‐specific quality of life (LAMP Study 2019). This study compared three doses of atropine to placebo and reported no difference between groups at 12 months.

248We found no economic evaluation studies comparing different methods of myopia control in children. The apparent shortage of relevant economic evaluations indicates that there is a paucity of evidence regarding the costs and consequences of measures of myopia control in children. Future research could consider economic as well as clinical evaluation of interventions for myopia control.

249Several factors limit the applicability of the evidence in our review. Although we were able to include evidence from 64 RCTs, approximately 80% of the studies followed participants for two years or less. A consensus report produced by the International Myopia Institute (IMI), guiding principles of myopia control clinical study design, recommend three years as the minimum length to assess the efficacy of a treatment for myopia control, since treatment needs to be applied over multiple years during the period of most rapid myopia progression (Wolffsohn 2019 ). Extrapolation of efficacy data for outcomes measured at one year is therefore likely to overestimate the effectiveness of treatment.

250A number of factors complicated the comparison of studies, including differences in the demographic characteristics of the participants, and variability in the parameters used within similar treatments (e.g. different add powers and lens designs for multifocal spectacles and soft contact lenses and variable doses of atropine). Although the majority of studies adopted similar eligibility criteria, recruiting children aged 6 to 13 years, other studies used a wider age range of up to 18 years. This is important since faster progression occurs in younger myopes and progression slows in older teenagers. Furthermore, studies were conducted in different ethnic groups, particularly in children from South East Asian countries that typically have faster progression of myopia (Morgan 2012; Morgan 2018).

251These factors may, at least in part, explain the considerable heterogeneity of treatment effects identified in the review for some comparisons.

252It was difficult to compare the incidence of adverse events across studies due to different methods used to classify and report them. Furthermore, the use of photochromic and multifocal spectacles in pharmacological studies to mitigate potential side effects of higher topical atropine doses (≥ 0.5%) may have underestimated the incidence of glare, photophobia and reading difficulties reported in these studies. Similarly, the evaluation of treatment adherence between studies was complicated by the use of different methods to measure compliance (e.g. retrospective self‐report by parents or children, questionnaires or diaries). Lastly, the short time frame of many studies may have overestimated compliance, since it is possible that compliance may reduce over time.

253The certainty of the evidence for the critical outcome 'Progression of myopia' at one and two years ranged from very low to moderate, depending on the intervention. The main reasons for downgrading the certainty of evidence was risk of bias (principally due to lack of reporting details) and unexplained heterogeneity or inconsistency in the results.

254We were unable to conduct a quantitative analysis for other outcomes and we summarised the results for these outcomes at study level in summary tables with an indication of the overall risk of bias for each of the included studies.

255We followed the Cochrane Handbook for Systematic Reviews of Interventions to conduct this systematic review (Higgins 2022c). We applied a broad search strategy to ensure that all relevant papers were included. Pairs of review authors independently extracted data and assessed risk of bias; we also followed prespecified methods for classical and network meta‐analyses. We therefore believe that there should be no bias in the review process with respect to study selection and analysis of available data.

256Whenever possible, we used random‐effects pairwise meta‐analyses to incorporate heterogeneity amongst studies. Since we were unable to carry out the planned subgroup and sensitivity analyses to explore heterogeneity, the presence of considerable unexplained heterogeneity for several comparisons reduces our confidence in effect estimates.

257We judged many of the RoB 2 assessments as ‘some concerns’ across the studies in our review, which often reflected an inadequate reporting of information by the study authors, for example, no information on allocation concealment and lack of a prespecified analysis plan. Consequently, we may have overestimated the impact of bias on our findings by downgrading the certainty of evidence of the critical and important outcomes due to risk of bias.

258A previous Cochrane systematic review on myopia control interventions in children (Walline 2020), reviewed evidence from 41 RCTs and concluded, similar to the current review, that there was moderate‐certainty evidence favouring antimuscarinic drugs to reduce myopia progression and axial elongation with inconclusive evidence for other interventions. By contrast, a NMA of 16 interventions for myopia control in children conducted in 2016 concluded that "a range of interventions can significantly reduce myopia progression when compared with single vision spectacle lenses or placebo" (Huang 2016). More recently, a systematic review and NMA comparing the efficacy and safety of different concentrations of topical atropine for myopia control reported that 1%, 0.5% and 0.05% atropine were the three most efficacious atropine concentrations (Ha 2022).

259Two systematic reviews and an Ophthalmic Technology Assessment by the American Academy of Ophthalmology (AAO) have considered safety outcomes of contact lens interventions for reducing myopia progression in children (Cheng 2020; VanderVeen 2019; Yu 2022). With respect to daily disposable soft contact lenses, a review of retrospective data of adverse events from six RCTs estimated an incidence of 4.5 adverse events per 100 patient years and suggested that soft contact lenses can be safely worn by children (Cheng 2020). Yu 2022 analysed data from three studies and found no difference in adverse events between MFSCLs and control single vision lenses. VanderVeen 2019 reviewed published evidence on orthokeratology treatment for an AAO Health Technology Assessment and identified a sparsity of evidence in paediatric populations; it was noted that orthokeratology carries a small but definite risk of sight‐threatening keratitis. Bullimore 2013 estimated the incidence of microbial keratitis associated with orthokeratology as 13.9 per 10,000 patient‐years (95% CI 1.7 to 50.4), which is similar to the overall incidence of microbial keratitis in overnight soft contact lens wear.

260In their NMA of efficacy and safety of topical atropine for myopia control, Huang 2016 considered the safety profiles of different atropine concentrations based on changes in pupil size and accommodation. The authors found that based on these proxies for photophobia and near vision difficulties, lower atropine concentrations had higher safety ranking probabilities.

261Based on the best available evidence, topical antimuscarinic agents and orthokeratology (ortho‐K) currently appear to be the most effective treatments for slowing childhood myopia progression. There is some uncertainty as to the optimal dose of atropine, the most studied antimuscarinic agent. Although higher doses slow overall axial elongation by approximately 0.5 mm over two years, corresponding to an approximate 1.00 D reduction in myopia, higher concentrations are more likely to cause adverse events and may increase the risk of rebound following cessation of treatment. The current review found limited evidence that rebound could potentially be reduced by tapering the treatment prior to termination. There are logistical difficulties in assessing change in refractive error in ortho‐K studies and therefore evidence of efficacy is based on slowing axial elongation. Ortho‐k may also require more specialised knowledge by the eye care practitioner, and therefore it may not be as available as some of the other treatment modalitiesEvidence on the efficacy of other treatments was limited by short study durations and considerable heterogeneity in treatment response. The finding that treatment efficacy reduces over time would add to the perception of greater efficacy in studies of short duration.Uncertainty remains regarding the risk‐benefit of ortho‐K and other contact lens interventions in children. Adverse events across the included studies were generally poorly described with a lack of standardisation of reporting. Although none of the included studies reported serious adverse events, the duration of follow‐up in trials may have been insufficient to capture long‐term or rare adverse events.Myopia control is a rapidly moving field, which emphasises the need for a living systematic review in this area that is underpinned by continual and active monitoring of new evidence.

262Based on the best available evidence, topical antimuscarinic agents and orthokeratology (ortho‐K) currently appear to be the most effective treatments for slowing childhood myopia progression. There is some uncertainty as to the optimal dose of atropine, the most studied antimuscarinic agent. Although higher doses slow overall axial elongation by approximately 0.5 mm over two years, corresponding to an approximate 1.00 D reduction in myopia, higher concentrations are more likely to cause adverse events and may increase the risk of rebound following cessation of treatment. The current review found limited evidence that rebound could potentially be reduced by tapering the treatment prior to termination.

263There are logistical difficulties in assessing change in refractive error in ortho‐K studies and therefore evidence of efficacy is based on slowing axial elongation. Ortho‐k may also require more specialised knowledge by the eye care practitioner, and therefore it may not be as available as some of the other treatment modalities

264Evidence on the efficacy of other treatments was limited by short study durations and considerable heterogeneity in treatment response. The finding that treatment efficacy reduces over time would add to the perception of greater efficacy in studies of short duration.

265Uncertainty remains regarding the risk‐benefit of ortho‐K and other contact lens interventions in children. Adverse events across the included studies were generally poorly described with a lack of standardisation of reporting. Although none of the included studies reported serious adverse events, the duration of follow‐up in trials may have been insufficient to capture long‐term or rare adverse events.

266Myopia control is a rapidly moving field, which emphasises the need for a living systematic review in this area that is underpinned by continual and active monitoring of new evidence.

267There are a number of research priorities in this field. Epidemiological evidence has shown that the age of onset and rate of myopia progression in children varies considerably. There is a need to develop better predictive models to identify children who are most likely to progress rapidly and will therefore potentially derive most benefit from treatment. The absence of long‐term data provides little evidence as to when myopia control interventions can be stopped or modified during treatment.Although topical atropine shows considerable promise as a treatment for myopia control, the optimal dose is yet to be established, which balances efficacy, safety and propensity to rebound. There are many ongoing trials investigating the efficacy of various doses of atropine, used either as monotherapy or in combination with another intervention, which may provide further data to determine the optimal drug dose for myopia control.The International Myopia Institute has developed a consensus set of principles on study design to guide the development of myopia control trial protocols (Wolffsohn 2019). Many of the included trials in this review did not meet these recommendations and researchers should be encouraged to adopt these principles to facilitate harmonised reporting of outcomes, including standardised reporting of AEs. There was also a tendency for authors to report a relative percentage reduction in myopia progression to express treatment effect, which can be misleading.To address uncertainty in the safety of myopia control interventions, particularly relating to rare and potentially sight‐threatening adverse events, it may be necessary to seek evidence from non‐randomised studies, since such events are unlikely to be seen in randomised controlled trials due to their small size and relatively short duration. Future systematic reviews considering safety could also, therefore, consider evidence from non‐randomised studies for a comprehensive evaluation of safety.Only one of the included studies evaluated the impact of myopia control interventions on quality of life. There is therefore a need for further studies using validated instruments to measure vision‐related and health‐related quality of life as an outcome of myopia control studies.There is also a lack of health economic studies that could inform policy‐makers and healthcare decision‐makers, enabling them to identify which interventions, policies or services provide the best value for money.

268There are a number of research priorities in this field. Epidemiological evidence has shown that the age of onset and rate of myopia progression in children varies considerably. There is a need to develop better predictive models to identify children who are most likely to progress rapidly and will therefore potentially derive most benefit from treatment. The absence of long‐term data provides little evidence as to when myopia control interventions can be stopped or modified during treatment.

269Although topical atropine shows considerable promise as a treatment for myopia control, the optimal dose is yet to be established, which balances efficacy, safety and propensity to rebound. There are many ongoing trials investigating the efficacy of various doses of atropine, used either as monotherapy or in combination with another intervention, which may provide further data to determine the optimal drug dose for myopia control.

270The International Myopia Institute has developed a consensus set of principles on study design to guide the development of myopia control trial protocols (Wolffsohn 2019). Many of the included trials in this review did not meet these recommendations and researchers should be encouraged to adopt these principles to facilitate harmonised reporting of outcomes, including standardised reporting of AEs. There was also a tendency for authors to report a relative percentage reduction in myopia progression to express treatment effect, which can be misleading.

271To address uncertainty in the safety of myopia control interventions, particularly relating to rare and potentially sight‐threatening adverse events, it may be necessary to seek evidence from non‐randomised studies, since such events are unlikely to be seen in randomised controlled trials due to their small size and relatively short duration. Future systematic reviews considering safety could also, therefore, consider evidence from non‐randomised studies for a comprehensive evaluation of safety.

272Only one of the included studies evaluated the impact of myopia control interventions on quality of life. There is therefore a need for further studies using validated instruments to measure vision‐related and health‐related quality of life as an outcome of myopia control studies.There is also a lack of health economic studies that could inform policy‐makers and healthcare decision‐makers, enabling them to identify which interventions, policies or services provide the best value for money.

273Protocol first published: Issue 4, 2021