Interventions to slow progression of myopia in children

Summary of findings 1. Interventions to slow progression of myopia in children

Outcome: change in refractive error, measured in diopters (D), from baseline to 1‐year follow‐up
Interventions to slow progression of myopia in children
Population: children with myopia (nearsightedness) Settings: ophthalmology or optometry clinics
Comparison (intervention vs comparator)	Mean difference (95% CI) Positive values represent slower progression of myopia in the treatment group than in the comparison group	No. of participants (studies)	Certainty of the evidence (GRADE)	Comments
Undercorrected vs fully corrected spectacles	‐0.15 D (‐0.29 to 0.00)	142 (2)	⊕⊕⊝⊝ low^a,b	A third study did not report this outcome at 1 year
Multifocal vs single vision lens spectacles	0.14 D (0.08 to 0.21)	1463 (9)	⊕⊕⊕⊝ moderate^b	Five studies not included in the meta‐analyses also showed mostly favorable effects of multifocal lenses for slowing myopia progression
Peripheral plus spectacles vs single vision lens spectacles	See comment	597 (3)	⊕⊕⊝⊝ low^b,c	No meta‐analysis was conducted because of clinical and methodological heterogeneity among the 3 studies; furthermore, the results from these studies were inconsistent
Bifocal vs single vision soft contact lenses	0.20 D (‐0.06 to 0.47)	300 (4)	⊕⊕⊝⊝ low^b,c	‐
Rigid gas permeable contact lenses vs spectacles or soft contact lenses	See comment	420 (2)	⊕⊝⊝⊝ very low^a,b,c	No meta‐analysis was conducted due to differences among 2 studies that reported inconsistent results
Orthokeratology contact lenses vs single vision lenses	See comment	‐	‐	Because orthokeratology contact lenses temporarily reduce myopia, their myopia control treatment effect can be measured only by axial elongation. We did not analyze the changes in refractive error for this comparison
Spherical aberration soft contact lenses vs single vision soft contact lenses	See comment	209 (2)	⊕⊕⊝⊝ low^b,d	No meta‐analysis was conducted because 1 of the studies did not provide effect estimates; however, 2 studies comparing spherical aberration SCLs with single vision SCLs reported no difference in myopia progression
Antimuscarinic agents vs placebo	Atropine: 1.00 D (0.93 to 1.07) Pirenzepine: 0.31 D (0.17 to 0.44) Cyclopentolate: 0.34 D (0.08 to 0.60)	629 (3) 326 (2) 64 (1)	⊕⊕⊕⊝ moderate^b	We stratified the analysis by types of antimuscarinic agents due to statistical inconsistency
Atropine vs tropicamide	Atropine 0.1%: 0.78 D (0.49 to 1.07) Atropine 0.25%: 0.81 D (0.57 to 1.05) Atropine 0.5%: 1.01 D (0.74 to 1.28)	196 (1)	⊕⊕⊕⊝ low^b	‐
Systemic 7‐methylxanthine vs placebo	0.07 D (‐0.09 to 0.24)	77 (1)	⊕⊕⊕⊝ moderate^a	‐
Timolol drops vs no drops	‐0.05 D (‐0.21 to 0.11)	95 (1)	⊕⊕⊝⊝ low^a,b	‐
Atropine plus multifocal spectacles vs placebo plus SVLs	0.78 D (0.54 to 1.02)	191 (2)	⊕⊕⊕⊝ moderate^b	‐
Atropine plus bifocal spectacles vs cyclopentolate plus SVLs	0.36 D (0.11 to 0.61)	64 (1)	⊕⊕⊕⊝ moderate^b	‐
Bifocal spectacles vs SVLs with timolol drops	0.19 D (0.06 to 0.32)	97 (1)	⊕⊕⊕⊝ moderate^b	‐
Tropicamide plus bifocal spectacles vs SVLs	See comment	50 (1)	‐	No estimate of effect was reported
Outcome: change in axial length, measured in millimeters (mm), from baseline to 1‐year follow‐up
Comparison (intervention vs comparator)	Mean difference (95% CI) Negative values represent less axial elongation in the treatment group than in the comparison group	No. of participants (studies)	Certainty of the evidence (GRADE)	Comments
Undercorrected vs fully corrected spectacles	0.05 mm (‐0.01 to 0.11)	94 (1)	⊕⊕⊝⊝ low^a,b	Two studies did not report this outcome at 1 year
Multifocal vs single vision lens spectacles	‐0.06 mm (‐0.09 to ‐0.04)	896 (4)	⊕⊕⊕⊝ moderate^b	Four studies (not included in the meta‐analysis) showed mostly favorable effects of multifocal lenses and 6 studies did not report this outcome
Peripheral plus spectacles vs single vision lens spectacles	See comment	597 (3)	⊕⊕⊝⊝ low^b,c	‐
Bifocal vs single vision soft contact lenses	‐0.11 mm (‐0.14 to ‐0.08)	300 (4)	⊕⊕⊝⊝ low^b,c	‐
Rigid gas permeable contact lenses vs spectacles or soft contact lenses	0.02 mm (‐0.05 to 0.10)	415 (2)	⊕⊕⊝⊝ low^a,b	‐
Orthokeratology contact lenses vs single vision lenses	‐0.28 mm (‐0.38 to ‐0.19)	106 (2)	⊕⊕⊕⊝ moderate^b	One other study reported this outcome; however, the study did not report sufficient data for analysis
Spherical aberration soft contact lenses vs single vision soft contact lenses	See comment	209 (2)	⊕⊝⊝⊝ very low^a,b,d	No meta‐analysis was conducted due to clinical, methodological, and statistical differences between the 2 studies; however, 2 studies comparing spherical aberration SCLs with single vision SCLs reported no difference in axial length
Antimuscarinic agents vs placebo	Atropine: ‐0.35 mm (‐0.38 to ‐0.31) Pirenzepine: ‐0.13 mm (‐0.14 to ‐0.12)	502 (2) 326 (2)	⊕⊕⊕⊝ moderate^c	We did not combine results for all antimuscarinic agents due to statistical inconsistency; outcome was not reported by 2 studies
Atropine vs tropicamide	See comment	196 (1)	‐	Outcome was not reported
Systemic 7‐methylxanthine vs placebo	‐0.03 mm (‐0.10 to 0.03)	77 (1)	⊕⊕⊕⊝ moderate^a	‐
Timolol drops vs no drops	See comment	95 (1)	‐	Outcome was not reported
Atropine plus multifocal spectacles vs placebo plus SVLs	‐0.37 mm (‐0.47 to ‐0.27)	127 (1)	⊕⊕⊕⊝ moderate^b	One study did not report this outcome
Atropine plus bifocal spectacles vs cyclopentolate plus SVLs	See comment	64 (1)	‐	Outcome was not reported
Bifocal spectacles vs SVLs with timolol drops	See comment	97 (1)	‐	Outcome was not reported
Tropicamide plus bifocal spectacles vs SVLs	See comment	50 (1)	‐	Outcome was not reported
Adverse effects
No serious adverse events were reported across all interventions. Two studies showed that participants receiving antimuscarinic topical medications (n=259) were more likely to experience accommodation difficulties (Risk Ratio 9.05, 95% CI 4.09 to 20.01), papillae and follicles (RR 3.22, 95% CI 2.11 to 4.90) than participants receiving placebo (n=128), but no difference in medication residue on the eyelids or eyelashes (RR 0.91, 95% CI 0.73 to 1.12). Certainty of a body of evidence was moderate, downgraded for imprecision of results (‐1).
GRADE Working Group grades of evidence. High certainty: further research is very unlikely to change our confidence in the estimate of effect. Moderate certainty: further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low certainty: further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low certainty: we are very uncertain about the estimate.
CI: confidence interval; D: diopters. ^aDowngraded for imprecision (i.e. wide confidence interval). ^bDowngraded for risk of bias among included trials. ^cDowngraded for inconsistency. ^dDowngraded for indirectness due to averaging values over time assuming linear change (e.g. reporting the change per year using data collected at baseline and at 2 years of follow‐up).

Background

Description of the condition

Myopia, also known as nearsightedness, occurs because the cornea or the lens is too powerful or the eyeball is longer than normal; this causes distant objects to be focused in front of the retina instead of on it, as occurs in nonmyopic individuals. In myopia, near objects are seen clearly but distant objects appear blurred.

Epidemiology

Myopia is an important cause of reduced vision in populations throughout the world and is one of the five immediate priorities for the "Vision 2020" initiative of the World Health Organization (WHO) (Pararajasegaram 1998). Approximately 33% of persons in the United States are myopic, reflecting an increase from approximately 25% in the early 1970s (Vitale 2009). It is estimated that half of the world’s population will be myopic by 2050 (Holden 2016). Racial and ethnic differences in the magnitude and prevalence of myopia have been observed (Garner 1999; Lin 1999; Maul 2000; Voo 1998; Zhan 2000), with both greater in Asia than in other parts of the world (Lin 1999; Zhan 2000).

Juvenile‐onset myopia in the United States typically develops at approximately six to eight years of age and progresses at a rate of approximately 0.50 D (diopters) per year through 15 to 16 years (COMET Study 2003; Fulk 2002; Goss 1987; Perrigin 1990). The progression of myopia is typically faster at younger ages (Braun 1996; Goss 1987; Goss 1990; Pärssinen 1989; Saw 2000), but myopia onset, progression, and stabilization vary widely among individuals (Braun 1996; Pärssinen 1989; Saw 2000). Similar proportions of boys and girls are affected by myopia, and the degree of myopia is similar between the two genders (Zadnik 2003).

Etiology and risk factors

Several factors have been suggested to have a role in the development of myopia. Many models estimate greater genetic effects than environmental effects for myopia (Chen 1985; Hammond 2001). Children with two myopic parents have greater axial lengths; this indicates higher risk of myopia than for children with one or no myopic parents (Zadnik 1994). Environmental influences are related to prolonged reading or near work, which has inconsistently been associated with increased myopia prevalence (Saw 2001; Young 1969). Fewer hours spent outdoors has also been associated with myopia (Dirani 2009; Guggenheim 2012; Guo 2013; Jones 2007; Rose 2008). Children randomly assigned to additional outdoor time exhibit a lower incidence of myopia onset but do not exhibit slowed progression of myopia after onset (He 2015; Wu 2010).

Presentation and diagnosis

The primary symptom of myopia is blurred distance vision. Children often present to an eye care practitioner after they have failed a vision screening at school or after a parent or teacher has noticed the child squinting or having difficulty seeing distant objects.

An eye care practitioner using autorefraction or retinoscopy may confirm the diagnosis of myopia objectively, or the practitioner can confirm the diagnosis by performing a subjective refraction, which requires responses from the child. To diagnose myopia in a child, cycloplegic drops should be placed in the child's eyes, hindering his or her ability to focus the eyes, so that an accurate prescription can be determined.

Description of the intervention

Spectacles are often the initial treatment for children with myopia because they provide clear vision with few potential side effects. Spectacles for myopia correction use concave lenses that focus light more posteriorly, resulting in a clear image focused on the retina.

Contact lenses are typically a secondary treatment option for children because they require greater dexterity and responsibility when compared to spectacles. They also bear greater risks than spectacles, which range from innocuous redness of the eyes to severe pain and vision loss due to corneal ulcers (Fonn 1988; MacRae 1991; Schein 1989). However, young children are at lower risk for problems associated with contact lens wear than are college‐age adults (Chalmers 2011; Wagner 2011). There are different types of contact lenses. Soft contact lenses are made of gel‐like, water‐containing, flexible plastics that allow oxygen to pass through the cornea. Spherical aberration soft contact lenses aims to correct an optical problem that occurs when incoming light rays end up focusing at different points after passing through a spherical surface (in this case the ocular system). Rigid gas permeable contact lenses (RGPCLs) are rigid, more durable and less likely to tear compared to soft contact lenses, and resistant to deposit buildup; however, they may be less comfortable to wear initially. Orthokeratology is a lens fitting procedures that uses specially designed RGPCLs to change the curvature of the cornea to temporarily improve the eye's ability to focus on objects. Most orthokeratology lenses are worn at night and then removed during the day. When orthokeratology is discontinued, the cornea will return to its original curvature and the eye to its original amount of nearsightedness.

Lastly, both spectacles and contact lenses can contain more than one power zone; they are called bifocal, multifocal, or progressive addition lenses.

There are currently no pharmaceutical agents approved by the US Food and Drug Administration for use as myopia treatments, although antimuscarinic agents, such as atropine, pirenzepine, tropicamide, and scopolamine, as well as 7‐methylxanthine (7‐mx), a non‐antimuscarinic agent, have been used off‐label and targeted in recent clinical trials.

Laser refractive surgery, such as laser in situ keratomileusis (LASIK) or photorefractive keratectomy (PRK), causes permanent flattening of the central corneal curvature resulting from removal of stromal tissue with a laser once myopia has developed (Duffey 2003; Shortt 2006), but it is not routinely performed in children.

Other forms of myopia correction, such as placement of a lens inside the eye and clear rings into the cornea, also are not used routinely in children because of the risk of potential myopia progression (Barsam 2010).

How the intervention might work

In terms of slowing myopia progression, use of multifocal spectacles and undercorrection of myopic refractive error are thought to reduce accommodative error, which may act as a stimulus for increased eye growth. Myopic patients exhibit greater accommodative lag than nonmyopic patients (COMET Study 2003; Mutti 2006). Accommodative lag results in light focused behind the retina during near work, which may act as a signal to increase eye growth and may result in myopia. If the accommodative error can be reduced with bifocals or myopic undercorrection, then the stimulus for eye growth will be reduced, and this may slow myopia progression.

Antimuscarinic agents were thought to reduce myopic progression by eliminating accommodation, but this has been shown to be a local retinal effect that slows myopia progression (Troilo 1987). Antimuscarinic receptor binding may lead to a biochemical change that slows eye growth, but the exact mechanism is unknown.

Multifocal contact lenses provide myopic defocus of light in the periphery while allowing clear vision by focusing light on the central retina (Charman 2006; Kang 2011; Moore 2017; Ticak 2013). The myopic defocus (light focused in front of the retina) may act as a signal to slow eye growth and reduce myopia progression (Smith 2009). Orthokeratology works by flattening the center cornea to temporarily improve the eye's ability to focus on objects.

Why it is important to do this review

Myopia has been reported to have reached epidemic proportions in parts of the world (Park 2004). Strategies to control progression of myopia gain importance in the context of the "Vision 2020" initiative by the WHO, which seeks to eliminate preventable causes of blindness, including risks associated with high myopia, by the year 2020 (Pararajasegaram 1998). Interventions that have been explored for this purpose include bifocal spectacles, cycloplegic eye drops, intraocular pressure–lowering drugs, muscarinic receptor antagonists, and contact lenses. In this review, we systematically assessed the effectiveness of strategies to control progression of myopia in children.

Objectives

To assess the effects of interventions, including spectacles, contact lenses, and pharmaceutical agents, such as muscarinic receptor antagonists, cycloplegic eye drops, and intraocular pressure–lowering medications, in slowing myopia progression in children.

Methods

Criteria for considering studies for this review

Types of studies

This review included randomized controlled trials (RCTs).

Types of participants

We included trials in which participants were treated with spectacles, contact lenses, or pharmaceutical agents for controlling progression of myopia. We excluded trials in which most participants were older than 18 years at the start of the trial. We also excluded trials that included participants with less than ‐0.25 D spherical equivalent myopia at baseline. (The spherical equivalent is an optical measurement based on a mathematical calculation: the sum of the spherical power plus half the cylindrical power of the refractive error.)

Types of interventions

We included trials in which any of the following interventions for slowing the progression of myopia were compared with a control treatment of single vision spectacle lenses, single vision soft contact lenses (SVSCLs), or placebo treatment, or with each other.

Undercorrection of myopia, bifocal lenses (spectacles), progressive addition lenses (PALs), and other modifications to spectacle lenses.
Bifocal soft contact lenses (BSCLs), RGPCLs, and corneal reshaping (orthokeratology) contact lenses.
Pharmaceutical agents (e.g. atropine, pirenzepine).

Types of outcome measures

Primary outcomes

Progression of myopia assessed as the mean change in refractive error (spherical equivalent) from baseline to each year of follow‐up and measured by any method

Secondary outcomes

Mean change in axial length, measured by any method
Mean change in corneal radius of curvature, measured by any method

We analyzed the secondary outcomes for each year of follow‐up when sufficient data were available.

Adverse effects

We summarized reported adverse effects related to the interventions as described in the included studies, including but not limited to blurry vision, red eyes, infection, and conjunctival reactions.

Economic data

We documented reported cost analyses and other data on economic outcomes when reported by the included trials.

Quality of life measures

We documented any quality of life information when reported by the included trials.

Follow‐up

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

Search methods for identification of studies

Electronic searches

The Cochrane Eyes and Vision Information Specialist searched the following electronic databases for RCTs and controlled clinical trials, with no language or publication year restrictions up to Febrary 2018 (and all relevant studies up to Febrary 2018 were included in the current version). A top up search was done on February 26, 2019. We listed potentially relevant studies from the top up search in the tables for "Characteristics of studies awaiting classification" and "Characteristics of ongoing studies".

Cochrane Central Register of Controlled Trials (CENTRAL; Issue 2, 2019) (which contains the Cochrane Eyes and Vision Trials Register), in the Cochrane Library (searched February 26, 2019) (Appendix 1).
MEDLINE Ovid (1946 to February 26, 2019) (Appendix 2).
Embase (1947 to February 26, 2019) (Appendix 3).
PubMed (1948 to February 26, 2019) (Appendix 4).
Latin American and Caribbean Health Science Information Database (LILACS) (1982 to February 26, 2019) (Appendix 5).
International Standard Randomized Controlled Trials Number (ISRCTN) registry (www.isrctn.com/editAdvancedSearch; searched February 26, 2019) (Appendix 6).
US National Institutes of Health Ongoing Trials Register ClinicalTrials.gov (www.clinicalTrials.gov; searched February 26, 2019) (Appendix 7).
World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP) (www.who.int/ictrp; searched February 26, 2019) (Appendix 8).

Searching other resources

We searched the reference lists of identified trial reports to find additional trials. We used the Science Citation Index (last assessed April 12, 2013) to find studies that had cited the identified trials. We contacted the primary investigators of identified trials for details of other potentially relevant trials not identified by the electronic searches, and of recently completed or ongoing trials. We did not conduct manual searches of abstracts of conference proceedings and optometry literature specifically for this review, as these sources are searched by the Cochrane Eyes and Vision Group and are listed in CENTRAL.

Data collection and analysis

Selection of studies

Two review authors, including at least one clinician and one methodologist, independently assessed the titles and abstracts of records identified by electronic and manual searches as per the Criteria for considering studies for this review. We classified records as (1) definitely relevant, (2) possibly relevant, or (3) definitely not relevant. We obtained and assessed the full‐text reports of records classified as (1) or (2) by at least one review author. After assessing the full‐text reports, we classified studies as (A) include, (B) awaiting assessment, or (C) exclude. A third review author resolved disagreements. Review authors were unmasked to report authors, authors' institutions, and trial results during this assessment. We included and further assessed studies identified as (A) for study design and risk of bias. We contacted the authors of studies classified as (B) for clarification and reassessed these studies as per the inclusion criteria, as further information became available. We excluded studies identified as (C) and documented the reasons for exclusion in this review.

We initially included Cheng 2010, but after data extraction and risk of bias assessment, we assessed this study to be quasi‐randomized and thus deemed it ineligible for the review. However, as we initially included the study, we did not exclude it post hoc but instead conducted sensitivity analyses for inadequate randomization when applicable.

Data extraction and management

Two review authors independently extracted the data for primary and secondary outcomes on two paper data collection forms developed by the Cochrane Eyes and Vision Group. We resolved discrepancies by discussion. We contacted primary investigators for data reported unclearly or incompletely. One review author entered the data into Review Manager 5 (RevMan 5) (Review Manager 2014), and a second review author verified the data entered.

Assessment of risk of bias in included studies

Two review authors independently assessed potential sources of bias in trials according to the methods described in Chapter 8 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2017). We resolved disagreements between authors through discussion.

We considered the following parameters.

Selection bias (random sequence generation, quality of allocation concealment).
Performance bias (masking of participants).
Detection bias (masking of outcome assessors and data analyzers).
Attrition bias (completeness of follow‐up, intention‐to‐treat [ITT] analysis).
Reporting bias (selective outcome reporting, incomplete reporting of results).
Other potential sources of bias (e.g. funding source).

For attrition bias, we considered whether or not reasons for losses to follow‐up were comparable between treatment arms, and whether or not all participants were analyzed as randomized. If studies reported that an ITT analysis was performed, we assessed whether (1) all randomized participants were included in the analysis, even when no outcome data were collected, and (2) participants were analyzed in the intervention groups to which they were randomized, regardless of the intervention they actually received. We interpreted a true ITT analysis to have been undertaken only when both of these criteria were fulfilled.

We classified the risk of bias for each parameter as "low risk of bias," "unclear risk of bias," or "high risk of bias." For example, we considered studies using allocation concealment by centralized randomization and use of sequential opaque envelopes (which provided reasonable confidence that participating eye care providers and patients were not aware of the randomization sequence) to be at low risk of bias. We contacted the authors of trials when we needed additional information to assess risk of bias. If trial authors did not respond within an eight‐week period, we classified the trial based on available information.

Measures of treatment effect

We reported mean differences (MDs) for continuous outcome measures and risk ratios (RRs) for dichotomous outcomes.

Unit of analysis issues

When only one eye per participant was randomized, the unit of analysis was the individual eye (and participant). When both eyes from the same participant were randomized (either to the same intervention or to different interventions), we used estimates that had accounted for the correlation between the two eyes. For cross‐over design and cluster‐randomized design, we analyzed only estimates that had accounted for the design.

Dealing with missing data

We contacted the authors of trial reports for any missing data. When we did not receive a response within eight weeks, we analyzed the studies based on available information. We will include any new information in future updates of the review.

Assessment of heterogeneity

We assessed methodological and clinical heterogeneity by examining the characteristics and design of included studies. We assessed statistical heterogeneity by using the Chi² test and the I² statistic. We considered a P value less than 0.1 as significant for the test of heterogeneity. We assessed the inconsistency of effect estimates across studies using the I² statistic. An I² value greater than 50% was an indication of substantial statistical heterogeneity.

Assessment of reporting biases

We assessed reporting biases based on communications with trial authors regarding any outcomes assessed but not reported.

Data synthesis

We used a fixed‐effect model for meta‐analyses including fewer than three studies, and a random‐effects model for meta‐analyses including three or more studies. Change‐from‐baseline data were combined in meta‐analyses with mean outcome data at annual measurement time points based on the generic inverse variance (unstandardized) MD method, as outlined in Chapter 9 of the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2017). When we assessed substantial clinical, methodological, or statistical heterogeneity, we did not combine individual trials in meta‐analysis but instead reported study results separately.

Subgroup analysis and investigation of heterogeneity

We undertook subgroup analyses for types of intervention modalities (i.e. bifocals, PALs, and specific pharmaceutical agents). In the future, if sufficient evidence becomes available, we will also conduct subgroup analyses according to age, degree of myopia at baseline, and type of contact lens (soft vs rigid gas permeable).

Sensitivity analysis

We conducted a sensitivity analysis for meta‐analyses in which more than three studies were included and when change‐from‐baseline outcomes were combined in analysis with mean outcomes at annual measurement time points. We combined studies using autorefraction in analysis with subjective refraction or when analyses included the Cheng 2010 study.

"Summary of findings"

We prepared a "Summary of findings" table including all comparisons for each of the following outcomes: change in refractive error, change in axial length, change in corneal curvature, and adverse effects. Additionally, we presented adverse effects by intervention in the Additional tables section, as the data were insufficient for quantitative analysis. We used the GRADE approach to assess the overall certainty of evidence for each outcome based on five criteria: risk of bias, imprecision, inconsistency, indirectness, and publication bias (Guyatt 2011).

Results

Description of studies

Results of the search

Details of results of the 2011 version of this review were published previously (Walline 2011). Briefly, we included 89 records (from 23 studies), excluded 82 records (from 61 studies), identified four records awaiting classification (from three studies: Anstice 2011; ATOM 2 Study 2012; COMET2 Study 2011), and identified one ongoing study (STAMP Study 2012).

In February 2018, we conducted an update of the electronic literature search, handsearched the reference lists of included studies, and used the Science Citation Index to identify additional studies. We identified 4064 additional records, 10 of which we identified by manual searching. After omitting duplicates and screening 4052 titles and abstracts, we excluded 3678 records and obtained full‐text reports of 374 records for further review. Upon full‐text review, we excluded 127 reports. Of them, six excluded reports belonged to a study previously assessed as awaiting classification because it did not include a single vision control group (ATOM 2 Study 2012). We also identified 66 reports for studies listed as ongoing. We listed seven reports as awaiting classification. We included the remaining 174 reports: 27 reported 15 newly included studies (Cambridge Anti‐Myopia Study 2013; Charm 2013; Cheng 2016; DISC Study 2011; Fujikado 2014; Han 2018; Hasebe 2014; Koomson 2016; Lu 2015; ROMIO Study 2012; Swarbrick 2015; Trier 2008; Wang 2005; Wang 2017; Yi 2015), six reported results for the previously assessed ongoing study (STAMP Study 2012), two reported results for studies previously assessed as awaiting classification (Anstice 2011; COMET2 Study 2011), 50 reported new results for studies already included in the review, and 89 reported results included in the previously published review (Walline 2011).

In an additional top‐up search conducted on February 26, 2019, we screened 724 titles and abstracts, of which we excluded 668 records. We excluded nine reports upon full‐text review. We identified 10 reports of 10 studies listed as ongoing and 18 records as awaiting assessment.

Overall, we included 41 studies (174 reports), excluded 91 studies (136 reports), identified 74 ongoing studies (76 reports), and 25 records awaiting classification (Figure 1).

Figure 1

Study flow diagram.

Included studies

We included 41 studies (6772 total participants) in this review. The studies evaluated varying interventions, including spectacles, contact lenses, and pharmaceutical agents (Table 1). With the exception of interventions, study characteristics and outcomes were comparable among the included studies. Except one study (Cambridge Anti‐Myopia Study 2013; n=147), all other studies included children 18 years or younger. No participant had myopia less than 0.25 D. Progression of myopia, measured as the change in refractive error, was assessed as the primary outcome in 37 studies, and as a secondary outcome in three studies (Charm 2013; Swarbrick 2015; Trier 2008). ROMIO Study 2012 was the only study that did not report refractive error as an outcome. Thirty‐eight studies measured refraction under cycloplegia, of which 33 used autorefraction. No study reported quality of life or economic outcomes. Outcomes by intervention are summarized in Table 2 Table 3 and Table 4.

Table 1. Interventions of included studies

Study	Spectacles					Contact lenses					Pharmaceutical agents		Combination of interventions
	Undercorrected SVLs	Multifocal lenses			Fully corrected SVLs	Soft bifocal lenses	RGP	Ortho‐k	SA‐SCL	SVSCL	Test group	Reference group
	Undercorrected SVLs	Bifocal lenses	PALs	Peripheral plus lenses	Fully corrected SVLs	Soft bifocal lenses	RGP	Ortho‐k	SA‐SCL	SVSCL	Test group	Reference group
Adler 2006; 2 study arms	X				X
Chung 2002; 2 study arms	X				X
Koomson 2016; 2 study arms	X				X
Cheng 2010; 3 study arms		+1.50 and +1.50 prism			X
Fulk 1996; 2 study arms		+1.25			X
Fulk 2002; 2 study arms		+1.50			X
Houston Study 1987; 3 study arms		+1.00 and +2.00			X
Jensen 1991; 3 study arms		+2.00			X								Timolol + SVLs
Pärssinen 1989; 3 study arms		+1.75			Continous use and distance only
COMET Study 2003; 2 study arms			+2.00		X
COMET2 Study 2011; 2 study arms			+2.00		X
Edwards 2002; 2 study arms			+1.50		X
Hasebe 2008; 2 study arms^a			+1.50		X
MIT Study 2001; 3 study arms			Plus placebo drops		Plus placebo drops								Atropine + PALs
STAMP Study 2012; 2 study arms			+2.00		X
Wang 2005; 2 study arms			Add NR		X
Yang 2009; 2 study arms			+1.50		X
Lu 2015; 2 study arms				+2.50	X
Hasebe 2014; 3 study arms				+1.00 and +1.50	X
Sankaridurg 2010; 4 study arms				+1.00, +1.90, and +2.00	X
Anstice 2011; 2 study arms^a						+2.00				X
CONTROL Study 2016; 2 study arms						Add NR				X
DISC Study 2011; 2 study arms						+2.50				X
Fujikado 2014; 2 study arms^a						+0.50				X
CLAMP Study 2004; 2 study arms							X			X
Katz 2003; 2 study arms					X		X
Charm 2013; 2 study arms					X			X
ROMIO Study 2012; 2 study arms					X			X
Swarbrick 2015; 2 study arms^a							X	X
Cambridge Anti‐Myopia Study 2013; 4 study arms									With and without vision training	With and without vision training
Cheng 2016; 2 study arms									X	X
ATOM Study 2006; 2 study arms											1% atropine	Placebo drops
Yi 2015; 2 study arms											1% atropine	Placebo drops
Yen 1989; 3 study arms											1% atropine + bifocals	Saline + SVLs	Cyclopentolate + SVLs
Shih 1999; 4 study arms											0.1%, 0.25%, and 0.5% atropine	0.5% tropicamide
PIR‐205 Study 2004; 2 study arms											2% pirenzepine gel	Placebo gel
Tan 2005; 3 study arms											2% pirenzepine gel once and twice daily	Placebo gel
Trier 2008; 2 study arms											Systemic 7‐methylxanthine	Placebo tablet
Schwartz 1981; 2 study arms					X								Tropicamide + bifocals

NR: not reported.
Ortho‐k: orthokeratology lenses.
PALs: progressive addition lenses.
RGP: rigid gas permeable contact lenses.
SA‐SCL: spherical aberration soft contact lenses.
SVLs: single vision lenses.
SVSCL: single vision soft contact lenses.

^aCross‐over trial.

Table 2. Outcomes reported by studies of spectacle interventionsa

Outcomes	Interventions studied
Outcomes	Undercorrected lenses: 3studies	Multifocal lenses: 14studies	Peripheral plus spectacles: 3studies
Primary outcome: change in refractive error	Analysis 1.1	Analysis 2.1; Analysis 2.2; Analysis 2.3	Analysis 3.1; Analysis 3.2
Secondary outcome: change in axial length	Analysis 1.2	Analysis 2.4; Analysis 2.5; Analysis 2.6	Analysis 3.3; Analysis 3.4
Secondary outcome: change in corneal radius of curvature	Not reported by 2 studies and reported only as nonsignificant by Chung 2002	Analysis 2.7	Not reported
Adverse effects	Two participants who were undercorrected complained of blurred vision (Adler 2006)	Three participants using PALs in 1 study had conjunctivitis, distance blur, or dizziness (COMET2 Study 2011)	Participants reported blurred side vision, visual distortion, dizziness, headaches, and falls (Sankaridurg 2010)

^aCompared with fully corrected single vision lenses.

Table 3. Outcomes reported by studies of contact lens interventionsa

Outcomes	Interventions studied
Outcomes	Soft bifocal contact lenses: 4studies	Rigid gas permeable contact lenses: 2 studies	Orthokeratology: 3 studies	Spherical aberration soft contact lenses: 2 studies
Primary outcome: change in refractive error	Analysis 4.1	Analysis 5.1	No data for analysis	Data reported by both studies, but not meta‐analyzable
Secondary outcome: change in axial length	Analysis 4.2	Analysis 5.2	Analysis 6.1	Data reported by both studies, but not meta‐analyzable
Secondary outcome: change in corneal radius of curvature	Analysis 4.3	Analysis 5.3	No data for analysis	Not reported
Adverse effects	Six children in 1 study withdrew from the study, 3 from each group (CONTROL Study 2016)	Not reported	Adverse effects reported from all 3 studies	One study reported 1 child with allergic conjunctivitis and 1 with contact dermatitis

^aCompared with fully corrected single vision lenses or contact lenses.

Table 4. Outcomes reported by studies of pharmaceutical interventionsa

Outcomes	Interventions studied
Outcomes	Antimuscarinic agents: 6studies	Atropine vs tropicamide: 1study	Systemic adenosine antagonists: 1study	Timolol: 1 study	Tropicamide (plus bifocals): 1 study
Primary outcome: change in refractive error	Analysis 7.1; Analysis 7.2	Analysis 8.1; Analysis 8.2	Analysis 9.1	Analysis 10.1	Control twins showed more progression in myopia than their co‐twins who received tropicamide and bifocals, but this difference was not statistically significant (Schwartz 1981)
Secondary outcome: change in axial length	Analysis 7.3; Analysis 7.4	Not reported	Analysis 9.2	Not reported	Not reported
Secondary outcome: change in corneal radius of curvature	Not reported	Not reported	Analysis 9.3	Not reported	Not reported

^aCompared with placebo or no drops.

The most common methods of handling unit of analysis issues were to use the average of both eyes (15 studies); to use data from the right eye only (15 studies); and to use data from the eye with more severe myopia (one study) (Table 5). Nine studies were funded primarily by industry, were conducted by employees of the manufacturer of the intervention, or both (Anstice 2011; Cheng 2010; Cheng 2016; CONTROL Study 2016; Fujikado 2014; Hasebe 2014; PIR‐205 Study 2004; Tan 2005; Trier 2008). An additional 14 studies were funded partially by industry or received materials from the manufacturer (Adler 2006; ATOM Study 2006; Charm 2013; CLAMP Study 2004; COMET Study 2003; COMET2 Study 2011; Edwards 2002; Hasebe 2008; ROMIO Study 2012; Sankaridurg 2010; Schwartz 1981; STAMP Study 2012; Swarbrick 2015; Yang 2009).

See: Summary of findings 1 Interventions to slow progression of myopia in children

Table 5. Unit of analysis for included studies

Unit of analysis	Studies reporting each type of unit of analysis
Average of both eyes	15 studies: Adler 2006; Chung 2002; COMET Study 2003^a; COMET2 Study 2011; CONTROL Study 2016; Fujikado 2014; Fulk 1996; Fulk 2002; Hasebe 2008^a; PIR‐205 Study 2004; Sankaridurg 2010; Schwartz 1981; Shih 1999; Tan 2005; Trier 2008
Right eye only	15 studies: Cambridge Anti‐Myopia Study 2013; Charm 2013; Cheng 2010; Cheng 2016; CLAMP Study 2004; DISC Study 2011; Edwards 2002; Houston Study 1987; Katz 2003; Koomson 2016; MIT Study 2001; ROMIO Study 2012; STAMP Study 2012; Yen 1989; Yi 2015
Right and left eyes reported as separate analyses	2 studies: Jensen 1991; Pärssinen 1989
One study eye randomized and treated per child	1 study: ATOM Study 2006
Child randomized and both eyes analyzed as independent units	2 studies: Hasebe 2014; Lu 2015
Paired‐eye design	2 studies: Anstice 2011; Swarbrick 2015
Eye with more severe myopia	1 study: Wang 2017
Not reported	3 studies: Han 2018; Wang 2005; Yang 2009

^aAverage values of both eyes were used if the correlation coefficient was > 0.85 between eyes and the mean difference (MD) was not statistically significant; otherwise the eye with more myopic change was used for each child (COMET Study 2003). Mean of both eyes or of right eye only (Hasebe 2008).

Spectacles

Undercorrected versus fully corrected spectacles

Three studies compared the use of undercorrected spectacles versus fully corrected spectacles. In two studies, one in Israel and one in Ghana, children up to 15 years old were randomized to receive spectacles blurred by +0.50 D or spectacles with full correction (Adler 2006; Koomson 2016). In the third study, 106 Malay and Chinese children were evenly randomized to receive spectacles undercorrected by approximately +0.75 D or fully corrected spectacles (Chung 2002). Study follow‐up periods were 18 months in Adler 2006 and two years in Chung 2002 and Koomson 2016.

Multifocal versus single vision lenses

Fourteen studies included in the review compared multifocal spectacles versus single vision lenses (SVLs) (spectacles) for slowing progression of myopia in children: six used bifocal lenses (Cheng 2010; Fulk 1996; Fulk 2002; Houston Study 1987; Jensen 1991; Pärssinen 1989), and eight used progressive addition lenses (PALs) (COMET Study 2003; COMET2 Study 2011; Edwards 2002; Hasebe 2008; MIT Study 2001; STAMP Study 2012; Wang 2005; Yang 2009). All studies enrolled children from 6 to 15 years of age, used a plus addition lens from +1.00 D to +2.00 D, and had at least 18 months of follow‐up (maximum three years). All bifocal studies were conducted outside of Asia (Canada, Denmark, Finland, or USA), although the Canadian study included only children of Chinese ancestry (Cheng 2010); five PAL studies were conducted in Asia (China, Hong Kong, Japan, or Taiwan), and three in the USA (COMET Study 2003; COMET2 Study 2011; STAMP Study 2012).

Of the six bifocal studies, two were two‐arm trials that directly compared bifocal spectacles to SVLs for slowing the progression of myopia in children. One study, conducted in Tahlequah, Oklahoma, USA, randomized 32 children to receive bifocals with +1.25 D addition or SVLs (Fulk 1996). The children were 6 to 13 years old and were followed for 18 months. Following this pilot study, study authors initiated a larger study with slight modifications to the study design (Fulk 2002). For their second study, study authors added another study center in Tulsa, Oklahoma, USA; enrolled 82 children aged 6 to 12 years; changed the bifocal addition to +1.50 D; and extended the follow‐up period to 30 months.

The remaining four bifocal studies were three‐arm trials with at least one bifocal group and one SVL group. In the Houston Myopia Control Study (Houston Study 1987), 207 children ages 6 to 15 years were randomized to one of three treatment groups and were followed for three years. Treatment groups included two intervention groups that received bifocals with either +1.00 D or +2.00 D addition and a standard treatment group that received SVLs. A three‐arm trial including interventions of bifocals, timolol maleate, and SVLs was completed in Odense, Denmark (Jensen 1991). For two years, 159 schoolchildren with a mean age of 10.9 years were followed after they were randomized to one of three treatment groups. The bifocal group received bifocal lenses with +2.00 D addition for constant wear. The timolol group received one drop of 0.25% timolol maleate (an intraocular pressure [IOP]‐reducing beta‐blocker) in each eye twice daily in addition to SVLs for constant wear. The control group received only SVLs for constant wear. Another study compared the effects of bifocal lenses (+1.50 D) with or without three‐prism diopters of base‐in prism in the near segment with single vision distance lenses for slowing the progression of myopia over two years in 150 Chinese Canadian children (aged 8 to 13 years) (Cheng 2010). A study from central Finland enrolled myopic schoolchildren referred by local doctors and nurses after routine vision check‐ups (Pärssinen 1989). In all, 240 children with a mean age of 10.9 years were randomized to one of three treatment groups and were followed for three years. The first intervention group, the distant use group, received full myopic correction and were advised to use glasses for distance vision only and to read at the greatest distance possible. The second intervention group, the bifocal group, received bifocal lenses with +1.75 D addition for continuous use. The third group was the control group and received minus lenses with full correction for continuous use.

All eight PAL studies directly compared use of PALs (multifocal lenses with gradual and progressive changes in power) to SVLs. The three USA‐based studies used +2.00 addition PALs, and four of the five Asia‐based studies used +1.50 addition PALs (the fifth Asian study did not specify the addition power). The Correction of Myopia Evaluation Trial (COMET) was a three‐year, multicenter trial conducted in four major US cities (COMET Study 2003). In all, 469 children aged 6 to 11 years were randomized to receive either PALs or SVLs. The COMET 2 study was conducted to evaluate effectiveness in slowing myopia progression among children (n = 118) aged 8 to 11 years with low baseline myopia, high accommodative lag, and near esophoria (COMET2 Study 2011). Follow‐up was provided for three years. In the third USA‐based study, 85 children aged 6 to 11 years between ‐0.75D and ‐4.50 D of myopia, high accommodative lag, and near esophoria wore either PALs or SVLs for one year; all children wore SVLs in the second year of the study (STAMP Study 2012). A Japanese cross‐over trial followed up children aged 6 to 12 years for 18 months after randomization to PALs or SVLs (Hasebe 2008). After 18 months, each child was switched to receive the alternate type of lens and was followed up for another 18 months. The Myopia Intervention Trial (MIT) included 227 Taiwanese children and investigated SVLs, PALs, and PALs in combination with atropine drops for controlling the progression of myopia (MIT Study 2001). The children, who were between 6 and 13 years of age, were randomized to one of three treatment groups and were followed up for 18 months: (1) SVLs and placebo eye drops; (2) PALs and placebo eye drops; and (3) PALs and 0.5% atropine instilled once a day at bedtime. Studies of 298 children from 7 to 10.5 years of age and of 178 children from 7 to 13 years of age were completed in Hong Kong and China (Edwards 2002; Yang 2009), respectively. The children in both studies were randomized to receive PALs or SVLs and were followed up for two years. Finally, another Chinese study, reported only in the form of a conference abstract, enrolled 104 children aged 6 to 15 years; the addition power used in the PAL lenses was not reported (Wang 2005).

Peripheral plus spectacles versus single vision lenses

Four studies compared various types of peripheral plus spectacles versus SVLs (Han 2018; Hasebe 2014; Lu 2015; Sankaridurg 2010). Peripheral plus spectacles are designed to reduce peripheral hyperopic defocus (peripheral vision farsightedness). As such they consist of lenses that correct for central vision as SVLs do, as well as for peripheral vision using positively aspherized and increasing peripheral power. The addition of the peripheral plus spectacles in these three trials ranged from +1.00 D to +2.50 D. All trials were conducted in China (Hasebe 2014 was a multicenter trial with additional sites in Japan and South Korea) and enrolled children aged 6 to 14 years. Hasebe 2014 randomized 197 children to one of three treatment groups: peripheral plus spectacles with +1.00 D addition, peripheral plus spectacles with +1.50 D addition, and SVLs. Lu 2015 randomized 80 children to either peripheral plus spectacles with up to +2.50 D addition or SVLs. Sankaridurg 2010 randomized 210 children to lens designs that had (1) a symmetrical, clear central aperture (20 mm) with increasing peripheral power to +1.00 D; (2) a symmetrical, clear central aperture (14 mm) with increasing peripheral power to +2.00 D; (3) an asymmetrical, clear central aperture with increasing peripheral power to +1.90 D; or (4) SVLs. The study was planned for two years of follow‐up but was terminated at year one because the older age of participants resulted in slower than expected myopia progression among all study participants. Study follow‐up periods were one year in Lu 2015 and two years in Hasebe 2014. Finally, Han 2018 included 240 children who were randomized to (1) peripheral defocus‐reducing spectacles, (2) single vision lenses, or (3) orthokeratology lenses.

Contact lenses

Bifocal soft contact lenses versus single vision soft contact lenses

Four studies compared bifocal soft contact lenses (BSCLs) to single vision soft contact lenses (SVSCLs) for controlling myopia progression; one each was conducted in China (DISC Study 2011), Japan (Fujikado 2014), New Zealand (Anstice 2011), and the USA (CONTROL Study 2016). The age of children included in all trials ranged from 6 to 18 years. The addition powers for BSCLs ranged from +0.50 to +2.50 D across trials.

The New Zealand study was a paired‐eye, cross‐over study in which one eye of each child aged 11 to 14 years was randomized to receive +2.00 D addition BSCLs or SVSCLs. Fellow eyes received the other type of lens. After 10 months, the types of lenses worn in each eye were switched and children were followed for another 10 months. The Japanese study also was a cross‐over trial in which children aged 6 to 16 years were randomized to wear +0.50 D addition BSCLs or SVSCLs in both eyes for one year, then were switched to the other type of lens for the second year. The remaining two studies were parallel‐group trials. In the first, children aged 8 to 13 years wore either +2.50 D addition BSCLs or SVSCLs for two years. In the second parallel‐group study, 78 children from California, USA, ages 8 to 18 years with eso (convergent) fixation disparity, were randomized to wear BSCLs or SVSCLs every day for one year; the BSCL power prescribed was that needed to eliminate the child's eso fixation disparity while looking at near.

Rigid gas permeable contact lenses versus single vision lenses

Two studies included in the review compared rigid gas permeable contact lenses (RGPCLs) to either SVSCLs or spectacles (SVLs). The Contact Lens and Myopia Progression (CLAMP) study was a three‐year trial that compared RGPCLs to SVSCLs for controlling myopia in school‐aged children (CLAMP Study 2004). All participants had to complete a run‐in period successfully before enrollment to exclude those who could not adapt to wearing rigid contact lenses. After the run‐in period, 116 children aged 8 to 12 were randomized to RGPCL or SVSCL treatment groups. A study of 564 children aged 6 to 12 years in Singapore compared RGPCLs to SVL spectacles for controlling myopia over a two‐year period (Katz 2003). After a three‐month adaptation period, 383 participants remained in the study.

Orthokeratology contact lenses versus single vision lenses

Four studies investigated overnight orthokeratology contact lenses for controlling myopia progression. Studies enrolled children from 6 to 16 years of age with East Asian ethnicity; two studies were conducted in Hong Kong (Charm 2013; ROMIO Study 2012), one in China (Han 2018), and one in Australia (Swarbrick 2015). Charm 2013 evaluated high myopia (‐5.00 D or worse), whereas ROMIO Study 2012 and Swarbrick 2015 included children with low to moderate myopia (no worse than ‐4.50 D). Axial length was the primary outcome in three studies (Charm 2013; ROMIO Study 2012; Swarbrick 2015).

The two studies from Hong Kong compared orthokeratology contact lenses worn overnight versus single vision spectacles. In Charm 2013, 26 of 52 children randomized were assigned to wear partial reduction orthokeratology contact lenses (target 4.00 D) and single vision spectacles during the daytime if needed. In ROMIO Study 2012, 51 of 102 children randomized were assigned to wear orthokeratology contact lenses (target not reported). The Australian study was a paired‐eye, cross‐over study in which one eye of each child was randomized to wear orthokeratology contact lenses during the night and the other eye to wear an RGPCL during the day. After six months, the type of contact lens worn in each eye was switched and children were followed for another six months. In Han 2018, 90 of 240 children were randomized to wear orthokeratology lenses with a "four‐district seven‐arc reverse geometric design."

Spherical aberration soft contact lenses versus single vision soft contact lenses

Two studies compared SVSCLs with or without an additional design to alter spherical aberration. The Cambridge Anti‐Myopia Study 2013 was a 2 × 2 factorial trial testing a spherical aberration design and vision training against SVSCLs in 147 British participants aged 14 to 22 years. Although this trial included adults, most participants were 18 years of age or younger, thus we decided to include it in this review. We did not include in the analysis the two groups with vision training as an intervention because the review is limited to devices and pharmaceuticals. In Cheng 2016, 127 children aged 8 to 11 years were randomized to receive soft daily disposable contact lenses either with positive spherical aberration or without positive spherical aberration. The trial was conducted in the USA and enrolled mostly Asian children (91%). Both studies were planned for two years, but Cheng 2016 was stopped early and reported only one‐year data.

Pharmaceutical agents

Antimuscarinic agents versus placebo

Use of three different topical antimuscarinic agents was compared with use of placebo for control of myopia progression in six studies: three studies evaluated atropine eye drops (ATOM Study 2006; MIT Study 2001; Yi 2015); two evaluated pirenzepine gel (PIR‐205 Study 2004; Tan 2005); and one evaluated both atropine and cyclopentolate eye drops (Yen 1989). All studies were conducted in Asia, except for PIR‐205 Study 2004, which was conducted in the USA. Studies included children from 6 to 14 years of age at enrollment who had low to moderate myopia (up to ‐6.00 D). The atropine studies used one of two doses, 0.5% or 1.0%; the pirenzepine studies used a 2% gel formulation; and the cyclopentolate study used a dosage of 1%.

Two studies compared daily 1% atropine with placebo. The Atropine in the Treatment of Myopia (ATOM) Study enrolled 400 Singaporean children aged 6 to 12 years (ATOM Study 2006). Once each child was randomized to a treatment group, one eye of each child was randomized to receive treatment and the other eye served as a natural control. Follow‐up for this study was two years. In Yi 2015, 140 children with low myopia (‐0.50 D to ‐2.00 D) instilled either 1% atropine or placebo drops in both eyes every night for one year.

One study compared daily 0.5% atropine with placebo (Wang 2017); 126 children aged 5 to 10 years with myopia ranging from ‐0.5 D to ‐2.00 D were randomized to the two interventions. In both groups, the intervention was administered once daily at night for one year.

Two three‐arm studies investigated using atropine in conjunction with wearing multifocal lenses. The Myopia Intervention Trial, described above in the "Spectacles" section, evaluated SVLs, PALs, and PALs in combination with 0.5% atropine drops for controlling progression of myopia (MIT Study 2001). Yen 1989 randomized 247 Taiwanese children aged 6 to 14 years to one of three treatment groups: (1) 1% atropine eye drops every other night and bifocal spectacles prescribed after two weeks of treatment; (2) 1% cyclopentolate eye drops every night and SVLs prescribed if necessary; and (3) normal saline eye drops every night and SVLs as prescribed if necessary. Follow‐up was at one year for Yen 1989 and at 18 months for the MIT Study 2001.

Two studies compared pirenzepine gel (an antimuscarinic) to placebo gel for control of myopia progression. The first was a multicenter US study that enrolled 174 children 8 to 12 years old and followed them up for one year (PIR‐205 Study 2004). Children were randomized at a 2:1 ratio to 2% pirenzepine ophthalmic gel or placebo gel twice a day. An additional year of follow‐up was provided for children who completed the first year. The final study was a three‐arm, multicenter trial from Singapore, Hong Kong, and Thailand (Tan 2005). For one year, 353 children aged 6 to 13 years were randomly treated with (1) 2% pirenzepine gel applied twice daily (gel/gel); (2) placebo once daily and 2% pirenzepine gel once daily (placebo/gel); or (3) placebo gel twice daily (placebo/placebo).

Antimuscarinic agents versus tropicamide

One study, completed in Taiwan, investigated the effectiveness of low concentrations of atropine for controlling progression of myopia in children aged 6 to 13 years (Shih 1999). Two hundred children were randomized to one of three atropine groups or to a control group: (1) daily drop of 0.5% atropine and advised to wear bifocal spectacles; (2) daily drop of 0.25% atropine and advised to wear slightly undercorrected SVLs; (3) daily drop of 0.1% atropine and advised to wear fully corrective SVLs; or (4) daily drop of 0.5% tropicamide.

Systemic adenosine antagonists versus placebo

One study investigated the effectiveness of systemic 7‐mx, an adenosine receptor antagonist, for slowing axial elongation and thus controlling the progression of myopia (Trier 2008). In the first year of this Danish study, 83 children aged 8 to 13 years were randomized to take a 7‐mx or placebo tablet once daily. After one year, all children had the option to receive 7‐mx and to continue follow‐up for two more years.

Combinations of interventions

Tropicamide plus bifocal spectacles versus single vision lenses

In a study of 26 twin pairs, the combined use of bifocal lenses and tropicamide ophthalmic solution for controlling myopia progression was compared to the use of SVLs over a 3½‐year period (Schwartz 1981). This Washington DC area study included monozygotic twin pairs between the ages of 7 and 14 with similar myopia. From each twin pair, one twin was randomized to receive combined treatment of bifocal spectacles with a +1.25 D addition and two drops of 1% tropicamide ophthalmic solution (a short‐acting cycloplegic) instilled into each eye nightly; the other twin received a standard myopic spectacle correction.

Other combinations of interventions

The following combinations of interventions were compared by studies described in the previous sections.

Atropine plus multifocal spectacles versus placebo plus SVLs (MIT Study 2001 Yen 1989).
Atropine plus bifocal spectacles versus cyclopentolate plus SVLs (Yen 1989).
Bifocal spectacles versus SVLs with timolol eye drops (Jensen 1991).

Ongoing studies

We identified 74 ongoing studies, which are described under Characteristics of ongoing studies. These studies compared multifocal contact lenses, orthokeratology lenses, progressive addition lenses, spectacle lenses, full correction, and undercorrection. We will incorporate their findings in future updates of this review.

Excluded studies

We excluded 91 studies from this review after full‐text assessment. The complete list of studies and reasons for exclusion are provided in the Characteristics of excluded studies table. Our reasons for exclusion were based on four categories: (1) the study was not an RCT (58 studies); (2) study interventions were not intended to control myopia progression (13 studies); (3) study interventions were not within the scope of this review (13 studies); and (4) the study population was not eligible for this review (seven studies).

We excluded from this review two RCTs comparing SVSCLs with spectacles in myopic children and adolescents because SVSCLs and spectacles are not meant to control the progression of myopia (ACHIEVE Study 2008; Horner 1999). The purpose of the ACHIEVE study was to compare the effects of contact lens wear versus spectacle wear on children’s self‐perception.

Risk of bias in included studies

Allocation

Thirty‐three (81%) of the 41 included studies described the randomization procedure used to allocate participants to treatment groups; we judged them as having adequate sequence generation and thus low risk of allocation bias (Figure 2). Methods employed for adequate sequence generation included using block randomization schemes, computer‐generated randomization lists, or independently prepared randomization lists or tables, and flipping coins or drawing lots. Seven studies stated that children were randomized but did not report other details on how randomization was implemented, and we judged unclear risks of bias for sequence generation (Charm 2013; Cheng 2016; Lu 2015; Wang 2005; Yang 2009; Yi 2015; Han 2018). This review included RCTs only; however, we included one study that was reported to be an RCT but was confirmed to be a quasi‐randomized study based on information provided by the study author (Cheng 2010). The first 50 numbers pulled out were assigned to the control group, the second 50 to the bifocal group, and the remaining 50 to the bifocal plus prism group. This method of sequence generation was inadequate because participants did not have equal chances of being assigned to all treatment groups once the first 50 numbers were drawn. In addition, because treatment assignments were consecutive, allocation concealment was inadequate. We judged this study as having inadequate sequence generation and allocation concealment because treatment assignment was determined by selecting from a container pieces of paper with patient numbers written on them.

Figure 2

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

We judged 13 studies to have adequate allocation concealment. Methods considered to be at low risk of bias for this domain included using sequentially numbered, sealed, and opaque envelopes, and calling a centralized coordinating center. Twenty‐five studies did not provide sufficient information on whether and how allocation was concealed. These studies were judged to be unclear risk of bias for allocation concealment. Two studies stated that the person assigning participants to treatment groups was aware of the allocation sequence (ROMIO Study 2012; Cambridge Anti‐Myopia Study 2013), and they were judged as high risk of bias for this domain.

Masking (performance bias and detection bias)

We assessed the use of masking (blinding) for three types of roles: study participants, outcome assessors, and data analysts (Figure 2). Furthermore, we considered separately the masking of outcome assessors for primary (change in refractive error) and secondary (changes in axial length and corneal radius of curvature) outcomes.

Due to the interventions under investigation, masking of participants was not feasible for many of the studies included in this review. Interventions from 34 (83%) of the 41 included studies had significant physical (e.g. contact lenses vs spectacles), functional (e.g. multifocal lenses vs SVLs), or performance (e.g. undercorrected vs fully corrected spectacles) differences between control interventions. Despite these differences, six studies reported masking participants, but we judged them as having unclear risk of bias because it was not clear whether masking was effective (Cambridge Anti‐Myopia Study 2013; Cheng 2016; CONTROL Study 2016; DISC Study 2011; Hasebe 2014; Sankaridurg 2010). Of six studies evaluating pharmaceutical agents exclusively, four studies masked participants adequately by distributing identically packaged and coded bottles or tablets (ATOM Study 2006; PIR‐205 Study 2004; Tan 2005; Trier 2008), and two did not implement masking of participants (Shih 1999; Yi 2015).

Adequate methods of masking outcome assessors involved having participants examined by an investigator who was unaware of treatment assignments. This method was implemented for spectacle or contact lens studies by having participants remove contact lenses and spectacles before they were examined or distributing SVLs for all participants to wear during office visits. Use of coded, identical packaging was considered adequate masking for pharmaceutical studies. Overall, masking of primary outcome assessors was done for 28 (72%) of the 39 included studies. Of the 28 studies that masked primary outcome assessors, 25 were masked similarly for secondary outcome assessors and three did not measure secondary outcomes related to this review. ROMIO Study 2012 did not assess change in refractive error as an outcome but masked assessments for our review's secondary outcomes. We assessed Charm 2013 as having unclear risk of bias for not reporting masking of primary outcome assessment and low risk for masking of secondary outcomes.

We judged five included studies as being at high risk of bias for not masking primary outcome assessors adequately. In a three‐armed study comparing bifocal lenses or timolol with SVLs, there was only one study investigator, who therefore could not be masked to treatment assignments (Jensen 1991). Refractive errors for this study were measured by cycloplegic autorefraction. In another three‐armed trial comparing bifocals or distance‐use spectacles versus continuous‐use single vision spectacles, it was reported that the examining ophthalmologist did not look at group assignments before the examination, but often for different reasons group assignments were revealed (Pärssinen 1989). However, three‐year follow‐up examinations were conducted by two different ophthalmologists, one of whom did not know the group assignments. Refractive errors for this study were measured by subjective cycloplegic refraction. The Houston Myopia Control Study included a team of masked observers (evaluation team) and a team of unmasked observers (patient care team) to measure outcomes in a trial of bifocal lenses versus SVLs (Houston Study 1987). Results presented in the final analysis of the primary outcome were derived from the nonmasked group; therefore we judged the study as having inadequate masking of primary outcome assessors. Refractive errors for this study's results were measured by subjective noncycloplegic refraction. Two included studies did not attempt to mask primary outcome assessors; one measured refractive errors by cycloplegic autorefraction (Cheng 2010), and the other measured refractive errors by subjective cycloplegic refraction (Katz 2003). With the exception of the Houston study, secondary outcome assessors were the same as primary outcome assessors. Data for secondary outcomes in the Houston study were collected by the masked evaluation team; therefore we considered these studies to have low risk of bias.

Five studies did not report masking of primary or secondary outcome assessors; we judged them to have unclear risk of bias (Han 2018; Lu 2015; Swarbrick 2015; Wang 2005; Yen 1989).

The final assessments for masking were applied to study data analysts. How data were handled and whether or not data analysts were masked to treatment groups were not reported in 26 (63%) of the 41 included studies. Two studies explicitly stated that masked investigators analyzed the data independently (Edwards 2002; Hasebe 2008). Additionally, study authors contacted for clarification replied that data analysts were masked for Cheng 2010, MIT Study 2001, and Shih 1999. Although three studies stated that data were analyzed independently after the conclusion of the trial, we considered masking of data analysts to be unclear because treatment assignments may have been accessible in the data (Adler 2006; Chung 2002; Yang 2009). One study could not be masked because only one investigator was involved (Jensen 1991). Study authors for five studies informed us that data analysts were not masked (via email communications) (CLAMP Study 2004; COMET Study 2003; Katz 2003; PIR‐205 Study 2004; Tan 2005). We assessed studies in which data analysts were not masked or in which masking of data analysts was not reported to have unclear risk of bias for this parameter.

Incomplete outcome data

Attrition rates reported by the included studies varied from 0% to 61%. Seven studies followed the intention‐to‐treat (ITT) analysis as defined by this review: (1) participants were analyzed in the intervention groups to which they were randomized, regardless of the intervention they actually received; and (2) all randomized participants were included in the analysis, even participants for whom no outcome data were collected. Three studies provided follow‐up data for all participants at the final follow‐up visit (CLAMP Study 2004; Lu 2015; Han 2018), and four used statistical methods to account for all randomized patients by imputing values for missing data (COMET Study 2003; COMET2 Study 2011; Fulk 2002; Koomson 2016).

A total of 14 studies analyzed participants in the intervention groups to which they were randomized but did not include all randomized participants in the analysis due to attrition. In none of these studies were participants excluded from the analysis due to noncompliance, switching of intervention groups, or failure to adhere to treatment protocols. In 11 of these studies, outcome data were missing for both intervention groups but dropouts were balanced across groups and participants who dropped out were similar to those who remained (Anstice 2011; Chung 2002; CONTROL Study 2016; Edwards 2002; Fujikado 2014; Fulk 1996; Hasebe 2008; Pärssinen 1989; STAMP Study 2012; Trier 2008; Yi 2015). The attrition rate for each of these studies was between 6.5% and 15%. For these considerations, we judged these 11 studies as having low risk of bias due to minimal quantities of incomplete outcome data. The other three studies had unclear risk of attrition bias due to unbalanced dropout rates between treatment groups, or because there were statistically significant differences between participants who dropped out compared with those who remained in the study (ATOM Study 2006; Katz 2003; Yang 2009). One additional study included all randomized participants in the analyses but did not report the methods used to address missing data.

One study published only as an abstract reported the number of patients included in the analyses, but it is not clear whether this number represents the total number initially enrolled in the study and randomized to treatment (Wang 2005).

We judged the remaining 18 studies to have high risk of bias due to incomplete outcome data. The percentage of missing data ranged from 4% to 61%. In all of these studies, a proportion of participants were excluded after randomization for not adhering to the treatment protocol, for having adverse events, or for withdrawing consent. In a study evaluating undercorrected spectacles with full correction spectacles, participants were excluded for not wearing spectacles continuously (Adler 2006). Two studies comparing bifocal lenses with SVLs excluded children from the analysis: one study excluded children who were randomized to receive SVLs but dropped out because their parents wanted them to receive bifocals (Cheng 2010), and another study dismissed noncompliant patients and patients who were fitted with contact lenses without informing study personnel, along with patients who were fitted with contact lenses without informing study personnel (Houston Study 1987). Two studies that investigated peripheral plus spectacles excluded participants for committing protocol violations, withdrawing because of an adverse event, or withdrawing consent (Hasebe 2014; Sankaridurg 2010). One study of BSCLs versus SVSCLs excluded 42% of participants because they did not want to wear the contact lenses (DISC Study 2011). All three orthokeratology studies excluded participants due to issues with wearing the orthokeratology lenses (Charm 2013; ROMIO Study 2012; Swarbrick 2015). Both studies of spherical aberration soft contact lenses excluded 17%—Cheng 2016—or 33%—Cambridge Anti‐Myopia Study 2013—of participants from analyses. The other seven studies evaluated a pharmaceutical agent in at least one treatment arm (seven of the eight pharmaceutical studies included in this review). Five of these studies excluded participants for not using the eye drops or gel, or for not using them consistently (MIT Study 2001; PIR‐205 Study 2004; Shih 1999; Tan 2005; Yen 1989). Another study evaluating timolol plus SVLs versus bifocals or SVLs excluded patients for switching to contact lenses, or because they could not adapt to bifocal lenses (Jensen 1991). The last study, a co‐twin study in which one twin received bifocal spectacles and 1% tropicamide ophthalmic solution and the other twin received SVLs, excluded one twin pair from the study for noncompliance (Schwartz 1981).

In addition to excluding participants for nonadherence, two studies reported an imbalance in dropout rates (PIR‐205 Study 2004; Tan 2005). The PIR‐205 Study 2004 reported that there were significantly more dropouts in the pirenzepine arm compared with the placebo arm, although three additional analytical methods used to impute missing values for those who discontinued the study revealed similar or more beneficial treatment effects for pirenzepine compared with the analysis censoring dropouts. Tan 2005 also reported more dropouts in the pirenzepine‐treated groups than in the placebo‐only group. Although differences were not statistically significant, all participants who dropped out because of adverse events were included in the pirenzepine group.

Two studies excluded participants for lack of efficacy of treatment (MIT Study 2001; Tan 2005). The MIT study excluded two children for having myopic progression greater than 2.00 D per year: one child was from the SVL group and the other child was from the PAL group (none were from the atropine plus PAL group). One child was dropped from the placebo group of the Tan 2005 study for inadequate efficacy.

Finally, the study with the highest percentage of missing data enrolled 247 children, but data were missing for 151 (61%) children (Yen 1989). Reasons for missing data were not reported. Study authors stated that "patients who used the eye drops continuously for one year received another complete ophthalmologic examination," and "96 such patients were collected for evaluation, 32 in each group." It is not clear whether the 96 patients analyzed included all children who were examined at one year or a subset of those examined.

Selective reporting

We assessed 31 (76%) studies as having low risk of bias for selective reporting: 16 studies reported results for study outcomes defined a priori (i.e. in a design and methods publication, baseline report, or clinical trial registry), and 15 studies reported results for outcomes described in the methods section of each paper (Figure 2).

We assessed three studies as having unclear risk of reporting bias. One study had unclear risk of bias due to inadequate reporting for one of the two outcomes measured (Fulk 1996). For this study, the refractive error outcome was reported by treatment assignment; however, the axial length outcome was presented only as it correlated with myopia progression and results by treatment groups were not given. One study was planned for two years but was terminated early and reported results for only one year of follow‐up; 18.9% of children completed two years of follow‐up before the study was terminated early (Cheng 2016). Another study was published as a conference abstract only (Wang 2005).

We considered seven studies to have high risk of reporting bias. One study reported results for only one intervention group (Charm 2013). Another study stated in its methods section that results would be discussed only if they were exceptional (Jensen 1991). In one study, not all outcomes described in the methods section were reported (Yen 1989), and in another study, all outcomes identified in the study methods were reported but the numbers of participants included in the analyses were not consistent between outcomes (Katz 2003). Two studies showed differences between outcomes specified in the clinical trial registration and in the journal publication: one study switched a secondary outcome in the clinical trial registration record to the primary outcome of the journal publication (Fujikado 2014), and the other study did not report results for secondary outcomes listed in the clinical trial registration record (Swarbrick 2015). The final study did not report results for evaluation team (masked observers) measurements nor for other secondary outcomes outlined in the design paper (Houston Study 1987). The methods paper stated that an evaluation team report would be based on (1) cycloplegic retinoscopy, (2) noncycloplegic autorefraction, and (3) cycloplegic autorefraction performed by masked examiners. However, findings from masked examinations were not reported in the outcomes paper; instead results from the patient team were reported (unmasked observers). Also, secondary outcomes such as change in axial length were not reported.

Other potential sources of bias

We assessed 18 (44%) studies as free of other potential sources of bias, 13 (33%) studies as having unclear risk of bias, and 10 (26%) studies as having high risk of bias (Figure 2).

We assessed all four cross‐over trials as having high risk of bias because data were not analyzed according to the cross‐over design, carry‐over effects were not investigated, and some participants who completed the first period dropped out during the second period (Anstice 2011; Fujikado 2014; Hasebe 2008; Swarbrick 2015). Thus, we used only data from the first period to estimate treatment effects. Additionally, three of these studies reported financial conflicts of interest.

We assessed two additional studies as having high risk of bias due to unit of analysis issues (i.e. not accounting for the nonindependence of eyes) (Hasebe 2014; Lu 2015), and two studies as having unclear risk of bias for not reporting the unit of analysis (Wang 2005; Yang 2009).

Two studies planned for two years' duration were terminated early. We assessed one study as having high risk of bias because it was the funder's decision to stop early based on its own interests (Cheng 2016), and the other study as having unclear risk of bias because the decision to stop was made by investigators because they observed lower than expected progression in myopia among all study participants (Sankaridurg 2010).

One study reported imbalances at baseline between treatment groups in gender, corneal curvature, and refractive error (Katz 2003). This study reported unequal losses to follow‐up between treatment groups and by gender. In addition to these considerations, and because 32% of participants dropped out of the study between randomization and the end of the adaptation period, we judged this study to have high risk of bias. Two other studies incorporated prerandomization administration of an intervention into their study design (ATOM Study 2006; CLAMP Study 2004). Run‐in periods may enhance or diminish the effects of a subsequent randomized intervention; thus we assessed these studies as having unclear risk of bias.

The remaining two of 10 studies that we judged to be at high risk of other sources of bias were Shih 1999 and Yen 1989. In Shih 1999, participants in different treatment groups were advised to wear different types of spectacle lenses depending on the concentration of atropine received. The rationale for recommending different types of spectacles for different atropine doses (bifocals in the 0.5% group; undercorrected lenses in the 0.25% group; and fully corrective lenses in the 0.1% group) was not explained. In Yen 1989, it was unknown why equal numbers of participants dropped out of each group, or how equal numbers of participants per group were selected for analysis ("96 such patients were collected for evaluation, 32 in each group").

Eight other studies were fully or partially funded by companies with financial interests in at least one of the interventions being studied. We considered these studies to have unclear risk of bias (Cheng 2010; COMET Study 2003; CONTROL Study 2016; PIR‐205 Study 2004; ROMIO Study 2012; STAMP Study 2012; Tan 2005; Trier 2008).

Effects of interventions

We compared several interventions to SVLs (spectacles) or SCLs to determine which treatments are effective in slowing the progression of myopia in children. We meta‐analyzed results for prespecified outcomes when appropriate; otherwise we reported study‐specific results. For the primary outcome of this review, progression of myopia assessed as the mean change in refractive error (spherical equivalent) from baseline for each year of follow‐up, negative mean differences (MDs) represented faster progression of myopia in the treatment group compared with progression in the control group. Thus, point estimates to the left of null on the forest plots favor the control group for this outcome. For axial length, negative MDs represent less axial elongation for treatment group participants compared with control group participants (point estimates to the left of null on the forest plots favor the treatment group for this outcome). The unit of analysis reported by each study is shown in Table 5.

Spectacles

1. Undercorrected versus fully corrected spectacles

Three studies with a total of 292 participants compared spectacles that undercorrected myopia by approximately ‐0.50 to ‐0.75 D with SVLs that fully corrected myopia (Adler 2006; Chung 2002; Koomson 2016).

Change in refractive error (Analysis 1.1)

At one year, two studies reported that 72 children who were undercorrected progressed, on average, by ‐0.15 D (95% CI ‐0.29 to 0.00) more than the 70 SVL wearers (Figure 3). At two years, progression of myopia from baseline for the undercorrection group was 0.02 D (95% CI ‐0.05 to 0.09) compared with the full correction group in two studies with 244 children. We graded the certainty of evidence for refractive error as low, downgrading for imprecision (‐1) and risk of bias (‐1).

Figure 3

$Forest plot of comparison: 1 Undercorrection vs full correction spectacles, outcome: 1.2 Change in refractive error from baseline (1 year).$

Forest plot of comparison: 1 Undercorrection vs full correction spectacles, outcome: 1.2 Change in refractive error from baseline (1 year).

Change in axial length (Analysis 1.2)

Changes in axial length were measured by Chung 2002 and Koomson 2016. The undercorrected group showed greater axial elongation than the fully corrected group in one study at one year (MD 0.05 mm, 95% CI ‐0.01 to 0.11), and the studies showed no difference at two years (MD ‐0.01 mm, 95% CI ‐0.06 to 0.03). We graded the certainty of evidence for axial length as low, downgrading for imprecision (‐1) and risk of bias (‐1).

Change in corneal radius of curvature

Changes in corneal radius of curvature were not measured by Adler 2006 nor Koomson 2016, and were reported to be statistically nonsignificant during the two‐year study by Chung 2002.

Adverse effects

Two participants who were undercorrected complained of blurred vision in the study by Adler 2006. No other adverse effects were reported by any study.

2. Multifocal spectacles versus single vision lens spectacles

Fourteen studies compared multifocal spectacles versus single vision lens spectacles to slow the progression of myopia in children. Six studies evaluated bifocal lenses (Cheng 2010; Fulk 1996; Fulk 2002; Houston Study 1987; Jensen 1991; Pärssinen 1989), and eight studies used progressive addition lenses (PALs) (COMET Study 2003; COMET2 Study 2011; Edwards 2002; Hasebe 2008; MIT Study 2001; STAMP Study 2012; Wang 2005; Yang 2009). Ten studies were included in the quantitative analysis, and four studies did not provide adequate data for meta‐analysis: two studies did not report data for each year of follow‐up (Hasebe 2008; Wang 2005), and two studies reported outcomes as rates of change per year based on varying follow‐up times (Fulk 1996; Houston Study 1987). Of the 10 studies that we analyzed quantitatively, eight reported mean changes from baseline and two reported only final values (Edwards 2002; MIT Study 2001). Because the studies were randomized with no significant imbalance in potential confounders between groups at baseline, we combined MDs based on changes from baseline with MDs based on final measurements, given the assumption that these measures address the same underlying between‐group effects. With the exception of Pärssinen 1989, which measured refractive error by subjective cycloplegic refraction, studies included in the analysis used cycloplegic autorefraction for refraction measurements. We included Cheng 2010 in the review following full‐text assessment, but we subsequently classified it as not adequately randomized; we examined the impact of excluding this study from meta‐analysis by performing a sensitivity analysis when appropriate.

Change in refractive error (Analysis 2.1; Analysis 2.2; Analysis 2.3)

At one‐year follow‐up, the average progression was 0.14 D slower (95% CI 0.08 to 0.21; I² = 40%) for 729 multifocal (+1.50 D to +2.00 D near addition) spectacle wearers than for 734 SVL wearers in nine studies (Figure 4). The effect, from subgroup analysis based on type of lens, was similar among PAL wearers (MD 0.15 D, 95% CI 0.09 to 0.21; five studies) and bifocal lens wearers (MD 0.16 D, 95% CI 0.01 to 0.32; four studies). One study with quantitative data did not report data at one year (Yang 2009). Excluding from the analysis the two studies with MDs based on final values did not influence the results substantively (MD 0.14 D, 95% CI 0.07 to 0.22). Excluding Pärssinen 1989, which used subjective refraction, from the analysis did not influence the result substantively (MD 0.16 D, 95% CI 0.10 to 0.23). Excluding Cheng 2010, which was not randomized adequately, from the analysis did not influence the overall result substantively (MD 0.13 D, 95% CI 0.08 to 0.18); however, when Cheng 2010 was excluded from the analysis, the I² was reduced from 40% to 0%.

Figure 4

$Forest plot of comparison: 2 Multifocal lenses vs single vision lenses, outcome: 2.1 Change in refractive error from baseline (1 year).$

Forest plot of comparison: 2 Multifocal lenses vs single vision lenses, outcome: 2.1 Change in refractive error from baseline (1 year).

Eight of the ten studies with meta‐analyzable data followed up participants for at least two years; of these, four evaluated bifocal lenses and four evaluated PALs. At two years, average progression was 0.19 D slower (95% CI 0.08 to 0.30; I² = 55%) for 696 multifocal (+1.50 to +2.00 near addition) spectacle wearers than for 705 SVL wearers. Excluding from the analysis Pärssinen 1989, which used subjective refraction and was the only study that favored SVLs, did not influence the result substantively (MD 0.22 D, 95% CI 0.15 to 0.29); however, when Pärssinen 1989 was excluded from the analysis, the I² was reduced from 55% to 0%.

Three of the ten studies with quantitative data followed up participants for three years (COMET Study 2003; COMET2 Study 2011; Pärssinen 1989). We did not combine these studies in an overall meta‐analysis due to statistical heterogeneity (I² = 84.4%). For the PAL subgroup, average progression was 0.21 D slower (95% CI 0.08 to 0.34; I² = 0%) for 287 multifocal (+2.00 D near addition) spectacle wearers than for 292 SVL wearers. Pärssinen 1989 reported a nonsignificant MD in the opposite direction for bifocal wearers compared with SVL wearers (MD ‐0.19, 95% CI ‐0.47 to 0.09).

Four studies not included in the meta‐analyses showed mostly favorable effects of multifocal lenses for slowing myopia progression. In a cross‐over study of +1.50 D PALs versus SVLs, children wearing PALs during the first 18‐month treatment period showed significantly less progression than children wearing SVLs (MD 0.31 D, 95% CI 0.11 to 0.51); however, no difference between groups was evident for the second 18‐month period (MD 0.02 D, 95% CI ‐0.17 to 0.21) (Hasebe 2008). In an 18‐month study reported only by a conference abstract, children wearing PALs showed significantly less progression than children wearing SVLs (MD 0.39 D, 95% CI 0.21 to 0.57; 104 children) (Wang 2005). In another 18‐month study of 14 children assigned to wear +1.25 D bifocal lenses and 14 children assigned to wear SVLs, bifocal wearers progressed by ‐0.39 D per year and SVL wearers progressed by ‐0.57 D per year (P = 0.26) (Fulk 1996). Trial authors noted that during the first year of the study, the rate of progression was equal between groups, but during the last six months of the study, the SVL group progressed more rapidly than the bifocal group. In a three‐arm trial of +1.00 D bifocals, +2.00 D bifocals, and SVLs, no significant differences were observed between groups after three years of follow‐up (Houston Study 1987). The reported average change in refraction error per year during the three‐year study for +1.00 D bifocals was ‐0.36 D per year (n = 41), for +2.00 D bifocals was ‐0.32 D per year (n = 44), and for SVLs was ‐0.34 D per year (n = 39).

We graded the certainty of evidence for refractive error as moderate, downgrading for risk of bias (‐1).

Change in axial length (Analysis 2.4; Analysis 2.5; Analysis 2.6)

Eight studies reported axial length outcomes, four of which reported results for one‐year follow‐up (Cheng 2010; COMET Study 2003; Edwards 2002; STAMP Study 2012). At one year, the summary MD was ‐0.06 mm (95% CI ‐0.09 to ‐0.04) for 445 PAL wearers compared with 451 SVL wearers. This was similar to the summary results after two years of follow‐up (MD ‐0.05 mm, 95% CI ‐0.10 to ‐0.01) for two of these studies (COMET Study 2003; Edwards 2002). After three years of follow‐up, participants in the COMET study wearing PALs continued to have less axial elongation compared with participants wearing SVLs (MD ‐0.11 mm, 95% CI ‐0.17 to ‐0.05).

The four studies that did not report one‐year data showed treatment effects in the same direction as the studies included in the meta‐analysis, although results were not significant in two studies. In a three‐arm trial of PALs with or without atropine compared with SVLs, a pairwise comparison showed that participants who wore PALs without atropine had on average 0.10 mm (95% CI 0.00 to 0.20) less axial elongation compared with participants who wore SVLs at 18‐month follow‐up (MIT Study 2001). In an 18‐month study reported only by a conference abstract, children wearing PALs showed significantly less axial elongation than children wearing SVLs (MD ‐0.21 D, 95% CI ‐0.34 to ‐0.08; 104 children) (Wang 2005). In the cross‐over trial Hasebe 2008, axial length was not measured at baseline; however, there was no significant difference in axial length between groups after the first 18‐month study period (MD ‐0.08 mm, 95% CI ‐0.41 to 0.25), and no significant change in axial length was reported between groups after the second 18‐month study period (MD ‐0.01 mm, 95% CI ‐0.09 to 0.07). In the third study, changes in axial length were not significantly different between bifocal wearers and SVL wearers after 30 months of follow‐up (MD ‐0.09 mm, 95% CI ‐0.24 to 0.06) (Fulk 2002).

We graded the certainty of evidence for axial length as moderate, downgrading for risk of bias (‐1).

Change in corneal radius of curvature (Analysis 2.7)

Changes in corneal radius of curvature outcomes were reported in four studies. Two studies stated only that differences were not significantly different between treatment and control groups (Edwards 2002; Hasebe 2008). In the COMET Study 2003, neither horizontal measurements nor vertical measurements differed between groups after three years of follow‐up (MD 0.00 D, 95% CI ‐0.15 to 0.15; MD 0.00 D, 95% CI ‐0.14 to 0.14, respectively). In an 18‐month study reported only by a conference abstract, children wearing PALs showed significantly greater change in horizontal corneal curvature when compared with children wearing SVLs (MD 0.03 D, 95% CI 0.01 to 0.05; 104 children) (Wang 2005).

Adverse effects

Only one study that compared multifocal lenses with SVLs reported adverse effects (COMET2 Study 2011). Three adverse effects were reported in the PAL group (one each of conjunctivitis, distance blur, and dizziness) and 14 in the SVL group (nine cases of dizziness, three of reduced visual acuity, and one each of floaters and eye pain).

3. Peripheral plus spectacles versus single vision lens spectacles

Three studies compared peripheral plus spectacles versus SVLs (Hasebe 2014; Lu 2015; Sankaridurg 2010). An additional study compared peripheral defocus‐reducing spectacles versus SVLs. Three studies reported outcomes at only one‐year follow‐up—Han 2018 Lu 2015 Sankaridurg 2010—and one at only two‐year follow‐up—Hasebe 2014. All but one study reported using cycloplegic autorefraction; Han 2018 did not specify the method of measurement of refractive error. Due to differences in lens design and in follow‐up across studies, we did not combine individual trial data in the meta‐analysis.

Change in refractive error (Analysis 3.1; Analysis 3.2)

At one year, there were no significant differences in myopia progression among three peripheral plus lens types when compared with each other or with SVLs, as reported by Sankaridurg 2010. At one year, Lu 2015 reported a nearly 1.00‐D difference between the peripheral plus group (mean change from baseline ‐0.35 D; SD 0.32; 80 eyes of 40 children) and the SVL group (mean change from baseline ‐1.32 D; SD 0.24; 80 eyes of 40 children) (MD 0.97, 95% CI 0.88 to 1.06). At two years, Hasebe 2014 reported similar mean changes from baseline for the peripheral plus +1.00 D group compared with the SVL group (MD 0.06, 95% CI ‐0.09 to 0.21) but less myopia progression for the peripheral plus +1.50 D group compared with the SVL group (MD 0.19, 95% CI 0.05 to 0.33). At one year, Han 2018 reported estimates of change from baseline within groups, at ‐0.43 (SD 0.14) with peripheral defocus‐reducing spectacles and ‐1.15 (SD 0.46) with SVLs. We graded the certainty of evidence for refractive error as low, downgrading for inconsistency (‐1) and risk of bias (‐1).

Change in axial length (Analysis 3.3; Analysis 3.4)

At one year, there were no significant differences in axial elongation among three peripheral plus lens types when compared with each other or with SVLs, as reported by Sankaridurg 2010, or between peripheral plus lenses compared with SVLs, as reported by Lu 2015 (MD 0.03, 95% CI ‐0.15 to 0.21). At two years, Hasebe 2014 reported similar mean changes from baseline for both the peripheral plus +1.00 D compared with SVL group (MD ‐0.05, 95% CI ‐0.16 to 0.06) and the peripheral plus +1.50 D compared with SVL group (MD ‐0.08, 95% CI ‐0.19 to 0.03). We graded the certainty of evidence for axial length as low, downgrading for inconsistency (‐1) and risk of bias (‐1).

Change in corneal radius of curvature

Corneal radius of curvature was not assessed by Hasebe 2014 Lu 2015 Sankaridurg 2010, or Han 2018.

Adverse effects

Sankaridurg 2010 conducted telephone questionnaires at one week post distribution of lenses. At this time, 2/50 participants in the type I group, 2/60 participants in the type II group, 5/50 participants in the type III group, and 1/50 participants in the SVL group reported blurred side vision. Three participants reported visual distortion—one in the type I group and two in the SVL group. Two participants in the type II group experienced dizziness; for one participant, dizziness resolved after one month; for the other participant, dizziness was accompanied by headaches causing the participant to withdraw from the study. Two falls were reported during the study period; both occurred in the type II lens group during the first weeks of the study. Hasebe 2014 reported that "no serious adverse events were reported during the 2‐year follow‐up," and that "children generally recognized the usability of wearing the study glasses as good (score 4) or very good (5). There was no difference in the median score in any of the questions among the study groups." Lu 2015 did not report adverse effects as an outcome.

Contact lenses

4. Bifocal soft contact lenses versus single vision soft contact lenses

Four studies compared bifocal soft contact lenses (BSCLs) versus single vision soft contact lenses (SVSCLs) (Anstice 2011; CONTROL Study 2016; DISC Study 2011; Fujikado 2014). Anstice 2011 was a paired‐eye, cross‐over study with two 10‐month periods, and Fujikado 2014 was a cross‐over study with two 12‐month periods. Because data were not reported appropriately for within‐person or cross‐over designs, we included data for only the first phase of each trial, considered as one‐year follow‐up, and acknowledged that between‐group estimates did not account for intraperson correlations for the paired‐eye study. CONTROL Study 2016 followed children with myopia and eso fixation disparity at near vision for one year, and DISC Study 2011 followed children for two years. All trials used cycloplegic autorefraction.

Change in refractive error (Analysis 4.1)

At one year, myopia in the BSCL group (n = 149) progressed slightly slower than in the SVSCL group (n = 151) (MD 0.20 D, 95% CI ‐0.06 to 0.47; I² = 86%). Only one trial assessed change in refractive error at two years; DISC Study 2011 reported a similar effect between the BSCL group (n = 65) and the SVSCL group (n = 63) (MD 0.20 D, 95% CI 0.02 to 0.38). We graded the certainty of evidence for refractive error as low, downgrading for inconsistency (‐1) and risk of bias (‐1).

Change in axial length (Analysis 4.2)

At one year, axial elongation in the BSCL group was significantly less than in the SVSCL group (MD ‐0.11 mm, 95% CI ‐0.14 to ‐0.08; I² = 67%). At two years, DISC Study 2011 reported a similar effect between the BSCL group (n = 65) and the SVSCL group (n = 63) (MD ‐0.12 mm, 95% CI ‐0.20 to ‐0.04). We graded the certainty of evidence for axial length as low, downgrading for inconsistency (‐1) and risk of bias (‐1).

Change in corneal radius of curvature (Analysis 4.3)

At one year, in CONTROL Study 2016, the corneal radius of curvature in the BSCL group showed similar change compared with the SVSCL group (MD ‐0.05 D, 95% CI ‐0.15 to 0.05). Corneal radius of curvature was not assessed by the other three trials.

Adverse effects

Six (15%) of 40 children in the Anstice 2011 study—three from each group—did not complete follow‐up; four children withdrew due to difficulties handling the contact lenses, one due to negative publicity on contact lens solutions, and one due to dislike of cycloplegia. The other three studies did not report any adverse effect.

5. Rigid gas permeable contact lenses versus spectacles or soft contact lenses

Two studies investigated the use of rigid gas permeable contact lenses (RGPCLs) in slowing the progression of myopia in children. RGPCLs were compared with soft contact lenses (SCLs) in one study (CLAMP Study 2004), and they were compared with SVLs in the other group (Katz 2003). The CLAMP Study 2004 followed up participants for three years, and the Katz 2003 study followed up participants for two years. We assessed the CLAMP Study 2004 as having generally low risk of bias, and the Katz 2003 study as having generally high risk of bias.

Change in refractive error (Analysis 5.1)

The CLAMP Study 2004 evaluated the use of RGPCLs to slow the progression of myopia in children compared with SCLs. At one (MD 0.40 D, 95% CI 0.19 to 0.61), two (MD 0.54 D, 95% CI 0.27 to 0.81), and three (MD 0.63 D, 95% CI 0.30 to 0.96) years of follow‐up, participants wearing RGPCLs had significantly less progression of myopia compared with participants wearing SCLs (Figure 5). After one year and two years of follow‐up, no difference in myopia progression was observed between RGPCL wearers and SVL wearers in Katz 2003 (MD ‐0.02, 95% CI ‐0.14 to 0.10; MD ‐0.05 D, 95% CI ‐0.25 to 0.15, respectively). Data were not pooled for these studies due to statistical heterogeneity (I² = 91% at one year and 92% at two years). We graded the certainty of evidence for refractive error as very low, downgrading for imprecision (‐1), inconsistency (‐1), and risk of bias (‐1).

Figure 5

$Forest plot of comparison: 5 Rigid gas permeable contact lenses vs control, outcome: 5.1 Change in refractive error from baseline [D].$

Forest plot of comparison: 5 Rigid gas permeable contact lenses vs control, outcome: 5.1 Change in refractive error from baseline [D].

Change in axial length (Analysis 5.2)

At one year, meta‐analysis of the two studies showed that axial elongation was 0.02 mm (95% CI ‐0.05 to 0.10) greater for the 176 RGPCL wearers than for 239 control participants. After two years, it was 0.03 mm greater (95% CI ‐0.05 to 0.12) for the 154 RGPCL wearers than for 240 control participants who were followed up by the two studies. After three years, it was 0.05 mm greater (95% CI ‐0.12 to 0.22) for the 59 RGPCL wearers than for 57 SCL participants in the CLAMP Study 2004. We graded the certainty of evidence for axial length as low, downgrading for impression (‐1) and risk of bias (‐1).

Change in corneal radius of curvature (Analysis 5.3)

Data from the CLAMP Study 2004 suggest that use of RGPCLs may prevent increases in the corneal radius of curvature compared with SCLs. At one, two, and three years of follow‐up, the MD between participants wearing RGPCLs and participants wearing SCLs was ‐0.24 D (95% CI ‐0.43 to ‐0.05), ‐0.38 D (95% CI ‐0.56 to ‐0.20), and ‐0.26 D (95% CI ‐0.48 to ‐0.04), respectively. After one year of follow‐up, Katz 2003 also suggested that RGPCLs may be beneficial compared with SCLs (MD ‐0.08 D, 95% CI ‐0.14 to ‐0.01); however, these results were not statistically significant after two years of follow‐up (MD ‐0.06 D, 95% CI ‐0.14 to 0.02). Data were not pooled for these studies due to statistical heterogeneity (I² = 60% for year one results and 90% for year two results).

Adverse effects

None were reported.

6a. Orthokeratology contact lenses versus single vision lenses

Four studies investigated orthokeratology contact lenses to slow the progression of myopia. Charm 2013 and ROMIO Study 2012 followed up participants wearing either orthokeratology contact lenses or SVLs for two years. Swarbrick 2015 compared orthokeratology contact lenses with RGPCLs in a paired‐eye, cross‐over study with two six‐month periods. Analysis for this study did not account for the within‐person, cross‐over design of the study; therefore no data were included in any meta‐analysis. While three studies had overall high risk of bias, one study had unclear risk of bias (Han 2018).

Change in refractive error

Because orthokeratology contact lenses temporarily reduce myopia, their myopia control treatment effect can be measured only by axial elongation. We did not analyze changes in refractive error for this comparison.

Change in axial length

Three of the four studies reported axial length as an outcome; however, as with outcomes of refractive error, Swarbrick 2015 did not report data eligible for analysis. A fourth study did not report data on axial length (Han 2018).

Summary estimates suggest that the change in axial length is significantly different in favor of the orthokeratology group at two‐year follow‐up (MD ‐0.28, 95% CI ‐0.38 to ‐0.19; Figure 6; Analysis 6.1).

Figure 6

Forest plot of comparison: 6 Orthokeratology contact lenses versus single vision lenses, outcome: 6.1 Change in axial length from baseline (2 years).

We graded the certainty of evidence for axial length as moderate, downgrading for risk of bias (‐1).

Change in corneal radius of curvature

Neither Charm 2013 nor ROMIO Study 2012 reported outcomes related to changes in corneal radius of curvature; Swarbrick 2015 reported inconsistent findings at the end of each six‐month cross‐over period.

Adverse effects

All three studies reported adverse effects; however, these effects were different across studies. Of 26 participants assigned to wear orthokeratology contact lenses in Charm 2013 five withdrew from the study because they could not achieve a proper lens fitting despite repeated modifications, one withdrew due to lens discomfort, and another withdrew due to not wearing the lenses as instructed. Of the remaining 19 participants, six reported issues with lens binding at the beginning of the study and six showed pigmented corneal arc formation at one‐month follow‐up. Additionally, all participants wore SVLs to correct for residual refractive error during the daytime.

In ROMIO Study 2012, mild rhinitis (3/51 participants), increased conjunctival hyperemia (1/51 participants), and chalazion (1/51 participants) were observed in the orthokeratology group and one case of recurrent corneal inflammation was observed in the SVL group during two years of follow‐up.

Swarbrick 2015 reported that one of 26 eyes in the orthokeratology group had lens adherence.

6b. Spherical aberration soft contact lenses versus single vision soft contact lenses

Two studies investigated spherical aberration soft contact lenses to slow the progression of myopia. The Cambridge Anti‐Myopia Study 2013 followed up participants for two years. Quantitative results for this study were reported for the combined cohort of participants (not by treatment group) or were graphically represented in figures only. Thus, we were not able to calculate between‐group effects for this study but instead describe the results as available. Cheng 2016 planned to follow participants for two years but ended the trial early and reported results for one year only. We identified issues impacting risk of bias in both studies.

Change in refractive error

Neither study reported clinically meaningful differences between treatment groups. Cheng 2016 reported that the least squares mean (LSM) difference in the mean change in refractive error was 0.137 D (95% CI ‐0.007 to 0.281) among 52 children in the spherical aberration group compared with 57 children in the SVSCL group at one‐year follow‐up. The Cambridge Anti‐Myopia Study 2013 reported, "The mean progression was found to be 0.33 Dioptres (D) over the 2 years of the study," and "There was no significant treatment effect of either Vision Training or Contact Lens Spherical Aberration control on myopia progression." We graded the certainty of evidence for refractive error as low, downgrading for risk of bias (‐1) and indirectness (‐1).

Change in axial length

At one year, Cheng 2016 reported that axial elongation was 0.143 mm (95% CI ‐0.188 to ‐0.098) less for children in the spherical aberration group compared with control participants. The Cambridge Anti‐Myopia Study 2013 reported, "Axial length increased steadily over the 2 years of the study by 0.15 mm (SD 0.14) in both right and left eyes," and "There were no significant differences between axial length increases in the different groups." We graded the certainty of evidence for axial length as very low, downgrading for imprecision (‐1), risk of bias (‐1), and indirectness (‐1).

Change in corneal radius of curvature

Corneal radius of curvature was not assessed by Cambridge Anti‐Myopia Study 2013 nor by Cheng 2016.

Adverse effects

Adverse effects were not reported by Cambridge Anti‐Myopia Study 2013. Cheng 2016 reported that one child in the spherical aberration group had allergic conjunctivitis and one child in the control group had contact dermatitis.

Pharmaceutical agents

7. Antimuscarinic agents versus placebo

Six studies assessed topical antimuscarinic agents for slowing the progression of myopia in children. Three studies evaluated an atropine ophthalmic solution, one at 0.5% (MIT Study 2001), and two at 1% (ATOM Study 2006; Yi 2015), versus placebo; two studies evaluated 2% pirenzepine gel versus placebo (PIR‐205 Study 2004; Tan 2005); and one study evaluated 1% cyclopentolate ophthalmic solution versus placebo (Yen 1989). Study participants included in these analyses from the MIT Study 2001 also were provided with PALs. With the exception of Yen 1989, which measured refractive error by subjective cycloplegic refraction, these studies used cycloplegic autorefraction for refraction measurements. We did not include data from one study because it did not report data eligible for meta‐analysis. Specifically, estimates of changes from baseline reported in the paper were of the same magnitude as those reported at baseline, suggesting that they were not, in fact, changes from baseline (Wang 2017).

Change in refractive error (Analysis 7.1; Analysis 7.2)

Due to statistical heterogeneity (I² = 98%), we did not combine results for all antimuscarinic agents but instead pooled the subgroups separately. At one‐year follow‐up, average progression was 1.00 D slower (95% CI 0.93 to 1.07) for participants treated with atropine, 0.31 D slower (95% CI 0.17 to 0.44) for participants treated with pirenzepine, and 0.34 D slower (95% CI 0.08 to 0.60) for participants treated with cyclopentolate (Figure 7). The difference in progression between groups continued among participants in the two studies with two years of follow‐up (MD 0.92 D, 95% CI 0.75 to 1.09 for atropine; ATOM Study 2006 MD 0.41 D, 95% CI 0.13 to 0.69 for pirenzepine; PIR‐205 Study 2004). We graded the certainty of evidence for refractive error as moderate, downgrading for risk of bias (‐1).

Figure 7

$Forest plot of comparison: 6 Antimuscarinic agents vs placebo, outcome: 6.1 Change in refractive error from baseline (1 year).$

Forest plot of comparison: 6 Antimuscarinic agents vs placebo, outcome: 6.1 Change in refractive error from baseline (1 year).

Change in axial length (Analysis 7.3; Analysis 7.4)

Four studies reported axial length outcomes; however, we did not combine results for all antimuscarinic agents due to statistical heterogeneity (I² = 99%). At one‐year follow‐up, the two atropine studies—ATOM Study 2006 and Yi 2015—reported significantly less axial elongation for participants assigned to atropine than for participants assigned to placebo (MD ‐0.35 mm, 95% CI ‐0.38 to ‐0.31). This effect persisted at the end of two years (MD ‐0.40 mm, 95% CI ‐0.48 to ‐0.32) in the ATOM Study 2006. In the pirenzepine gel studies, Tan 2005 reported that after one year, the mean increase in axial length was greatest in the placebo/placebo‐treated group (0.33 mm) than in the placebo/gel (0.30 mm) and gel/gel (0.20 mm) groups. Although standard deviations (SDs) for mean changes in axial length were shown only on a graph, the paper reported that there was a statistically significant treatment effect at one year (repeated‐measures analysis of variance; P = 0.008). No significant changes in axial length were observed at one year in the PIR‐205 Study 2004 (MD ‐0.04 mm, 95% CI ‐0.15 to 0.07). We graded the certainty of evidence for axial length as moderate, downgrading for inconsistency (‐1).

Change in corneal radius of curvature

Corneal radius of curvature outcomes were not assessed by studies comparing topical antimuscarinic agents versus placebo.

Adverse effects (Analysis 7.5)

Both of the studies evaluating pirenzepine documented ocular and systemic adverse events that occurred during the trials (Table 6). Both studies used a significance level of P < 0.15 for reporting 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. In the PIR‐205 Study 2004, events of common cold, rhinitis, and sinusitis differed statistically between groups (P < 0.15) and occurred more frequently in the placebo group than in the pirenzepine group. Tan 2005 reported more complaints of abdominal pain in the gel/gel group than in the placebo/placebo group (P = 0.065) and more incidents of rash in the placebo/gel group than in the placebo/placebo group (P = 0.104). The three ocular adverse events most frequently reported by both studies (n=387) were symptoms of decreased accommodation (RR 9.05, 95% CI 4.09 to 20.01), papillae/follicles (RR 3.22, 95% CI 2.11 to 4.90), and medication residue on the eyelids or eyelashes (RR 0.91, 95% CI 0.73 to 1.12). Six ocular adverse events differed significantly (P < 0.15) between groups in the PIR‐205 Study 2004; symptoms of decreased accommodation, papillae/follicles, decreases in visual acuity, eye discomfort, and mydriasis occurred more frequently in the pirenzepine‐treated group, and medication residue on the eyelids or eyelashes occurred more frequently in the placebo group. Four ocular adverse events differed significantly (P < 0.15) between groups in Tan 2005: symptoms of decreased accommodation, papillae/follicles, and decreases in visual acuity occurred more frequently in the gel/gel and placebo/gel groups, and itchy eyes occurred more frequently in the placebo/gel group than in the placebo group. We graded the certainty of evidence for adverse effects as moderate, downgrading for imprecision of results (‐1).

Table 6. Adverse effects reported by studies of pharmaceutical interventions

Study	Interventions studied	Details
PIR‐205 Study 2004	Pirenzepine gel vs placebo gel	Reported 6 ocular adverse events with P ≤ 0.15 Accommodation abnormality symptoms: 40% vs 7% Papillae and follicles: 40% vs 18% Medication residue: 38% vs 53% Visual acuity decreased: 15% vs 2% Eye discomfort: 10% vs 4% Mydriasis: 9% vs 2%
Tan 2005	Pirenzepine gel and placebo gel PIR/PIR PLC/PIR PLC/PLC	Reported 4 ocular adverse events with P ≤ 0.15 (compared to PLC/PLC) Papillae and follicles: 1 = 58.5%; 2 = 51.4%; 3 = 14.1% Abnormality of accommodation: 1 = 44.4%; 2 = 22.1%; 3 = 2.8% Eye itching: 2 = 10.0%; 3 = 18.3% Visual acuity decreased: 1 = 16.9%; 2 = 14.3%; 3 = 1.4%
ATOM Study 2006	Atropine 1% vs placebo eye drops	No serious adverse events reported, but reasons for withdrawal among atropine users included allergic or hypersensitivity reactions or discomfort (4.5%), glare (1.5%), blurred near vision (1%), and logistical difficulties (3.5%)
Yen 1989	Atropine 1% + bifocals vs cyclopentolate + SVLs vs placebo + SVLs	All atropine users reported photophobia; most reported that they stopped gym classes and did not like going outdoors. No other systemic or ocular complications were observed
Shih 1999	Atropine 0.5%, 0.25%, 0.1%, and tropicamide 0.5%	Three events reported in the atropine 0.5% group: 2 patients complained of photophobia, 1 with allergic blepharitis

PIR: pirenzepine gel.
PLC: placebo gel.
SVLs: single vision lenses.

Five studies included in this review evaluated atropine in at least one study arm (ATOM Study 2006; MIT Study 2001; Shih 1999; Yen 1989; Yi 2015); however, only three studies compared atropine with placebo directly (ATOM Study 2006; MIT Study 2001; Yi 2015), and four studies reported adverse effect data. In the ATOM Study 2006, no serious adverse events were reported, although the four most common reasons for study withdrawal in the atropine group were allergic or hypersensitivity reactions or discomfort (4.5%), logistical difficulties (3.5%), glare (1.5%), and blurred near vision (1%). No instances of decreased visual acuity; intraocular pressure changes over 5.5 mmHg; or lenticular, optic disc, or macular changes were reported. Shih 1999 reported three adverse events, all of which occurred in the highest‐dose atropine group (0.5%); two participants complained of photophobia and one participant had allergic blepharitis. Yen 1989 reported that all patients in the atropine (plus bifocal lenses) group had photophobia, which was not reported in the cyclopentolate (plus SVLs) or placebo (plus SVLs) groups. Yi 2015 reported that no adverse effects were observed for children in either atropine or placebo groups.

8. Antimuscarinic agents versus tropicamide

One study compared three doses of atropine versus tropicamide (Shih 1999). In the four‐arm trial, participants were assigned to receive 0.5% atropine drops plus bifocals, 0.25% atropine drops plus slightly undercorrected lenses, 0.1% atropine drops plus fully corrected SVLs, or 0.5% tropicamide drops (control group).

Change in refractive error (Analysis 8.1; Analysis 8.2)

At one‐year follow‐up, myopia progression measured with cycloplegic autorefraction was significantly slowed for each atropine group compared with the tropicamide group, with the highest atropine dose showing the least progression (MD 0.78 D, 95% CI 0.49 to 1.07 for 0.1% atropine; MD 0.81 D, 95% CI 0.57 to 1.05 for 0.25% atropine; and MD 1.01 D, 95% CI 0.74 to 1.28 for 0.5% atropine). This effect was also observed at two‐year follow‐up for each atropine group compared with the tropicamide group (MD 1.95, 95% CI 1.60 to 2.30 for 0.1% atropine; MD 1.98, 95% CI 1.68 to 2.28 for 0.25% atropine; and MD 2.42, 95% CI 2.16 to 2.68 for 0.5% atropine). We graded the certainty of evidence as moderate, downgrading for risk of bias (‐1).

Shih 1999 did not report on axial length, corneal radius of curvature, or adverse effects.

9. Systemic adenosine antagonists versus placebo

One study compared systemic 7‐methylxanthine (7‐mx), an adenosine receptor antagonist, versus placebo for one year (Trier 2008). Participants in both groups (35 in the 7‐mx group and 42 in the placebo group) wore SVLs. Refractive error was measured by cycloplegic autorefraction.

Change in refractive error (Analysis 9.1)

At one‐year follow‐up, the mean difference in myopia progression when 7‐mx was compared with placebo was 0.07 D (95% CI ‐0.09 to 0.24). We graded the certainty of evidence as moderate, downgrading for imprecision (‐1).

Change in axial length (Analysis 9.2)

The 7‐mx group showed less or the same amount of change in axial length compared with the placebo group at one‐year follow‐up (MD ‐0.03 mm, 95% CI ‐0.10 to 0.03). We graded the certainty of evidence as moderate, downgrading for imprecision (‐1).

Change in corneal radius of curvature (Analysis 9.3)

The mean change in corneal radius of curvature between 7‐mx and placebo groups was not significantly different at one‐year follow‐up (MD 0.02 D, 95% CI ‐0.03 to 0.07).

Adverse effects

Trial authors reported that "no subjective side effects were reported."

10. Timolol drops versus no drops

One study compared 0.25% timolol drops versus no drops for slowing the progression of myopia in children (Jensen 1991). Participants in both groups wore SVLs. Refractive error was measured by cycloplegic autorefraction.

Change in refractive error (Analysis 10.1)

There were no statistically significant differences in myopia progression for 46 participants who used timolol compared with 49 participants who did not, at one year (MD ‐0.05 D, 95% CI ‐0.21 to 0.11) and at two years (MD ‐0.04 D, 95% CI ‐0.30 to 0.22). We graded the certainty of evidence as low, downgrading for imprecision (‐1) and risk of bias (‐1).

Jensen 1991 did not report on axial length, corneal radius of curvature, or adverse effects.

Comparisons of combinations of interventions

11. Atropine plus multifocal spectacles versus placebo plus SVLs

Two studies compared atropine drops plus multifocal lenses versus placebo drops plus SVLs to slow the progression of myopia in children. The MIT Study 2001 used 0.5% atropine plus PALs, and Yen 1989 used 1% atropine plus bifocal lenses.

Change in refractive error (Analysis 11.1)

At one year, both studies showed less progression among atropine plus multifocal lens users compared with placebo plus SVL users (MD 0.78 D, 95% CI 0.54 to 1.02). We graded the certainty of evidence as moderate, downgrading for risk of bias (‐1).

Change in axial length (Analysis 11.2)

At the end of the 18‐month MIT Study 2001, participants in the atropine plus multifocal lens group had significantly less axial elongation compared with participants in the placebo plus SVL group (MD ‐0.37 mm, 95% CI ‐0.47 to ‐0.27). We graded the certainty of evidence as moderate, downgrading for risk of bias (‐1).

Change in corneal radius of curvature

Neither the MIT Study 2001 nor Yen 1989 reported outcomes for corneal radius of curvature.

Adverse effects

Yen 1989 reported that all participants in the atropine plus bifocal lenses group had photophobia, which was not reported in the placebo plus SVLs group. The MIT Study 2001 did not report on adverse effects.

12. Atropine plus bifocal spectacles versus cyclopentolate plus SVLs

One study compared 1% atropine drops plus bifocal lenses versus 1% cyclopentolate drops plus SVLs (Yen 1989).

Change in refractive error (Analysis 12.1)

At one year, participants in the atropine plus bifocal lens group had significantly less myopia progression compared with participants in the cyclopentolate plus SVLs group (MD 0.36 D, 95% CI 0.11 to 0.61). We graded the certainty of evidence as moderate, downgrading for risk of bias (‐1).

Axial length and corneal radius of curvature were not measured by Yen 1989. It was reported that all participants in the atropine plus bifocal lenses group had photophobia, which was not observed in the cyclopentolate plus SVLs group.

13. Bifocal spectacles versus SVLs with timolol drops

Change in refractive error (Analysis 13.1)

In a three‐arm trial of +2.00 D bifocal lenses, 0.25% timolol drops plus SVLs, and SVLs (Jensen 1991), a pairwise comparison of bifocal and SVL plus timolol groups suggested that use of bifocals slowed the progression of myopia more effectively than SVLs plus timolol drops at one year (MD 0.19 D, 95% CI 0.06 to 0.32) and at two years (MD 0.23 D, 95% CI 0.00 to 0.46). Neither intervention when compared with the SVL‐only group was statistically significant for this study (see Analysis 2.1; Analysis 2.2; Analysis 9.1). We graded the certainty of evidence as moderate, downgrading for risk of bias (‐1).

Jensen 1991 did not report on axial length, corneal radius of curvature, or adverse effects.

Tropicamide plus bifocal spectacles versus SVLs

In a co‐twin study, one twin from each twin pair was randomized to receive either 1% tropicamide once per day and +1.25 D bifocals or SVLs; follow‐up was provided for 3.5 years (Schwartz 1981). No numerical results were presented in the paper. Study authors stated that control twins showed more progression in myopia than their co‐twins who received tropicamide and bifocals, but that this difference was not statistically significant.

Discussion

Summary of main results

Our review included 41 studies that investigated 15 comparisons of interventions to slow progression of myopia in children.

Our findings suggest that there is limited evidence favoring full correction of myopia over undercorrection. Trials have also shown a statistically significant but clinically unimportant benefit of multifocal spectacle lenses compared with single vision lenses (SVLs) for both myopia progression and axial elongation.

We found consistent evidence favoring antimuscarinic drugs compared with placebo for reducing progression of myopia and elongation of axial length in children with myopia. Atropine resulted in an effect of larger magnitude than was seen with pirenzepine or cyclopentolate. No trial directly compared the three different antimuscarinic drugs. These drugs are associated with frequent side effects, such as accommodation difficulties, papillae and follicles, sensitivity to light, and eye discomfort, which may lead to approximately 15% of children quitting therapy (ATOM Study 2006). One study directly compared atropine versus tropicamide and found a significant beneficial effect with atropine; however, the concentration of atropine (0.1% and 0.25%) was much lower than that used in other trials.

Studies investigating peripheral plus spectacle lenses, bifocal soft contact lenses, rigid gas permeable contact lenses, overnight orthokeratology contact lenses, spherical aberration soft contact lenses, systemic adenosine antagonists (7‐methylxanthine), and topical timolol eye drops provided limited or inconclusive evidence as to the effectiveness of these interventions for slowing the progression of myopia compared with single vision lenses alone.

In summary, we found consistent evidence of meaningful benefit when antimuscarinic drugs were used for slowing progression of myopia in children. Neither the optimal dose of antimuscarinic drug nor the additional value of using an antimuscarinic drug along with spectacles or contact lenses has been adequately answered by available evidence. Evidence regarding beneficial effects of the other interventions included in this review is neither consistent nor confirmatory.

Overall completeness and applicability of evidence

Several interventions have been investigated by more than one reporting source (journal publication, conference abstract, trial registry, etc.), which provided sufficient evidence to determine the applicability of treatment for slowing myopia progression. However, reporting of results was inconsistent among studies, so grouping of findings was difficult. Antimuscarinic pharmaceutical agents hold the greatest promise for slowing myopia progression in children, but not all studies provided complete data for inclusion in the meta‐analysis.

The included trials have been conducted across diverse geographic locations. The effects that we observed for antimuscarinic drugs were consistent across studies conducted in Caucasian populations as well as in Asian populations.

Evidence regarding the beneficial effects of various myopic control agents may be related to the ethnicity of participants and/or the comparator intervention in the included trials. For example, Asian children are more likely to be myopic and their myopia progresses faster than that in Caucasian children (Lin 1999; Zhan 2000), so any myopia control agent may be more or less effective for Asian children than for Caucasian children because the cause of their myopia may be different.

The primary outcome for myopia progression studies typically has been change in refractive error over time; however, as new methods of assessing and treating myopia have become available, the primary outcome has been switched to axial growth of the eye. For example, all three studies evaluating orthokeratology lenses—Charm 2013 ROMIO Study 2012 Swarbrick 2015—and one study assessing systemic 7‐methylxanthine—Trier 2008—defined axial length as their primary outcome. In some studies, both methods have been measured and reported, which may enhance confusion if the two methods yield differing information. For example, the rigid gas permeable contact lenses (RGPCLs) trial by Walline and colleagues found that RGPCLs significantly slowed myopia progression but did not slow axial eye growth (CLAMP Study 2004). The change in the primary outcome of myopia control studies also may lead to reporting bias, as was observed in one study that planned to report refractive error as the primary outcome but switched the primary outcome to axial length in the journal publication based on significant findings for axial length but not for refractive error (Fujikado 2014).

Quality of the evidence

Overall, the certainty of evidence ranged from very low to high across all comparisons and outcomes. In terms of the primary outcome for this review—change in refractive error—we assessed two comparisons (multifocal vs single vision lens spectacles and atropine plus multifocal spectacles vs placebo plus SVLs) to provide high‐certainty evidence in favor of the treatment group, finding no reason to downgrade. We downgraded for imprecision, inconsistency, or risk of bias for most analyses. Imprecision and inconsistency may reflect comparisons with only one or two small trials and underlying differences among studies, respectively.

This review was limited to randomized controlled trials (RCTs), minimizing the chance of treatment selection bias based on participants' desires for a specific correction. However, not all biases were completely eliminated, and many children dropped out of studies due to dissatisfaction with the intervention received. For many interventions, participants could not be masked with regard to treatment when they were assigned randomly to spectacles versus contact lenses or to one of two types of contact lenses. Although it is unlikely that participants could influence the outcome of myopic eye growth, they may have been more likely to halt participation in a study if they received a treatment that did not interest them, which could potentially increase the risk of bias. Thus, high risk of performance bias, attrition bias, or both was the most common reason for downgrading the certainty of evidence when risk of bias was an issue.

Potential biases in the review process

We reduced the risk of bias during the review process by utilizing a thorough literature search and by not limiting reviewed studies on the basis of language or dates. Two review authors, including at least one clinician and one methodologist, independently assessed the search results for eligibility and extracted data. There is little reason to believe that investigations would have been missed by the search methods unless the study results were never reported.

Overall, despite improvement seen in trials following the CONSORT statement for RCTs (Schulz 2010), many studies still lack the required rigor of reporting necessary to allow the reader to assess the risk of bias in individual trials. The vast majority of studies either did not mask the person analyzing the study data or did not report whether this occurred. It is important for statisticians to make decisions based solely on available data that should not include treatment allocation to reduce or eliminate the potential for reporting bias.

Of the 25 records awaiting for classification, eight studies with published results are likely to be included in the future update of the review. Five of these eight studies are unlikely to contribute any quantitative data for synthesis because outcomes and timepoints for outcomes are out of the scope of the review (Cheung 2018; Lam 2018; Tan 2019; Tilia 2018; Wu 2018). The remaining three studies are likely to contribute quantitative data (Pärssinen 2017; Ren 2017; Wei 2017); however, because the findings in these three studies are consistent with what we report herein, we anticipate they will not change the effect sizes in any meaningful way.

Agreements and disagreements with other studies or reviews

Saw 2002a, Saw 2002b, and Gwiazda 2009 did not include a systematic and comprehensive literature search nor any meta‐analyses. The conclusions of these three reviews are consistent with our observations in this systematic review. The 2017 report by the American Academy of Ophthalmology also concluded that there is "high‐level evidence" to support the use of atropine to prevent myopia progression, although reports have described rebound of myopia after treatment is discontinued (Pineles 2017). Other treatments, such as undercorrection of myopia, multifocal spectacles, and RGPCLs, do not slow growth of the eye in a clinically meaningful manner (i.e. slowing growth of the eye by 50% or more).

Figure 1

Study flow diagram.

Figure 2

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

$Forest plot of comparison: 1 Undercorrection vs full correction spectacles, outcome: 1.2 Change in refractive error from baseline (1 year).$

Figure 3

Forest plot of comparison: 1 Undercorrection vs full correction spectacles, outcome: 1.2 Change in refractive error from baseline (1 year).

$Forest plot of comparison: 2 Multifocal lenses vs single vision lenses, outcome: 2.1 Change in refractive error from baseline (1 year).$

Figure 4

Forest plot of comparison: 2 Multifocal lenses vs single vision lenses, outcome: 2.1 Change in refractive error from baseline (1 year).

$Forest plot of comparison: 5 Rigid gas permeable contact lenses vs control, outcome: 5.1 Change in refractive error from baseline [D].$

Figure 5

Forest plot of comparison: 5 Rigid gas permeable contact lenses vs control, outcome: 5.1 Change in refractive error from baseline [D].

Figure 6

Forest plot of comparison: 6 Orthokeratology contact lenses versus single vision lenses, outcome: 6.1 Change in axial length from baseline (2 years).

$Forest plot of comparison: 6 Antimuscarinic agents vs placebo, outcome: 6.1 Change in refractive error from baseline (1 year).$

Figure 7

Forest plot of comparison: 6 Antimuscarinic agents vs placebo, outcome: 6.1 Change in refractive error from baseline (1 year).

$Comparison 1: Undercorrection vs full correction spectacles, Outcome 1: Change in refractive error from baseline$

Analysis 1.1

Comparison 1: Undercorrection vs full correction spectacles, Outcome 1: Change in refractive error from baseline

Analysis 1.2

Comparison 1: Undercorrection vs full correction spectacles, Outcome 2: Change in axial length from baseline

$Comparison 2: Multifocal lenses vs single vision lenses, Outcome 1: Change in refractive error from baseline (1 year)$

Analysis 2.1

Comparison 2: Multifocal lenses vs single vision lenses, Outcome 1: Change in refractive error from baseline (1 year)

$Comparison 2: Multifocal lenses vs single vision lenses, Outcome 2: Change in refractive error from baseline (2 years)$

Analysis 2.2

Comparison 2: Multifocal lenses vs single vision lenses, Outcome 2: Change in refractive error from baseline (2 years)

$Comparison 2: Multifocal lenses vs single vision lenses, Outcome 3: Change in refractive error from baseline (3 years)$

Analysis 2.3

Comparison 2: Multifocal lenses vs single vision lenses, Outcome 3: Change in refractive error from baseline (3 years)

Analysis 2.4

Comparison 2: Multifocal lenses vs single vision lenses, Outcome 4: Change in axial length from baseline (1 year)

Analysis 2.5

Comparison 2: Multifocal lenses vs single vision lenses, Outcome 5: Change in axial length from baseline (2 years)

Analysis 2.6

Comparison 2: Multifocal lenses vs single vision lenses, Outcome 6: Change in axial length from baseline (3 years)

Analysis 2.7

Comparison 2: Multifocal lenses vs single vision lenses, Outcome 7: Change in corneal radius of curvature from baseline, horizontal (3 years)

$Comparison 3: Peripheral plus spectacles vs single vision lenses, Outcome 1: Change in refractive error from baseline (1 year)$

Analysis 3.1

Comparison 3: Peripheral plus spectacles vs single vision lenses, Outcome 1: Change in refractive error from baseline (1 year)

$Comparison 3: Peripheral plus spectacles vs single vision lenses, Outcome 2: Change in refractive error from baseline (2 years)$

Analysis 3.2

Comparison 3: Peripheral plus spectacles vs single vision lenses, Outcome 2: Change in refractive error from baseline (2 years)

Analysis 3.3

Comparison 3: Peripheral plus spectacles vs single vision lenses, Outcome 3: Change in axial length from baseline (1 year)

Analysis 3.4

Comparison 3: Peripheral plus spectacles vs single vision lenses, Outcome 4: Change in axial length from baseline (2 years)

$Comparison 4: Bifocal soft contact lenses vs single vision soft contact lenses, Outcome 1: Change in refractive error from baseline (1 year)$

Analysis 4.1

Comparison 4: Bifocal soft contact lenses vs single vision soft contact lenses, Outcome 1: Change in refractive error from baseline (1 year)

Analysis 4.2

Comparison 4: Bifocal soft contact lenses vs single vision soft contact lenses, Outcome 2: Change in axial length from baseline (1 year)

Analysis 4.3

Comparison 4: Bifocal soft contact lenses vs single vision soft contact lenses, Outcome 3: Change in corneal radius of curvature from baseline (1 year)

$Comparison 5: Rigid gas permeable contact lenses vs control, Outcome 1: Change in refractive error from baseline$

Analysis 5.1

Comparison 5: Rigid gas permeable contact lenses vs control, Outcome 1: Change in refractive error from baseline

Analysis 5.2

Comparison 5: Rigid gas permeable contact lenses vs control, Outcome 2: Change in axial length from baseline

Analysis 5.3

Comparison 5: Rigid gas permeable contact lenses vs control, Outcome 3: Change in corneal radius of curvature from baseline

Analysis 6.1

Comparison 6: Orthokeratology contact lenses vs single vision lenses, Outcome 1: Change in axial length from baseline (2 years)

$Comparison 7: Antimuscarinic agents vs placebo, Outcome 1: Change in refractive error from baseline (1 year)$

Analysis 7.1

Comparison 7: Antimuscarinic agents vs placebo, Outcome 1: Change in refractive error from baseline (1 year)

$Comparison 7: Antimuscarinic agents vs placebo, Outcome 2: Change in refractive error from baseline (2 years)$

Analysis 7.2

Comparison 7: Antimuscarinic agents vs placebo, Outcome 2: Change in refractive error from baseline (2 years)

Analysis 7.3

Comparison 7: Antimuscarinic agents vs placebo, Outcome 3: Change in axial length from baseline (1 year)

Analysis 7.4

Comparison 7: Antimuscarinic agents vs placebo, Outcome 4: Change in axial length from baseline (2 years)

Analysis 7.5

Comparison 7: Antimuscarinic agents vs placebo, Outcome 5: Incidence of adverse events

$Comparison 8: Atropine vs tropicamide, Outcome 1: Change in refractive error from baseline (1 year)$

Analysis 8.1

Comparison 8: Atropine vs tropicamide, Outcome 1: Change in refractive error from baseline (1 year)

$Comparison 8: Atropine vs tropicamide, Outcome 2: Change in refractive error from baseline (2 years)$

Analysis 8.2

Comparison 8: Atropine vs tropicamide, Outcome 2: Change in refractive error from baseline (2 years)

$Comparison 9: Systemic 7‐methylxanthine vs placebo, Outcome 1: Change in refractive error from baseline (1 year)$

Analysis 9.1

Comparison 9: Systemic 7‐methylxanthine vs placebo, Outcome 1: Change in refractive error from baseline (1 year)

Analysis 9.2

Comparison 9: Systemic 7‐methylxanthine vs placebo, Outcome 2: Change in axial length from baseline (1 year)

Analysis 9.3

Comparison 9: Systemic 7‐methylxanthine vs placebo, Outcome 3: Change in corneal radius of curvature from baseline (1 year)

$Comparison 10: Timolol eye drops vs no eye drops, Outcome 1: Change in refractive error from baseline$

Analysis 10.1

Comparison 10: Timolol eye drops vs no eye drops, Outcome 1: Change in refractive error from baseline

$Comparison 11: Atropine + multifocal lenses vs placebo + single vision lenses, Outcome 1: Change in refractive error from baseline (1 year)$

Analysis 11.1

Comparison 11: Atropine + multifocal lenses vs placebo + single vision lenses, Outcome 1: Change in refractive error from baseline (1 year)

Analysis 11.2

Comparison 11: Atropine + multifocal lenses vs placebo + single vision lenses, Outcome 2: Change in axial length from baseline (1 year)

$Comparison 12: Atropine + multifocal lenses vs cyclopentolate + single vision lenses, Outcome 1: Change in refractive error from baseline (1 year)$

Analysis 12.1

Comparison 12: Atropine + multifocal lenses vs cyclopentolate + single vision lenses, Outcome 1: Change in refractive error from baseline (1 year)

$Comparison 13: Bifocal spectacles vs single vision lenses + timolol, Outcome 1: Change in refractive error from baseline$

Analysis 13.1

Comparison 13: Bifocal spectacles vs single vision lenses + timolol, Outcome 1: Change in refractive error from baseline

Summary of findings 1. Interventions to slow progression of myopia in children

Outcome: change in refractive error, measured in diopters (D), from baseline to 1‐year follow‐up
Interventions to slow progression of myopia in children
Population: children with myopia (nearsightedness) Settings: ophthalmology or optometry clinics
Comparison (intervention vs comparator)	Mean difference (95% CI) Positive values represent slower progression of myopia in the treatment group than in the comparison group	No. of participants (studies)	Certainty of the evidence (GRADE)	Comments
Undercorrected vs fully corrected spectacles	‐0.15 D (‐0.29 to 0.00)	142 (2)	⊕⊕⊝⊝ low^a,b	A third study did not report this outcome at 1 year
Multifocal vs single vision lens spectacles	0.14 D (0.08 to 0.21)	1463 (9)	⊕⊕⊕⊝ moderate^b	Five studies not included in the meta‐analyses also showed mostly favorable effects of multifocal lenses for slowing myopia progression
Peripheral plus spectacles vs single vision lens spectacles	See comment	597 (3)	⊕⊕⊝⊝ low^b,c	No meta‐analysis was conducted because of clinical and methodological heterogeneity among the 3 studies; furthermore, the results from these studies were inconsistent
Bifocal vs single vision soft contact lenses	0.20 D (‐0.06 to 0.47)	300 (4)	⊕⊕⊝⊝ low^b,c	‐
Rigid gas permeable contact lenses vs spectacles or soft contact lenses	See comment	420 (2)	⊕⊝⊝⊝ very low^a,b,c	No meta‐analysis was conducted due to differences among 2 studies that reported inconsistent results
Orthokeratology contact lenses vs single vision lenses	See comment	‐	‐	Because orthokeratology contact lenses temporarily reduce myopia, their myopia control treatment effect can be measured only by axial elongation. We did not analyze the changes in refractive error for this comparison
Spherical aberration soft contact lenses vs single vision soft contact lenses	See comment	209 (2)	⊕⊕⊝⊝ low^b,d	No meta‐analysis was conducted because 1 of the studies did not provide effect estimates; however, 2 studies comparing spherical aberration SCLs with single vision SCLs reported no difference in myopia progression
Antimuscarinic agents vs placebo	Atropine: 1.00 D (0.93 to 1.07) Pirenzepine: 0.31 D (0.17 to 0.44) Cyclopentolate: 0.34 D (0.08 to 0.60)	629 (3) 326 (2) 64 (1)	⊕⊕⊕⊝ moderate^b	We stratified the analysis by types of antimuscarinic agents due to statistical inconsistency
Atropine vs tropicamide	Atropine 0.1%: 0.78 D (0.49 to 1.07) Atropine 0.25%: 0.81 D (0.57 to 1.05) Atropine 0.5%: 1.01 D (0.74 to 1.28)	196 (1)	⊕⊕⊕⊝ low^b	‐
Systemic 7‐methylxanthine vs placebo	0.07 D (‐0.09 to 0.24)	77 (1)	⊕⊕⊕⊝ moderate^a	‐
Timolol drops vs no drops	‐0.05 D (‐0.21 to 0.11)	95 (1)	⊕⊕⊝⊝ low^a,b	‐
Atropine plus multifocal spectacles vs placebo plus SVLs	0.78 D (0.54 to 1.02)	191 (2)	⊕⊕⊕⊝ moderate^b	‐
Atropine plus bifocal spectacles vs cyclopentolate plus SVLs	0.36 D (0.11 to 0.61)	64 (1)	⊕⊕⊕⊝ moderate^b	‐
Bifocal spectacles vs SVLs with timolol drops	0.19 D (0.06 to 0.32)	97 (1)	⊕⊕⊕⊝ moderate^b	‐
Tropicamide plus bifocal spectacles vs SVLs	See comment	50 (1)	‐	No estimate of effect was reported
Outcome: change in axial length, measured in millimeters (mm), from baseline to 1‐year follow‐up
Comparison (intervention vs comparator)	Mean difference (95% CI) Negative values represent less axial elongation in the treatment group than in the comparison group	No. of participants (studies)	Certainty of the evidence (GRADE)	Comments
Undercorrected vs fully corrected spectacles	0.05 mm (‐0.01 to 0.11)	94 (1)	⊕⊕⊝⊝ low^a,b	Two studies did not report this outcome at 1 year
Multifocal vs single vision lens spectacles	‐0.06 mm (‐0.09 to ‐0.04)	896 (4)	⊕⊕⊕⊝ moderate^b	Four studies (not included in the meta‐analysis) showed mostly favorable effects of multifocal lenses and 6 studies did not report this outcome
Peripheral plus spectacles vs single vision lens spectacles	See comment	597 (3)	⊕⊕⊝⊝ low^b,c	‐
Bifocal vs single vision soft contact lenses	‐0.11 mm (‐0.14 to ‐0.08)	300 (4)	⊕⊕⊝⊝ low^b,c	‐
Rigid gas permeable contact lenses vs spectacles or soft contact lenses	0.02 mm (‐0.05 to 0.10)	415 (2)	⊕⊕⊝⊝ low^a,b	‐
Orthokeratology contact lenses vs single vision lenses	‐0.28 mm (‐0.38 to ‐0.19)	106 (2)	⊕⊕⊕⊝ moderate^b	One other study reported this outcome; however, the study did not report sufficient data for analysis
Spherical aberration soft contact lenses vs single vision soft contact lenses	See comment	209 (2)	⊕⊝⊝⊝ very low^a,b,d	No meta‐analysis was conducted due to clinical, methodological, and statistical differences between the 2 studies; however, 2 studies comparing spherical aberration SCLs with single vision SCLs reported no difference in axial length
Antimuscarinic agents vs placebo	Atropine: ‐0.35 mm (‐0.38 to ‐0.31) Pirenzepine: ‐0.13 mm (‐0.14 to ‐0.12)	502 (2) 326 (2)	⊕⊕⊕⊝ moderate^c	We did not combine results for all antimuscarinic agents due to statistical inconsistency; outcome was not reported by 2 studies
Atropine vs tropicamide	See comment	196 (1)	‐	Outcome was not reported
Systemic 7‐methylxanthine vs placebo	‐0.03 mm (‐0.10 to 0.03)	77 (1)	⊕⊕⊕⊝ moderate^a	‐
Timolol drops vs no drops	See comment	95 (1)	‐	Outcome was not reported
Atropine plus multifocal spectacles vs placebo plus SVLs	‐0.37 mm (‐0.47 to ‐0.27)	127 (1)	⊕⊕⊕⊝ moderate^b	One study did not report this outcome
Atropine plus bifocal spectacles vs cyclopentolate plus SVLs	See comment	64 (1)	‐	Outcome was not reported
Bifocal spectacles vs SVLs with timolol drops	See comment	97 (1)	‐	Outcome was not reported
Tropicamide plus bifocal spectacles vs SVLs	See comment	50 (1)	‐	Outcome was not reported
Adverse effects
No serious adverse events were reported across all interventions. Two studies showed that participants receiving antimuscarinic topical medications (n=259) were more likely to experience accommodation difficulties (Risk Ratio 9.05, 95% CI 4.09 to 20.01), papillae and follicles (RR 3.22, 95% CI 2.11 to 4.90) than participants receiving placebo (n=128), but no difference in medication residue on the eyelids or eyelashes (RR 0.91, 95% CI 0.73 to 1.12). Certainty of a body of evidence was moderate, downgraded for imprecision of results (‐1).
GRADE Working Group grades of evidence. High certainty: further research is very unlikely to change our confidence in the estimate of effect. Moderate certainty: further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low certainty: further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low certainty: we are very uncertain about the estimate.
CI: confidence interval; D: diopters. ^aDowngraded for imprecision (i.e. wide confidence interval). ^bDowngraded for risk of bias among included trials. ^cDowngraded for inconsistency. ^dDowngraded for indirectness due to averaging values over time assuming linear change (e.g. reporting the change per year using data collected at baseline and at 2 years of follow‐up).

Summary of findings 1. Interventions to slow progression of myopia in children

Table 1. Interventions of included studies

Study	Spectacles					Contact lenses					Pharmaceutical agents		Combination of interventions
	Undercorrected SVLs	Multifocal lenses			Fully corrected SVLs	Soft bifocal lenses	RGP	Ortho‐k	SA‐SCL	SVSCL	Test group	Reference group
	Undercorrected SVLs	Bifocal lenses	PALs	Peripheral plus lenses	Fully corrected SVLs	Soft bifocal lenses	RGP	Ortho‐k	SA‐SCL	SVSCL	Test group	Reference group
Adler 2006; 2 study arms	X				X
Chung 2002; 2 study arms	X				X
Koomson 2016; 2 study arms	X				X
Cheng 2010; 3 study arms		+1.50 and +1.50 prism			X
Fulk 1996; 2 study arms		+1.25			X
Fulk 2002; 2 study arms		+1.50			X
Houston Study 1987; 3 study arms		+1.00 and +2.00			X
Jensen 1991; 3 study arms		+2.00			X								Timolol + SVLs
Pärssinen 1989; 3 study arms		+1.75			Continous use and distance only
COMET Study 2003; 2 study arms			+2.00		X
COMET2 Study 2011; 2 study arms			+2.00		X
Edwards 2002; 2 study arms			+1.50		X
Hasebe 2008; 2 study arms^a			+1.50		X
MIT Study 2001; 3 study arms			Plus placebo drops		Plus placebo drops								Atropine + PALs
STAMP Study 2012; 2 study arms			+2.00		X
Wang 2005; 2 study arms			Add NR		X
Yang 2009; 2 study arms			+1.50		X
Lu 2015; 2 study arms				+2.50	X
Hasebe 2014; 3 study arms				+1.00 and +1.50	X
Sankaridurg 2010; 4 study arms				+1.00, +1.90, and +2.00	X
Anstice 2011; 2 study arms^a						+2.00				X
CONTROL Study 2016; 2 study arms						Add NR				X
DISC Study 2011; 2 study arms						+2.50				X
Fujikado 2014; 2 study arms^a						+0.50				X
CLAMP Study 2004; 2 study arms							X			X
Katz 2003; 2 study arms					X		X
Charm 2013; 2 study arms					X			X
ROMIO Study 2012; 2 study arms					X			X
Swarbrick 2015; 2 study arms^a							X	X
Cambridge Anti‐Myopia Study 2013; 4 study arms									With and without vision training	With and without vision training
Cheng 2016; 2 study arms									X	X
ATOM Study 2006; 2 study arms											1% atropine	Placebo drops
Yi 2015; 2 study arms											1% atropine	Placebo drops
Yen 1989; 3 study arms											1% atropine + bifocals	Saline + SVLs	Cyclopentolate + SVLs
Shih 1999; 4 study arms											0.1%, 0.25%, and 0.5% atropine	0.5% tropicamide
PIR‐205 Study 2004; 2 study arms											2% pirenzepine gel	Placebo gel
Tan 2005; 3 study arms											2% pirenzepine gel once and twice daily	Placebo gel
Trier 2008; 2 study arms											Systemic 7‐methylxanthine	Placebo tablet
Schwartz 1981; 2 study arms					X								Tropicamide + bifocals
NR: not reported. Ortho‐k: orthokeratology lenses. PALs: progressive addition lenses. RGP: rigid gas permeable contact lenses. SA‐SCL: spherical aberration soft contact lenses. SVLs: single vision lenses. SVSCL: single vision soft contact lenses. ^aCross‐over trial.

Table 1. Interventions of included studies

Table 2. Outcomes reported by studies of spectacle interventionsa

Outcomes	Interventions studied
Outcomes	Undercorrected lenses: 3studies	Multifocal lenses: 14studies	Peripheral plus spectacles: 3studies
Primary outcome: change in refractive error	Analysis 1.1	Analysis 2.1; Analysis 2.2; Analysis 2.3	Analysis 3.1; Analysis 3.2
Secondary outcome: change in axial length	Analysis 1.2	Analysis 2.4; Analysis 2.5; Analysis 2.6	Analysis 3.3; Analysis 3.4
Secondary outcome: change in corneal radius of curvature	Not reported by 2 studies and reported only as nonsignificant by Chung 2002	Analysis 2.7	Not reported
Adverse effects	Two participants who were undercorrected complained of blurred vision (Adler 2006)	Three participants using PALs in 1 study had conjunctivitis, distance blur, or dizziness (COMET2 Study 2011)	Participants reported blurred side vision, visual distortion, dizziness, headaches, and falls (Sankaridurg 2010)
^aCompared with fully corrected single vision lenses.

Table 2. Outcomes reported by studies of spectacle interventionsa

Table 3. Outcomes reported by studies of contact lens interventionsa

Outcomes	Interventions studied
Outcomes	Soft bifocal contact lenses: 4studies	Rigid gas permeable contact lenses: 2 studies	Orthokeratology: 3 studies	Spherical aberration soft contact lenses: 2 studies
Primary outcome: change in refractive error	Analysis 4.1	Analysis 5.1	No data for analysis	Data reported by both studies, but not meta‐analyzable
Secondary outcome: change in axial length	Analysis 4.2	Analysis 5.2	Analysis 6.1	Data reported by both studies, but not meta‐analyzable
Secondary outcome: change in corneal radius of curvature	Analysis 4.3	Analysis 5.3	No data for analysis	Not reported
Adverse effects	Six children in 1 study withdrew from the study, 3 from each group (CONTROL Study 2016)	Not reported	Adverse effects reported from all 3 studies	One study reported 1 child with allergic conjunctivitis and 1 with contact dermatitis
^aCompared with fully corrected single vision lenses or contact lenses.

Table 3. Outcomes reported by studies of contact lens interventionsa

Table 4. Outcomes reported by studies of pharmaceutical interventionsa

Outcomes	Interventions studied
Outcomes	Antimuscarinic agents: 6studies	Atropine vs tropicamide: 1study	Systemic adenosine antagonists: 1study	Timolol: 1 study	Tropicamide (plus bifocals): 1 study
Primary outcome: change in refractive error	Analysis 7.1; Analysis 7.2	Analysis 8.1; Analysis 8.2	Analysis 9.1	Analysis 10.1	Control twins showed more progression in myopia than their co‐twins who received tropicamide and bifocals, but this difference was not statistically significant (Schwartz 1981)
Secondary outcome: change in axial length	Analysis 7.3; Analysis 7.4	Not reported	Analysis 9.2	Not reported	Not reported
Secondary outcome: change in corneal radius of curvature	Not reported	Not reported	Analysis 9.3	Not reported	Not reported
^aCompared with placebo or no drops.

Table 4. Outcomes reported by studies of pharmaceutical interventionsa

Table 5. Unit of analysis for included studies

Unit of analysis	Studies reporting each type of unit of analysis
Average of both eyes	15 studies: Adler 2006; Chung 2002; COMET Study 2003^a; COMET2 Study 2011; CONTROL Study 2016; Fujikado 2014; Fulk 1996; Fulk 2002; Hasebe 2008^a; PIR‐205 Study 2004; Sankaridurg 2010; Schwartz 1981; Shih 1999; Tan 2005; Trier 2008
Right eye only	15 studies: Cambridge Anti‐Myopia Study 2013; Charm 2013; Cheng 2010; Cheng 2016; CLAMP Study 2004; DISC Study 2011; Edwards 2002; Houston Study 1987; Katz 2003; Koomson 2016; MIT Study 2001; ROMIO Study 2012; STAMP Study 2012; Yen 1989; Yi 2015
Right and left eyes reported as separate analyses	2 studies: Jensen 1991; Pärssinen 1989
One study eye randomized and treated per child	1 study: ATOM Study 2006
Child randomized and both eyes analyzed as independent units	2 studies: Hasebe 2014; Lu 2015
Paired‐eye design	2 studies: Anstice 2011; Swarbrick 2015
Eye with more severe myopia	1 study: Wang 2017
Not reported	3 studies: Han 2018; Wang 2005; Yang 2009
^aAverage values of both eyes were used if the correlation coefficient was > 0.85 between eyes and the mean difference (MD) was not statistically significant; otherwise the eye with more myopic change was used for each child (COMET Study 2003). Mean of both eyes or of right eye only (Hasebe 2008).

Table 5. Unit of analysis for included studies

Table 6. Adverse effects reported by studies of pharmaceutical interventions

Study	Interventions studied	Details
PIR‐205 Study 2004	Pirenzepine gel vs placebo gel	Reported 6 ocular adverse events with P ≤ 0.15 Accommodation abnormality symptoms: 40% vs 7% Papillae and follicles: 40% vs 18% Medication residue: 38% vs 53% Visual acuity decreased: 15% vs 2% Eye discomfort: 10% vs 4% Mydriasis: 9% vs 2%
Tan 2005	Pirenzepine gel and placebo gel PIR/PIR PLC/PIR PLC/PLC	Reported 4 ocular adverse events with P ≤ 0.15 (compared to PLC/PLC) Papillae and follicles: 1 = 58.5%; 2 = 51.4%; 3 = 14.1% Abnormality of accommodation: 1 = 44.4%; 2 = 22.1%; 3 = 2.8% Eye itching: 2 = 10.0%; 3 = 18.3% Visual acuity decreased: 1 = 16.9%; 2 = 14.3%; 3 = 1.4%
ATOM Study 2006	Atropine 1% vs placebo eye drops	No serious adverse events reported, but reasons for withdrawal among atropine users included allergic or hypersensitivity reactions or discomfort (4.5%), glare (1.5%), blurred near vision (1%), and logistical difficulties (3.5%)
Yen 1989	Atropine 1% + bifocals vs cyclopentolate + SVLs vs placebo + SVLs	All atropine users reported photophobia; most reported that they stopped gym classes and did not like going outdoors. No other systemic or ocular complications were observed
Shih 1999	Atropine 0.5%, 0.25%, 0.1%, and tropicamide 0.5%	Three events reported in the atropine 0.5% group: 2 patients complained of photophobia, 1 with allergic blepharitis
PIR: pirenzepine gel. PLC: placebo gel. SVLs: single vision lenses.

Table 6. Adverse effects reported by studies of pharmaceutical interventions

Comparison 1. Undercorrection vs full correction spectacles

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 Change in refractive error from baseline Show forest plot	3		Mean Difference (IV, Fixed, 95% CI)	Subtotals only

1.1.1 At 1 year	2	142	Mean Difference (IV, Fixed, 95% CI)	‐0.15 [‐0.29, ‐0.00]
1.1.2 At 2 years	2	244	Mean Difference (IV, Fixed, 95% CI)	0.02 [‐0.05, 0.09]
1.2 Change in axial length from baseline Show forest plot	2		Mean Difference (IV, Fixed, 95% CI)	Subtotals only

1.2.1 At 1 year	1	94	Mean Difference (IV, Fixed, 95% CI)	0.05 [‐0.01, 0.11]
1.2.2 At 2 years	2	244	Mean Difference (IV, Fixed, 95% CI)	‐0.01 [‐0.06, 0.03]

Comparison 1. Undercorrection vs full correction spectacles

Comparison 2. Multifocal lenses vs single vision lenses

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
2.1 Change in refractive error from baseline (1 year) Show forest plot	9	1463	Mean Difference (IV, Random, 95% CI)	0.14 [0.08, 0.21]

2.1.1 Bifocal lenses	4	421	Mean Difference (IV, Random, 95% CI)	0.16 [0.01, 0.32]
2.1.2 Progressive addition lenses	5	1042	Mean Difference (IV, Random, 95% CI)	0.15 [0.09, 0.21]
2.2 Change in refractive error from baseline (2 years) Show forest plot	8	1401	Mean Difference (IV, Random, 95% CI)	0.19 [0.08, 0.30]

2.2.1 Bifocal lenses	4	416	Mean Difference (IV, Random, 95% CI)	0.20 [‐0.09, 0.49]
2.2.2 Progressive addition lenses	4	985	Mean Difference (IV, Random, 95% CI)	0.20 [0.12, 0.28]
2.3 Change in refractive error from baseline (3 years) Show forest plot	3		Mean Difference (IV, Fixed, 95% CI)	Subtotals only

2.3.1 Bifocal lenses	1	158	Mean Difference (IV, Fixed, 95% CI)	‐0.19 [‐0.47, 0.09]
2.3.2 Progressive addition lenses	2	579	Mean Difference (IV, Fixed, 95% CI)	0.21 [0.08, 0.34]
2.4 Change in axial length from baseline (1 year) Show forest plot	4	896	Mean Difference (IV, Random, 95% CI)	‐0.06 [‐0.09, ‐0.04]

2.5 Change in axial length from baseline (2 years) Show forest plot	2	723	Mean Difference (IV, Fixed, 95% CI)	‐0.05 [‐0.10, ‐0.01]

2.6 Change in axial length from baseline (3 years) Show forest plot	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

2.7 Change in corneal radius of curvature from baseline, horizontal (3 years) Show forest plot	2		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

2.7.1 At 3 years, horizontal	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
2.7.2 At 3 years, vertical	2		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

Comparison 2. Multifocal lenses vs single vision lenses

Comparison 3. Peripheral plus spectacles vs single vision lenses

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
3.1 Change in refractive error from baseline (1 year) Show forest plot	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

3.1.1 Type I	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
3.1.2 Type II	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
3.1.3 Type III	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
3.2 Change in refractive error from baseline (2 years) Show forest plot	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

3.2.1 +1.0 Diopters	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
3.2.2 +1.5 Diopters	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
3.3 Change in axial length from baseline (1 year) Show forest plot	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

3.3.1 Type I	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
3.3.2 Type II	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
3.3.3 Type III	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
3.4 Change in axial length from baseline (2 years) Show forest plot	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

3.4.1 +1.0 Diopters	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
3.4.2 +1.5 Diopters	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

Comparison 3. Peripheral plus spectacles vs single vision lenses

Comparison 4. Bifocal soft contact lenses vs single vision soft contact lenses

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
4.1 Change in refractive error from baseline (1 year) Show forest plot	4	300	Mean Difference (IV, Random, 95% CI)	0.20 [‐0.06, 0.47]

4.2 Change in axial length from baseline (1 year) Show forest plot	4	300	Mean Difference (IV, Fixed, 95% CI)	‐0.11 [‐0.14, ‐0.08]

4.3 Change in corneal radius of curvature from baseline (1 year) Show forest plot	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

Comparison 4. Bifocal soft contact lenses vs single vision soft contact lenses

Comparison 5. Rigid gas permeable contact lenses vs control

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
5.1 Change in refractive error from baseline Show forest plot	2		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

5.1.1 At 1 year	2		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
5.1.2 At 2 years	2		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
5.1.3 At 3 years	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
5.2 Change in axial length from baseline Show forest plot	2		Mean Difference (IV, Fixed, 95% CI)	Subtotals only

5.2.1 At 1 year	2	415	Mean Difference (IV, Fixed, 95% CI)	0.02 [‐0.05, 0.10]
5.2.2 At 2 years	2	394	Mean Difference (IV, Fixed, 95% CI)	0.03 [‐0.05, 0.12]
5.2.3 At 3 years	1	116	Mean Difference (IV, Fixed, 95% CI)	0.05 [‐0.12, 0.22]
5.3 Change in corneal radius of curvature from baseline Show forest plot	2		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

5.3.1 At 1 year	2		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
5.3.2 At 2 years	2		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
5.3.3 At 3 years	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

Comparison 5. Rigid gas permeable contact lenses vs control

Comparison 6. Orthokeratology contact lenses vs single vision lenses

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
6.1 Change in axial length from baseline (2 years) Show forest plot	2	106	Mean Difference (IV, Fixed, 95% CI)	‐0.28 [‐0.38, ‐0.19]

Comparison 6. Orthokeratology contact lenses vs single vision lenses

Comparison 7. Antimuscarinic agents vs placebo

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
7.1 Change in refractive error from baseline (1 year) Show forest plot	6		Mean Difference (IV, Fixed, 95% CI)	Subtotals only

7.1.1 Atropine eye drops	3	629	Mean Difference (IV, Fixed, 95% CI)	1.00 [0.93, 1.07]
7.1.2 Pirenzepine 2% gel	2	326	Mean Difference (IV, Fixed, 95% CI)	0.31 [0.17, 0.44]
7.1.3 Cyclopentolate eye drops	1	64	Mean Difference (IV, Fixed, 95% CI)	0.34 [0.08, 0.60]
7.2 Change in refractive error from baseline (2 years) Show forest plot	2		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

7.2.1 Atropine eye drops	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
7.2.2 Pirenzepine 2% gel	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
7.3 Change in axial length from baseline (1 year) Show forest plot	4		Mean Difference (IV, Fixed, 95% CI)	Subtotals only

7.3.1 Atropine eye drops	2	502	Mean Difference (IV, Fixed, 95% CI)	‐0.35 [‐0.38, ‐0.31]
7.3.2 Pirenzepine 2% gel	2	326	Mean Difference (IV, Fixed, 95% CI)	‐0.13 [‐0.14, ‐0.12]
7.4 Change in axial length from baseline (2 years) Show forest plot	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

7.5 Incidence of adverse events Show forest plot	2		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only

7.5.1 Accomodation abnormality symptoms	2	387	Risk Ratio (M‐H, Fixed, 95% CI)	9.05 [4.09, 20.01]
7.5.2 Papillae/follicles	2	387	Risk Ratio (M‐H, Fixed, 95% CI)	3.22 [2.11, 4.90]
7.5.3 Medication residue on the eyelids/eyelashes	2	387	Risk Ratio (M‐H, Fixed, 95% CI)	0.91 [0.73, 1.12]

Comparison 7. Antimuscarinic agents vs placebo

Comparison 8. Atropine vs tropicamide

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
8.1 Change in refractive error from baseline (1 year) Show forest plot	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

8.1.1 Atropine 0.5%	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
8.1.2 Atropine 0.25%	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
8.1.3 Atropine 0.1%	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
8.2 Change in refractive error from baseline (2 years) Show forest plot	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

8.2.1 Atropine 0.5%	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
8.2.2 Atropine 0.25%	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
8.2.3 Atropine 0.1%	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

Comparison 8. Atropine vs tropicamide

Comparison 9. Systemic 7‐methylxanthine vs placebo

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
9.1 Change in refractive error from baseline (1 year) Show forest plot	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

9.2 Change in axial length from baseline (1 year) Show forest plot	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

9.3 Change in corneal radius of curvature from baseline (1 year) Show forest plot	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

Comparison 9. Systemic 7‐methylxanthine vs placebo

Comparison 10. Timolol eye drops vs no eye drops

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
10.1 Change in refractive error from baseline Show forest plot	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

10.1.1 At 1 year	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
10.1.2 At 2 years	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

Comparison 10. Timolol eye drops vs no eye drops

Comparison 11. Atropine + multifocal lenses vs placebo + single vision lenses

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
11.1 Change in refractive error from baseline (1 year) Show forest plot	2	191	Mean Difference (IV, Fixed, 95% CI)	0.78 [0.54, 1.02]

11.2 Change in axial length from baseline (1 year) Show forest plot	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

Comparison 11. Atropine + multifocal lenses vs placebo + single vision lenses

Comparison 12. Atropine + multifocal lenses vs cyclopentolate + single vision lenses

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
12.1 Change in refractive error from baseline (1 year) Show forest plot	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

Comparison 12. Atropine + multifocal lenses vs cyclopentolate + single vision lenses

Comparison 13. Bifocal spectacles vs single vision lenses + timolol

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
13.1 Change in refractive error from baseline Show forest plot	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

13.1.1 At 1 year	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
13.1.2 At 2 years	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

Comparison 13. Bifocal spectacles vs single vision lenses + timolol