Myopia and Near Work: A Systematic Review and Meta-Analysis

Background: Myopia is a global public health problem affecting quality of life and work productivity. Data is scarce regarding the effects of near work on myopia. Providing a larger meta-analysis with life-long perspective, including adults and occupational exposure seemed needed. Methods: We searched PubMed, Cochrane Library, Embase and Science Direct for studies reporting myopia prevalence in near work. Myopia was defined as a mean spherical equivalent ≤ −0.50 diopter. We performed a meta-analysis using random-effects model on myopia prevalence, myopia progression per year, and odds ratio (OR) of myopia in near work, completed by subgroup analyses and meta-regressions on patients’ characteristics, type of work in adults, geographic zones, time and characteristics of near work. Results: We included 78 studies, representing a total of 254,037 participants, aged from 6 to 39 years. The global prevalence of myopia in near work was 35% (95% CI: 30 to 41%), with a prevalence of 31% (95% CI: 26 to 37%) in children and 46% (95% CI: 30 to 62%) in adults. Myopia progression was −0.39 diopters per year (−0.53 to −0.24 D/year), ranging from −0.44 (−0.57 to −0.31) in children to −0.25 D/year (−0.56 to 0.06) in adults. The odds of myopia in workers exposed vs. non-exposed to near work were increased by 26% (18 to 34%), by 31% (21 to 42%) in children and 21% (6 to 35%) in adults. Prevalence of myopia was higher in adults compared to children (Coefficient 0.15, 95% CI: 0.03 to 0.27). Conclusions: Near work conditions, including occupational exposure in adults, could be associated with myopia. Targeted prevention should be implemented in the workplace.


Introduction
Myopia is the most common type of refractive error [1] and constitutes a rising global health issue causing significant impact on visual function [2,3] with potential consequences on quality of life and work productivity [4,5] and other sequelae (glaucoma, macular degeneration, retinal detachment and cataract) [6,7]. Myopia most often involves the axial elongation of the eyeball [8,9], and can be caused by genetic [10,11] and environmental

Literature Search
The present research was carried out in accordance with the PRISMA Guidelines. We reviewed all cohort studies involving myopia in near work. Specifically, the inclusion criteria for the search strategy were cross-sectional and cohort studies (minimal number of 10 individuals), without a case-study design. We used the following keywords: "myopia" or "short-sight" for myopia, and "near work", "display" or "users" for near work. A display is hereafter defined as a computer output surface and projecting mechanism that shows text and images onto a screen, using a cathode ray tube, liquid crystal display, light-emitting diode, gas plasma, or other image projection technology. To be included, articles needed to describe the outcomes of interest of the present research, i.e., the number of workers with myopia, myopia progression, or odds ratio for myopia. The following databases were searched up until 30 May 2022: PubMed, Cochrane Library, Science Direct and Embase. The search was not limited to a specific time period. In addition, reference lists of all publications meeting the inclusion criteria were manually searched to identify any further studies not found through electronic research on the databases stated above. Articles written in Chinese or Japanese language were excluded. The search strategy is described in Text S1. Two authors (J. Castanon and T. Oueslati) conducted all literature searches, collected the abstracts, separately reviewed the abstracts and rated the suitability of the articles for inclusion, based on the criteria mentioned above. Another author (F. Dutheil) reviewed articles for which consensus on suitability was not met by the first 2 raters.

Quality Assessment
We used the Scottish Intercollegiate Guidelines Network (SIGN) for cohort studies to evaluate the risk of bias of the included articles [21]. Not designed for quantifying the internal validity of studies [22], the "STrengthening the Reporting of OBservational studies in Epidemiology" (STROBE) criteria [23] were also used to evaluate the quality of reporting. For both checklists, one point was attributed per completed item, then converted into a percentage to give a score of risk of bias and reporting quality for each included study ( Figure S2).

Data Collection
The collected data included first author's name, publication year, study design, country, aims and outcomes, sample size, characteristics of individuals (age, sex, education level, parental history of myopia, occupation for adults), myopia outcomes (prevalence of myopia, myopia progression, i.e., diopter changes per year, risk of myopia or odds ratio of myopia), use of cycloplegia or not for the diagnosis of myopia and characteristics of near work (computer, paper reading, indoor and outdoor activities, diopter-hour per week, visual display terminal exposure).

Statistical Considerations
Baseline characteristics were summarized for each study and reported as mean ± standard-deviation (SD) for continuous variables, and number (%) for categorical variables. Due to the observational nature of the included studies, we chose to using randomeffect models for the present meta-analysis (DerSimonian and Laird approach) [24,25] to assess myopia prevalence, myopia progression, and risk of myopia following exposure to near work, with results expressed as percentage, diopter change per year, and odds ratio with their 95% confidence interval (95% CI), respectively. We conducted subgroup analysis by age (adults above 18 years old vs. children/teenagers), type of work in adults, continents, and time periods (before 2005, between 2005 and 2015, and after 2015). Statistical heterogeneity between results was assessed by examining forest plots, confidence intervals and I-squared (I 2 ). Heterogeneity was considered low for I 2 < 25%, modest for 25 < I 2 < 50%, and high for I 2 > 50%. The robustness of our results was assessed by conducting a sensitivity analysis after exclusion of studies that were not evenly distributed around the funnel plot. When sample size was sufficient, meta-regressions were used to study the relationships between our outcome variables (prevalence, myopia progression, and risk for myopia) and clinically relevant parameters such as characteristics of individuals (age, sex, education level, parental history of myopia), type of occupation for adults, continents, time periods (before year 2005, years 2005 to 2015, after 2015), duration and characteristics of near work (computer, paper reading, indoor and outdoor activity, diopter-hour per week, VDT exposure). Results were expressed as regression coefficients and 95% CI. All p values were two-tailed. Statistical significance was defined by a p value < 0.05. Statistical analysis was conducted using Stata software (v16, StataCorp, College Station, TX, USA).

Quality of Articles
Using the SIGN checklist, overall quality of the 78 included studies was good, with a mean score of 72.2 ± 8.5%, ranging from 40.0 [12,85] to 86.0% [69]. The main limitation was a confusion bias with few studies considering parental history of myopia. For the studies reporting myopia progression, the main bias was the number of dropouts that was seldom reported (Figure 2 and Figure S2). The use of the STROBE checklist confirmed the overall good reporting quality of the included articles with a mean score of 81.2 ± 11.1%, ranging from 54.7 [57] to 96,7% [69,99]. All articles either did not report the description of confounding factors or did not consider the use of a flow chart. Only half of studies reported an ethical approval. a confusion bias with few studies considering parental history of myopia. For the studies reporting myopia progression, the main bias was the number of dropouts that was seldom reported (Figures 2 and S2). The use of the STROBE checklist confirmed the overall good reporting quality of the included articles with a mean score of 81.2 ± 11.1%, ranging from 54.7 [57] to 96,7% [69,99]. All articles either did not report the description of confounding factors or did not consider the use of a flow chart. Only half of studies reported an ethical approval.

Quantification of Exposure to Near Work
Twenty-six studies reported the duration of exposure to near work, ranging from few weeks [67] to 23 years [13].
Seven studies quantified the exposure to near work in Diopter-hours per week [26,41,53,74,75,84,95] ranging from 20 [26] to 134 [75]. The duration of the different kinds of near work was estimated by questionnaire. Four studies attributed a specific accommodative effort of 3D to reading or writing papers and 2D for computer work [14,41,74,75]. Then, the accommodative effort was multiplied by the duration of the activity in hours i.e., 3 h of computer work per day during 5 days equals to 30 Diopter-hours per week. Two studies did not specify the measurement of Diopter-hours [59,75] and one study quantified Diopter-hours as the duration of near work multiplied by the reciprocal of the distance from which the activity was performed [82]. All the studies reporting Diopter-hours per week also reported the prevalence of myopia except one [53].

Meta-Analysis on Odds Ratio
Within the 22 studies estimating an OR for myopia when exposed to near work [1-3,9, [32]. More specifically, the odds of myopia in case of exposure to near work are increased by 21% for adults (OR = 1.21, 95% CI: 1.06 to 1.34)

Meta-Regressions
There was a greater prevalence of myopia in adults than in children (coefficient 0.15, 95% CI: 0.03 to 0.27), in women compared to men (0.60, 95% CI: 0.30 to 0.90), and in Asia compared to Europe (0.21, 95% CI: 0.10 to 0.33) and Oceania (0.28, 95% CI: 0.09 to 0.47) (p < 0.001). There was a trend for a greater prevalence of myopia with the increase of indoor activity (0.11, −0.03 to 0.24, p = 0.090). Except a trend for a decreased prevalence of myopia in recent years, we did not find any other variables linked with prevalence of myopia ( Figure 6).
Regarding the progression of myopia, there was a trend for a greater myopia progression in Asia compared with North America (0.25, −0.13 to 0.63, p = 0.095) ( Figure S3). We did not find any other variables linked with myopia progression, as well as no significant variables influencing the odds of myopia ( Figure S4).

Discussion
The main findings were that near work exposure, including occupational exposure in adults, could be associated with myopia. The odds of myopia in the population exposed to near work is increased by 31% in children, and by 21% in adults. Some regions such as Asia may be more susceptible to the risk of myopia in case of near work exposure. We then discussed the factors that may contribute to this association between the risk of myopia and near work exposure.

Myopia and Near Work: Also in Adults
This study is the first meta-analysis that studied myopia and near work in adults. We demonstrated that occupational exposure to near work in adults is associated with a 20% increase in the odds of myopia. The increasing prevalence of myopia also implies untreatable complications that might occur even in the working-age population, such as myopic maculopathy, more frequently observed in high myopia [7]. In addition to its impact on visual function [1] myopia can lead to fatigue, anxiety [30,85] and distress at work [16], affecting the patient's quality of life. Occupations at risk of myopia are those with an important accommodative effort [100] such as clinical microscopy [26], law student [67] or medical students [76], among others. These studies suggest that near work might participate to myopia onset and progression in young adults, but the mechanisms at stake are not clearly understood. The data regarding accommodative effort by occupation is scarce in the literature, therefore limiting the possible explorations of the relationship between near work and myopia in the workplace. While we regret that the comparisons between the prevalence of myopia and types of occupation were not feasible in our meta-analysis, myopia as a public health concern seems overlooked, despite economic implications both for employees and employers [101][102][103][104][105]. Therefore, preventative actions at the workplace should target screening for myopia and may influence management strategies [101]. The literature being very scarce in terms of studies regarding myopia in adults and its risk factors, prospective longitudinal studies in adult on the matter are needed.

Effect of Time and Influence of Geographic Zones
Myopia prevalence in our study was similar to those previously reported globally [1] and in Europe [16]. Our results did not support the suspected increased prevalence of myopia. However, nearly all included studies were conducted after 2000, whereas the increased prevalence was suggested by comparing prevalence of myopia between 1930 and now [16]. Moreover, diagnosis of myopia during the 20th century may have become more and more systematic, potentially explaining this increased prevalence. Our meta-analysis highlighted heterogeneous results, with wide variations of the prevalence of myopia from 1.5% [83] to 88.1 [81]. Our results in the regional myopia prevalence (USA 29%, Europe 24%, Asia 48 %, Oceania 18%) are in accordance with those reported in the literature [106]. The role of genetic factors in geographic variations have been discussed, especially in East-Asia [101]. Nevertheless, a large number of genes found to be potentially linked with myopia with low penetrance [107,108] suggests a common predisposition towards myopia in the environments with higher amount of near work and lower outdoor light exposure [109][110][111][112], a feature especially frequent in Asia.

Other Factors Influencing Myopia
Factors promoting myopia are also specific to some cultural or subcultural characteristics, such as urbanization, economy, social class, education and lifestyle [19,37,41,113]. Despite potential additive effects of education and age were suggested as increasing the risk of myopia [16], the link between these factors (reading, age and education) makes assessment of separate effects difficult [19,57,97]. The presence of myopia in one parent multiplies by two the risk for a child to be myopic [114]. In our meta-analysis, the adult patients were rather young-mostly under 40 years old. Myopia seems to be more present in young adults, and seems to decrease secondarily [16] due to physiological aging or to the onset of a corticonuclear or cortical cataract [115]. The duration of the post-secondary school education could be in itself a risk factor of myopia more than the cognitive performance [116]. On the contrary, despite regular outdoor activities have been shown to have a beneficial effect in reducing the onset of myopia [117], our data could not confirm this result.

Myopia Progression in Near Work and Dose Response Relationships
Our meta-analyses suggest that myopia progression may be associated with the exposure to near work [8,9,11,13,19,20,29,31,35,36,40,42,44,45]. Despite being vaguely defined in the literature, indoor activities are generally considered as a risk factor for myopia [41,42]. Our results confirm this association. Although myopia progression is generally linked with an increasing accommodative effort [9,20,32,39], we did not retrieve significant differences between paper reading (accommodation at 50 cm) and computer work (accommodation at 80 cm). To our knowledge, no previous studies have specifically compared the influence of paper reading and computer work on myopia. According to some studies, myopia is significatively associated with continuous reading, longer duration of reading and shorter reading distance [51,70,118], due to the promotion of eye growth by sustained accommodation. Moreover, myopia may also be linked to the issue of contrast between support and text, as suggested by Aleman et al. [119] and Wang et al. [120]. One controversial hypothesis is that computer work might prevent myopia thanks to the release of dopamine induced by the brightness of digital screens, which could prevent eye growth (i.e., axial length), and therefore myopia [121]. Taking breaks after 30 min of reading may protect against myopia [50], leading to innovative device developed to monitor near work, triggering an alert if a risky behaviour is detected [122].

Limitations
Our study has some limitations. Several biases could have been introduced via the literature search and selection procedure. We conducted the meta-analyses on published articles only; thus potentially exposed to publication bias. We observed that a large part of the variability in out meta-analysis is secondary to between-studies heterogeneity. Conclusions from subgroup analysis and meta-regressions were also affected by this high heterogeneity. The populations investigated in our meta-analysis appear to be from varied origin, predominantly from the developed and developing countries, favouring the external validity of our results toward the population exposed to near work. One potential confounding factor could be the amount of exposure to outdoor light one is exposed to, a parameter not reported in every included study. We could hypothesize that outdoor light exposure may be inversely correlated with indoor light exposure and near work, and thus could also be associated with myopia, but our results do not support this hypothesis ( Figure 6). Moreover, meta-analyses are designed to take into account heterogeneous conditions [123]. Non-significant relationships between myopia and characteristics of population might be due to an insufficient sample size and other confounding factors. All included studies used an autorefractometer to measure the degree of myopia, however, some studies did not use cycloplegia which can induce detection bias. Finally, within our meta-analysis, most studies specified computer work as the VDT use. However, VDT may now relate to various screen distance (such as desktop computer, laptop, tablet, or smartphone) and distance is key factor of accommodation. Further studies should describe more precisely the distance of VDT exposure to allow for more precise estimation of the role of distance between the eye and object of focus in near work.

Conclusions
In conclusion, near work seems to be associated with myopia, including occupational exposure in adults. The prevalence of myopia following near-work exposure was significantly higher in adults (46%) compared to children (31%). Furthermore, our results suggest that the odds of myopia in adults with occupational exposure to near work is 21% higher (95% CI: 6 to 35%) compared to the non-exposed. Considering the consequences of myopia both for employees and employers, targeted prevention strategies should be implemented in the workplace.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/ijerph20010875/s1, Text S1: Details of search strategy; Figure S2: Quality of articles as risk of bias for each included article using the SIGN checklist and as reporting quality using the STROBE checklist; Figure S3 Data Availability Statement: All relevant data are within the paper and its Supporting Information files.