Prevalence of myopia in Indian school children: Meta-analysis of last four decades

Background India is the second most populated country in the world with 41% of the population (492 million) under 18 years of age. While numerous studies have shown an increasing prevalence of myopia worldwide, there continues to be uncertainty about the magnitude of myopia in Indian school going population. Design Systematic review and meta-analysis. Methods We systematically identified published literature of last four decades from 1980 to March 2020 and assessed them for methodological quality. Data were gathered into 5-year age groups from 5–15, in urban or rural populations, and standardized to definition of myopia as refractive error ≥ -0.50 dioptre. Random effects meta-analysis was done. Results We included data from 59 quality assessed studies, covering nearly 1,66,000 urban and 1,20,000 rural children. The overall crude prevalence of myopia over last four decades is 7.5% (95% CI, 6.5–8.5%) in 5-15-year age group. The prevalence of myopia is 8.5% (95% CI, 7.1–9.9%) in urban and 6.1% (95% CI, 4.5–7.7%) in rural children, with highest prevalence in urban 11-15-year age group [15.0% in last decade]. A significant increment in prevalence is noted in the last decade in rural children from 4.6% to 6.8%, reflecting changing rural environment. Conclusion Myopia is an emerging public health problem in both urban and rural school going adolescents in India requiring urgent efforts.


Introduction
Rising prevalence of myopia is a major challenge worldwide, giving rise to an epidemic in certain regions. It is the most common refractive error and an important cause of ocular morbidity especially affecting school going children and young adults. Uncorrected myopia has huge social, economic, psychological and developmental implications [1]. Various studies in the literature have predicted dramatic rise of myopia in the coming years causing a great concern among stakeholders and is projected to affect 50% of world population by 2050 [2]. There is a large regional variation in the myopia prevalence with the dominance of East Asian countries that report a far greater prevalence as compared to other countries [2,3].
India is the second most populated country in the world, with around 41% of its population (492 million) being less than 18 year age group [4]. This young population is an important asset for development of the country and their challenges must be addressed in time. While rising myopia is a cause of concern in most of the countries, it is not given due importance in India due to lack of adequate nationwide prevalence data and prospective studies comparing the trend of myopia over decades [5]. Due to this, the representation of India is poor in studies predicting global trends of myopia [6]. Previous studies by the authors have reported a prevalence of myopia of only 13.1% among school going children in north India with an annual incidence of 3.4% [7,8]. However, due to the large regional differences in culture, habits, socioeconomic status, educational levels and urbanisation, there continues to be an uncertainty about the exact magnitude of myopia burden in Indian school going children and its trend over time. The study was undertaken to fill up this lacuna which can help in understanding the prevalence of myopia, regional variations and prediction of trend, using all the published literature of the last four decades from India.

Methods
The study followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines for the purpose of this review.

Search strategy
We performed a systematic search and review of the prevalence of myopia in India using published data of the last four decades. We searched PubMed, Medline, Embase, OVID, Web of Science, CINAHL and Cochrane library databases from 1 st January 1980 to 31 st March, 2020. Many research articles from India are not available in PubMed search. Thus, we also searched other indexing systems-Index Copernicus and Google Scholar to make our search more inclusive. The search was restricted to all the online available articles mentioning prevalence of myopia in any region of India and published till March 2020. We searched these databases using the following MeSH (Medical Subject Heading) terms and keywords: myopia AND prevalence AND India and refractive error AND prevalence AND India. Broad search strategy also used terms related to epidemiology like epidemiology, incidence, rates, proportion and prevalence, terms related to disease (including medical subject headings search using exp refractive error � , exp myopia � and keyword search using the terms refractive error, myopia and shortsight � ) and terms related to population (including medical subject headings search using exp India � and keyword search using the word India). We also identified and included relevant studies by manually searching the reference lists of eligible studies. Further details about search strategy are available in S1 File.
India. Later, the search was restricted to age group of 5-15 years for this systematic review. Prevalence was defined as the number of individuals in a population that have myopia at a given point of time divided by those at risk. Myopia was defined as spherical equivalent of -0.5 Dioptre or worse [2,[7][8][9]. This standard definition was applied to most of the studies to shortlist them for data abstraction. We covered both urban and rural settings, making our search more representative as majority of Indian population resides in rural villages. Cross-sectional studies including population-based as well as school-based studies were included. Qualitative studies, review articles, articles published in languages other than English and articles which did not have relevant information available online were excluded. A data extraction form was later developed to include all the studies which met our inclusion criteria. Various study characteristics like study design, study population, study location/ region, demographic details (age, gender), screening tools, case definitions used and epidemiological data were compiled from the above studies. We extracted separate urban and rural myopia prevalence rates and gender-based rates, wherever possible. The data was combined and later stratified in each 5-year age groups-5-10; 11-15 years wherever possible. A detailed uniform methodological quality assessment of each of the included study was done by three independent observers using the critical appraisal checklist developed for prevalence studies by Hoy et al. (2012) [10]. Those studies which obtained aggregate score more than six were labelled as 'high risk' studies. Those studies which obtained aggregate score less than four and between four-six were labelled as 'low risk' and 'moderate risk' studies, respectively. Final score was decided based on consensus among the three observers.

Statistical methods
Meta-analysis was carried out using Stata 12.0 (StataCorp LP, Texas, USA). The random effects model using DerSimonian and Laird method was used to calculate pooled effect sizes and its 95% confidence interval (CI) limit [11]. Forest plots were generated displaying prevalence of myopia with corresponding 95% CI. The variation in the magnitude of the effect was examined and heterogeneity was quantified using I 2 statistic. The funnel plot was used to detect potential reporting bias and small/large study effects and Egger method was used to assess asymmetry.
Studies which were categorised as 'high risk' based on assessment of methodological quality described above, were excluded from the final analysis. All studies (low, moderate and high risk) were included in a sensitivity analysis. Urban and rural data was analysed separately. The studies which represented both urban and rural population were later subdivided into separate datasets based on study settings (urban or rural) for detailed analysis. Rural-urban and timestratified estimates of prevalence of myopia across included studies were obtained. For time stratified estimates, year of publication of study was taken for subgroup analysis unless study period was mentioned in the study. Decadal variation was assessed by subgroup analysis of 2009-2019 studies with those of previous decades. Results of rural were compared with urban studies, and studies conducted during 2009-2019 were compared with older studies, by computing z-scores. A sub-group separate analysis for children aged 11-15 years was also done for urban and rural studies. No such analysis can be carried out in 5-10-year age group due to limited number of available studies.

Results
Using the above described search strategy, 469 potentially relevant titles/ abstracts were identified, 165 relevant articles were assessed for eligibility, and 77 studies were found to be eligible . The detailed quality assessment of eligible studies is reported in S1 Table. 18 studies were found to be "high risk". Thus, 59 studies were included in the main analysis, while data from all 77 eligible studies were included in sensitivity analysis. The summary of review strategy is presented in S1 Fig. Out of 59 studies included in main analysis, 37 showed representation of only urban data, 12 showed only rural data and 10 studies showed both urban and rural data. Region wise representation of studies is as follows: North India (12), North East India (4), Central India (6), East India (7), West India (8) and South India (22). All studies were cross sectional in nature. Most of the studies were school-based, with only 4 being population-based. Most of the studies were conducted in the last decade (2009-2019) with only few studies being available before 2009. Gender-based data was available only in few studies, precluding a gender stratified analysis. Overall, the review included around 1,66,000 urban school going children and 1,20,000 rural children over the last 4 decades. The studies were stratified into urban and rural settings and separate analysis was done for the same. Those studies which represented both urban and rural study settings were subdivided into separate datasets as urban or rural subset depending on study setting and data availability. Nine additional datasets were created to represent nine studies where data was available separately for urban and rural setting. The total number of datasets included in analysis were 68 (59 original studies + 9 additional datasets). The details of studies which were included for final statistical analysis are presented in Table 1, along with their study coverage area (urban/ rural/ both). One study did not give rural-urban data separately and was excluded from rural-urban sub-group analysis [22].

Sensitivity analysis
Eighteen studies classified as "high risk" were included in the sensitivity analysis (S2 Table). Similar trends were obtained as main analysis, with 7.4% (95% CI, 7-7.8%) overall pooled prevalence of myopia over last four decades in 5-15 year age group. The overall pooled prevalence of myopia in urban school going children (5-15 years age) was 9.2% (95% CI, 8.2-10.2%) and in rural settings was 7% (95% CI, 5.5-8.5%) in past four decades. Additional details of sensitivity analysis are available in S3 Table.

Discussion
Myopia is emerging as a major public health problem worldwide [2,5]. School going children are one of the most important risk group who constitute a large part of the Indian population [4,5]. The current systematic review estimates the pooled prevalence of myopia, with a focus on studying rural-urban differences and time trends, and included fifty-nine quality assessed studies ensuring adequate rural-urban representation over different time intervals in the main analysis. Results show that the crude prevalence of myopia over last four decades is 7.5% in 5-15-year age group, being 8.5% and 6.1% in urban and rural school going children respectively. The prevalence has increased in rural India from 4.6% in 1980-2008 to 6.8% in 2009-2019, compared to a change from 7.9% to 8.9% in urban India during the same period.
Our results show that there is an increasing trend of myopia in India over the last four decades. Other meta-analyses from different parts of the world have also shown similar trends [2,6,[88][89][90][91]. The prevalence of myopia is much less in Indian school going children as compared to other Asian countries where it could be as high as 70-80% [2,88,89]. While the prevalence may not be as high as that of East Asian countries, the actual numbers of myopes will be large considering our huge population and that 29% of the population consists of children less  [88][89][90][91][92]. This epidemiological variation also holds great importance as it pertains to world's second most populated country which has more than 40% of young population who are at risk of developing myopia. Holden et al has estimated the prevalence of myopia in South Asia region (which includes India) to be around 20% in 2010, 38% in 2030 and 53% in 2050 [2]. We have found a lower prevalence of myopia in school going children in India over the last four decades as compared to other Asian countries where myopia is far more prevalent. Rudnicka et al has also found that increment in myopia prevalence in South Asian countries is less as compared to East Asian countries [90]. Thus, various meta-analyses which predicts global myopia trends fail to bring out this regional variation due to under representation of Indian studies [2,6,90]. This study has shown for the first time that there appears to be a significant rise in the prevalence of myopia in rural school going children. The percentage increment in myopia prevalence among rural school going children was 4 times more as compared to their urban counterparts, in the last decade (48% vs 12%). This is a novel epidemiological finding challenging the previous notion that myopia was less prevalent in rural areas in India as compared to urban areas [5,91]. Systematic review by Sheeladevi et al. showed very low prevalence of myopia in rural settings as compared to urban settings in Indian children (3.5% vs 10.8%) [91]. While this might be a result of a demographic transition, their study assessed only eight school based and four population based studies.
There could be multiple reasons for the increase observed in rural school children. For the past few years, many Indian villages have become developed with access to basic amenities just like their urban counterparts. India is also witnessing a digital revolution starting from the past decade with increasing number of televisions, mobiles, laptops and computers. Internet usage has increased dramatically owing to reduced data tariff, low cost smartphones and improved telecom connectivity in Indian villages. This might have resulted in decreased outdoor activities, increased near work, and computer-related visual stress and fatigue [5,90]. Changing schooling pattern to high pressure education system can also be another contributory factor [93,94]. While direct causal relationship may be difficult to prove, but the rapidly changing environment (nurture) especially the ongoing urbanisation of rural environment in India could be implicated as a potential factor for this rising myopic prevalence.
This study confirms the findings of existing literature that urban adolescents (11)(12)(13)(14)(15) year age group) constitute an important 'at risk' subset of the general population requiring immediate Table 2. Meta-analysis of prevalence of myopia in Indian school-age children, overall and stratified by time-periods and rural-urban population during 1980-2019.

PLOS ONE
attention and intervention where the prevalence of myopia increased to more than double in the last decade. Rural adolescents are also achieving the similar growth rate. Similar trend was

PLOS ONE
obtained in other countries as well because myopia tends to develop after the natural curve of emmetropisation is over [3,88,89,95]. India is geographically and demographically a large country with distinct regional identity and characteristics. Lack of studies reflecting the myopia prevalence from different regions of India and long-time gap between these studies were some important limitations of the study. The studies using inappropriate methodology, not published in English or where the relevant details in study text was unavailable, were excluded. We could not evaluate the prevalence and increment of high myopia which is important to prevent myopia related complications. Although numerous studies have shown an effect of gender on the myopia prevalence, gender-

PLOS ONE
based variations could not be assessed due to limited data availability. Despite great heterogeneity in the results of the studies, we tried to address the differences and bring out some meaningful trends by using stratification, subgroup analysis and random effects model. Similar trends were noted even after including eighteen high risk studies in the sensitivity analysis. Most of the studies which had poor methodological quality were conducted in the last decade. By excluding high risk studies, we adopted a conservative approach. The sensitivity analysis reaffirms the possibility of definite change in epidemiology of myopia in India over time. This large database is also one of the strengths of the present study which has helped to  (11-15 years). The datasets which represented urban and rural data are separately denoted as 'u' and 'r' respectively. Those studies in which urban/rural segregated data was not available are denoted as 'r/u'. https://doi.org/10.1371/journal.pone.0240750.g004

PLOS ONE
predict better trends and highlight subtle variations in epidemiology of myopia. Assessment of the time trends can be accurately done by observing the same population though repeated surveys at definite time intervals but these are rarely collected. Thus, the trend analysis from compiling available data might help in planning policies and setting priorities.
Myopia control programs require consistent efforts to increase awareness about risk factors, encourage lifestyle modification and changes in the school curriculum and education policy of the country. Therefore, this review should help stimulate the initiation of various preventive and corrective measures for myopia control, resource planning and infrastructure augmentation especially targeting the school going children [90,95,96].

Conclusion
To conclude, this is the first Indian study to show and compare the prevalence of myopia in urban and rural settings over the last four decades. It has shown for the first time the rapidly rising trend of myopia in rural school going children compared to their urban counterparts. This should result in adoption of urgent preventive and curative measures among various stakeholders to tackle this menace on time. Future prospective studies should be planned among various diverse regions of India to elucidate the trend of myopia and study various local epidemiological risk factors involved.