Prevalence of Refractive Errors and Accuracy of School-Based Visual Acuity Screening Programs Among School Children In Vietnam


 Background

School-based visual acuity screening programs can help identify students with possible visual impairments, inform their parents of their status and the need for further examination and other conditions to assist students with visual impairments.
Objectives

This study aimed to calculate the prevalence of refractive errors among students and to describe the accuracy of visual acuity screening practices of school staff in some selected provinces in Viet Nam.
Methods

Visual acuity assessments were conducted by trained teachers and nurses at all schools participating in the survey, covering a sample size of 4,838 students. Around 30% of these students were then randomly selected for a close-up examination by ophthalmologists, making up a sub-sample of 1,404 students. The number of cases of low vision based on a set of criteria. The accuracy of visual acuity screening activities by comparing school staff’s and ophthalmologists’ results.
Results

The proportion of students with untreated eye problems and that of students with uncorrected refractive error are currently quite high (at respectively 18.52% and 24.64%), especially in urban areas in Da Nang and Hai Duong provinces. Among every three children with refractive errors (RE), two of them have not received appropriate correction. The accuracy of the visual acuity screenings performed by teachers and school nurses in the project’s target area is demonstrated by a sensitivity of 60.92% and a specificity of 93.76%.
Conclusions

This indicates that efforts toward detecting school students with poor vision at school needs further investments to mitigate the current problems. This outcome suggests that teachers and school nurses need to be trained and equipped further with better skills and knowledge in conducting screening to improve their screening performance.

Vietnam's National Prevention of Blindness Strategic Plan (2015-2020) identi ed refractive errors (RE) and childhood blindness as among key areas to be addressed to reduce the burden of avoidable blindness throughout the country. However, eye health has not been yet o cially included in the school health curriculum. Besides, there is the lack of guidelines for implementing vision screening for students during annual school health checks. Additionally, there exists the shortcomings in the eye health awareness among students and teachers, as well as the lack of resources and training opportunities for school nurses and teachers to provide eye care and appropriate referrals (1).
The Fred Hollows Foundation (The Foundation) is a developmental organization aligned with Vietnam's Strategic Plan. The Foundation established the Vietnam Child Eye Care (VNCEC) and Better Eyes for Better Education (BEBE) projects. These two projects are piloted in three provinces Hai Duong, Da Nang and Tien Giang given these provinces are well equipped with eye health infrastructure and human resources by previous works of international non-governmental organisations.
School-based visual acuity screening programs can help identify students with possible visual impairments, inform their parents of their status and the need for further examination and other conditions to assist students with visual impairments (2), (3)(4)(5). Several developing countries have undertaken such studies on RE screening to determine the level of suitability of the programs in the school system as different countries have different infrastructure, systems and methods ,(6), (7). As the VNCEC project is targeting to reduce visual issues, including RE, among schoolchildren in Vietnam, it is critical that Vietnam has a similar study on school RE screening, which should focus on both primary and lower-secondary schoolchildren. This part aims to review RE screening practices of other programs in the world, especially developing countries and key issues to be noted in carrying out such programs.
Firstly, the cut-off points or a threshold for RE screening adopted may affect the cost and effectiveness of RE screening activities. Murthy (2000) suggests that it is possible to use the < 6/12 (equivalent to 5/10) in the better eye threshold to increase the cost effectiveness of the school program (8). Gianni et al argues that decreasing the cut-off point would lead to higher number of false negatives, which are the main cases to be avoided, and increasing the cut-off might lead to a larger number of false positive cases, which may affect the cost of the program (9). Secondly, the level of accuracy of the screenings conducted by school staff varies sharply in previous studies in the literature. In studies in Asian countries which adopted the 5/10 (or equivalent) cut-off point, the level of sensitivity, which demonstrates the proportion of correctly identi ed positive cases among actual positive cases, could be as high as 93.5% in China (10) and as low as 59.0% (11) in Thailand. However, it can be noticed that there are few studies that examined the accuracy of RE screenings conducted on primary school students, and the studies involving these subjects tend to feature lower rates of accuracy, such as those done by Teerawattananon et al. (2014) and Ostadi Moghaddam et al. (2012) (11), (12). A study in Vietnam in Vung Tau presents a relatively high level of accuracy, at 86.7% sensitivity and 95.7% speci city, however without the participation of primary school students (13). Lastly, external factors such as parents' co-operation and training of school staff can have impacts on the implementation and effectiveness of such RE screening activities in school. In the study in Thailand, only 470 out of 624 students who were referred to ophthalmologists at provincial hospitals actually went there because of not obtaining consent from their parents or they did not show up (11).
There were only a few studies on the prevalence of RE in Vietnam, which mostly focus on adults and students in secondary schools in speci c areas. In Ho Chi Minh City in 2009, Le Thi Thanh Xuyen et al.
(2009) reported a very high rate of RE of 39.4%, mostly consisting of myopia, among students in lower and upper-secondary schools (14). The authors also noticed a signi cantly higher prevalence of RE in urban areas compared with urban ones. In Ha Noi, in the same year, the rate of myopia was 33.7% (15). Signi cantly, this rate rose to as high as 40.0% the urban area. In Vung Tau, Paudel et al. (2014) observed that the rate of RE was 21.5% on average, and 27.5% in urban areas (13). There was, however, a lack of studies that measure the prevalence of RE in other provinces, and especially among younger schoolchildren, including those in primary schools.
From the above literature review, it has pointed out that there is currently a lack of studies in Vietnam on the feasibility of school-based screenings for RE. There is an urgent need to address these gaps in the literature, especially in the context of Hai Duong, Da Nang and Tien Giang, given the fact that VNCEC and BEBE are being implemented in these provinces. The results of the study will serve as an input for the design of school eye health activities. This study would ultimately provide the rationale to inform the development of programmatic approaches and advocacy efforts. This baseline study's objectives are to identify the prevalence of RE among students as well as describe the accuracy of visual acuity screening practices of school staff in selected provinces.

Study Design
The baseline study covers two main components: 1. Visual acuity (VA) screening performed by teachers and school nurses (commonly referred to as school staff); 2. Follow-up RE screening performed by ophthalmologists;

Study Setting
As the VNCEC and BEBE projects are delivered in Hai Duong, Da Nang and Tien Giang, these three provinces were chosen for the study by The Foundation. The Foundation provided a list of schools participating in the projects, and a list of schools in the non-target area in Da Nang to Mekong Development Research Institute (MDRI). Based on this list, MDRI's sampling experts chose 26 schools in the target area of the projects, and 9 schools in the non-target area for the study.
As the ratio of the number of students in primary schools and lower-secondary schools participating in the project's area was around 1.2, school selection was done using the Probability Proportional to Size (PPS) approach, with the target of maintaining this ratio of students. Under this sampling approach, the schools with larger number of students were more likely to be chosen in the sample. The nal number of schools chosen in the target area was 9 schools in Da Nang (comprising 5 primary schools), 9 schools in Tien Giang (comprising 6 primary schools), and 8 schools in Hai Duong (comprising 4 primary schools). In the non-target area in Da Nang, 9 schools were chosen, among which 5 were at the primary level.

Participants and Recruitment
With regards to VA screening by school staff and ophthalmologists, initially, students in all grades were selected to participate in this activity; however, during the rst day of the eldwork, The Foundation and MDRI realised that VA screening using alphabetical visual charts was not suitable for grade 1 students. Therefore, o cially, only those from grade 2 to grade 9 took part in this part of the study.

Recruitment Approach
To ensure the representativeness of the survey, the multistage strati ed sampling, Probability Proportional to Size sampling and random sampling techniques were used to select the sample. At rst, the Probability Proportional to Size approach was used to select 26 target schools and 9 non-target schools, as discussed above in the Study Setting section. After the list of all classes and all students in selected school was available, the random sampling technique was used to identify one class per grade in each school, and all students from the selected class were chosen to participate in VA assessment. Among the students screened by school staff, 25% of them were randomly chosen to take part in a follow-up RE screening performed by ophthalmologists.

Sample Size
The sample size was estimated separately for the two groups: target and non-target. Visual acuity (VA) assessments were conducted by trained teachers and nurses at all schools participating in the survey, covering a sample size of 4,838 students. Around 30% of these students were then randomly selected for a close-up examination by ophthalmologists, making up a sub-sample of 1,404 students.

Data Collection
There is a variety of materials used in the baseline study to collect the data. Each component has its unique material and technique. In term of VA screening by school staff, we used training documents for nurses and other school staff on knowledge of RE and VA assessment procedure (Alphabetical) Visual chart Hard tape for measuring distance template to record VA screening result. We also refered to the supporting documents -VA screening guideline. Regarding to follow-up RE screening by ophthalmologis, we used (Alphabetical) Visual chart; Pinhole glasses and lens kit with trial lens set; Autorefractometer Ophthalmoscope; retinoscope Jackson Crossed Cylinder Auto lensmeter; Dilating eye drops; Template to record RE screening result. This is the initial list of equipment. During the implementation, due to the parents not approving the use of dilating eye drops, the ophthalmologists just screened for low VA and eye diseases and therefore utilised fewer materials.

Data Synthesis and Analysis
Analysis of the prevalence of RE and visual issues Based on ophthalmologists' results, the research team calculated the number of cases of low vision based on a set of criteria. Sampling weights were used to re ect the representativeness of the sample.
Some indicators for analysis as untreated vision problems (VA with current glasses of one or both eyes < 7/10 or presenting eye diseases in one or both eyes), refractive error (Concluded as having RE in one or both eyes by ophthalmologists), uncorrected refractive error (Concluded as having RE in one or both eyes by ophthalmologists and VA with current glasses of one or both eyes < 7/10), corrected refractive error (Concluded as having RE in one or both eyes by ophthalmologists and VA with current glasses of both eyes >= 7/10), mild cases of refractive error (that do not require correction) (Concluded as having RE in one or both eyes by ophthalmologists and VA without glasses of both eyes >= 7/10).
Untreated vision problems was de ned as VA with current glasses of one or both eyes < 7/10 or presenting eye diseases in one or both eyes Analysis of school staff's performance The research team assessed the accuracy of VA screening activities by comparing school staff's and ophthalmologists' results. From ophthalmologists' results, the students who were concluded by ophthalmologists as having a vision of lower than 7/10 in either eyes (with current pair of glasses, if they already had one), or having abnormalities in either eyes, were marked as those with "positive" results. From school staff's results, the research team based solely on their nal column in the result forms (about whether the student was referred for further examination) to identify those with "positive" results. This was because the study is more interested in the accuracy of the referrals that school staff made than their detailed VA results.
School staff's results were then compared to those of ophthalmologists by a set of indicators, consisting of sensitivity, speci city, positive predictive value and negative predictive value. Sensitivity was calculated as the ability of school staff to identify low VA (The proportion of actual cases of low VA that are correctly identi ed by school staff). Meanwhile, speci city was analysed as The ability of school staff to identify normal VA (The proportion of actual cases of normal VA that are correctly by school staff). Department in each province also provided strong support to arrange logistics, in order to help the enumerators to approach the schools more easily. Consent forms were sent to parents of students participating in the examination. The study was only conducted with children's parent consent agreement.

Prevalence of RE and visual issues
As demonstrated in Table 3.1, the overall prevalence of RE was 24.64% in the target area, of this, 17.94% were corrected, 11.54% were mild cases, and 70.53% were uncorrected cases. RE affected 32.65% of the student population in the non-target districts of Da Nang, 44.27% of those in Da Nang's targeted districts, 35.60% of those in Hai Duong and only 6.42% of students in the project's target areas in Tien Giang, where the prevalence of RE was the lowest.
Breaking down these results by classi cation criteria also revealed some interesting trends. There was no proven difference between male and female students in terms of eye diseases and RE, however, there was a signi cant difference (at 99% of con dence) of 27.59% in the prevalence of RE (in one or both eyes) in rural (14.26%) compared to urban (41.85%) areas, proving once again a well-observed fact in the eye care community. In addition, lower-secondary students were more likely to have untreated eye diseases and RE than their primary counterparts (at 99% of con dence).
While there were several notable differences in the prevalence of eye issues and RE between provinces, area types and school levels, there was no striking difference in terms of corrected, mild or uncorrected RE proportions across those classi cation criteria. The general trend was that only 15 to 20% of RE cases have been corrected, together with around 10% of mild cases requiring no treatment and leaving behind 60% to 70% of uncorrected cases.
It has also been shown from the results that the prevalence of untreated vision problems (including RE) in one or both eyes varies among areas. Speci cally, Tien Giang had a signi cantly lower proportion of untreated cases of visual issues (around 4.49%) than in other provinces. The non-target area in Da Nang also showed a lower level of prevalence compared to the targeted area in the same province (23.66% compared to 31.94%, veri ed by a t-test at 95% of con dence). Meanwhile, there were no statistically signi cant difference between the target areas in Da Nang and Hai Duong. The average prevalence of untreated eye issues was around 28 to 32% of in these two areas. Performance of school staff in conducting VA assessment The performance of school staff plays a key role in the effectiveness of school eye health intervention because their initial stage of VA screening will help determine the students who need further examination by ophthalmologists. It is important to maximise correctly speci ed positive cases and avoid false negative cases as these would be against both the students' and the project's interests. Table 2 below shows some diagnostic statistics demonstrating the accuracy of VA screening performed by school staff. Target and non-target areas are separated and the "Total" value only considers target areas as some interventions have only been undertaken there. In the target area, school staff's overall performance was re ected by a sensitivity measure of 60.92%. It means that only 60.92% of students with either eye abnormalities or whose vision was lower than 7/10 in one or both eyes were identi ed by school staff. The speci city level demonstrated that 93.76% of students with good vision were recognised by school staff, leaving a false positive rate of approximately 6.24%. Among the students who were referred to ophthalmologists, 74.36% were actual cases of low vision (as shown in the positive predictive value). This means that ophthalmologists would need to spend extra time and efforts on the 25.64% of students who were incorrectly referred to them.
In the non-target area in Da Nang, the sensitivity, speci city, positive predictive and negative predictive values were respectively 69.01%, 90.09%, 68.06% and 90.48%. In comparison with those of the target area in the same province, these statistics demonstrated a better coverage of correct referrals, however they also indicated a slightly higher level of false positive cases, as shown by a lower speci city measure.
Across the surveyed provinces in the target areas, Hai Duong performed signi cantly better than Da Nang in terms of sensitivity, with 9.53% more cases of low vision being identi ed. Between Tien Giang and two other provinces, although the sensitivity measure seems to be higher, statistical tests proved that they only featured a signi cantly higher level of speci city, lower level of positive predictive value and higher level of negative predictive value, without any rm evidence backing the difference in sensitivity. These did not mean that school staff in Tien Giang performed better, but were most likely the results of the low prevalence of low vision and RE in Tien Giang.
Statistically signi cant differences have been revealed between urban and rural areas. Statistical test results had demonstrated that in rural areas, school staffs were better at identifying low vision, with a sensitivity measure of 74.00%, compared to 57.45% in urban areas. Their positive predictive value was also signi cantly lower and they also demonstrated a higher negative predictive value. This implied that rural school staff had fewer cases of false negative results.

Discussion
This part discusses the results of VA screenings, including guring out the proportion of school children with untreated vision problems, uncorrected RE and corrected RE in the surveyed areas and assessing the accuracy of VA screening at the school level. Regarding the prevalence of presenting low VA and RE among school children, the results of this study con rm previously investigated trends in surveys in other Vietnamese locations. The accuracy of school staff's screening activities, however, is lower than previously observed in other provinces, with some noticeable differences across location and area type. Tien Giang (6.42%) can be considered consistent, with an overall rate of 24.64% in all three provinces (13). Discrepancies between these provinces can be due to the unbalanced proportion of rural and urban schools, as well as of primary and lower-secondary schools among them. Speci cally, all schools in Da Nang's target area are urban schools while most of the schools in Tien Giang are located in rural areas.
A limitation of this study is that due to parents' disapproval with the ophthalmologists' use of dilating eye drops when conducting eye examination at the schools, students' speci c RE condition (myopia, hyperopia or astigmatism) was not collected. Instead, the ophthalmologists examined the students' VA like the school staff did to assess their accuracy. If any abnormalities were noticed, a result form has been sent to the parents to inform them so they can have their child further checked by an eye specialist. Therefore, without such information, comparison with existing studies in terms of each condition is not possible.
In terms of school staff's performance, the overall performance in the target area is much weaker than previously reported by Paudel et al. (2014) in Vung Tau, where the accuracy is observed at 86.7% sensitivity, 95.7% speci city, 86.7% positive predictive value and 95.7% negative predictive value (13).
When compared to studies in other Asian countries (Table 3.2.3), the sensitivity and speci city measures in this study are only higher than those of Iran. Nevertheless, those studies mostly approached secondary school students. When it comes to screening primary school children's VA, most previous studies feature lower detection rates. One factor affecting the screening performance of school staff which has not been frequently discussed is the quality of the training given to them. Paudel et al. (2014) only mentioned that they conducted a halfday training session with practice and provision of necessary equipment (13). This is very similar with the training sessions that were held at the three provinces before the eld work of this study. However, it has been noticed during The Foundation and MDRI's supervision that the school staff (especially in Da Nang's target area) were initially not very con dent when implementing VA screening at their schools, and were not clear about how to record results on the VA form. This was improved by The Foundation and MDRI in the survey in the non-target area by providing school staff more time to practice before the eld work. That might explain why school staff in the non-target area performed signi cantly better than their colleagues in targeted schools.
The fact that school staff in rural areas performed signi cantly better than those in urban schools might be attributed to two factors. Firstly, in rural areas, eye issues are less prevalent, and thus school staff had more time to examine the students who showed signs of low vision. In urban areas, with a high rate of untreated visual issues at around 30%, school staff's attention was more diverted. In addition, teachers could only conduct VA screening during a xed amount of time allocated for this activity in the timetable and very few schools allowed extra time. This xed amount of time and the large proportion of students with low vision might have affected the quality of their screening activities.
A small number of unidenti ed cases of eye problems (8/118 -in the whole target and non-target sample) such as conjunctivitis, blepharitis, conjunctiva stones and retinal detachment were encountered by school staff. They were often overlooked as school staff have not been trained to identify such abnormalities.
Although the proportion of students with these diseases is small, the project may consider training school staff on identifying these abnormalities and how to correctly refer these students for treatment.

Limitations
This baseline study has some limitations. First is the disagreement of the parents with the use of dilating eye drops in visual acuity assessment. Second is the waiting time between the teacher's assessment and doctor's follow-up examination. Speci cally, ophthalmologists are unable to conduct the eye examinations thoroughly as permission has not been granted by parents for the use of dilating eye drops on the children, although the research team has tried to convince them. However, as the safety of the children and the cooperation of the parents are highly valued by the research team, the research's design has been modi ed to remove this component. The children identi ed as having visual issues by teachers and con rmed by ophthalmologists have therefore been referred to local hospitals for further examination. This has in turn made this study unable to analyse in more detailed indicators for various forms of RE. In addition, visual acuity screenings performed by school staff and doctors take place in separate days for most schools.
This may lead to inconsistencies in the evaluation of students' visual acuity between two days of screening, especially when the eyes of children possess great adjustment capacity during school age.

Conclusions
The proportion of students with untreated eye problems and that of students with uncorrected refractive error are currently quite high (at respectively 18.52% and 24.64%), especially in urban areas in Da Nang and Hai Duong provinces. The accuracy of the visual acuity screenings performed by teachers and school nurses in the project's target area is demonstrated by a sensitivity of 60.92% and a speci city of 93.76%. This indicates that efforts toward detecting school students with poor vision at school needs further investments to mitigate the current problems.This outcome suggests that teachers and school nurses need to be trained and equipped further with better skills and knowledge in conducting screening to improve their screening performance.

Declarations
Funding: This research received no external funding.
Acknowledgements: This survey is a project supported by funding from Standard Chartered Bank via The Fred Hollows Foundation. The information and opinions contained in it do not necessarily re ect the views or policy of either Standard Chartered Bank or The Fred Hollows Foundation. We are also grateful to the Project Management Board of the School Eye Heath projects, as well as 35 primary and lower-secondary schools in Hai Duong, Da Nang and Tien Giang for their strong support during the project's eld work. We owe thanks to the great team of senior ophthalmologists from Hai Duong Hospital of Ophthalmology and Dermatology, Da Dang Hospital of Ophthalmology and Tien Giang Hospital of Ophthalmology for providing a thorough training to 175 teachers and school nurses in the three provinces, and for conducting visual acuity tests for over 1,400 students. Last but not least, the implementation of this project would not have been possible without the contribution of 12 enthusiastic enumerators whose professional experience has helped ensure that our collected data is truthful and always of the highest quality.
CompetingInterest: The authors declare no con ict of interest.
Ethics approval: Subjects were adequately informed about the purpose of the study and participated voluntarily. Information on the study subjects is con dential and data collected in this study were only used for research purposes. The study did not affect the rights, health, and economy of the study subjects.
Consent to participate: Verbal informed consent was obtained prior to the interview.
Consent for publication: Not applicable