Distribution of Eyeglasses in Developing Countries

What is the problem?

The World Health Organization (WHO) reports that 1.3 billion people are estimated to live with some form of distance or near vision impairment.1 Of these, 826 million have near vision impairment, or presbyopia.2 The WHO considers about 80% of vision impairment to be avoidable.3

Visual acuity is reported as the ratio of the distance at which the patient can distinguish a fixed detail (e.g., a letter) to the distance at which a person with 'normal' vision can distinguish the same detail. In the metric system, normal visual acuity is reported as 6/6;4 in the metric system, moderate or severe visual impairment (MSVI) is defined as a visual acuity score in the better eye of less than 6/18 but at least 3/60.5

What is the program?

Corrective eyeglasses can be provided to people with MSVI, either for free, or at a subsidized rate. Beneficiaries are screened to determine what prescription they need; a trained professional may be needed to conduct the screenings.6

Does the program have strong evidence of effectiveness?

Does distributing eyeglasses improve test scores?

There is some evidence that distributing free corrective eyeglasses to children has a positive effect on test scores. Two randomized controlled trials (RCTs) have been conducted in China. These are the only RCTs of this program that we found. They are:

• Glewwe, Park and Zhou 2016, a cluster-RCT7 of a program offering free eyeglasses to school children (25 clusters, 2,529 students with poor vision)8 found that being offered free eyeglasses led to a 0.11 or 0.16 (depending on the statistical method used)9 standard deviation improvement in test scores after one year.10 Results were heterogeneous between the two different counties in which the intervention was carried out.11 Wearing eyeglasses was estimated to lead to an improvement of 0.16 or 0.22 standard deviations (depending on the statistical method used).12 Both were statistically significant, although p-values are not reported.13 We note some questions we have about limitations of this study in this footnote.14
• Ma et al 2014, a preregistered cluster-RCT of an eyeglass program in the same province of China as Glewwe, Park and Zhou 2016, found that being offered free eyeglasses improved test scores by 0.11 standard deviations (95% confidence interval 0.01 to 0.21, P=0.03) compared to the control group (which received a prescription but no free glasses).15 In contrast, children who received vouchers for free glasses at a local facility did not have significantly higher test scores than control (95% CI -0.05 to 0.14, P=0.35) (details in footnote).16

Does distributing eyeglasses increase economic productivity?

We reviewed quantitative evidence from one RCT, which found that distributing eyeglasses markedly improves productivity among tea pickers with impaired near vision in India. Given the simple nature of the intervention, we believe that this evidence is likely to at least partly generalize to other workers who rely heavily on their manual dexterity. We discuss this RCT in more depth on this page.

Glewwe and Schaffner 2014, an RCT investigating the economic impact of providing reading glasses to Rwandan women in weaving cooperatives, concluded there was qualitative and self-reported evidence that eyeglasses increased productivity and were valued.17 The authors state the quantitative results were ambiguous.18

How cost-effective is the program?

We have not attempted to model the cost-effectiveness of distributing eyeglasses to students.

We modeled the cost-effectiveness of distributing eyeglasses to improve workers' manual dexterity. See the cost-effectiveness section of that page for detail.

Does the program appear to have room for more funding?

We have not attempted to estimate the room for more funding for programs that distribute eyeglasses to students.

Our initial analysis suggests that there are a large number of people with vision problems who could potentially benefit from programs that distribute eyeglasses to improve workers’ manual dexterity, but we have not yet assessed what funding gaps exist for particular charities working in this area. See the room for more funding section of our page on eyeglasses to improve workers’ manual dexterity for more detail.

Organizations that implement this program

We have not attempted to identify all organizations that distribute eyeglasses. Organizations that do so include:

• VisionSpring19
• Eyes on Africa20
• Lions International21
• Vision for a Nation22
• Sightsavers23

Our Process

We conducted a brief literature review of results on Google Scholar for the search term “eyeglasses randomized controlled trial,” and reviewed World Health Organization information on MSVI.

Sources

• 1

  "Globally, it is estimated that approximately 1.3 billion people live with some form of distance or near vision impairment." WHO Vision Impairment Fact Sheet, 2018.

• 2

  "With regards to near vision, 826 million people live with a near vision impairment." WHO Vision Impairment Fact Sheet, 2018.

• 3

  "Approximately 80% of vision impairment globally is considered avoidable." WHO Vision Impairment Fact Sheet, 2018.

• 4

  In other words, visual acuity of 6/6 means that a person can see detail from six meters to the same degree that a person with 'normal' eyesight can see from six meters.

  For background, see WHO Consultation on Development of Standards for Characterization of Vision Loss and Visual Functioning, 2003

• 5
  • "The table below gives a classification of severity of visual impairment recommended by the Resolution of the International Council of Ophthalmology (2002) and the Recommendations of the WHO Consultation on "Development of Standards for Characterization of Vision Loss and Visual Functioning" (September 2003)" In the table referenced,
    • The category "0 Mild or no visual impairment" is defined as presenting distance visual acuity equal to or better than 6/18.
    • The category "1 Moderate visual impairment" is defined as presenting distance visual acuity worse than 6/18 and equal to or better than 6/60.
    • The category "2 Severe visual impairment" is defined as presenting distance visual acuity worse than 6/60 and equal to or better than 3/60.
    • The category "3 Blindness" is defined as presenting distance visual acuity worse than 3/60 and equal to or better than 1/60 (or counts fingers (CF) at 1 metre).
    • The category "4 Blindness" is defined as presenting distance visual acuity worse than 1/60 (or counts fingers (CF) at 1 metre) and equal to or better than light perception.
    • The category "5 Blindness" is defined as no light perception.

  WHO International Statistical Classification of Diseases, Injuries and Causes of Death, 2016, H54.

• 6

  "In each county, an optometrist hired for the summer visited all townships to conduct formal eye exams for students who accepted the glasses. If poor vision was confirmed, they were prescribed appropriate lenses." Glewwe, Park and Zhou 2016, Pg 6.

• 7

  “Random assignment was conducted in 2004 as follows. In each county, all participating townships were ranked by 2003 per capita income. Starting with the two wealthiest, one was randomly assigned to be a treated township and the other to the control group; this was repeated for all subsequent township pairs. In County 1, the 19th township (the poorest) was not paired with another township; it was randomly assigned to the treatment group. In each township primary schools were either all assigned to the treated group or all to the control group” Glewwe, Park and Zhou 2016, P 6.

• 8
  • “Of the 2,529 students with poor vision in the compliant sample, 1,528 were in the program schools. Of these, 1,066 (69.8%) accepted, while the other 462 declined, the eyeglasses.” Glewwe, Park and Zhou 2016, Pg 9.
  • The study was originally intended to include 37 clusters. Twelve of these were excluded on the basis that the randomization was incorrectly implemented. Results are reported for the 25 compliant clusters.
    • ”Unfortunately, in 5 of the 18 control townships some students were given eyeglasses; after providing eyeglasses in the treatment townships, local officials used the remaining funds to buy them for students with poor vision in the paired control township. This occurred in two control townships in County 1,3 and three in County 2. In another township pair in County 1, there was a “role reversal”; no one in the treatment township was offered glasses, while many children with poor vision in the control township were offered glasses. In contrast, the randomization was correctly implemented for six pairs of townships in County 2 and six pairs (plus the poorest township, the one randomly assigned to be treated) in County 1. To check for selection bias due to the exclusion of township pairs that deviated from random assignment, we conducted several robustness checks; they provide very little evidence of such bias.” Glewwe, Park and Zhou 2016, Pg 7.
    • ”Table 1 presents descriptive statistics for the 25 “compliant” townships (township pairs for which both townships complied with their random assignment) and for all 37 townships.” Glewwe, Park and Zhou 2016, Pg 8.

• 9
  • Glewwe, Park and Zhou 2016 uses two different methods to estimate the impact of being offered eyeglasses for children with poor vision.
    • ”The simplest estimate of the program’s impact on students with poor vision is a t-test that compares the mean test scores of students with poor vision in the program schools with the same mean for their counterparts in the control schools. This estimates the impact of offering eyeglasses (intent to treat effect), not the impact of receiving them.” Glewwe, Park and Zhou 2016, P.10
    • ”Equation (1) uses only students with poor vision. More precise estimates of β can be
      obtained by adding students with good vision. This “double difference” method compares the
      difference in test scores between students with good vision and poor vision in treatment schools to the same difference in control schools.
  • The ‘full sample’ equation (equation 1) found a 0.11 standard deviation improvement in average test scores for children with poor vision. The ‘poor vision only’ equation found a 0.16 standard deviation improvement in average test scores for children with poor vision. Both estimates are statistically significant at the 5% level using conventional standard errors. Using wild bootstrapped standard errors means only the 0.11 standard deviation improvement is statistically significant at the 5% level.
    • ”Equation (1): Compliant Sample, Only Students with Poor Vision; Treatment Township (β): 0.158” Glewwe, Park and Zhou 2016, P 35. Table 4
    • ”Equation (2): Compliant Sample, All Students; Poor Vision x Treatment Township (β): 0.109” Glewwe, Park and Zhou 2016, P 35. Table 4
    • “Both estimates are statistically significant using the non-bootstrapped standard errors, but the wild bootstrap yields significance at conventional levels only for the sample of all children (p-value of 0.046, compared to 0.118 for the sample of poor vision children).” Glewwe, Park and Zhou 2016, Pg 176.

• 10

  “Baseline data were collected in June of 2004 (end of the 2003-04 school year) on student
  characteristics, exam scores, and visual acuity. Data were collected from treatment and control
  schools for all students finishing grades 1-5 in June of 2004. Treatment school students slated to enter grades 4-6 in the fall of 2004 who had poor vision were offered free eyeglasses. In each
  county, an optometrist hired for the summer visited all townships to conduct formal eye exams
  for students who accepted the glasses. If poor vision was confirmed, they were prescribed
  appropriate lenses. Students had a limited choice of colors and styles for their eyeglasses. All the eyeglasses were ordered from a reputable company. The 2004 fall semester began on August 26; most students who accepted the offer received glasses by mid-September. Teacher monitoring and field visits by project staff found high rates of wearing eyeglasses. At the end of the 2004-05 school year (late June/early July of 2005), fall and spring semester exam scores were collected.” Glewwe, Park and Zhou 2016, P 6.

• 11

  “We also find evidence of heterogeneous treatment effects. Most notably, in one county the
  intent-to-treat effect on test scores was 0.26 to 0.39 standard deviations, while there was little or
  no impact in the other county.” Glewwe, Park and Zhou 2016, P 2.

• 12
  • Equation (1): Compliant Sample, Only Students with Poor Vision; Treatment Township (β): 0.156 (0.071)
  • Equation (2): Compliant Sample, All Students; Poor Vision x Treatment Township (β):

  Glewwe, Park and Zhou 2016, Appendix Table A6.

• 13

  “Pooling both counties’ compliant samples, the impacts on test scores are 0.22 (0.16) standard deviations for the sample of poor vision children (all children), and significant at the 5% level.” Glewwe, Park and Zhou 2016, P 18.

• 14
  • Glewwe, Park and Zhou 2016 uses both conventional standard errors and wild bootstrap standard errors to assess statistical significance. Using wild bootstrap errors gives statistically significant results at a 5% level for the sample of all children, but does not give statistically significant results for the sample of children with poor vision. We are unsure about why this would be the case.
    • "...covariance matrices that allow for clustered errors are valid only as the number of clusters (townships) goes to infinity. Our estimates that drop township pairs with a township that did not follow its random assignment are based on 25 townships. Such covariance matrices can be misleading if there are 30 or fewer clusters (Cameron et al., 2008). For robustness, we present p-values estimated using the wild bootstrap, as Cameron et al. (2008) suggest." Glewwe, Park and Zhou 2016, P 174.
    • "Both estimates are statistically significant using the non-bootstrapped standard errors, but the wild bootstrap yields significance at conventional levels only for the sample of all children (p-value of 0.046, compared to 0.118 for the sample of poor vision children)." Glewwe, Park and Zhou 2016, P 176.
  • Glewwe, Park and Zhou 2016 was published in June 2016, 12 years after the intervention was carried out. We would like to better understand the reasons for the delay.

• 15

  “Table 3 gives the baseline, endline, and change in the mathematics score by study group. Intention to treat analyses showed a significant difference in endline score adjusted for baseline score of 0.11 SD (95% confidence interval 0.01 to 0.21, P=0.03) for the free glasses group and 0.04 (−0.05 to 0.14, P=0.35) for the voucher group compared with controls. The unadjusted intraclass correlation coefficient for endline math score was 0.12; adjusted for baseline scores it was 0.07.” Ma et al 2014, P 4.

• 16
  • Study description:
    “Design Cluster randomized, investigator masked, controlled trial.
    Setting 252 primary schools in two prefectures in western China, 2012-13.
    Participants 3177 of 19 934 children in fourth and fifth grades (mean age 10.5 years) with visual acuity 6/12 in either eye without glasses correctable to >6/12 with glasses. 3052 (96.0%) completed the study.
    Interventions Children were randomized by school (84 schools per arm) to one of three interventions at the beginning of the school year: prescription for glasses only (control group), vouchers for free glasses at a local facility, or free glasses provided in class.
    Main outcome measures Spectacle wear at endline examination and end of year score on a specially designed mathematics test, adjusted for baseline score and expressed in standard deviations.
    Results Among 3177 eligible children, 1036 (32.6%) were randomized to control, 988 (31.1%) to vouchers, and 1153 (36.3%) to free glasses in class. All eligible children would benefit from glasses, but only 15% wore them at baseline. At closeout glasses wear was 41% (observed) and 68% (self reported) in the free glasses group, and 26% (observed) and 37% (self reported) in the controls. Effect on test score was 0.11 SD (95% confidence interval 0.01 to 0.21) when the free glasses group was compared with the control group. The adjusted effect of providing free glasses (0.10, 0.002 to 0.19) was greater than parental education (0.03, −0.04 to 0.09) or family wealth (0.01, −0.06 to 0.08). This difference between groups was significant, but was smaller than the prespecified 0.20 SD difference that the study was powered to detect.” Ma et al 2014
  • ”Intention to treat analyses showed a significant difference in endline score adjusted for baseline score of 0.11 SD (95% confidence interval 0.01 to 0.21, P=0.03) for the free glasses group and 0.04 (−0.05 to 0.14, P=0.35) for the voucher group compared with controls. The unadjusted intraclass correlation coefficient for endline math score was 0.12; adjusted for baseline scores it was 0.07.” Ma et al 2014

• 17

  “According to qualitative data gathered during the follow-up study sessions, of the 47 women who received glasses immediately after the baseline sessions, 46 reported that the glasses allowed them to weave more quickly or more accurately, and the great majority of them (41 of 47) reported their willingness to pay at least 2,000 Rwandan francs (Rwf) to replace their glasses. This is more than the 1,300 Rwf cost of a pair of near vision glasses. Moreover, nearly half of them were willing to pay much more (at least 10,000 Rwf).” Glewwe and Schaffner 2014, Pg 1.

• 18

  “Quantitative evidence derived from the randomized control trial (RCT) design proved less useful than anticipated because, for reasons examined in the full report, the results suffer from statistical weakness and ambiguity in interpretation. It is likely that a follow-up study employing a larger number of study groups, and a compensation scheme that provides participants with stronger and more balanced motivation to pursue quality as well as quantity, would yield more clear-cut results.” Glewwe and Schaffner 2014, Pg 1.

• 19

  VisionSpring website

• 20

  Eyes on Africa website

• 21

  Lions International website

• 22

  Vision for a Nation website

• 23

  Sightsavers website