London School of Hygiene & Tropical Medicine — RCT of Eyeglass Provision (May 2024)

Note: This page summarizes the rationale behind a GiveWell grant to the London School of Hygiene & Tropical Medicine. London School of Hygiene & Tropical Medicine staff reviewed this page prior to publication.

Published: August 2024

The organizations

The London School of Hygiene & Tropical Medicine (LSHTM) is a center for research and postgraduate education in public and global health. It is based in the United Kingdom, with offices in The Gambia and Uganda.1

This trial will be led by members of LSHTM’s International Centre for Eye Health (ICEH), which works to improve vision and eye health in low- and middle-income countries through research, capacity-building, education, and technology.2 The study team includes Professor Matthew Burton and Professor Andrew Bastawrous (principal investigators), Dr. David Macleod (trial statistician), Dr. Patricia Marques and Professor John Cairns (health economists), Dr. Elliot Taylor (research fellow), and Dr. Priya Morjaria and Ian McCormick (research optometrists).3

Professor Bastawrous is the Founder and CEO of Peek Vision and its parent charity The Peek Vision Foundation, which was spun out of LSHTM.4 (Professor Burton is a Trustee of Peek Vision.5 ) Peek Vision is a social enterprise that develops smartphone technology and public health measures to support eyecare services in low- and middle-income countries.6 Peek Vision will supply screening tests and a software platform for the trial.

The trial itself will be implemented by Dr. Shroff’s Charity Eye Hospital in India, and by Kisii Eye Hospital in conjunction with a local non-profit in Kenya.7

The intervention

Estimates for the number of people globally with near vision impairment caused by unaddressed presbyopia range between 510 million and 826 million people.8

Those suffering from visual impairment may experience a loss of productivity at work, which may, in turn, affect their employment opportunities and incomes. These economic effects are likely to be especially pronounced for individuals who rely on their manual dexterity (fine motor control) and/or visual discrimination for their work.

This study is designed to inform our understanding and cost-effectiveness estimates of programs that involve distributing eyeglasses (either for free or at a subsidized rate) to people whose ability to be productive at work is constrained by their impaired near vision. While programs typically screen individuals to determine their vision-related needs, they vary in the extent to which they target individuals according to other characteristics, such as their occupations. Programs may also vary in the types of eyeglasses they distribute. Our full report on this intervention is forthcoming.

At present, we estimate that eyeglass provision may be around 15 times as cost-effective as unconditional cash transfers for workers in rural Bangladesh, and around 11 times as cost-effective as unconditional cash transfers for tea pickers in Assam, India.9 These cost-effectiveness estimates are based on two randomized controlled trials, one for each area.10 At face value, this would place eyeglass provision above our funding bar, at least in these specific contexts.

Based on this evidence, we would consider recommending grants for comparable programs. However, we currently have a number of uncertainties around the existing evidence for the intervention, which limits the amount of money we would be willing to direct. In particular, our main uncertainties around the intervention’s impact are:

• Income and consumption:
  • We are not fully confident in our estimate of the effect of eyeglass provision on income, since the effect on income in the Bangladesh study was self-reported,11 and therefore subject to social desirability bias, while the India study measured workplace productivity, rather than income directly.12 As a result, our current best guess is that across both studies, the 25th-75th percentile interval for the effect of eyeglasses on income is an increase of between 0% and 14% per year. This would imply a cost-effectiveness range of between 2 and 41 times cash transfers.13
  • Furthermore, income may not be the best way to capture the impact of these programs. GiveWell’s moral weights ultimately value consumption, rather than income. We believe consumption is a more reliable proxy for material well-being in low-income countries, primarily because there are more established methodologies for estimating consumption in this context,14 and consumption helps to aggregate the different potential sources of income and offsetting effects.
• Generalizability: We are uncertain about the extent to which the existing Bangladesh and India studies are informative for programs we might fund, since (a) the trials targeted specific occupations that stood to gain the most from eyeglasses, rather than a broader share of the population,15 the latter of which we believe would be more representative of a scaled-up program,16 (b) we assume that, in general, few income-generating programs successfully replicate in new contexts, and (c) we are unsure whether individuals targeted by programs that GiveWell might fund would have access to affordable, quality eyeglasses in the absence of funding.
• Duration of impact: We are unsure how long eyeglasses may last. We currently assume a lifespan for the glasses of two years, but the longest follow-up study currently available is 8 months after provision, so our assumption remains highly speculative.17

The grant

Grant activities

This grant will fund a randomized controlled trial of eyeglasses for presbyopia (near vision impairment) for adults aged between 35 and 65 across a range of occupations in India and Kenya.18

The trial would randomly assign approximately 10,000 individuals in each country to either a treatment group, where participants will receive free near-vision glasses, or to a control group, where participants will be provided with near-vision glasses at the end of the trial period (i.e. two years after random assignment).19 The trial would screen for those who have bilateral presbyopia, do not already have glasses for near vision, are normally resident in the locality, and provide informed consent.20 The trial will not screen for occupation, but the researchers are planning to conduct a subgroup analysis on occupations that are particularly dependent on near-vision, such as farming, manufacturing, shop workers, teachers, and office workers.21

The primary outcomes of the study are household consumption and vision-related quality of life.22 These will be measured at the individual level at 12- and 24-month follow-up. The study’s secondary outcomes are individual income, and both individual income and workplace productivity for those with presbyopia working in particularly near-vision dependent occupations that yield easily quantifiable units of production (e.g., garments).23

How might this grant affect our funding behavior?

GiveWell typically assesses grants based on the direct impact that we believe the program will have. However, in funding research projects, we think in terms of ‘value of information’. The purpose of these grants is to learn more about a program in order to recommend more highly cost-effective grants in future. If new evidence suggests that a program is less effective than we previously thought, we will scale back our grantmaking; if it reveals that a program is more effective, we think of this as unlocking new, cost-effective funding opportunities, improving the cost-effectiveness of our grantmaking at the margin.

We model the value of information of this grant as equivalent to 13 times the cost-effectiveness of cash transfers. Absent the trial, we think that we would have made relatively small grants for the provision of eyeglasses. Therefore, we think the study will prove valuable for GiveWell whether it provides a negative or a positive update relative to our current thinking:

• In the event that the trial provides a negative update and leads us to believe that the program is below our cost-effectiveness bar, we would redirect this money away from eyeglasses, increasing the cost-effectiveness of our grantmaking;
• If the trial leads us to believe with greater confidence that the program is above our bar, we would direct relatively more funding to this intervention. This is particularly important for this intervention, as we believe that eyeglasses programs have a large amount of ‘room for more funding’.24 A positive update from this trial could unlock an especially large amount of highly cost-effective new grantmaking in the future.

Our case for recommending this grant is partly based on our value-of-information model, and partly on more qualitative factors, such as our sense that the study is well-suited to addressing our uncertainties (see here).

Budget for grant activities

We estimate that the trial will cost around $4.8 million. The budget is broken down as follows:

• $1,821,814.24 to LSHTM for personnel, travel costs, and indirect costs;
• $1,091,358.87 to Dr. Shroff's Charity Eye Hospital in Uttar Pradesh for implementation and data collection;
• $1,873,640.20 to Kenya Medical Research Institute for implementation and data collection.25

GiveWell is co-funding this grant with the Livelihood Impact Fund (LIF). LIF is contributing $1 million toward the overall cost of the grant.

Timeline

The above budget covers three years and nine months of fieldwork time. This includes six months of preparation work and piloting, one year for baseline data collection, and one year each for the 12- and 24-month follow-up. A three month buffer was included in case of data collection overruns.26

We currently expect preparation work, including project set-up and the study's pilot phase, to begin in October 2024, baseline data collection to begin by June 2025, baseline data collection to conclude in June 2026, and the 12- and 24-month follow-ups to conclude in June 2027 and June 2028 respectively. We expect exact dates to be confirmed soon after the grant award.

The case for the grant

We are recommending this grant because:

• This study is well-positioned to address several of our core uncertainties around eyeglass provision, and will therefore influence our funding decisions in the future. At the moment, our best guess is that the provision of eyeglasses for workers with a vision impairment is between 11 and 15 times as cost-effective as cash transfers.27 However, we have several uncertainties around key factors in our estimate, and so would currently only consider small grants for this intervention. By funding this trial, we expect to significantly reduce our uncertainty around the program, which may increase or decrease our grantmaking in this area (as above). For detail on how the trial will address our uncertainties, see below.
• The study will occur in Kenya and India, two countries where we could foreseeably fund direct delivery of eyeglasses in the future. This increases our confidence that the evidence this trial generates will be well-suited to our grantmaking decisions.
• The trial will have the direct benefit of providing free eyeglasses to around 20,000 working-age adults. The 10,000 people across the two treatment groups in India and Kenya will receive free eyeglasses during the trial, and the 10,000 in the control group will be given them after the trial ends.

Which of our uncertainties will the trial address?

This RCT would specifically address the following core uncertainties:

• The effect of eyeglass provision on household consumption and recipient income. This is our largest uncertainty around the cost-effectiveness of the intervention.28 The study will:
  • Take household consumption as its primary outcome, and measure recipient income as a secondary outcome for both the full sample and the sub-group in vision-intensive occupations. The study is powered to detect a 3% effect on household consumption at 24-month follow-up, which is the effect size that we currently believe is required in order for eyeglasses to meet our cost-effectiveness funding bar in Bangladesh.29
  • Include two different countries (Kenya and India), and measure the effects of eyeglass provision on a broader share of the population of working adults than in the existing studies, which screened based on specific occupations (those requiring more manual dexterity). We think this will be more representative than the existing literature of the effect of the intervention when distributed at scale.
• The duration of eyeglasses. We model eyeglasses as lasting for two years. However, eight months is the longest follow-up in the existing evidence (as above). This trial will include follow-up at 12 and 24 months, which should allow us to reach a more confident estimate of eyeglasses’ lifespan.

In addition, we think the study may provide valuable information on several of our other uncertainties, although these are not the primary outcomes of the trial:

• The cost per person receiving glasses. The trial may include a rough cost analysis, addressing our second-biggest uncertainty about the cost-effectiveness of the intervention.30 This analysis would only estimate the cost of implementation in the context of this RCT, rather than the cost of implementation at scale, where we would anticipate that unit costs may be lower, as the RCT requires changes to the workflow of the program. However, this could still provide a reasonable ‘upper-bound’ cost estimate for programs that GiveWell could fund in the future.
• The extent to which the program’s benefits differ by occupational group. The researchers plan to conduct an occupational sub-group analysis. This will involve supplementing the core sample of working age people with presbyopia by including specific occupational groups that are likely to benefit especially strongly from eyeglasses. (These groups will be excluded from the primary analysis.)31 Depending upon the final size of this sub-group, it is possible that this analysis may not be sufficiently powered to identify statistically significant changes in income, but it could still identify a suggestive effect.32
• The baseline availability of glasses among this population. As we mention above, we are unsure about the extent to which the populations that these programs target already have access to affordable, quality eyeglasses. The researchers are currently considering embedding questions into the baseline survey that would allow us to better understand why program recipients do not already have eyeglasses. This may include asking a sub-sample of study participants if they know where they can obtain eyeglasses, if they know how much they cost, and if they know anyone who has them.33 The researchers plan to determine this during the pilot phase of the trial.34 The study will also measure eyeglass take-up among the control group.

Risks and reservations

Our main reservations about this grant are:

• This evaluation might have been funded without us. This study addresses important research questions around a promising intervention, so, intuitively, it seems plausible that other funders might have funded it in our absence. However, we think other funders would have been less likely to opt to fund a study large enough to be powered to defect the relatively small effect size on consumption that is decision-relevant for GiveWell (as above), so even if another funder may have stepped in in our absence, our decision to fund the study has likely led to a more informative trial for GiveWell.
• There may be a conflict of interest. The two lead investigators on the study team are affiliated with Peek Vision, a social enterprise that would provide software to help identify those who may be eligible for enrolment in the trial. Andrew Bastawrous is Peek Vision CEO and Matthew Burton is a Trustee and Scientific Advisor to Peek Vision.35 Our understanding is that the remainder of the study team is not affiliated with Peek. Peek’s software would not be used in the trial itself, and some overlap in staffing is not unprecedented for studies in this area: the two RCTs that we currently rely upon for our cost-effectiveness estimate had senior VisionSpring leadership as coauthors.36 The RCT will also have a pre-registered analysis plan, which we believe should mitigate concerns about bias or conflict of interest.
• We are not sure whether this study involves implementers we would be likely to fund at scale in the future. This study has two implementers, Dr Shroff’s Charity Eye Hospital and Kisii Eye Hospital.37 LSHTM believes the former has a track record for scaling interventions in Northern India, but we have not vetted this claim. Meanwhile, our understanding is that Peek Vision typically focuses on innovation and technology, rather than program implementation and eyeglass distribution at scale.38 It may have been preferable to fund an RCT with implementers that we are more confident that we could fund for this work at scale in the future.

Plans for follow up

• We plan to check in with LSHTM and LIF every few months. These check-ins will be timed around key study milestones, such as registering the pre-analysis plan, wrapping up the pilot, launching random assignment, and after data analysis.
• As mentioned above, we expect to receive outcome data for the first-year follow up around June 2027, and second-year follow up around June 2028.

Internal forecasts

For this grant, we are recording the following forecasts:

Our process

• We learned about this opportunity through a discussion about eyeglass provision with LIF. In that conversation, we discussed our key uncertainties about the intervention, and our interest in funding a trial to resolve these. LIF made us aware that it was in the early stages of reviewing a draft protocol from LSHTM that would address our uncertainties.
• Prior to recommending the grant, we had several calls with LIF and the LSHTM researchers. We also received feedback on the grant from researchers at the Development Innovation Lab at the University of Chicago.

Sources

• 1

  “The London School of Hygiene & Tropical Medicine (LSHTM) is one of the world’s leading public health universities. … LSHTM has two UK sites in the Bloomsbury area of London at Keppel Street and Tavistock Place, and units in The Gambia and Uganda. Originally set up to serve a colonial mission, today we are a global network of specialist centres and partners, working together to improve health worldwide” London School of Hygiene & Tropical Medicine, “Introducing LSHTM”.

• 2

  “The International Centre for Eye Health (ICEH) works to improve vision and eye health worldwide. We engage in a large variety of activities, spanning research, education, capacity development and technology. All of our work is done in partnership with colleagues in low- and middle-income countries, and our work is focused on populations with the greatest needs. We are based at the London School of Hygiene & Tropical Medicine.” International Centre for Eye Health, “About us”.

• 3

  EARN trial protocol, (unpublished).

• 4

  “Professor Andrew Bastawrous OBE is an Ophthalmologist (Eye Surgeon), and Professor in Global Eye Health at the London School of Hygiene & Tropical Medicine where he teaches on public health for eye health and leads a research portfolio. Andrew is also the Co-Founder & CEO of Peek Vision Ltd and the parent charity The Peek Vision Foundation, a spin-out from LSHTM. The company is a social enterprise which develops smartphone technology and public health tools to help eye health organisations increase their impact.” London School of Hygiene & Tropical Medicine, “Prof. Andrew Bastawrous”.

• 5

  “Professor Matthew Burton is a Trustee of The Peek Vision Foundation and Dr. Andrew Bastawrous is Chief Excecutive Officer (CEO) of The Peek Vision Foundation and Peek Vision Ltd. HR is an advisor to Peek Vision Ltd” Rono et al. 2019.

• 6

  “Peek Vision, a social enterprise that provides technological solutions to eyecare services in lower and middle-income countries. Their solutions, including smartphone vision screening tools and health system software, allow services to design their programmes effectively and optimise them with the data they create. By democratising the tools needed to improve services, Peek helps to overcome a lack of resources or reduced capacity to create better eye health for all.” International Centre for Eye Health, “Peek Vision”.

• 7

  LSHTM confirmed this in review of a draft of this page.

• 8
  • "For the year 2020 (the most recent VLEG/GBD analysis) it was estimated that there were 510 million people worldwide living with unaddressed near vision impairment due to presbyopia (hereafter “presbyopia”) who needed glasses to enable them to perform near vision dependent activities.” EARN trial protocol, (unpublished).
  • “510 million people had uncorrected near vision impairment, simply because of not having reading spectacles. A large proportion of those affected (90%), live in low-income and middle-income countries (LMICs).” Burton et al. 2021.
  • "Globally, at least 2.2 billion people have a near or distance vision impairment. In at least 1 billion – or almost half – of these cases, vision impairment could have been prevented or has yet to be addressed. Among this 1 billion people, the main conditions causing distance vision impairment or blindness are cataract (94 million), refractive error (88.4 million), age-related macular degeneration (8 million), glaucoma (7.7 million), diabetic retinopathy (3.9 million). The main condition causing near vision impairment is presbyopia (826 million)." World Health Organization, “Blindness and vision impairment”.
  • For more on this, see Bastawrous, Kassalow and Watts, “Global uncorrected refractive error and presbyopia: the size of the problem” (2024).

• 9

  Note that a) our cost-effectiveness analyses are simplified models that are highly uncertain, and b) our cost-effectiveness threshold for directing funding to particular programs changes periodically. As of April 2024, our bar for directing funding is programs that are about 10 times as cost-effective as unconditional cash transfers. See GiveWell’s Cost-Effectiveness Analyses webpage for more information about how we use cost-effectiveness estimates in our grantmaking.
  Note that unconditional cash transfers are GiveWell’s benchmark for comparing different programs.

• 10

  See Sehrin et al. 2024 (Bangladesh), and Reddy et al. 2018 (Assam).

• 11

  "Weaknesses must also be acknowledged: the trial’s main and secondary outcomes were both self-reported, and the investigators did not feel that participant masking with sham glasses was ethical in this setting. Thus, the potential for placebo effects cannot be excluded." Sehrin et al 2024.

• 12

  Between the baseline and evaluation periods, mean productivity in the intervention group increased from 25·0 kg per day to 34·8 kg per day (an increase of 9·84 kg per day), a significantly higher increase than in the control group (from 26·0 kg per day to 30·6 kg per day; an increase of 4·59 kg per day), corresponding to a between-group difference of 5·25 kg per day (95% CI 4·50–5·99; 21·7% relative productivity increase; effect size 1·01 [95% CI 0·86–1·16]; p<0·0001). Reddy et al. 2018

• 13

  We outline our uncertainty ranges for key parameters, and implications for cost-effectiveness, in our separate report on programs that provide eyeglasses to improve income. This report is forthcoming.

• 14

  “Practical guidelines have been developed to assist researchers and analysts computing broadly comparable and theoretically consistent consumption aggregates and poverty measures from household surveys (Deaton and Zaidi 2002), but much less information is available for low-income countries in terms of looking at income data.” Carletto, Tiberti, and Zezza (2021), p. 2.

• 15
  • "Inclusion criteria were: lack of apparent ocular conditions under gross, external examination under natural light; usual participation in paid occupations requiring intensive close work (such as tailors, artisans, mechanics, carpenters, farmers and shopkeepers); uncorrected distance visual acuity 6/12 or better in each eye; presence of presbyopia (defined as inability to see the 6/13 optotype at 40 cm with both eyes together, correctable with near vision glasses), and never having used near or distance glasses." Sehrin et al 2024.
  • "This randomised trial was done in tea pickers aged 40 years or older in Assam, India, with unaided near visual acuity (NVA) lower than 6/12 in both eyes, correctable to 6/7·5 with near glasses; unaided distance vision 6/7·5 or greater; and no eye disease." Reddy et al. 2018.

• 16

  This is because (i) we heard that some implementers have reservations about turning people away if e.g. they don’t meet certain employment or occupation criteria, and (ii) we’re skeptical that implementers can ex ante predict which communities have significant unmet demand for eyeglasses among the targeted subset of the population whose economic productivity is tied to near vision.

• 17

  The 8 month follow-up was in the Bangladesh study (Sehrin et al. 2024): "A follow-up home survey in March 2018 (8 months later) gathered information on income, NVRQOL and self-reported frequency of glasses wear among both study groups."

• 18

  Note that the trial specifically addresses those with presbyopia but no other eye health issues. The trial will not involve more complex prescriptions, such as bi-focials or varifocals, which correct for distance vision and presbyopia simultaneously.

  • "Inclusion criteria:... Adults aged between 35 and 65 years." EARN trial protocol, (unpublished).
  • Source for the countries included in the study: EARN trial protocol, (unpublished).
  • Note that presbyopia is not the only cause of near-vision impairment (hyperopia, also known as hypermetropia or farsightedness, is the other). However, presbyopia is estimated to account for the majority of near vision impairment globally: “The main condition causing near vision impairment is presbyopia (826 million).” World Health Organization, “Blindness and vision impairment”.

• 19
  • "Trial participants who are randomly allocated to the intervention arm will be given free near vision glasses (spherical, non-astigmatic correction), appropriate for their degree of presbyopia." EARN trial protocol, (unpublished).
  • "Trial participants who are randomly allocated to the control arm will not be given free near vision glasses at baseline. They will be given free near vision glasses (spherical, non-astigmatic correction), appropriate for their degree of presbyopia at the end of the trial follow-up period (2 years)." EARN trial protocol, (unpublished).
  • "A sample of 10,000 participants (5,000 per arm), in each trial, would have 90% power to detect a 3% increase in household consumption between baseline and 1 year" EARN trial protocol, (unpublished).

• 20

  "Inclusion criteria:

  • Bilateral presbyopia and do not currently own/use have glasses for near vision.
  • Adults aged between 35 and 65 years.
  • Normally resident in the locality where the study is being conducted, defined as:
  • lived there >6 months and intend to continue living there for the next two years.
  • Provide informed consent and agreement to be randomly allocated to one of
  • the two study arms."

  EARN trial protocol, (unpublished).

• 21
  • Individuals who are not working (and otherwise meet the eligibility criteria) are eligible for the study.
  • “Which occupations are likely to make up the occupational subgroup analysis? … Within the study summary we had previously listed three groups of near vision dependent occupations for the subgroup analysis. We have updated this to more specifically align with the International Standard Classification of Occupations (from ILO). Part of the pilot work will include refining these subgroups to the local context where the study will be taking place.
    • Group 1: Agriculture, forestry and fisheries.
    • Group 2: Crafts, repairs and operation of machinery
    • Group 3: Clerical, sales and service occupations
    • Group 4: Managerial, professional and technical occupations.”

  GiveWell, ‘Questions for EARN trial’, April 12, 2024 (unpublished).

• 22

  “We will use the well-established and widely used WHOPBD/VF20 questionnaire… it has 20 vision related questions, six of which are relevant to near vision functioning.” EARN trial protocol, (unpublished).

• 23

  "Workplace Productivity. For certain sub-groupings of the working population, we will apply occupation specific measures to quantify workplace productivity, limited mainly to people in jobs with easily quantifiable units of production (e.g. garments made)." EARN trial protocol, (unpublished).

• 24
  • See this tab of our cost-effectiveness analysis for our room for more funding estimates for this program.
  • See this row of our value-of-information analysis for our rough estimate of the annual amount of funding we would direct toward eyeglasses programs in the future if we decided to fund them.

• 25

  EARN budget, April 05, 2024 (unpublished).

• 26

  EARN trial budget summary, April 05, 2024 (unpublished).

• 27

  Note that a) our cost-effectiveness analyses are simplified models that are highly uncertain, and b) our cost-effectiveness threshold for directing funding to particular programs changes periodically. As of April 2024, our bar for directing funding is programs that are about 10 times as cost-effective as unconditional cash transfers. See GiveWell’s Cost-Effectiveness Analyses webpage for more information about how we use cost-effectiveness estimates in our grantmaking.

• 28

  See GiveWell, “Eyeglasses for workers CEA”. Our current best guess is that across both studies we reviewed, the 25th-75th percentile interval for the effect of eyeglasses on income is an increase of between 0% and 14% per year. This would imply a cost-effectiveness range of between 5 and 41 times cash transfers. We outline our uncertainty ranges for key parameters, and implications for cost-effectiveness, in our separate report on programs that provide eyeglasses to improve income. This report is forthcoming.

• 29

  “A sample of 10,000 participants (5,000 per arm), in each trial, would have 90% power to detect a 3% increase in household consumption between baseline and 1 year, if we exclude the households that are in the top 30% by consumption and assume a correlation in household consumption between one year and the next of 0.6. This sample size would also have 80% power to detect a 3% increase in household consumption between baseline and 2 years, if we exclude the households that are in the top 30% by consumption.” EARN trial protocol, (unpublished).

• 30

  "An additional cost analysis could be performed to quantify the costs involved in delivering the screening and near vision in these two programmes." EARN trial protocol, (unpublished).
  See GiveWell, “Eyeglasses for workers CEA”.

• 31

  "While recruiting the core sample of people of working age population we will supplement the sample for several specific occupation groups to provide sufficient power to examine changes in work-place productivity (and income) by arm for these subgroups. These additional recruits would be excluded from the primary analysis of the population-based sample." EARN trial protocol, (unpublished).

• 32

  "Depending on the size of this supplementary recruitment, it is possible that the sub-group analysis would be underpowered to identify subtle changes in income, but would be indicative of an effect, and will have greater power to pick up effects on productivity." EARN trial protocol, (unpublished).

• 33

  LSHTM staff confirmed this in review of a draft of this page.

• 34

  Source: comment from LSHTM staff in review of a draft of this page.

• 35
  • “Professor Matthew Burton is a Trustee of The Peek Vision Foundation and Dr. Andrew Bastawrous is Chief Excecutive Officer (CEO) of The Peek Vision Foundation and Peek Vision Ltd. HR is an advisor to Peek Vision Ltd” Rono et al. 2019.

• 36

  Comment from LSHTM staff in review of this page.

• 37

  LSHTM confirmed this in review of a draft of this page.

• 38

  "What Peek does… Our technology optimises eye health services in communities, schools and workplaces. Our partners see greater screening coverage, increased attendance at follow-up appointments, and services that cost less per person." Home, Peek Vision.