Pure Earth — Support for Reducing Lead Exposure in Low- and Middle-Income Countries

Note: This page summarizes the rationale behind a GiveWell Incubation Grant to Pure Earth. Pure Earth staff reviewed this page prior to publication.

This page is an explanation of why we recommended this grant to Pure Earth. This page is intended to be transparent about GiveWell's process, judgment calls, and uncertainties. It is not intended to be an authoritative review of the relevant literature on lead exposure.

We spoke to Pure Earth for an update on this grant on October 20, 2021.

Published: August 2021; Last updated: February 2024 (August 2021 version)

The organization

Pure Earth works on reducing the harm to human health caused by pollution in low- and middle-income countries.1 In 2019, it spent $4.3 million.2

Pure Earth was founded in 1999.3 In the past, it has worked on a number of different pollutants but now focuses primarily on reducing lead exposure.4

Much of its past work has focused on remediating toxic sites.5 We do not expect that GiveWell's funding will be used for this purpose. Pure Earth expects future work to focus more on interventions that could affect a larger proportion of the population.6

The grant

Activities and budget

Pure Earth's planned activities vary by geography. They are structured around:7

1. Testing baseline levels of exposure to lead, measured by blood lead levels (BLLs).
2. Identifying likely sources of lead exposure through a combination of testing products in markets for lead and identifying potential sources of exposure in the home.
3. Designing and implementing an intervention to reduce lead exposure from the most important sources.
4. Measuring blood lead levels to assess the reduction in blood lead levels over the course of the intervention.

We expect this grant to fund lead reduction interventions in six locations:

1. Bangladesh. Testing the effect of a previous intervention to reduce lead in spices on blood lead levels (approximately $860,000).8
2. Bihar (India). Conducting and testing an intervention to reduce lead in spices (approximately $1.3 million).9
3. Jharkhand, Uttar Pradesh, West Bengal, or Tamil Nadu (India). Assessment of baseline blood lead levels, identifying sources, designing and implementing an intervention, and monitoring its effect (approximately $1.3 million).10
4. The Philippines. Identifying sources, designing and implementing an intervention, and monitoring its effect (approximately $1 million).11
5. Two additional locations. To be decided on the basis of the rapid market assessments described below (approximately $1.2 million).12

In addition to location-specific work, we expect the grant to be used to:

1. Conduct "Regional Rapid Marketplace Screenings" in 25 locations to test levels of lead in products sold in marketplaces, and develop communications materials to help disseminate the results of the screenings (approximately $920,000).13
2. Build a monitoring, evaluation, and learning team (approximately $870,000).14
3. Support the Global Alliance on Health and Pollution to advocate for reducing lead exposure in global forums and convene a global working group on lead (approximately $530,000).15

For more details on these activities, see Pure Earth's budget, timeline, and project proposal.

Pure Earth plans to work with Vital Strategies and Stanford University on some of the activities above.16

Given that Pure Earth will be gathering information over the course of the grant period,17 which might affect how it should prioritize its activities, we would not be surprised if some funding was used differently from the activities outlined above.

Case for the grant and reservations

We recommended this grant because:

1. Lead exposure appears to receive little philanthropic attention relative to the burden it imposes and may be a tractable problem to make progress on. (more)
2. We think Pure Earth is the most promising giving opportunity we have found to address lead exposure (more).

The majority of Pure Earth's past work on reducing lead exposure has focused on reducing harms from lead-acid battery recycling, which will not be a focus of this grant.18 We recommended this grant primarily based on our assessment of the activities outlined in Pure Earth's proposal, and our view that lead exposure is a promising problem to work on, rather than because of a deep review of Pure Earth's track record.

Why we think reducing lead exposure is a promising focus area

Reducing lead exposure is one of the promising areas we identified as part of GiveWell's work investigating giving opportunities to assist governments in low- and middle-income countries (LMICs) in the design and implementation of effective public health regulations.

We chose lead exposure as one of GiveWell's focus areas primarily based on our assessment of the harm it imposes, the amount of philanthropic funding it receives, and how tractable it is to make progress on. Our assessment of different cause areas to focus on led us to believe that lead exposure causes approximately 1/3 as much harm as tobacco use, and that efforts to address the issue receive approximately 1/10 as much philanthropic funding. More information on these estimates is on this page (a table from this page is also included below).19

Problem	Total burden (units of value, millions)	Annual philanthropic spend ($ millions)	Spend per unit of burden ($ millions)
Alcohol use	132	$3	$0.02
Pesticide self-harm	11	$0.4	$0.04
Exposure to lead	94	$7	$0.07
Diet high in sodium	62	$6	$0.10
Tobacco	288	$70	$0.24
Transport injuries	130	$32	$0.25
Drug use	40	$10	$0.25
Micronutrient deficiencies (iron, zinc, vitamin A)	118	Unclear	Unclear
Ambient air pollution	162	Unclear	Unclear

We believe lead exposure receives little philanthropic funding relative to the harm it causes. Progress on reducing blood lead levels in the U.S. suggests it is possible to make substantial progress on this problem, although we are less certain how this factor will translate to other contexts (more below).

On the other hand, we believe lead exposure causes less harm than other areas we believe are also neglected, such as alcohol policy. Lead exposure is also caused by a variety of sources, meaning it may be less easy to make progress on, compared to areas such as pesticide regulation to prevent suicide (though we believe it likely imposes a higher burden).20

See below for more details on why we think reducing lead exposure is a high-priority problem and this page for a description of how we decided to prioritize lead exposure through comparisons with other areas.

What are sources of lead exposure?

Our impression is that there is limited evidence to inform an estimate of the proportion of global lead exposure that comes from different sources. We understand that the most important sources of lead exposure likely vary by context.21

Ericson et al. 2021 finds that studies on blood lead levels have been conducted investigating populations considered to be at risk from sources including:22

• lead-acid battery recycling and manufacture,
• metal mining and processing,
• electronic waste,
• the use of lead as a food adulterant, including in spices,
• paint,
• bullets,
• ceramics,
• industry, and
• tobacco products.

We think this review gives a rough indication of potential exposure sources.23 However, many of the studies included in the review were conducted to assess average blood lead levels in populations where a particular source of exposure was already suspected (rather than evaluating the contribution of different sources of exposure). That suggests that the proportion of studies identifying a particular source is a better proxy for academic attention than the proportion of exposure that comes from that source.24 While we see this review as indicative of potential sources of exposure, we do not believe there is sufficient data to confidently rank different sources of exposure.25

In Bangladesh and India, which Pure Earth will work in, we believe there is suggestive evidence that spices are an important source of exposure. (see below).

How high is lead exposure in low- and middle-income countries?

Lead exposure is typically measured by measuring the concentration of lead in the blood.26 For reference, in the U.S., the Centers for Disease Control and Prevention (CDC) estimates that geometric mean blood lead levels declined from 12.8 to 0.8 micrograms per deciliter (µg/dL) between 1976-80 and 2015-16.27 The CDC blood lead "reference value" (used to identify "children with blood lead levels that are much higher than most children’s levels") is 5 µg/dL.28

Ericson et al. 2021, a systematic review of studies on blood lead levels in low- and middle-income countries, notes that there is limited population-representative data.29 In the 34 countries that the authors considered to have representative data for children, 49% of children were estimated to have blood lead levels over 5 µg/dL, and 32% were estimated to have blood lead levels over 10 µg/dL.30

We have not reviewed these estimates in detail, but our broad takeaways are that (i) representative data on blood lead levels in low- and middle-income countries is limited, (ii) the data that does exist suggests that roughly 50% of children in low- and middle-income countries have blood lead levels over five times as high as the average in the U.S., and (iii) blood lead levels in low- and middle-income countries are likely lower than they were in the U.S. in 1976-80 (when leaded gasoline was still in use).31

How much harm does lead exposure cause?

According to the World Health Organization, high levels of lead exposure can lead to death, and can cause a range of neurological disorders. Lower levels of exposure can affect children's brain development, leading to reductions in cognitive development and behavioral problems.32

We model the benefits of reducing lead exposure by estimating the losses in cognitive ability and lifetime earnings that would otherwise occur absent the intervention but could be avoided by reducing lead exposure among children.

We have not deeply reviewed estimates of mortality, morbidity, or social harm caused by lead exposure because the grant already seemed cost-effective to us without considering these additional benefits.

Details follow.

The effect of lead exposure on cognitive ability and earnings

Our best guess is that reducing blood lead levels by 1 µg/dL throughout childhood increases intelligence quotient (IQ) by 0.15 points, leading to a 0.1% increase in lifetime earnings. These estimates are necessarily uncertain but are based on:

Effect of high blood lead levels on cognitive ability:

• Human observational studies. The strongest evidence we found on the effect of lead exposure on cognition is Lanphear et al. 2005, a meta-analysis combining data from seven observational studies on 1,333 children. After controlling for a set of potential confounders, it estimates a concave relationship between blood lead levels and cognitive ability (i.e., incremental increases in blood lead levels are associated with decreasing marginal harms at higher levels).33 It estimates that:
  • Over the range of 2.4-10 µg/dL, each additional 1 µg/dL in blood lead levels causes a 0.5 point decrease in IQ on average.
  • Over the range of 10-20 µg/dL, each additional 1 µg/dL in blood lead levels causes a 0.2 point decrease in IQ on average.
  • Over the range of 20-30 µg/dL, each additional 1 µg/dL in blood lead levels causes a 0.1 point decrease in IQ on average.34

  Our biggest uncertainties are whether (i) unobserved confounding variables, such as nutrition status or parenting ability, may have led to upward bias in the estimate, (ii) error in blood lead level measurement may have caused a downward attenuation bias in the estimate, and (iii) whether diminishing marginal harms from blood lead is the best functional form to approximate actual harm at different levels of exposure.35

• Experimental non-human animal studies. Our review of the experimental non-human animal literature led us to believe that (i) there is strong evidence that lead exposure over 10 µg/dL impairs cognitive ability (learning and memory) in non-human animals, including rodents and macaque monkeys, and (ii) there is weaker and more mixed evidence that lead exposure between 5 µg/dL and 10 µg/dL impairs cognition in rodents. This informed our prior that lead exposure likely has cognitive impacts in humans.36
• Synthesis. We use a linear estimate (0.2 decrease in IQ points per 1 µg/dL increase of BLL) with no minimum threshold rather than the log-linear estimate from Lanphear et al. 2005 (which uses a minimum threshold of 2.4 µg/dL) because (i) the non-human animal studies we reviewed suggest stronger evidence that higher blood lead levels lead to impaired cognition than lower blood lead levels, (ii) it seems plausible to us that the steep slope at low blood lead levels in the log-linear model may be driven by the sampling strategy of two of the included studies,37 and (iii) a linear estimate is more practical for our purposes, where we have limited information on the distributional impact of interventions to reduce lead exposure.

  GiveWell routinely makes adjustments to effect sizes taken from published literature. In this case, we reduce this estimate by 25% because we believe observational studies are generally more likely to overstate than understate true effect sizes. That yielded a highly uncertain best guess that each 1 µg/dL increase in blood lead levels throughout childhood reduces IQ by 0.15 points.38

Effect of cognitive ability on earnings:

• Our best guess, based on a shallow review and synthesis of eight observational studies in low- and middle-income countries, is that each one point increase in intelligence quotient leads to a 0.67% increase in future earnings (or a 0.1% increase in earnings for an increase in IQ of 0.15 points).39 This is in line with the estimates we use to estimate the impact of iron fortification here. This estimate is uncertain and relies on judgment calls about how to weigh studies with different methodologies. It is also possible that we have not seen other relevant studies that may affect our conclusions. However, the three largest studies (which together comprise 72% of the sample we reviewed) estimate effects between 0.4% and 0.8%, which gives us some confidence that this estimate is reasonable.40

We have not reviewed other studies that may be relevant to this question, such as studies related to educational outcomes and the lead-crime hypothesis.41

The effect of lead exposure on mortality and morbidity

The Institute for Health Metrics and Evaluation (IHME) estimates that lead exposure caused approximately 900,000 deaths in 2019, and 4.1 million years lived with disability (YLDs). Of these, 830,000 deaths and 3.9 million YLDs were in low- and middle-income countries.42 Based on IHME data, we estimate the average age of death from lead exposure in 2019 was approximately 71.43 We have not thoroughly reviewed the calculations behind these estimates.

We did not include estimates of morbidity or mortality in our main cost-effectiveness analysis but include a "best guess adjustment" as a forecast of what we will conclude after further investigation. We decided not to investigate this further because the grant already looked sufficiently cost-effective on the basis of the modeled gains in cognitive ability. We plan to revisit this question in the future.

Is lead exposure a tractable problem to work on?

We think that reducing lead exposure may be an unusually tractable problem relative to other focus areas we considered. However, we feel quite uncertain in this assessment, and it plays a smaller role in our prioritization than our assessment of the neglectedness of lead relative to its burden (although it was a reason to prioritize further investigation into lead rather than sodium reduction).

We think that reducing lead exposure may be an unusually tractable problem because:

• Between 1976-1980 and 2015-2016, average blood lead levels in the U.S. (the country we know of with the best available data) have declined 94% from 12.8 to 0.8 µg/dL.44 We take this as evidence that similar reductions in low- and middle-income countries are feasible.
• The elimination of leaded gasoline appears to be an example of a regulatory approach that had a large impact on blood lead levels.45 Leaded gasoline has now been banned in every country apart from Algeria.46 We take this as evidence that global regulatory work to reduce lead use is feasible.

On the other hand, we note that:

• The U.S. is a high-income country with more resources to address lead exposure than the countries we expect to fund work in.
• It seems unlikely that any single source of exposure contributes as much to global blood lead levels as leaded gasoline did in the U.S. before it was banned.47

We did not deeply review the evidence that particular interventions could successfully reduce lead exposure as part of our decision to choose lead exposure as a focus area because we believed that the appropriate interventions would be context-specific and that we could more efficiently assess this through investigation of particular grants. As part of the grant to Pure Earth, we reviewed the evidence that reducing lead exposure from spices could reduce blood lead levels (see below).

Other organizations and funders

We roughly estimate that activities to reduce lead exposure in low- and middle-income countries receive about $7 million of philanthropic funding each year.48 Our understanding is that most of that funding is used to reduce lead exposure from lead-acid battery recycling and lead paint.

Organizations working to reduce lead exposure include:

• International Pollutants Elimination Network (IPEN) and its member organizations
• Lead Exposure Elimination Project (LEEP)
• Global Alliance to Eliminate Lead Paint
• Global Alliance on Health and Pollution (GAHP)
• Occupational Knowledge International
• Pure Earth
• Vital Strategies

Major funders of work to reduce lead exposure include:

• USAID49
• The Swedish International Development Agency (Sida).50
• The Global Environment Facility51
• Clarios Foundation52

Why we think Pure Earth is a promising giving opportunity​​

We believe Pure Earth is a promising giving opportunity because:

• We expect to be able to estimate the impact of Pure Earth's work based on its monitoring plans. We place particular weight on monitoring because we expect that it will inform potential future investment from GiveWell to reduce lead exposure. (more)
• We think there is a somewhat compelling (but uncertain) case that reducing lead exposure from spices could be one of the highest priorities in locations where spices have been identified as a likely source of exposure. (more)
• We think Pure Earth has answered our questions transparently, which increases our confidence that we will learn relevant information from this grant.

Generating useful evidence

We expect this grant to contribute evidence on:

• Which commonly used products are contaminated with lead in the 25 locations in which Pure Earth plans to conduct rapid market surveys?
• What are the most likely sources of lead exposure in people with elevated blood lead levels in locations where Pure Earth plans to work?
• How high are blood lead levels in locations where Pure Earth plans to work?
• Did blood lead levels decline after interventions to reduce lead exposure in the locations where Pure Earth plans to work?

We expect that this evidence will be useful for prioritizing future activities to reduce lead exposure.

We have some uncertainty around whether the studies that measure blood lead levels before and after an intervention will give a precise estimate of the effect of the intervention because of the risk of confounding, and the difficulty of extrapolating results to the wider population.53 We discussed some questions we had about how well the studies will answer these questions with Pure Earth and their colleague at Vital Strategies who will be assisting with this work, and we believe that this evidence will be informative.54 We also understand that Pure Earth and Vital Strategies have a track record of completing similar research in the past, although we have not reviewed that work in depth.55

Lead in spices

A portion of the grant will be used to assess and address contamination of spices with lead in states in northeast India, and collect data on the effect of an intervention to reduce spice contamination in Bangladesh.56 We believe this is a promising source of exposure to target.

There is strong evidence that spices in Bangladesh were contaminated with lead, and weaker evidence that this source of exposure contributed to high blood lead levels. An intervention in Bangladesh appears to have reduced the prevalence of lead in spices, and so we think it may be a tractable source of exposure to address. However, we have not yet seen strong evidence that reducing contamination of spices with lead led to a reduction in population-level blood lead levels, so we remain uncertain about this conclusion. Part of the grant will involve gathering that evidence, and we expect that Pure Earth's prioritization of activities (including those funded by this grant) will depend on the results.

Because of our uncertainties about the relative contribution of different sources of exposure, we are not trying to make an authoritative claim that lead in spices is the most important source of global lead exposure. Instead, we are making the weaker claim that there is a somewhat compelling case that spices contribute substantially to blood lead levels, and may be cost-effective to address in the locations Pure Earth plans to work in. Importantly, we expect to get evidence about whether this work did in fact contribute to lower blood lead levels at the population level.

Are spices contaminated with lead?

• Forsyth et al. 2019a collected 340 samples of turmeric in Bangladesh and tested them for contamination with lead. It found that 11% of samples had elevated lead (defined as >2.5 µg/g) in major turmeric-producing districts, and 26% were contaminated in minimally-producing districts.57 The turmeric powder with the highest lead concentration was over 450 times higher than the regulatory limit of 2.5 µg/g.58 Interviews with wholesalers suggested that lead chromate was added to poor-quality roots to make turmeric more yellow, and therefore more desirable to consumers.59 There appears to be other supporting evidence (we have not reviewed past the abstracts) suggesting that spices in Bangladesh are contaminated with lead.60
• Pure Earth told us that spice samples collected in Bihar were also contaminated with lead.61 This evidence has not been published, and we have not reviewed it. Pure Earth suspects other Northeast Indian states may also have contaminated spices and plans to collect samples to test this hypothesis.62

Do spices contribute to high blood lead levels?

• Most lead exposure is related to ingesting or inhaling lead.63 Spices are intended for human consumption, and so our prior is that lead in spices will contribute to elevated blood lead levels, although we are unsure about the size of the relative contribution.
• Lead from different sources have different isotopic compositions that can assist with distinguishing between different sources of lead exposure.64 Forsyth et al. 2019b compared the isotopic composition of different potential sources and 45 blood samples from pregnant women.65 The isotopic composition of lead in the blood samples more closely matched those found in spices than other suspected sources of exposure (lead-soldered cans, clay, soil, ash).66 In conversation, the lead author told us that drawing strong conclusions from isotopic analysis is challenging, and that this study should be interpreted as contributing to an evidence base rather than as a definitive finding. For example, the isotopic signature of lead chromate may be linked to both spices and paint. She told us that her modeling suggested an intervention to reduce spice contamination could reduce blood lead levels by 1-3 µg/dL in Bangladesh, but that this estimate is highly uncertain.67 We have not reviewed that model.

Is lead in spices a tractable source of exposure to address?

• After the release of the results above, the researchers from Forsyth et al. 2019 supported the Bangladesh Food Safety Authority to reduce lead exposure in turmeric. The package of interventions included monitoring lead levels in spices in Dhaka (which is a hub for spice sales), imposing fines, and educating consumers and suppliers.68 In a presentation to the International Symposium on Spices as Flavors, Fragrances & Functional Foods, the lead author claimed that measured prevalence of lead in spices declined by 90% in the period the intervention was conducted, between September 2019 and September 2020.69 We have not reviewed this work.
• We have not seen strong evidence that blood lead levels declined over this period. Part of this grant will be used to test blood lead levels in samples collected before and after the intervention, to estimate the effect on blood lead levels.70

We have not yet reached conclusions on whether we should recommend funding to giving opportunities to address other potential sources of lead exposure.

Risks and reservations

Risks and reservations include:

• A substantial proportion of this funding will be allocated to gathering evidence on likely sources of lead exposure in particular locations. But we are unsure whether the identified sources of lead exposure will be able to be addressed cost-effectively. (more)
• One of the most likely ways this work could fail to achieve the impact we expect is if reductions in exposure fail to be sustained over time. (more)
• While we believe that lead exposure likely causes substantial harm, we are uncertain about the likely effect size because the evidence base (in humans) relies on observational rather than experimental evidence. (more)

In addition, we note that Pure Earth receives funding from Clarios Foundation, which is funded by a lead-acid battery manufacturer. We think the risk that this funding relationship will negatively influence the activities funded by this grant is low. We would like to better understand if there are promising approaches that could be taken to reduce lead exposure which might be better suited to an organization that does not receive funding from industry.71

We are unsure whether the identified sources of lead exposure will be able to be addressed cost-effectively

It seems likely to us that the evidence generated by Pure Earth's activities will be broadly useful to highlight potential context-specific sources of exposure. However, we are unsure whether these sources of exposure will be able to be addressed at a level of cost-effectiveness competitive with GiveWell's top charities. It therefore seems plausible to us that GiveWell will not recommend funding to address sources of exposure identified by this work, meaning the evidence may not be acted on. We recommended funding for this work because of the low cost of collecting this information relative to our assessment of the likelihood of GiveWell or other funders acting on this evidence.72

Relatedly, Pure Earth's proposed theory of change involves first collecting data to assess likely sources of lead exposure, and then designing an intervention to address those sources (except in Bihar, India and Bangladesh, where spices have already been identified as a likely source of exposure).73 While we find this theory of change compelling, it makes assessing cost-effectiveness relative to our top charities particularly challenging.

Reductions in exposure may fail to be sustained over time

One of the major reasons we recommended this grant was the possibility of sustained reductions in lead exposure over time, with limited need for ongoing activities by Pure Earth. This is highly uncertain, and it is possible that behavior like adding lead chromate to spices could revert over time without high levels of continued funding. More details here.

Uncertainty about effect size

We believe that lead causes substantial harm when ingested, based on our (non-exhaustive) review of the experimental literature on the effects of lead in non-human animals and observational literature in humans. However, we have some uncertainties about the size of the effect of reduced lead exposure on cognitive ability, which we expect to be the primary benefit of this work. More details here.

Cost-effectiveness

We completed a back-of-the-envelope calculation on Pure Earth's potential future cost-effectiveness to weigh the considerations above. That rough estimate suggested Pure Earth's work over the next grant period could be more cost-effective than additional funding to our top charities.

We completed this back-of-the-envelope calculation to discipline ourselves to make our assumptions explicit and quantitatively weigh different considerations. In general, there are limitations to all of our cost-effectiveness analyses, and we believe they should not be taken literally, due to the significant uncertainty around them. However, this estimate in particular relies on a number of difficult best-guess assumptions and judgment calls about modeling structure. It therefore contains less information-value than cost-effectiveness estimates for our top charities, which limits its comparability.

As such, we see the activity of conducting a cost-effectiveness estimate as contributing to synthesizing and evaluating the qualitative case for this grant, rather than as the primary argument for the grant.

Our estimates of a given program's cost-effectiveness tend to go down as we gain more information. However, we would not be surprised if our estimate of the cost-effectiveness of Pure Earth's activities at the end of the grant period was either substantially more or less than we estimate here.

Our back-of-the-envelope estimate is available here. We outline some of the more important inputs below.

Scope of modeling

Benefits

• We model the potential benefits of three funding gaps that aim to directly reduce blood lead levels: in Bihar (a state in India), neighboring states in Northeast India, and the Philippines. Those gaps comprised $3.6 million (45%) of the grant.74
• We did not explicitly model the potential benefits of funding gaps comprising $4.4 million (55%) of the grant either because we believed their benefits were more diffuse or we did not yet know which countries this funding would be allocated to (and so we thought explicit modeling would be less helpful).75

Costs

• We include costs for all direct activities, apart from the funding gap in two additional countries that have not yet been identified (a total of $3.6 million).76 We then apply a multiplier of 1.7x to account for the cost of indirect activities (making the total cost $6 million).77 We did this because we thought that including the costs of "indirect activities" (which we expect to contribute to future work to directly reduce blood lead levels) would give us a more accurate understanding of the total cost associated with reducing blood lead levels. We did not include costs for the two additional countries that have not yet been identified because we expect those activities to contribute to directly reducing blood lead levels (and we have not modeled those benefits because of our uncertainty about the locations in which these activities will take place).
• We also include a multiplier of 1.5x to account for the possibility that additional activities and funding will be required in the future to sustain any potential reduction in blood lead levels over the modeled period (making the final cost $9 million).78 We are highly uncertain about this estimate and consider it a rough guess (more below).

Expected reduction in blood lead levels in the target population

In our cost-effectiveness analysis, we assumed that Pure Earth's activities would reduce blood lead levels in children by 0.2-0.4 µg/dL in expectation in the target locations (the range refers to differences between target locations). For context, in the U.S., average blood lead levels appear to have declined by 0.1 µg/dL each year between 1990 and 2016.79

We have a limited understanding of baseline blood lead levels in these specific locations. Pure Earth is planning to collect nationally representative estimates as part of this grant.80 The most recent systematic review we identified, Ericson et al. 2021, estimates that in low- and middle-income countries where nationally representative data was available, average blood lead levels in children ranged from 1.66 μg/dL in Ethiopia to 9.30 μg/dL in Palestine.81

We are very uncertain about these estimates and consider them a rough guess.

• We estimated a 35-50% chance that the intervention would reduce blood lead levels by less than 0.5 μg/dL in each location.82 This estimate was based on a generally skeptical prior, and the lack of conclusive direct evidence that spices contribute to high blood lead levels (although we find the broader evidence base compelling).83 We estimated a 35% chance of a less than 0.5 μg/dL reduction in Bihar, where we are more confident that spices contribute to blood lead levels, and a 50% chance in other Indian states and the Philippines, where we are less confident that spices contribute to blood lead levels.84
• We assigned the rest of the probability mass between 0.5 and 3 µg/dL based on modeling (that we have not reviewed) suggesting that an intervention to reduce spice contamination could reduce blood lead levels by 1-3 µg/dL in Bangladesh.85 We placed most of the weight on the lower end of that estimate because our intuition is that modeling is more likely to overestimate than underestimate the effect of interventions, and because it is possible that the intervention does not completely eliminate lead in spices.
• We assumed that this decrease in blood lead levels could be applied to 30-40% of the population in expectation depending on the location (the weighted average of a wide credible interval).86 Pure Earth told us that it expects that an intervention to reduce spices would have nationwide impact.87 However, we are uncertain about whether some regions might have lower exposure to spices (or be less affected by the intervention) than the population of Bangladesh that the model was applied to.

This yielded an estimate of a 0.2-0.4 µg/dL decline in average nationwide blood lead levels in expectation.88 We decided not to spend more time refining these estimates because we expect to find out more from the data that Pure Earth plans to collect. While these estimates are guesses, we thought it was important to make them explicit to make it clear what we would consider to be a better outcome than expected.

Extent to which reductions in blood lead levels are counterfactually sustained over time

Our best guess is that reductions in lead exposure Pure Earth contributes to will be accelerated by seven years, and so we model seven years of benefit.89 This assumption is very uncertain and subjective, which makes it difficult to compare the cost-effectiveness of this grant with that of GiveWell's top charities. This factor conceptually includes both (i) the chance that particular sources of lead exposure would be addressed without Pure Earth's involvement, and (ii) the chance that behavior might revert over time.

We selected ten years as a reasonable default for evaluating the impact of implementing regulations in general, and subjectively adjusted downwards to seven years because this grant is primarily about enforcing existing regulations rather than implementing new regulations (and we expect improvements in enforcement to be at higher risk of reversion than changes in regulations).90

We believe it is unlikely that lead would be removed from spices in the near term without Pure Earth's involvement. We spoke with two people familiar with food regulation policy in India who told us that regulations on maximum lead levels in spices were poorly enforced, and that the public was generally not aware of the risk.91 We have not spoken with anyone familiar with the regulatory status of lead in the Philippines.

What are the benefits of reducing blood lead levels?

Our best guess is that reducing blood lead levels by 1 µg/dL throughout childhood increases intelligence quotient (IQ) by 0.15 points, leading to a 0.1% increase in lifetime earnings. These estimates are necessarily uncertain (though less speculative than our guesses about the expected effect size above). They are based on a non-exhaustive review of the academic literature linking blood lead levels to cognitive ability and cognitive ability to earnings. More details above.

We applied the expected benefit of a population-wide reduction in lead exposure to a single annual cohort in each location, approximated by the number of births each year in those locations.92

Other considerations

There were a number of considerations we did not investigate fully because we thought it was unlikely they would change our decision to recommend this grant. We include rough guesses about how they would affect the cost-effectiveness of the grant here. Those considerations include:

• Government costs of enforcement to maintain reductions in lead exposure.
• Morbidity from lead exposure.
• Mortality from lead exposure.
• Potential social harms from lead exposure (such as crime or increased aggression).

Plan for follow-up and open questions

We expect to have conversations with Pure Earth every few months to check in on progress. We expect to consider whether to recommend renewing or extending this grant when Pure Earth has one year of funding remaining on this grant, or if Pure Earth identifies additional promising opportunities to reduce lead exposure through its evidence-generating activities.

Open questions about the grant we plan to follow up on

• How are Pure Earth's activities progressing against planned milestones?
• Did Pure Earth and partners successfully collect national representative blood lead level estimates, and what do those imply about potential countries to work in?
• Did the Rapid Market Surveys identify other promising sources of lead exposure to address?
• Did blood lead levels decline in locations funded by this grant?
• Did the prevalence of lead in spices in Bangladesh remain low in the years after the initial intervention?

Other open questions we may investigate further

• Would deeper review of the literature on the relationship between lead exposure and cognitive function meaningfully change our conclusions?
• Would deeper review of the literature on the relationship between cognitive function and earnings meaningfully change our conclusions?
• How much would including estimates of mortality, morbidity, and potential social harm caused by lead exposure increase our estimates of cost-effectiveness?
• What is our best guess of the relative contribution of different sources of lead exposure in different locations, and how should that affect our prioritization? Are there promising opportunities to get more evidence on this question?
• How does the global supply chain for lead work? Are there opportunities to support activities to reduce lead exposure by focusing at a higher point on the supply chain (rather than focusing on the most proximate cause of exposure)?

Forecasts

We made a number of forecasts related to this grant.

Prediction	Resolution date	Credence
We recommend another grant to Pure Earth	3 years after funding is received	James: 75% Julie: 60%
We recommend another grant to Pure Earth	2 years after funding is received	James: 40% Julie: 40%
We estimate that Pure Earth's cost-effectiveness based on its direct contributions to reducing blood lead levels over the course of this grant period was <5x as cost-effective as cash transfers	3 years after funding is received, conditional on having an estimate that we think is a reasonable approximation.	James: 40% Julie: 30%
We estimate that Pure Earth's cost-effectiveness based on its direct contributions to reducing blood lead levels over the course of this grant period was 5-10x as cost-effective as cash transfers	3 years after funding is received, conditional on having an estimate that we think is a reasonable approximation.	James: 20% Julie: 30%
We estimate that Pure Earth's cost-effectiveness based on its direct contributions to reducing blood lead levels over the course of this grant period was 10-20x as cost-effective as cash transfers	3 years after funding is received, conditional on having an estimate that we think is a reasonable approximation.	James: 20% Julie: 25%
We estimate that Pure Earth's cost-effectiveness based on its direct contributions to reducing blood lead levels over the course of this grant period was >20x as cost-effective as cash transfers	3 years after funding is received, conditional on having an estimate that we think is a reasonable approximation.	James: 20% Julie: 15%
Pure Earth identifies at least two other location-specific funding gaps that we believe are >10x as cost-effective as cash transfers (in expectation)	3 years after funding is received	James: 65% Julie: 50%
Forecasts of blood lead level surveys identifying different levels of reduction in blood lead levels in the target population	3 years after funding is received	Available here.

Our process

• GiveWell identified lead exposure as a promising area to work on through its work to identify giving opportunities to assist with public health regulation. We started investigating a potential grant to Pure Earth in April 2021. At that time, we also started investigating grants to organizations working to reduce lead exposure from paint.
• We had a number of conversations with senior management at Pure Earth to better understand its model, and how Pure Earth plans to spend additional funding.
• We spoke with past funders of Pure Earth's work and in-country partners to better understand Pure Earth's capabilities.
• We spoke with two people who were familiar with the regulatory landscape for lead exposure in India and Bangladesh.
• We completed shallow reviews of the evidence on lead exposure's effect on cognitive ability, and the relationship between cognitive ability and future earnings (this work was largely completed before we started investigating Pure Earth).
• We reviewed the landscape of other potential funders and grant opportunities to address lead exposure.
• We did desktop research and spoke to people with relevant expertise to understand the case for focusing on lead in spices.
• We built a rough cost-effectiveness model to synthesize the case and decide which parts of Pure Earth's initial proposal would be a good fit for funding from GiveWell.

Sources

Document	Source
American Academy of Pediatrics 2016	Source (archive)
Arain et al. 2015	Source
Bernard and Schukraft 2021	Source (archive)
CDC, "Blood Lead Reference Value"	Source (archive)
Clarios, "Products"	Source (archive)
Dignam et al. 2019	Source
Ericson et al. 2021	Source (archive)
Ericson et al. 2021, Supplementary appendix 1	Source (archive)
Forsyth et al. 2019a	Source (archive)
Forsyth et al. 2019b	Source (archive)
Gazze, Persico, and Spirovska 2021 (working paper)	Source (archive)
Ghiasvand et al. 2016	Source (archive)
GiveWell, Back-of-the-envelope cost-effectiveness analysis of grant to Pure Earth, 2021	Source
GiveWell, Calculations for the effect of IQ on income, 2021	Source
GiveWell, Crowdedness calculations for public health regulation, 2021	Source
GiveWell, Literature review for the effect of IQ on income, 2020	Source
GiveWell, Public health regulation prioritization dashboard, 2021	Source
GiveWell, Public health regulation update, 2021	Source
GiveWell, Report on lead toxicity, 2019	Source
GiveWell, Shallow review of evidence on the relationship between blood lead levels and IQ, 2019	Source
GiveWell's non-verbatim summary of a conversation with Dr. Kamalesh Sarkar, June 28, 2021	Source
GiveWell's non-verbatim summary of a conversation with Dr. Stephen Luby, November 13, 2017	Source
GiveWell's non-verbatim summary of a conversation with IPEN, June 24, 2019	Source
GiveWell's non-verbatim summary of a conversation with IPEN, October 23, 2017	Source
GiveWell's non-verbatim summary of a conversation with Jenna Forsyth, May 28, 2021	Source
GiveWell's non-verbatim summary of a conversation with Misbah Wasi, July 2, 2021	Source
GiveWell's non-verbatim summary of a conversation with Pure Earth, April 7, 2021	Source
GiveWell's non-verbatim summary of a conversation with Pure Earth, July 6, 2021	Source
GiveWell's non-verbatim summary of a conversation with Pure Earth, May 26, 2021	Source
Gleason et al. 2014	Source (archive)
Gottesfeld and Ismawati 2021	Source (archive)
Government of India, Office of the Registrar General & Census Commissioner, Primary Census Data Highlights: Chapter 1	Source (archive)
Government of Mexico, National Institute of Statistics and Geography, Mexico in Figures: Michoacán de Ocampo	Source (archive)
Government of Mexico, National Institute of Statistics and Geography, Mexico in Figures: United Mexican States	Source (archive)
Higney, Hanley, and Moro 2021 (working paper)	Source (archive)
IHME, "Frequently Asked Questions"	Source (archive)
IHME, Global Burden of Disease, GBD Results Tool, 2019 deaths and YLDs due to lead exposure	Source (archive)
IHME, Global Burden of Disease, GBD Results Tool, 2019 deaths for ages 25-80 plus due to lead exposure in Mexico, Peru, Bangladesh, India, the Philippines, and globally	Source (archive)
IHME, Live births retrospective estimates and forecasts, all scenarios: 1950-2100	Source (archive)
Knoema, World Data Atlas, 2018 crude birth rate for Bihar, India	Source (archive)
Lanphear et al. 2005	Source (archive)
Lin et al. 2010	Source
Macrotrends, India Birth Rate 1950-2021 (accessed August 9, 2021)	Source (archive)
Macrotrends, Mexico Birth Rate 1950-2021 (accessed August 9, 2021)	Source (archive)
Pure Earth, "Research"	Source (archive)
Pure Earth, "Who We Are"	Source (archive)
Pure Earth, FY2019 Form 990	Source (archive)
Pure Earth, Project budget, June 7, 2021	Source
Pure Earth, Project proposal, June 8, 2021	Source
Pure Earth, Project timeline, June 7, 2021	Source
Pure Earth, Responses to GiveWell's questions, May 26, 2021	Source
Rees and Fuller 2020	Source (archive)
Syed, Qadiruddin, and Khan 1987	Source (archive)
The Agency for Toxic Substances and Disease Registry, Case Studies in Environmental Medicine: Lead Toxicity, 2017	Source (archive)
U.S. EPA, America's Children and the Environment, Appendix A: Data tables on lead biomonitoring 2019	Source (archive)
UNICEF, "Pure Earth and Clarios Foundation," 2020	Source (archive)
Van Landingham et al. 2020	Source (archive)
Wani et al. 2017	Source
Wilson and Wilson 2016	Source
World Health Organization, "Lead poisoning and health," 2019	Source (archive)

• 1

  "Pure Earth is an international non-profit organization dedicated to solving pollution problems in low- and middle-income countries, where human health is at risk." Pure Earth, "Who We Are"

• 2

  See Line 18 ("Total expenses") on Pg. 5 of Pure Earth, FY2019 Form 990.

• 3

  "Pure Earth, formerly known as the Blacksmith Institute, is a leader in global toxic pollution cleanup. Since its inception in 1999, Pure Earth has completed 110 environmental remediation projects in 27 countries, improving the lives of millions of people, especially children, who are most at risk from the threat of toxic pollution." Pure Earth, "Who We Are"

• 4

  Conversation with Richard Fuller and Marc Weinreich, October 15, 2020 (unpublished).

• 5

  "Pure Earth, formerly known as the Blacksmith Institute, is a leader in global toxic pollution cleanup. Since its inception in 1999, Pure Earth has completed 110 environmental remediation projects in 27 countries, improving the lives of millions of people, especially children, who are most at risk from the threat of toxic pollution." Pure Earth, "Who We Are"

• 6

  "A little over a year ago, Pure Earth underwent a strategic shift from working on many different pollutants to focusing mostly on reducing lead exposure. Its lead work has primarily consisted of remediating toxic sites contaminated by lead-acid battery recycling, under the assumption that governments would then scale these projects. However, this has not been the case, and Pure Earth is now focused instead on upstream interventions that impact larger populations (e.g., influencing regulatory policy)." GiveWell's non-verbatim summary of a conversation with Pure Earth, April 7, 2021, Pg. 1.

• 7
  1. "Data Collection: Collect information on the scope, severity, demographics, impacts and sources of lead exposure in selected geographies to increase the efficacy and efficiency of subsequent intervention design and implementation. This component includes conducting baseline and endline BLL assessments, conducting rapid baseline and endline marketplace surveys or home-based source analyses to identify main sources of lead, conducting value or supply chain tracking of lead containing products, reviewing current regulatory frameworks and gaps, creating a case studies, and other activities to monitor progress and outcomes.
  2. Influence: Share results of BLL, marketplace/source analyses, case studies, and relevant guidance with local, national, and international stakeholders to: develop educational materials for consumers, develop education and/or training materials for producers, sensitize governments to potential intervention strategies and develop joint action plans, activate resources and partners, and increase the effectiveness of lead exposure reduction programs.
  3. Intervention/Impact: Implement lead exposure prevention and source control measures for priority sources with local partners to reduce lead exposures from identified sources and decrease baseline blood lead levels (BLLs): provide technical assistance (training, communications, joint investigations) to governments to enforce regulations and roll out public/consumer education and producer training programs.

  Pure Earth’s activities and interventions are intended to significantly reduce the BLLs of children. Intervention design is informed by an understanding of the main, or contributory, sources of exposure in a country or region. More details are available in Appendix A and B." Pure Earth, Project proposal, June 8, 2021, Pgs. 1-2.

• 8
  • "Work is needed to validate a recently completed intervention that addressed turmeric adulteration in Bangladesh. That intervention reduced the prevalence of highly contaminated spice samples from a pre-intervention rate of 50% highly contaminated samples, to a post-intervention rate of 5% (90% reduction). However, the endline analysis only analyzed effects on lead concentrations in turmeric samples and did not analyze the effects on BLLs. To validate this intervention and understand the impacts on beneficiaries, an analysis of the endline BLLs is required." Pure Earth, Project proposal, June 8, 2021, Pg. 6.
  • See Column V ("Total") for row "Bangladesh" in Pure Earth, Project budget, June 7, 2021.

• 9
  • "In 2018, Pure Earth implemented a project to remediate a lead-contaminated neighborhood in suburban Patna, Bihar, and did not see the anticipated reduction in BLLs [...] Pure Earth now believes that the neighborhood remediation project in Patna described above did not reduce BLLs as much as expected because much of the children’s exposure was from spices.

    Scope of Work

    • Representative baseline BLL assessment by Vital Strategies
    • Spice intervention.
    • Endline BLL assessment by Vital Strategies and spice sampling." Pure Earth, Project proposal, June 8, 2021, Pg. 7.
  • See Column V ("Total") for row "India/Bihar" in Pure Earth, Project budget, June 7, 2021

• 10
  • "[...] To evaluate the need for interventions, Pure Earth will conduct Rapid Market Screenings in Tamil Nadu (pop. ), Jharkhand (pop. 39M), Uttar Pradesh (pop. 231M), and West Bengal (pop. 100M). Depending on the results of the screenings, Pure Earth, with implementing partner Vital Strategies, will then conduct a baseline assessment of BLLs and lead sources in the homes of highly exposed children and design spice interventions likely similar to the approach implemented in Bihar (see Attachment B) in the selected state, implement the intervention, and then carry out an endline BLL survey and product sampling" Pure Earth, Project proposal, June 8, 2021, Pg. 8.
  • See Column V ("Total") for row "India/other" in Pure Earth, Project budget, June 7, 2021.

• 11
  • "Scope of Work [in the Philippines]:
    • Analysis of government-conducted BLL survey with support from Vital Strategies
    • Home-based source analysis among high BLL homes.
    • Intervention(s) to be determined.
    • National endline BLL survey with support from Vital Strategies, and product sampling." Pure Earth, Project proposal, June 8, 2021, Pg. 9.
  • See Column V ("Total") for row "Philippines" in Pure Earth, Project budget, June 7, 2021.

• 12
  • "Specifically, Pure Earth will:

    [...]

    • Design and launch interventions in an additional 2-3 countries, based upon the data collected from the RMS (Years 2 and 3)." Pure Earth, Project proposal, June 8, 2021, Pg. 3.
  • See Column V ("Total") for row "Additional 2 countries" in Pure Earth, Project budget, June 7, 2021.

• 13
  • "Specifically, Pure Earth will:
    • Conduct Regional Rapid Marketplace Screenings (RMS) in 20+ countries to identify lead-containing products and prioritize countries for future interventions (Year 1)." Pure Earth, Project proposal, June 8, 2021, Pg. 3.
  • "A regional public health communications consultant will be part of the eight Regional Rapid Marketplace Screening teams. This person will capture video and photos of all the project activities. They will observe and create documentation of relevant socio-economic, cultural, gender-norms, and other contextual factors that will be useful in when [sic] creating awareness-raising, outreach and educational materials in Phase 2 - Influence, and Phase 3 - Intervention/Impact of the project [...] The media assets gathered can be crafted in multiple ways to deliver messages to a range of project stakeholders, targeting messages for different audience segments." Pure Earth, Project proposal, June 8, 2021, Pg. 11.
  • Pure Earth's project timeline indicates that 25 geographies will be chosen for the Regional Rapid Marketplace Screenings.
  • See Column V ("Total") for rows "RMS" and "Comms" in Pure Earth, Project budget, June 7, 2021.
    • $441,826 + $476,530 = $918,356

• 14
  • "Pure Earth will undertake vigorous monitoring, evaluation and learning (M&E) during all three years of the proposed project, including developing metrics, standards and protocols; identifying and adopting best practices; M&E training for staff and partner organizations; monitoring key activities, budgets and project documentation, and evaluation of outcomes, impacts and achievements of the project overall. To build its capacity in MEL [(monitoring, evaluation, and learning)], Pure Earth is proposing to add full time M&E staff, engage consultants and create a working group to establish robust M&E protocols to support the proposed project." Pure Earth, Project proposal, June 8, 2021, Pg. 10.
  • See Column V ("Total") for row "MEL" in Pure Earth, Project budget, June 7, 2021.

• 15
  • "Pure Earth will also work with GAHP [(The Global Alliance on Health and Pollution)] to develop a robust global response to lead pollution and poisoning, with the aim to raise not only attention and awareness to the issue, but also to elevate it on the international development agenda and increase allocation, provision and access to resources to solve lead pollution at source." Pure Earth, Project proposal, June 8, 2021, Pg. 9.
  • "Pure Earth will also work with the GAHP to develop robust responses from other agencies and NGOs on this issue. A Global Lead Working Group will help to spread knowledge and interest in this problem within the UN system, multi-lateral development banks, NGOs, and affected country ministries of health and environment." Pure Earth, Project proposal, June 8, 2021, Pg. 4.
  • See Column V ("Total") for row "GAHP" in Pure Earth, Project budget, June 7, 2021.

• 16

  GiveWell's non-verbatim summary of a conversation with Pure Earth, May 26, 2021

  • "Pure Earth has a small office and country director in Bangladesh. The country director will work closely with Stanford researchers who will be conducting additional blood lead testing in Bangladesh; Stanford researchers have also done some advocacy work with local officials." Pg. 2.
  • "Pure Earth would like to receive funding to support a collaborative effort with Vital Strategies [...] If Pure Earth worked on blood lead level assessments with Vital Strategies instead of with the government, the cost would be higher." Pg. 5.

• 17

  "Pure Earth will undertake vigorous monitoring, evaluation and learning (M&E) during all three years of the proposed project [...]" Pure Earth, Project proposal, June 8, 2021, Pg. 10.

• 18

  "A little over a year ago, Pure Earth underwent a strategic shift from working on many different pollutants to focusing mostly on reducing lead exposure. Its lead work has primarily consisted of remediating toxic sites contaminated by lead-acid battery recycling, under the assumption that governments would then scale these projects. However, this has not been the case, and Pure Earth is now focused instead on upstream interventions that impact larger populations (e.g., influencing regulatory policy)."GiveWell's non-verbatim summary of a conversation with Pure Earth, April 7, 2021, Pg. 1.

• 19

  The second table in this section of our "Public health regulation update" page presents our burden and philanthropic spending estimates. We estimate a burden of 94 for lead exposure and 288 for tobacco, and we estimate philanthropic spending of $7 million for lead exposure and $70 million for tobacco.

  • 94/288 = ~33%
  • 7/70 = 10%

  Note that to compare cause areas, these burden estimates were converted into GiveWell's moral weights, based on data from the Institute of Health Metrics and Evaluation on the total deaths and disability attributed to each cause area. Our total burden estimate for lead exposure also includes an estimate of its financial burden due to lost IQ and earnings, a result of the negative developmental effects of lead exposure. Each unit of burden is equal to the value of doubling economic consumption for one person for one year. This means that, for example, we would weigh the benefits of preventing all deaths, disability, and financial burden from lead exposure for one year equally to doubling the economic consumption of approximately 94 million people. For more details, see this page.

• 20

  The second table in this section of our "Public health regulation update" page presents our burden and philanthropic spending estimates. Our burden estimates for lead exposure, alcohol, and pesticide regulation are 94, 132, and 11 respectively, and our philanthropic spending estimates are $7 million, $3 million, and $400,000 respectively.

• 21

  "Fifth, learn which products may contain lead and avoid using them. These products vary considerably by country and context; however, products that frequently contain lead include artisanal ceramics, some spices, traditional remedies and some cosmetics." Rees and Fuller 2020, Pg. 17.

• 22
  • "In this systematic review, we searched PubMed for studies published between Jan 1, 2010, and Oct 31, 2019, that reported blood lead levels in the 137 countries in World Bank LMIC groupings [...] , the final dataset for analysis comprised 520 studies covering 1100 sampled populations (ie, subsamples) with a total population of 1 003 455 people from 49 countries (table 1) [...] Probable sources of exposure were identified for 478 (43%) of 1100 subsamples." Ericson et al. 2021, Pgs. e145 and e149.
  • "Instead, key sources of lead exposure in LMICs identified in this systematic review include informal lead acid battery recycling and manufacture, metal mining and processing, electronic waste, and the use of lead as a food adulterant, primarily in spices. In one country in particular, Mexico, lead-based ceramic glazes were a significant source of exposure." Ericson et al. 2021, Pgs. e151-e152.
  • See Table 2, Pg. e148 of Ericson et al. 2021 for the full list of sources of lead exposure identified.

• 23

  "For this systematic review, we searched PubMed between Nov 11 and Nov 29, 2019, using the search terms “[country name]” (all fields, Medical Subject Heading [MeSH] terms, abstract text), “blood” (subheading, all fields, MeSH terms), and “lead” (all fields, MeSH terms) for studies published between Jan 1, 2010, and Oct 31, 2019. LMIC names (n=137) were taken from World Bank groupings.17 Studies were reviewed for inclusion if they contained blood lead level data from human populations residing in any given country; comprised at least 30 participants; presented blood lead level data derived from venous, capillary, or umbilical cord samples of whole blood (serum and plasma samples were excluded); had data that were collected after Dec 31, 2004; and were published in English. At least two reviewers independently searched PubMed with the defined search terms.

  Titles were reviewed for relevance as a first screening step and selected studies were then subject to review of associated abstracts. When a study did not meet the inclusion criteria, it was excluded and a justification was provided by the reviewer. Each reviewer independently completed all steps in the review process, with all unique studies being combined in a comprehensive list. Any potential conflicts were resolved between reviewers through discussion with a third collaborator.

  Data analysis

  A bias assessment was done following guidance provided by the US Office of Health Assessment and Translation Handbook.18 The tool is comprised of 11 possible questions addressing different types of bias. Studies are assessed against the applicable questions and ranked according to tier, with tier one indicating a low risk of bias and tier three indicating an unacceptable (ie, high) risk of bias. Each study was reviewed against seven relevant questions to assess study design, data analysis and interpretation, and the presentation of results. The full text of these seven questions is attached in appendix 1 (p 3). Only studies in tier one and tier two were included in the analysis.

  The following information was extracted by individual reviewers in duplicate from studies that met the inclusion criteria: title, author, year, location, population characteristics (eg, sex and age), statistics on blood lead level (eg, central tendency, dispersion, sample size), sources of exposure, analysis method, and the nature of exposure (ie, background, occupational, non-occupational).

  Population subsamples were coded to different subgroups on the basis of the severity of their exposure. Subsamples with an identifiable high risk of exposure not representative of the general population (eg, living near a known hazardous waste site or applying contaminated cosmetics) were coded to the non-occupational subgroup. Those subsamples drawn from general populations or used as controls in case-control studies were coded to the background subgroup and were assumed to be representative of general exposure. Subsamples drawn from worker populations exposed to lead were coded to the occupational subgroup.

  Where possible, blood lead levels were separated into adult or child subsamples, and coded as such. Where this was not possible, the subsample was coded as both. A child was defined as aged 18 years or younger, consistent with the UN Convention on the Rights of the Child.19 Subsamples were disaggregated by occupation and source of exposure where those data were available. Subsamples were disaggregated by sex only when a sample with both sexes was unavailable. If sex was not specified, subsamples were coded to the so-called both subgroup.

  Studies were also flagged as to whether or not probable sources of exposure were identified, such as communities living near contamination hotspots or adjacent to industrial areas, occupational exposures, or those with significant associations between blood lead level and an environmental assessment or questionnaire. Studies carried out of populations in cities with significant industrial activity that the authors linked to lead exposure were also categorised as probable. Studies that evaluated blood lead level in a given population and that provided possible sources of exposure on the basis of a review of the literature, rather than an assessment of exposure in situ, were coded to the so-called possible subgroup. Studies that did not define an exposure source were coded to the undefined subgroup. To deal with data duplication, we first compared study titles, then the unique PubMed identifier (ie, PMID or EntrezUID)." Ericson et al. 2021, Pgs. e146-e147.

• 24
  • "Subsamples with an identifiable high risk of exposure not representative of the general population (eg, living near a known hazardous waste site or applying contaminated cosmetics) were coded to the non-occupational subgroup. Those subsamples drawn from general populations or used as controls in case-control studies were coded to the background subgroup and were assumed to be representative of general exposure."Ericson et al. 2021, Pg. e147.
    • See column AJ ("elevated or background") on sheet "included" of Ericson et al. 2021, Supplementary appendix 1. Subsamples coded as "exposed" in this column are not representative of the general population.
  • We randomly selected and read the abstracts of three studies cited in the review that identified a source of exposure (Arain et al. 2015, Ghiasvand et al. 2016, and Wani et al. 2017). Each of these studies identified a source of exposure in a population already known to be exposed to that particular source.
    • "The present study was aimed to evaluate the cadmium (Cd) and lead (Pb) levels in the blood samples of adolescent boys, chewing different smokeless tobacco (SLT) products in Pakistan. For comparative purpose, boys of the same age group (12-15 years), not consumed any SLT products were selected as referents." Arain et al. 2015, Abstract.
      • The subsamples representing the treatment group in this study were coded as identifying "tobacco products" as the source of exposure. See column AJ ("elevated or background") for rows 838-840 on sheet "included" of Ericson et al. 2021, Supplementary appendix 1.
    • "This study was undertaken with a view to investigate the relationship between blood lead level (BLL) and hearing loss in workers in a lead-acid battery manufacturing plant in Tehran, Iran [...] In a cross-sectional study, 609 male workers were recruited from different locations in the factory." Ghiasvand et al. 2016, Abstract.
      • The subsamples in this study were coded as identifying "battery manufacture or recycling" as the source of exposure. See column AJ ("elevated or background") for rows 572-575 on sheet "included" of Ericson et al. 2021, Supplementary appendix 1.
    • "Blood samples were collected from 92 workers as participants occupationally exposed to lead or lead and zinc and 38 comparison participants having no history of such exposure." Wani et al. 2017, Abstract.
      • The subsamples representing the treatment group in this study were coded as identifying "battery manufacture or recycling" as the source of exposure. See column AJ ("elevated or background") for rows 520-521 on sheet "included" of Ericson et al. 2021, Supplementary appendix 1.
  • We agree with Rethink Priorities' assessment: "A recent systematic review in The Lancet (Ericson et al., 2021) analyzes 478 studies of blood lead levels in low- and middle-income countries.[5] The review finds that battery manufacture or recycling was the primary source of lead exposure in more studies (118) than any other source. Somewhat surprisingly, lead-based paint is only reported as the primary source in 7 studies. However, the review only tracks primary exposure sources. It would be naive to extrapolate from the fact that 7/478 studies are from contexts where lead paint is the primary exposure source to the claim that 1.5% of the total burden is from paint because (1) paint is likely to be a non-primary but significant exposure source in many contexts, and (2) there’s no reason to expect the distribution of studies across contexts to be representative since studies are more likely to take place in localities with high lead exposure. In other words, we are concerned that if you only did studies near hotspots and you only looked at primary exposure pathways, you might come to believe that a majority of the burden is attributable to (e.g.) informal battery recycling sites. If you looked instead, at the population as a whole, it might turn out that (e.g.) paint causes a higher (though more diffuse) burden." Bernard and Schukraft 2021

• 25

  "Furthermore, exposures from lead paint are not limited to home interiors, given that these coatings are also applied to bridges, boats, roadways, automobiles, and other products. Even homes with lead-based paint on the exterior only generally have high dust and soil lead contamination.4 Para-occupational or take-home exposures from workers manufacturing lead paint and from those working in construction and painting are known to be a substantial source of exposure. Furthermore, lead paint is most problematic when it is disturbed during renovations and repainting, which is the reason why the USA has a specific rule to address these practices in homes built before 1978.5 Generally, the worst exposures occur in older homes with lead-based paint that deteriorates over many decades." Gottesfeld and Ismawati 2021, Pg. e859.

• 26

  "Lead poisoning usually is detected by measuring the level in blood. Many screening procedures use capillary blood as point-of-care testing. While finger-prick samples are appropriate for screening tests, all elevated capillary levels should have confirmation with a venous blood draw since capillary tests can yield frequent false positives. An elevated capillary lead level indicates that lead is in the child's environment even if the venous level is low, however, so primary prevention and education should be initiated." American Academy of Pediatrics 2016

• 27

  "From 1976-1980 to 2015-2016, the overall estimated geometric mean BLL of the US population aged 1 to 74 years decreased from 12.8 to 0.82 μg/dL, representing a decline of 93.6% (CDC, unpublished data, 2018)." Dignam et al. 2019, Pg. 2.

• 28

  "In 2012, CDC updated its recommendations on children’s blood lead levels. CDC now uses a blood lead reference value of 5 micrograms per deciliter (µg/dL) to identify children with blood lead levels that are much higher than most children’s levels. This new level is based on the U.S. population of children ages 1-5 years who are in the highest 2.5% of children when tested for lead in their blood.

  CDC’s Blood Lead Reference Value

  The blood lead reference value is based on the 97.5th percentile of the National Health and Nutrition Examination Survey (NHANES) blood lead distribution in children ages 1-5 years. The current reference value is based on NHANES data from 2007-2008 and 2009-2010.

  NHANES is a population-based survey to assess the health and nutritional status of adults and children in the United States and determine the prevalence of major diseases and risk factors for diseases.

  CDC will assess the NHANES data every 4 years using the two most recent survey cycles of available data to determine if the blood lead reference value should be updated." CDC, "Blood Lead Reference Value"

• 29

  "This systematic review of published values of blood lead level in LMICs had three major findings. The first finding is that there is a paucity of rigorous data on lead exposure in the general populations of LMICs. In the 137 countries classified as LMICs by the World Bank, only 44 countries (32%) had data on background blood lead level that were judged to be of adequate quality to be included in this analysis. Fewer countries had data on blood lead levels of adequate quality for children (34 countries [25%]) and adults (37 countries [27%]) from the general population, and fewer still had such data based on more than one subpopulation (23 countries [17%] for children and 27 countries [20%] for adults). In terms of geographical distribution, most of the LMICs in Africa and many in central Asia and Latin America did not have any studies that met our criteria." Ericson et al. 2021, Pg. e150.

• 30

  "The second finding is that, among the 44 countries with background data that were analysed, the average blood lead level of the general population appears to be higher in LMICs than in HICs, most importantly in children. Of the 34 countries with background data for children, more than 632 million (95% CI 394 million–780 million; 48·5%) of 1·28 billion children were estimated to have a blood lead level exceeding the CDC reference value of 5 μg/dL, among whom 413 million children (236 million–551 million; 31·7%) were estimated to have a blood lead level above 10 μg/dL. These figures might be underestimates of global LMIC exposure because they represent only 34 of 137 LMIC countries. Additionally, the subsamples used here were chosen in studies as control populations specifically because they represented an absence of identifiable sources of exposure. By excluding non-occupational exposures, this study presents a more conservative picture of lead exposure in LMICs." Ericson et al. 2021, Pg. e150.

• 31
  • The data we have seen suggests that 50% of children in low- and middle-income countries have blood lead levels five times as high as the average blood lead level for individuals in the U.S.:
    • "From 1976-1980 to 2015-2016, the overall estimated geometric mean BLL of the US population aged 1 to 74 years decreased from 12.8 to 0.82 μg/dL, representing a decline of 93.6% (CDC, unpublished data, 2018)." Dignam et al. 2019, Pg. 2.
    • "Of the 34 countries with background data for children, more than 632 million (95% CI 394 million–780 million; 48·5%) of 1·28 billion children were estimated to have a blood lead level exceeding the CDC reference value of 5 μg/dL, among whom 413 million children (236 million–551 million; 31·7%) were estimated to have a blood lead level above 10 μg/dL." Ericson et al. 2021, Pg. e150.
    • 5/0.82 = ~5
  • The data we have seen suggests that the blood lead levels of children in low- and middle-income countries are lower than the average blood lead level for individuals in the U.S. during a time when leaded gasoline was still partially in use:
    • "From 1976-1980 to 2015-2016, the overall estimated geometric mean BLL of the US population aged 1 to 74 years decreased from 12.8 to 0.82 μg/dL, representing a decline of 93.6% (CDC, unpublished data, 2018)." Dignam et al. 2019, Pg. 2.
    • "Of the 34 countries with background data for children, more than 632 million (95% CI 394 million–780 million; 48·5%) of 1·28 billion children were estimated to have a blood lead level exceeding the CDC reference value of 5 μg/dL, among whom 413 million children (236 million–551 million; 31·7%) were estimated to have a blood lead level above 10 μg/dL." Ericson et al. 2021, Pg. e150.
    • "Nonetheless, in 1972, the Environmental Protection Agency (EPA) put forward a health-based regulation to remove lead from gasoline.14 An EPA cost-benefit analysis showed that tightening lead standards would lead to health benefits far outweighing the industrial costs of attaining those standards.15 In addition, the CDC released data demonstrating the close correlation between BLLs among the general population and gasoline lead concentrations during the late 1970s.16 Over time, lead in gasoline was reduced from an average 2.5 g per gallon in 1971 17 to 0.1 g per gallon in 1986.18 The Clean Air Act Amendments of 1990 resulted in a final ban on leaded gasoline for most motor vehicle use, effective January 1, 1996.19 Figure 2 shows the amount of leaded gasoline used in the United States during 1975 to 1996 and the quarterly average maximum concentrations of lead in air at EPA-monitoring sites during 1977 to 2016. During the period, the quarterly average maximum amount of lead in air decreased by 97.7% from 1.35 μg/m3 in 1977 to 0.03 μg/m3 in 2016." Dignam et al. 2019, Pg. 3.

• 32

  "Lead exposure can have serious consequences for the health of children. At high levels of exposure, lead attacks the brain and central nervous system to cause coma, convulsions and even death. Children who survive severe lead poisoning may be left with mental retardation and behavioural disorders. At lower levels of exposure that cause no obvious symptoms lead is now known to produce a spectrum of injury across multiple body systems. In particular lead can affect children’s brain development resulting in reduced intelligence quotient (IQ), behavioural changes such as reduced attention span and increased antisocial behavior, and reduced educational attainment. Lead exposure also causes anaemia, hypertension, renal impairment, immunotoxicity and toxicity to the reproductive organs. The neurological and behavioural effects of lead are believed to be irreversible." World Health Organization, "Lead poisoning and health," 2019

• 33

  "The objective of this study was to examine the association of intelligence test scores and blood lead concentration, especially for children who had maximal measured blood lead levels < 10 µg/dL. We examined data collected from 1,333 children who participated in seven international population-based longitudinal cohort studies, followed from birth or infancy until 5–10 years of age. The full-scale IQ score was the primary outcome measure. The geometric mean blood lead concentration of the children peaked at 17.8 µg/dL and declined to 9.4 µg/dL by 5–7 years of age; 244 (18%) children had a maximal blood lead concentration < 10 µg/dL, and 103 (8%) had a maximal blood lead concentration < 7.5 µg/dL. After adjustment for covariates, we found an inverse relationship between blood lead concentration and IQ score." Lanphear et al. 2005, Abstract.

• 34

  “Using a log-linear model, we found a 6.9 IQ point decrement [95% confidence interval (CI), 4.2–9.4] associated with an increase in concurrent blood lead levels from 2.4 to 30 μg/dL. The estimated IQ point decrements associated with an increase in blood lead from 2.4 to 10 μg/dL, 10 to 20 μg/dL, and 20 to 30 μg/dL were 3.9 (95% CI, 2.4–5.3), 1.9 (95% CI, 1.2–2.6), and 1.1 (95% CI, 0.7–1.5), respectively.” Lanphear et al. 2005, Abstract.

  • 3.9 IQ points / (10 - 2.4) μg/dL = ~ 0.5 IQ points / μg/dL
  • 1.9 IQ points / (20 - 10) μg/dL = ~ 0.2 IQ points / μg/dL
  • 1.1 IQ points / (30 - 20) μg/dL = ~ 0.1 IQ points / μg/dL
  • Note that, for simplicity, our calculations above assume there is a linear relationship between blood lead level and IQ within each of the three blood lead level ranges.

• 35
  • "The evidence from economic studies is affected by two key sources of bias: a) omitted variables bias, which will lead us to overestimate the effect of lead on IQ, and b) attenuation bias from measurement error in BLL measures, which will lead us to underestimate the effect of lead on IQ [...] Parenting ability and nutrition are the omitted variables that I am most concerned about." GiveWell, Shallow review of evidence on the relationship between blood lead levels and IQ, 2019, Pg. 1. More information on our reasoning is available in the report.
  • We are concerned that the steep slope at low blood lead levels in the log-linear model in Lanphear et al. 2005 may be driven by the sampling strategy of two of the included studies:
    • A casual visual examination of Figure 2 of Lanphear et al. 2005 seems to indicate that the Boston and Rochester cohorts had participants with the lowest blood lead levels. A separate review piece we skimmed noted: “The Rochester study (Canfield et al. 2003, Figure 2) shows an apparent lack of highly intelligent families living where lead exposure is more prevalent. [...] The international pooled analysis (Lanphear et al. 2005) has included these biased data-sets and added to the problem by including the high IQ and relatively low BLL Boston cohort. The Boston study (Bellinger et al. 1992) tried to control potential confounding by selecting children from a narrow range of high SES families. However, the three groups (differentiated by umbilical cord Pb levels) were not rigorously matched, resulting in sampling bias. Figure 2 shows how the biased maternal IQ in these groups is reflected in the unadjusted child IQ and how this effect was removed for the low and medium cord Pb groups by covariance adjustments (Bellinger et al. 1987).” Wilson and Wilson 2016, Pgs. 5-7.

• 36

  "The evidence of cognitive impairments with blood lead levels above 10 ug/dL is fairly strong and consistent in nonhuman primates and rodents, particularly as blood lead levels exceed 20 ug/dL [...] The evidence of cognitive impairments with blood lead levels between approximately 5 and 10 ug/dL are less abundant, less consistent, and only exists in rodents."GiveWell, Report on lead toxicity, 2019, Pg. 2. More information on our reasoning is available in the report.

• 37
  • A casual visual examination of Figure 2 of Lanphear et al. 2005 seems to indicate that the Boston and Rochester cohorts had participants with the lowest blood lead levels. A separate review piece we skimmed noted:
    • “The Rochester study (Canfield et al. 2003, Figure 2) shows an apparent lack of highly intelligent families living where lead exposure is more prevalent. [...] The international pooled analysis (Lanphear et al. 2005) has included these biased data-sets and added to the problem by including the high IQ and relatively low BLL Boston cohort. The Boston study (Bellinger et al. 1992) tried to control potential confounding by selecting children from a narrow range of high SES families. However, the three groups (differentiated by umbilical cord Pb levels) were not rigorously matched, resulting in sampling bias. Figure 2 shows how the biased maternal IQ in these groups is reflected in the unadjusted child IQ and how this effect was removed for the low and medium cord Pb groups by covariance adjustments (Bellinger et al. 1987).” Wilson and Wilson 2016, Pgs. 5-7.
  • The average linear estimate of 0.2 IQ points per 1 µg/dL increase of BLL comes from a paper that reanalyzed the data in Lanphear et al. 2005 using a linear model: “Our work with the full database of cohort data used by Lanphear et al. (2005) and Crump et al. (2013) in this paper presents a method to identify the confounding variables and a reanalysis of the data using the interaction terms identified [...] Table 11 shows the estimate for the change in IQ based on four different levels of BPb. For 1 µg/dL BPb, the estimate for change in IQ is –0.203 for the linear model and –2.36 for the log-linear model when the confounders are set at an average (both indicate a decrease in the child’s IQ)." Van Landingham et al. 2020, Pgs. 816 and 821-822.
    • Note that this paper’s authors disclosed support from a lead industry group: “This work was supported by a contract awarded to Ramboll US Consulting (Ramboll) by the International Lead Association (ILA).” Pg. 824.

• 38

  0.2 * (1 - 0.25) = 0.15

• 39

  "I go through each of the studies of the effect of cognitive skills on income from low and middle income countries [...] The bottom-line best guess is based on estimates from eight Mincerian-style LMIC studies [...] Bottom-line: a one point increase in IQ causes a 0.67 percent increase in income." GiveWell, Literature review for the effect of IQ on income, 2020, Pgs. 1-2. More information on our reasoning is available in the report.

  • 0.15 * 0.67 = ~0.1%

• 40

  See columns T ("Sample size") and N ("Converting to per IQ point (i.e. each increase in IQ is associated with what % increase in income)?") for rows "Diaz et al. 2013," "Acosta et al. 2015," and "Jolliffe 1998," in GiveWell, Calculations for the effect of IQ on income, 2021, sheet "Literature review." Additional details on our reasoning are available in GiveWell, Literature review for the effect of IQ on income, 2020.

• 41

  For example, see the following studies:

  • "Children exposed to pollutants like lead are more disruptive and have lower achievement. However, little is known about whether lead-exposed children affect the long-run outcomes of their peers. We estimate these spillover effects using new data on preschool blood lead levels (BLLs) matched to education data for all students in North Carolina public schools." Gazze, Persico, and Spirovska 2021 (working paper), Abstract.
  • "Does lead pollution increase crime? We perform the first meta-analysis of the effect of lead on crime by pooling 529 estimates from 24 studies." Higney, Hanley, and Moro 2021 (working paper), Abstract.

• 42
  • IHME, Global Burden of Disease, GBD Results Tool, 2019 deaths and YLDs due to lead exposure
    • 47,866 deaths in low-income countries + 405,675 deaths in lower-middle-income countries + 378,131 deaths in upper-middle-income countries = ~830,000 deaths in low- and middle-income countries
    • 319,015 YLDs in low-income countries + 2,525,812 YLDs in lower-middle-income countries + 1,045,775 YLDs in upper-middle-income countries = ~3.9 million YLDs in low- and middle-income countries
  • "YLD is an abbreviation for years lived with disability, which can also be described as years lived in less than ideal health. This includes conditions such as influenza, which may last for only a few days, or epilepsy, which can last a lifetime. It is measured by taking the prevalence of the condition multiplied by the disability weight for that condition. Disability weights reflect the severity of different conditions and are developed through surveys of the general public." IHME, "Frequently Asked Questions"

• 43

  See our calculation in this spreadsheet.

• 44

  "From 1976-1980 to 2015-2016, the overall estimated geometric mean BLL of the US population aged 1 to 74 years decreased from 12.8 to 0.82 μg/dL, representing a decline of 93.6% (CDC, unpublished data, 2018)." Dignam et al. 2019, Pg. 2.

• 45

  For comparisons of blood lead levels, air lead concentration, and leaded gas consumption over time in the U.S., see Figures 1 and 2, Dignam et al. 2019, Pgs. 15-16.

• 46

  "The United Nations Environment Programme Partnership for Clean Fuels and Vehicles reported in May 2019 that only Algeria has not enacted legislation banning leaded gasoline." Rees and Fuller 2020, Pg. 40.

• 47

  We have not thoroughly investigated this question, but our view is based on blood lead levels being substantially higher in the U.S. before the ban of leaded gasoline than more recent studies in low and middle income countries.

  • "From 1976-1980 to 2015-2016, the overall estimated geometric mean BLL of the US population aged 1 to 74 years decreased from 12.8 to 0.82 μg/dL, representing a decline of 93.6% (CDC, unpublished data, 2018)." Dignam et al. 2019, Pg. 2.
  • "Nonetheless, in 1972, the Environmental Protection Agency (EPA) put forward a health-based regulation to remove lead from gasoline.14 An EPA cost-benefit analysis showed that tightening lead standards would lead to health benefits far outweighing the industrial costs of attaining those standards.15 In addition, the CDC released data demonstrating the close correlation between BLLs among the general population and gasoline lead concentrations during the late 1970s.16 Over time, lead in gasoline was reduced from an average 2.5 g per gallon in 1971 17 to 0.1 g per gallon in 1986.18 The Clean Air Act Amendments of 1990 resulted in a final ban on leaded gasoline for most motor vehicle use, effective January 1, 1996.19 Figure 2 shows the amount of leaded gasoline used in the United States during 1975 to 1996 and the quarterly average maximum concentrations of lead in air at EPA-monitoring sites during 1977 to 2016. During the period, the quarterly average maximum amount of lead in air decreased by 97.7% from 1.35 μg/m3 in 1977 to 0.03 μg/m3 in 2016." Dignam et al. 2019, Pg. 3.
  • "Pooled mean blood lead concentrations in children ranged from 1·66 μg/dL (SD 3·31) in Ethiopia to 9·30 μg/dL (11·73) in Palestine, and in adults from 0·39 μg/dL (1·25) in Sudan to 11·36 μg/dL (5·20) in Pakistan." Ericson et al. 2021, Abstract.

• 48

  "Our best guess is that approximately $7 million of philanthropic funding is spent on lead elimination each year." GiveWell, Crowdedness calculations for public health regulation, 2021, Pg. 4. More details are available in the document.

• 49

  "Recently, Pure Earth successfully advocated with the government of the Philippines to integrate BLL testing into its national health survey and secured funding from USAID and the Clarios Foundation, among others, and has engaged technical partners to facilitate this survey enhancement." Pure Earth, Project proposal, June 8, 2021, Pg. 8.

• 50

  "Swedish Government – The Swedish Government, via Sweden’s international aid agency, Sida, is the other of IPEN's two largest donors and provides an unrestricted grant that includes targets IPEN must meet."GiveWell's non-verbatim summary of a conversation with IPEN, June 24, 2019, Pg. 1.

• 51

  "From 2018 to 2020, IPEN will be part of a lead paint elimination project funded with $2 million from the Global Environment Facility and $8 million in co-financing." GiveWell's non-verbatim summary of a conversation with IPEN, October 23, 2017, Pg. 6.

• 52

  "Recently, Pure Earth successfully advocated with the government of the Philippines to integrate BLL testing into its national health survey and secured funding from USAID and the Clarios Foundation, among others, and has engaged technical partners to facilitate this survey enhancement." Pure Earth, Project proposal, June 8, 2021, Pg. 8.

• 53

  GiveWell's non-verbatim summary of a conversation with Pure Earth, July 6, 2021, Pg. 2.

  • "Pure Earth plans to evaluate the effectiveness of its interventions through studies that measure children's BLLs before and after the interventions. It intends to select nationally representative samples for these studies, although it may not in cases where it decides to target its intervention to specific geographic areas."
  • "The distribution of lead exposure in the target population is an important factor in evaluating Pure Earth's impact. It expects its interventions to reduce average BLLs by some amount but that it will have a significantly larger effect on the BLLs of children with very high lead exposure."
  • "Although some level of confounding is always possible in before-and-after assessments, the decline in BLLs expected to occur organically in LMICs (due to the ban of leaded gasoline many years ago) is very small relative to the impact Pure Earth expects to have."

• 54

  GiveWell's non-verbatim summary of a conversation with Pure Earth, July 6, 2021, Pg. 2.

  • "Pure Earth plans to evaluate the effectiveness of its interventions through studies that measure children's BLLs before and after the interventions. It intends to select nationally representative samples for these studies, although it may not in cases where it decides to target its intervention to specific geographic areas."
  • "The distribution of lead exposure in the target population is an important factor in evaluating Pure Earth's impact. It expects its interventions to reduce average BLLs by some amount but that it will have a significantly larger effect on the BLLs of children with very high lead exposure."
  • "Although some level of confounding is always possible in before-and-after assessments, the decline in BLLs expected to occur organically in LMICs (due to the ban of leaded gasoline many years ago) is very small relative to the impact Pure Earth expects to have."
  • "In Bangladesh, Pure Earth expects to detect at least a 1-2 µg/dL decline and potentially a much larger decline."

• 55
  • Pure Earth, "Research"
  • "Mr. Kass, now at Vital Strategies, noted that in his former work tracking mercury exposure with the New York City Department of Health and Mental Hygiene, follow-up assessments were conducted for individuals found to have abnormally high BLLs in a baseline survey, and the assessments were able to track the elevated BLLs to a single source of exposure. Pure Earth plans to implement a similar approach." GiveWell's non-verbatim summary of a conversation with Pure Earth, July 6, 2021, Pg. 2.

• 56

  For more details, see the section above on Pure Earth's proposed grant activities.

• 57

  "In total, we collected 140 turmeric samples from the 9 major turmeric-producing districts, and an additional 200 turmeric samples from the 2 minimally-producing districts, Dhaka and Munshiganj (Fig. 2,Table S4). On average, Pb concentrations were lower in the major turmeric-producing districts, with 11% of samples containing Pb in excess of the Bangladesh Standards and Testing Institution's limit of 2.5 μg/g Pb in turmeric (Bangladesh Standards and Testing Institution, 2001), compared to 26% in Dhaka and Munshiganj." Forsyth et al. 2019a, Pg. 5.

• 58

  "In total, we collected 140 turmeric samples from the 9 major turmeric-producing districts, and an additional 200 turmeric samples from the 2 minimally-producing districts, Dhaka and Munshiganj (Fig. 2, Table S4). On average, Pb concentrations were lower in the major turmeric-producing districts, with 11% of samples containing Pb in excess of the Bangladesh Standards and Testing Institution's limit of 2.5 μg/g Pb in turmeric (Bangladesh Standards and Testing Institution, 2001), compared to 26% in Dhaka and Munshiganj. Polished bulbs in all districts contained the highest Pb concentrations, followed by polished finger roots. The maximum Pb concentration of loose powder in major turmeric-producing districts was 690 μg/g compared to 1152 μg/g Pb in Dhaka and Munshiganj." Forsyth et al. 2019a, Pg. 5.

  • 1152 / 2.5 = ~460

• 59

  "Turmeric wholesalers reported that the practice of adding yellow pigments to dried turmeric root during polishing began more than 30 years ago and continues today, primarily driven by consumer preferences for colorful yellow curries. Farmers stated that merchants are able to sell otherwise poor-quality roots and increase their profits by asking polishers to adulterate with yellow pigments." Forsyth et al. 2019a, Abstract.

• 60
  • Background. During the conduct of a cohort study intended to study the associations between mixed metal exposures and child health outcomes, we found that 78% of 309 children aged 20–40 months evaluated in the Munshiganj District of Bangladesh had blood lead concentrations ≥5 µg/dL and 27% had concentrations ≥10 µg/dL. Hypothesis. Environmental sources such as spices (e.g., turmeric, which has already faced recalls in Bangladesh due to high lead levels) may be a potential route of lead exposure. Methods. We conducted visits to the homes of 28 children randomly selected from among high and low blood lead concentration groups. During the visits, we administered a structured questionnaire and obtained soil, dust, rice, and spice samples. We obtained water samples from community water sources, as well as environmental samples from neighborhood businesses. Results. Lead concentrations in many turmeric samples were elevated, with lead concentrations as high as 483 ppm. Analyses showed high bioaccessibility of lead. Conclusions. Contamination of turmeric powder is a potentially important source of lead exposure in this population." Gleason et al. 2014, Abstract.
  • "Background: Significant lead poisoning has been associated with imported nonpaint products.

    Objectives: To describe cases of pediatric lead intoxication from imported Indian spices and cultural powders, determine lead concentrations in these products, and predict effects of ingestion on pediatric blood lead levels (BLLs).

    Patients and methods: Cases and case-study information were obtained from patients followed by the Pediatric Environmental Health Center (Children's Hospital Boston). Imported spices (n = 86) and cultural powders (n = 71) were analyzed for lead by using x-ray fluorescence spectroscopy. The simple bioaccessibility extraction test was used to estimate oral bioavailability. The integrated exposure uptake biokinetic model for lead in children was used to predict population-wide geometric mean BLLs and the probability of elevated BLLs (>10 μg/dL).

    Results: Four cases of pediatric lead poisoning from Indian spices or cultural powders are described. Twenty-two of 86 spices and foodstuff products contained >1 μg/g lead (for these 22 samples, mean: 2.6 μg/g [95% confidence interval: 1.9-3.3]; maximum: 7.6 μg/g). Forty-six of 71 cultural products contained >1 μg/g lead (for 43 of these samples, mean: 8.0 μg/g [95% confidence interval: 5.2-10.8]; maximum: 41.4 μg/g). Three sindoor products contained >47% lead. With a fixed ingestion of 5 μg/day and 50% bioavailability, predicted geometric mean BLLs for children aged 0 to 4 years increased from 3.2 to 4.1 μg/dL, and predicted prevalence of children with a BLL of >10 μg/dL increased more than threefold (0.8%-2.8%).

    Conclusions: Chronic exposure to spices and cultural powders may cause elevated BLLs. A majority of cultural products contained >1 μg/g lead, and some sindoor contained extremely high bioaccessible lead levels. Clinicians should routinely screen for exposure to these products." Lin et al. 2010, Abstract.

• 61

  "During our recent national turmeric sampling effort in India in the past 6 months, we only identified alarmingly high lead levels in samples from Patna, Bihar [...] We also conducted a prior study of lead exposure among 136 children in Patna. This study found a high prevalence of adulterated spices from children’s homes, particularly turmeric, but also coriander and chili powder. Among turmeric samples, 53% were above 100 ppm, and 28% were above 1,000 ppm. The source of lead was confirmed to be lead chromate and not unintentional contamination." Pure Earth, Responses to GiveWell's questions, May 26, 2021, Pgs. 3-4.

• 62

  "These three northern states neighbor Bihar, and have been identified by Stanford as states likely to have lead chromate contaminated spices. The quality of spices in the south is generally higher, because of the export markets, and there is little adulteration per our recent study. Our hypothesis is that cities in neighboring states may also be affected by spice adulteration because of their proximity to Bihar. These nearby cities may either trade directly with vendors in Patna or may have local spice processors who adulterate with lead chromate. However, we do not have direct evidence of excessive lead levels in spices in these states as of now. If we do not find contaminated spices we will pivot accordingly." Pure Earth, Responses to GiveWell's questions, May 26, 2021, Pgs. 1-2.

• 63

  "Most human exposure to lead occurs through ingestion or inhalation." The Agency for Toxic Substances and Disease Registry, Case Studies in Environmental Medicine: Lead Toxicity, 2017, Pg. 37.

• 64

  "There are four different naturally occurring isotopes of lead, and the mix of isotopes in a sample of lead can be determined using mass spectrometry. Lead from different mines has different isotopic signatures, so if it is the case that different industries and manufacturers use lead from different mines, it might be possible for scientists to draw blood samples from people, look at the isotopic signature of the blood lead, and thereby approximate how much of the lead in each person’s blood came from different sources. This method is currently only hypothetical, but Dr. Luby believes that it is worth exploring because if it works it might solve many problems simultaneously. He estimates that such an experiment would cost on the order of hundreds of thousands of dollars." GiveWell's non-verbatim summary of a conversation with Dr. Stephen Luby, November 13, 2017, Pgs. 1-2.

• 65

  Forsyth et al. 2019b

  • "We measured the Pb isotopic composition of both potential Pb sources and 45 blood samples in order to understand which of three sources predominate: (1) food from Pb-soldered cans, (2) turmeric, or (3) geophagous materials (clay, soil, or ash)." Abstract.
  • "Forty-five pregnant women were selected for inclusion in this isotope study based on district of residence, BLL, and remaining sample blood volume. These women resided in three rural districts in Bangladesh: Mymensingh, Kishoreganj, and Tangail (Figure S3)." Pg. 11430.

• 66

  "We measured the Pb isotopic composition of both potential Pb sources and 45 blood samples in order to understand which of three sources predominate: (1) food from Pb-soldered cans, (2) turmeric, or (3) geophagous materials (clay, soil, or ash). The Pb isotope ratios of the three sources are distinct (p = 0.0001) and blood isotope ratios are most similar to turmeric. Elevated lead and chromium (Cr) concentrations in turmeric and a yellow pigment used in turmeric processing are consistent with reported consumption behavior that indicated turmeric as a primary contributor to blood Pb." Forsyth et al. 2019b, Abstract.

• 67

  GiveWell's non-verbatim summary of a conversation with Jenna Forsyth, May 28, 2021

  • "The Stanford research team conducted a study that correlated the lead isotopes in blood samples in some of the districts they visited with lead isotopes in the turmeric. This supported the conclusion that lead-adulterated turmeric was a significant contributor to blood lead levels [...] However, isotope studies on their own may not provide enough information to draw conclusions [...] Lead chromate is also added to yellow craft and artistic paint, so it is possible that the matching lead isotopes in the blood samples resulted from paint exposure instead of turmeric [...] The research team felt more confident in the results of the isotope study because it was paired with a case-control study and environmental sampling effort." Pg. 5.
  • "How much this intervention [to reduce contamination of spices with lead in Bangladesh] could contribute to a reduction in blood lead levels is uncertain. Ms. Forsyth has modeled a reduction of roughly 2 micrograms/deciliter (µg/dl) from the intervention and guesses the real effect is between 1 and 3 µg/dl." Pg. 3.

• 68

  GiveWell's non-verbatim summary of a conversation with Jenna Forsyth, May 28, 2021

  • "Turmeric production and processing in Bangladesh is not centered in a single area. According to the country's Department of Agriculture, 50% of turmeric production takes place in nine districts. [...] Much of the turmeric produced in these districts is shipped to Dhaka, the capital and largest city, then redistributed." Pg. 1.
  • "The intervention took a three-pronged approach: [...] Nationwide, revisiting the major turmeric processors whose products had high levels of lead chromate and educating them on the dangers of lead. [...] In Dhaka, working with the national Food Safety Authority on developing better monitoring and enforcement. [...] The Stanford researchers did some of the monitoring themselves at the major wholesale markets in Dhaka where turmeric shipments are aggregated. [...] The Food Safety Authority also issued warnings and large fines to wholesalers selling lead-adulterated turmeric roots. [...] In Dhaka, awareness-raising and promoting behavior change among wholesalers/distributors and consumers." Pg. 2.

• 69

  Jenna Forsyth, Presentation at the International Symposium on Spices as Flavors, Fragrances & Functional Foods, February 12, 2021 (unpublished)

• 70

  For more details, see the section above on Pure Earth's proposed grant activities.

• 71
  • Clarios Foundation is funded by Clarios LLC, a manufacturer of lead-acid batteries.
    • "Clarios Foundation is funded by Clarios LLC and is a 2020 signatory of the United Nations Global Compact, committed to aligning its strategies and operations with universal principles focused on human rights, labour, environment and anti-corruption. Clarios Foundation supports three main focus areas: children’s health and the environment, circular economy innovation and entrepreneurship, and sustainable communities." UNICEF, "Pure Earth and Clarios Foundation," 2020
    • "Clarios has a robust portfolio of innovative lead-acid and lithium-ion battery technologies designed to power nearly every type of passenger, commercial and recreational vehicle – from conventional to fully electric."Clarios, "Products"
  • We considered whether a funding relationship with Clarios Foundation could cause a conflict of interest for Pure Earth after concerns were raised to us in confidence regarding the funding relationship.
    • We spoke with three people in our network who worked on advocacy campaigns for regulations that directly opposed corporate interests and we understood to have strong views on the importance of not soliciting funding from those corporations. They believed that the risk was low in this case because (a) reducing lead exposure does not necessarily directly oppose the interests of lead-acid battery recyclers in the formal sector, (b) Clarios Foundation is constituted as a 501c(3) and so is technically independent of Clarios LLC (although we would guess this does not rule out the possibility of influence in practice), and (c) As a major donor to UNICEF, Clarios was subject to UNICEF's due diligence partnership evaluation, and was determined to be suitable.
    • We raised this question with Pure Earth, who told us:
      • "This issue has been recently a subject of board discussions. While no concerns regarding conflict of interest have arisen under the current Clarios contract, we have decided to define a donor conflict-of-interest policy broadly, and it’s currently under development. The policy will establish grant boundaries such that there is no perceived conflict of interest specifically where our work might be seen to provide a competitive benefit for the donor, or where our work (especially research) might be criticized as being compromised.

        Further, our current contract with Clarios Foundation has Ethical Standard clauses including, 'Both parties confirm that no official of the other party has been offered or received (and will not in the future receive) any benefit as a result of this collaboration.'

        'Because each Party is committed to achieving the goals of the collaboration, and the collaboration between Parties is not designed to afford commercial advantage, it is confirmed that this is not an exclusive arrangement between the parties.'" Pure Earth, Responses to GiveWell's questions, May 26, 2021, Pg. 4.

    • Overall, we believe that this funding relationship has some sub-optimal characteristics because our experience of funder-grantee power dynamics suggests that we can not rule out that funding from Clarios Foundation either has a subtle influence on Pure Earth's agenda, or could be perceived to do so. However, we decided not to investigate this issue further before making this grant because (i) we believe there is a reasonable case that accepting funding from foundations funded by industry will be net positive in situations where interests are aligned or unopposed, and (ii) this grant will focus on reducing lead in spices, and it seems unlikely to us that other funders will influence this work. We would think differently about an organization working to regulate alcohol or tobacco that accepted funding from those industries.

• 72

  Pure Earth estimates that rapid market surveys to identify potential sources of lead exposure in 25 locations will cost $442,000. See this spreadsheet.

• 73

  See this section above for details on Pure Earth's theory of change, and this section for details on the evidence for lead in spices in Bihar, India and Bangladesh.

• 74

  See this spreadsheet, sheet "Pure earth BOTEC (more detailed)," row "Accounting cost ($m)."

  • 1.3 + 1.3 + 1 = 3.6
  • 3.6 / 8 = 45%

• 75
  • We did not explicitly model the benefits from the following funding gaps:
    • Bangladesh (where the intervention has already been conducted, but funding will be used to evaluate declines in blood lead levels due to the intervention)
    • Rapid Market Surveys
    • An additional two countries to be determined
    • Support to the Global Alliance on Health and Pollution
    • Building a monitoring, evaluation and learning function
    • Communications.
  • For details on these activities, see above.
  • For details on costs, see this spreadsheet, sheet "Pure earth BOTEC (more detailed)," section "Funding gaps we assessed qualitatively."
    • 0.9 + 0.4 + 1.2 + 0.5 + 0.9 + 0.5 = 4.4
    • 4.4 / 8 = 55%

• 76

  Direct activities include funding gaps to reduce blood lead levels in Bihar, other states in Northeast India (Jharkhand, Uttar Pradesh, West Bengal, or Tamil Nadu), the Philippines, and two additional countries. Excluding the funding gap in two additional countries, these activities total $3.6 million.

  • See this spreadsheet, sheet "Pure earth BOTEC (more detailed)," row "Total costs of funding gaps we included in this analysis (and funded)."

• 77

  See this spreadsheet, sheet "Pure earth BOTEC (more detailed)," row "Multiplier for supporting activities."

  • 3.6 * 1.7 = ~$6 million
  • This total is lower than $6.8 million, which is the total of all funding gaps other than "two additional countries". See this spreadsheet, sheet "Pure earth BOTEC (more detailed)," rows "Total costs of funding gaps we included in this analysis (and funded)," "Bangladesh," "RMS," "GAHP," "MEL," and "Comms."
    • 3.6 + 0.9 + 0.4 + 0.5 + 0.9+ 0.5 = $6.8 million

    That is because the 1.7x multiplier apportions the costs of "indirect activities" (Bangladesh, Rapid Market Surveys, support to the Global Alliance on Health and Pollution, hiring a monitoring and evaluation lead, and communications) to "direct activities" in our cost-effectiveness analysis. Part of those costs are apportioned to the two additional countries Pure Earth plans to work in with this grant, which we do not include in the cost-effectiveness analysis.

• 78

  See this spreadsheet, sheet "Pure earth BOTEC (more detailed)," row "Multiplier for sustained reduction)."

  • 6 * 1.5 = $9 million

• 79

  See this spreadsheet, sheet "US BLLs historical," row "Absolute reduction per year since last measurement" for section "1990-2016."

• 80

  "The M&E protocols will include, but not limited to, the following key elements:

  • Establishing data quality objectives for all facets of the project, including appropriate training of staff; proper collection and testing of representative samples; monitoring and maintenance of equipment; and proper data management and reporting.
  • Ensuring appropriate design and implementation of statistically representative and replicable baseline and/or endline kitchen source and blood lead level data, including obtaining Institutional Review Board approval." Pure Earth, Project proposal, June 8, 2021, Pg. 10.

• 81
  • "Those subsamples drawn from general populations or used as controls in case-control studies were coded to the background subgroup and were assumed to be representative of general exposure." Ericson et al. 2021, Pg. e147.
  • "Our search yielded 12 695 studies, of which 520 were eligible for inclusion (1100 sampled populations from 49 countries comprising 1 003 455 individuals). Pooled mean blood lead concentrations in children ranged from 1·66 μg/dL (SD 3·31) in Ethiopia to 9·30 μg/dL (11·73) in Palestine, and in adults from 0·39 μg/dL (1·25) in Sudan to 11·36 μg/dL (5·20) in Pakistan. Background values for blood lead level in children could be pooled in 34 countries and were used to estimate background distributions for 1·30 billion of them. 632 million children (95% CI 394 million–780 million; 48·5%) were estimated to have a blood lead level exceeding the US Centers for Disease Control's reference value of 5 μg/dL. Major sources of lead exposure were informal lead acid battery recycling and manufacture, metal mining and processing, electronic waste, and the use of lead as a food adulterant, primarily in spices." Ericson et al. 2021, Abstract.
    • See the full range of average blood lead concentrations in countries where background (i.e., nationally representative) data was available in Table 3, Pg. e149 of Ericson et al. 2021.

• 82

  See this spreadsheet, sheet "Pure earth BOTEC (more detailed)," row "Probability of decreasing average BLLs by <0.5 ug/dl."

• 83

  See more details on the evidence base for the contribution of spices to high blood lead levels in the section above.

• 84

  See more details on the evidence base for the presence of lead in spices in the section above.

• 85
  • See this spreadsheet, sheet "Pure earth BOTEC (more detailed)," section "Probability of different levels of impact of intervention in covered population."
  • "How much this intervention [to reduce contamination of spices with lead in Bangladesh] could contribute to a reduction in blood lead levels is uncertain. Ms. Forsyth has modeled a reduction of roughly 2 micrograms/deciliter (µg/dl) from the intervention and guesses the real effect is between 1 and 3 µg/dl." GiveWell's non-verbatim summary of a conversation with Jenna Forsyth, May 28, 2021, Pg. 3.

• 86

  See this spreadsheet, sheet "Pure earth BOTEC (more detailed)," row "Expected proportion of population covered."

• 87

  "Pure Earth does not consider its project in Bangladesh as regional because the intervention’s effects were likely nationwide." GiveWell's non-verbatim summary of a conversation with Pure Earth, May 26, 2021, Pg. 2.

• 88

  See this spreadsheet, sheet "Pure earth BOTEC (more detailed)," row "Expected reduction in country average BLL (ug/dl)."

• 89

  See this spreadsheet, sheet "Pure earth BOTEC (more detailed)," row "Number of years benefits are counterfactually sustained for (assuming additional funding)."

• 90

  For a grant to the Centre for Pesticide Suicide Prevention (CPSP), we used a best guess of nine years for policy changes we believe CPSP contributed to (adjusting down from a 10 year "default"). We used a more pessimistic estimate of seven years for this grant.

  The most important reason is that it seems more plausible to us that behavior like adding lead to spices would return to pre-intervention levels than would the use of newly-banned pesticides. Adding lead to spices is already illegal in Bangladesh and India, suggesting lack of enforcement is the key barrier rather than lack of regulation. Our best guess is that enforcement is more likely to revert to pre-intervention levels than new regulation (which CPSP is pursuing). We include a separate adjustment for CPSP for the chance that any regulation is not enforced. For our grant to Pure Earth, we apply a subjective downward adjustment to the total years of benefit for this consideration, from a ten year default to seven years.

  • "FSSAI's regulations on contaminants, toxins, and residues include specific restrictions on the level of lead and other toxic metals permitted in different food items. As per section 20 of the FSS Act 2006, all food businesses have to adhere to the limits provided in the Food Safety and Standards (Contaminants, Toxins, and Residues) Regulations, 2011." GiveWell's non-verbatim summary of a conversation with Misbah Wasi, July 2, 2021, Pg. 1.
  • "It is illegal to adulterate spices with lead in India, and the government is sometimes able to identify and take legal action against producers of adulterated spices, although broad enforcement is difficult." GiveWell's non-verbatim summary of a conversation with Dr. Kamalesh Sarkar, June 28, 2021, Pg. 2.

  A factor pushing against a more pessimistic adjustment is that the earliest study we could find suggested that turmeric in Bangladesh may have been contaminated with lead as early as 1987, meaning the problem may have persisted for more than 30 years. We have not reviewed this study in depth but take it as a data point suggesting that it is unlikely that this source of exposure would be addressed without external intervention.

  • "Table-1 shows the analysis for lead in pure samples of curcuma longa bulbs (turmeric) obtained from N.W.F.P. area of Pakistan and imported sample from Bangla–Desh. In order to find out the distribution of lead with in the roots, the inside matrix and the cuticle were separately analysed by atomic absorption spectrophotometry. The matrix showed the presence of 1.2 ppm of lead where as the cuticle contained 2.8 ppm in the turmeric from N.W.F.P. The samples from Bangla-Desh gave almost similar results (Table-1)." Syed, Qadiruddin, and Khan 1987, Pg. 388.

• 91
  • "Key sources of lead exposure in India are PVC pipes (leaching lead into water), informal battery recycling, adulteration of spices (e.g., the addition of lead chromate to turmeric to improve color), and lead paint. Although existing regulations prohibit lead contamination, they have not been well-enforced [...] Public awareness of the harms of lead exposure in India is low, although there have been high-profile incidents of lead contamination that raised awareness (e.g., media stories in 2015 on the presence of lead in a popular brand of instant noodles)." GiveWell's non-verbatim summary of a conversation with Misbah Wasi, July 2, 2021, Pg. 1.
  • "Public awareness of the harms of lead is low in India [...] It is illegal to adulterate spices with lead in India, and the government is sometimes able to identify and take legal action against producers of adulterated spices, although broad enforcement is difficult." GiveWell's non-verbatim summary of a conversation with Dr. Kamalesh Sarkar, June 28, 2021, Pg. 2.

• 92

  See this spreadsheet, sheet "Pure earth BOTEC (more detailed)," section "Absolute population covered."