Salt Iodization | GiveWell

# In a nutshell

This page discusses salt iodization, or the fortification of salt with iodine, to prevent iodine deficiency disorders. The thyroid, an organ in the neck that produces hormones, requires the element iodine to function properly. Some diets contain insufficient amounts of iodine.

Several small-scale randomized controlled trials (RCTs) have examined the effect of iodine supplementation of children on their mental function. Each study uses a different suite of mental tests, which makes it difficult to interpret the clinical significance of treatment effects. The studies also find inconsistent results and we do not have enough information to conduct a formal meta-analysis. Considering all the evidence available, our best guess is that iodine supplementation improves the mental function of deficient children with greater improvements the more severe the deficiency.

Negative effects of salt iodization may arise due to rapid increases in iodine intake, chronic iodine excess, or a sensitivity to higher levels of iodine intake in certain subpopulations. The number of serious cases of iodine-induced thyroid dysfunction seems very low compared to the population with impaired mental function, but some programs, especially those that rapidly increase iodine intakes in older individuals with severe deficiency, may result in very slightly increased mortality.

We believe there is a reasonably strong case that countrywide salt iodization efforts have successfully reduced iodine deficiency, though the evidence has some weaknesses. Many observational studies show increases in urinary iodine and reductions in the prevalence of goiter after salt iodization, though these studies may not be nationally representative. Nationally representative surveys generally show a high proportion of households with iodized salt, but they do not reveal whether salt contains a sufficient amount of iodine or how much salt households actually consume, and we have remaining questions about the significance of these measurements.

The cost-effectiveness of a salt iodization program depends on salt intake, the iodine content in salt once it reaches consumers (poor storage, for instance, can result in iodine loss), and the prevalence of iodine deficiency before implementation of the program. We have limited data on the costs of iodization, but one source estimates $0.02-$0.10 per person per year. Salt iodization appears to be within the range of cost-effectiveness of our priority programs.

Published: December 2014

## What is iodine deficiency?

The thyroid is an organ in the neck that makes hormones that control the rate of many activities in the body, including growth and development, the metabolic rate, and how fast the heart beats.1 The hormones produced by the thyroid contain the element iodine.2 Food grown in different regions contains different amounts of iodine.3 Iodine deficiency disorders arise when the body does not get a sufficient amount of iodine for the thyroid to properly function.4 For example, iodine deficiency can cause goiter, an enlargement of the thyroid gland that can result from an underproduction of thyroid hormones (i.e. hypothyroidism).5

The severity of iodine deficiency in a population is determined primarily through school-based surveys of urinary iodine concentration (UIC).6

## What is the program?

Salt iodization, or the fortification of salt with iodine by salt manufacturers, is a major approach in the prevention of iodine deficiency.7 Other iodized foods, iodine tablets, and iodized oils (given orally or by injection) can also be used to prevent iodine deficiency, but salt is useful because it is more likely to be widely consumed in consistent amounts throughout the year, its production is usually limited to a few facilities that can be monitored, its taste and smell are not affected by iodization, and it is cheap.8

## What are the benefits of salt iodization?

Correcting iodine deficiency may have a wide range of benefits.9 In this report, we focus on improved cognition, which we perceive to be the most important benefit of salt iodization programs. In particular, we discuss the two cognitive benefits supported by RCTs or quasi-RCTs of iodine supplementation:10

• Improving mental function in iodine deficient children (including mildly deficient children)
• Preventing cretinism (a condition of profoundly stunted physical and mental development) in neonates born to women with chronic, severe deficiency11

We would guess that iodine supplementation of mild-to-moderately deficient children increases their IQ by somewhere around 4 points on average.12 We place more weight on the most recent studies - Gordon et al 2009 and Zimmermann et al 2006 - that appear to be of similar or higher quality than previous studies and that substantially increased iodine intakes in the treatment group without increasing iodine intakes in the control.13

We would also guess that the more severe the deficiency, the greater the impairment, so severely deficient children might see larger improvements in IQ. We do not have high confidence in these conclusions, because of the inconsistent results across studies and the challenges of interpreting the variety of mental tests used.

### Improving mental function

Two meta-analyses (Angermayr and Clar 2004 and Taylor et al 2013) identify 6 RCTs evaluating the effect of iodine supplementation of children on their mental function.14

Study characteristics:

We provide a summary of the results from the studies in the table below. For more detail on the specific tests used and the data available from the studies, see this spreadsheet.

Table 1: Summary of results from 6 RCTs evaluating the effect of iodine on children’s mental function26

Study Control change in median UIC (urinary iodine concentration) (µg/L) Treatment change in median UIC (µg/L) Follow-up (months) Treatment effect (Units of standard deviation)
Gordon et al 2009 62 to 8127 66 to 14528 6.529 0.19 (95% CI: 0.04-0.34)30
Zimmermann et al 2006 44 to 4931 42 to 17232 5.533 0.2934
Shrestha et al 1994 19 to "No change" (11 months)35 19 to 53 (3 months); 19 to 23 (11 months)36 1037 0.8838
Huda, Grantham-McGregor and Tomkins 2001 36 to 42 (geometric mean)39 35 to 87 (geometric mean)40 441 0.0142
Untoro et al 1999 48 to 8943 46 to 18544 11.545 Not enough information to calculate46
Bautista et al 1982 15 to 32 (mean)47 14 to 54 (mean)48 2249 Not enough information to calculate50

We note some potential limitations of these studies:

• Increase in iodine intakes in the control group. Urinary iodine substantially increased in the control group in Gordon et al 2009, Bautista et al 1982 and Untoro et al 1999 perhaps due to increasing availability of iodized salt in the study areas though it is not clear.51 We would expect an increase in iodine intakes in the control group to attenuate the effect of iodine supplementation on mental function.
• Insufficient iodine supplementation. The treatment group's iodine status improved only temporarily in Shrestha et al 1994, which does not appear to provide an explanation, and remained mildly iodine deficient in Huda, Grantham-McGregor and Tomkins 2001 based on the WHO's classification, perhaps because of the efficacy of the iodine preparation used.52
• Publication bias. Our general concern about the possibility of publication bias is heightened with respect to these studies because of the number of cognitive tests available to researchers to choose from and because finding an effect on IQ may be a particularly publishable result.53
• Fade-out. We are not sure whether the positive effects found in the trials with shorter follow-ups would persist over a longer time horizon.54 It appears that the most prominent discussion of fade-out comes from several long-term evaluations of early childhood education programs, which found that IQ increases fade out over time, though these findings are the subject of dispute.55 We are uncertain to what extent this literature generalizes to iodine supplementation programs. Even if IQ increases fade out, other related benefits may still persist.56

Ideally, we would like to combine the results of the trials in a meta-analysis, but many of the studies do not report the information necessary to easily derive standard errors for the overall treatment effects.57 Taking an unweighted average of the 4 studies for which we have a standardized treatment effect yields an overall treatment effect of 0.34 standard deviations.58 Taking the average of the treatment effects in Gordon et al 2009 and Zimmermann et al 2006, which appear to be of similar quality or higher than previous studies and that substantially increased iodine intakes in the treatment group without increasing iodine intakes in the control group, yields a standardized treatment effect of 0.24 standard deviations.59 This standardized treatment effect equates to roughly 4 IQ points.60

#### IQ and life outcomes

We have not thoroughly examined the extensive literature on the significance and predictive importance of IQ. We have spoken with Dr. Jelte Wicherts, an associate professor in the Department of Methodology and Statistics at Tilburg University, and Dr. James Flynn, Emeritus Professor of Political Studies and Psychology at the University of Otago, and tentatively taken away the following points:

• IQ is defined by performance on a battery of cognitive tests.61 These cognitive tests tend to correlate with one another, so the battery of tests used can vary.62 The tests used in the RCTs appear to be the types of test included in the traditional battery of tests for IQ or at least the type of tests that correlate with those tests.63
• Many studies have explored the correlation between IQ and life outcomes, particularly in education and in job performance.64 The stability of IQ throughout one's life, the wide array of outcomes for which it appears predictive, and the fact that IQ generally has predictive power even when trying to adjust for a variety of potential confounders suggest (though don't necessarily conclusively establish) a causal role for the sort of cognitive ability measured by IQ tests.65
• From our conversation with Dr. Flynn: "The impact of increases in IQ on quality of life might vary in different countries depending on the opportunities available. However, because IQ predicts a wide variety of positive outcomes (including job performance on a wide variety of jobs), it seems likely that an increase in IQ would be valuable."66
• If we define "intellectual disability" as having an IQ under 70, a 4 point increase in IQ in a population with a mean of 100 and standard deviation of 15 (normally distributed) would decrease the prevalence of intellectual disability from 2.3% to 1.2%.67 However, the change in prevalence would be much greater for a population with lower average starting IQ; for example, for a population with an average IQ of 80 (which limited evidence suggests may be common in sub-Saharan Africa), a 4-point increase in mean IQ would decrease the prevalence of intellectual disability from ~25% to ~18%.68

### Preventing cretinism

A severe potential consequence of iodine deficiency is cretinism, a condition characterized by profound intellectual disability, stunted physical growth, deafness, the inability to speak, and muscle spasticity.69 Cretinism now seems to be extremely rare;70 we discuss cretinism here, because there may still be a small number of cases of cretinism in some severely deficient areas and because the consequences of severe deficiency may be relevant to understanding the consequences of mild-to-moderate deficiency.

Mahomed and Gulmezoglu 1997, a Cochrane review, and Zhou et al 2013, a more recent, non-Cochrane review with a systematic search and well-defined inclusion criteria, identify one quasi-randomized controlled trial, Pharoah, Buttfield and Hetzel 1971, in an area with severe iodine deficiency and high levels of cretinism, that found a large reduction in cretinism in children born to women receiving iodine supplementation compared to the control group.71 Following up with the families four years later, the authors found 7 cases of cretinism in the 412 children born to mothers receiving iodine compared to 26 cases out of 406 children in the control group, giving a relative risk of 0.27 (95% CI: 0.12 to 0.60).72 The next follow-up study of children born between 1966 and 1972 reported 3 cases of cretinism out of 274 children in the treatment group and 16 cases out of 248 in the control, giving a relative risk of 0.17 (95% CI: 0.05 to 0.58).73

However, the study had some methodological weaknesses. The study was quasi-randomized; alternating families were injected with iodized oil and a placebo saline solution, presumably in the order of households visited along a walking patrol of the villages though it is not clear,74 and follow-ups were limited to a subset of the original included villages.75

## What are the negative effects of salt iodization?

In some cases, there may be substantial negative effects of implementing a salt iodization program:

• Rapid increases in iodine intake in older people with nodular goiter and chronic deficiency causing iodine-induced hyperthyroidism (IIH) and death. Rapid increases in iodine intake in iodine deficient individuals may cause hyperthyroidism, a condition in which the thyroid overproduces hormones.76 Hospitals in several countries have observed increases in admissions for hyperthyroidism after the introduction of salt iodization programs.77 Most cases seem to occur in older people with nodular goiter and chronic iodine deficiency.78 Stanbury et al 1998, the narrative review cited by WHO Assessment of Iodine Deficiency Disorders and Monitoring their Elimination 2007 in addressing the risks of salt iodization, considers the experience of Tasmania starting in the late 1960s as "by far the best documented epidemic of IIH."79 Two of the state's principal hospitals found that the incidence rate of hyperthyroidism in patients 40 years and older increased from a baseline of around 20 per 100,000 per year to a peak of over 120 per 100,000 per year after the introduction of an iodine deficiency control program and then gradually returned to baseline levels about 15 to 20 years later; the incidence rate for patients under 40 did not appear to increase.80 49 deaths had IIH listed as one of the causes of death in a population of about 370,000 over a period of 21 years after the increase in iodine intake, implying a mortality rate of 0.63 per 100,000 per year.81 Hospitals and death registration systems may capture only a fraction of cases though. Stanbury et al 1998 notes that the more severe and longstanding the deficiency and the greater and more rapid the increase in iodine intake, the greater the increase in IIH.82 Over the long term, full correction of iodine deficiency may lead to a decreased incidence of IIH in a population.83
• Chronic iodine excess in children causing cognitive impairment. Salt iodization programs may result in excessive iodine intake, which could plausibly cause impairments in mental function analogous to those caused by iodine deficiency.84 We have not found evidence documenting this effect, but Zimmermann et al 2005 has reported an association between excessive iodine intake and goiter in children.85 Most countries do not seem to have levels of iodine intake that would cause negative effects, but this threshold is also difficult to determine.86
• Standard levels of iodine causing negative effects in highly sensitive subpopulations. Salt iodization programs may cause negative effects in certain subpopulations with a sensitivity to higher levels of iodine intake, for instance in people with autoimmune thyroid diseases, even at levels of intake generally considered safe.87

## What is the track record of large-scale salt iodization?

The first salt iodization programs began in the early 1920s in Switzerland and the United States in order to combat endemic goiter.88 In 1991, the World Health Assembly resolved to eliminate iodine deficiency as a public health problem globally.89 Since then, many countries have introduced salt iodization programs.90 As of 2012, the International Council for the Control of Iodine Deficiency Disorders (ICCIDD) classified 23 countries in the world as mildly iodine deficient, 9 as moderately deficient and none as severely iodine deficient.91

The primary indicator of impact for salt iodization programs has shifted from goiter to urinary iodine, because palpation (the traditional method of assessing goiter) does not perform well in areas of mild-to-moderate iodine deficiency and because urinary iodine responds much more rapidly than goiter to changes in iodine intake.92 The WHO still recommends goiter as an indicator, but only for monitoring the long-term impact of programs and for assessing the baseline level of iodine deficiency.93 We discuss the evidence on these outcomes below. We also examine evidence from several national household surveys that measure the iodine content in table salt in households.94

### Urinary iodine

Abudou et al 2014, a systematic review, identifies 38 studies evaluating the impact of salt iodization programs on urinary iodine.95 Nearly all of them showed increases in urinary iodine after salt iodization.96

We haven't thoroughly vetted all of these studies. Instead, we examined studies with the following criteria:

• Reported median UIC (µg/L), the indicator recommended by the WHO97
• Evaluated programs that did not directly provide iodized salt to households (i.e., those that evaluated large-scale fortification programs of the type we understand international aid agencies support)
• In areas where the baseline median UIC indicated inadequate iodine intakes98
• Published in English

We identified 4 studies that met the above criteria:99

• Zimmermann et al 2003: In 1997, Cote d'Ivoire passed legislation mandating universal salt iodization.100 Zimmermann et al 2003 conducted a survey in 1997 in 2 schools in different villages in the Danane district of Cote d'Ivoire before iodized salt became available.101 Over the course of a month, the researchers measured urinary iodine in all children aged 5 to 14 in attendance at each school.102 They then returned to the same schools again in 1998 after iodized salt became available.103 They added 4 schools in 4 other villages to the sample in 1999 and measured urinary iodine in the 6 schools annually until 2001.104 They found that the median UIC increased from 28 µg/L in 1997 to 86 µg/L in 1998 to 161 µg/L in 1999 and stayed above 100 µg/L for the remainder of the years.105
• Jooste, Weight and Lombard 2000: In 1995, South Africa passed legislation mandating iodization of household salt.106 The study conducted a survey in primary schools in 4 communities, where endemic goiter had been studied in the past, in 1995 before the implementation of USI and in 1996 afterwards.107 The study found the overall median UIC increased from 22 µg/L to 186 µg/L.108
• Baczyk et al 2007: Poland introduced mandatory salt iodization in 1996.109 Baczyk et al 2007 sampled around 400 children in 1992, 2000 and 2005.110 It reported that median UIC increased from 49 µg/L to 104 µg/L to 107 µg/L.111 It does not provide much detail on the methodology of the surveys.
• Tazhibayev et al 2008: In 2001, the Commonwealth of Independent States established a uniform standard of salt iodization in a high-level policy agreement.112 Sentinel sites in several countries in Central Asia were selected in order to track the impact of the program.113 In each sentinel site, 40 households were randomly sampled from a clinical registry. Median UIC was measured in the sentinel sites between 2002 and 2003 before the implementation of salt iodization and in the same households again in 2007.114 Azerbaijan, Kazakhstan and Uzbekistan all had median UICs indicating adequate iodine intakes before salt iodization began.115 In Mongolia, the median UIC increased from 69 µg/L to 119 µg/L and in Tajikistan from 29 µg/L to 120 µg/L.116

Some key potential biases of these studies:117

• Representativeness of study sites: None of the studies appear to have taken nationally representative samples.118 The severity of deficiency and the effect of the program may differ in different parts of the country.119 In some cases, study sites may get additional attention, because people know that they are the focus of evaluation.120
• Changes in iodine intake unrelated to the program: None of these studies had control groups.121 In general, we'd attribute a large change in urinary iodine shortly after salt iodization to the program, but changes in diet unrelated to the program could account for some of the difference in studies with longer follow-ups, such as Baczyk et al 2007.122

Though nearly all the studies on urinary iodine have shown increases, some salt iodization programs seem not to have had their intended effect.123 For instance, Ghana passed legislation for salt iodization in 1996 after surveys in the early 1990s showed a national goiter prevalence of about 24%.124 In 2007, a district-level survey found a median UIC of 33 µg/L indicating moderate iodine deficiency based on the WHO's classification.125 It seems that smaller producers supply inadequately iodized salt in some areas.126

#### How reliable and predictive is measuring urinary iodine content?

The median UIC reflects very recent iodine intake, because most consumed iodine appears in the urine within a day or two; it does not directly measure the iodine status of the population.127 The WHO relates a median UIC to iodine status largely based on Ascoli and Arroyave 1970, which randomly sampled approximately 20 families to survey from each of 186 localities in Central America, finding higher goiter rates in localities with a lower mean estimated urinary iodine excretion (UIE), or the µg of iodine excreted in urine per day.128 The WHO set a median UIC of 100 µg/L (a healthy child produces about 1 liter of urine per day on average) as the threshold for adequate iodine nutrition, because Ascoli and Arroyave 1970 found that most localities with a mean UIE of more than 100 µg/day had goiter rates below 10%.129 Some small-scale RCTs of iodine supplementation in children have also shown increases in cognition associated with increases in the median UIC.130 Because urine volume and iodine intake vary considerably day-to-day, UIC surveys do not produce reliable estimates of individual intakes or iodine status.131

It also surprises us that New Zealand has a median UIC of 66 µg/L, the same as Sudan, and the United Kingdom, Finland, and Ireland have median UICs that indicate mild iodine deficiency based on the WHO's classification; we wouldn't rule out the possibility of iodine deficiency in these countries, but it also heightens our concerns about the reliability of this indicator.132

### Goiter

Abudou et al 2014 identifies 54 studies evaluating the effect of salt iodization programs on goiter.133 Nearly all of the studies report reductions in goiter prevalence after salt iodization.134 28 of the studies took place in China.135 20 of the 26 studies outside of China evaluate programs broadly similar to those that ICCIDD promotes and assists.136 Of those 20 studies, 11 took place in developing countries, of which 2 took place in Africa since the year 2000.137

The studies varied in their sampling methods and measurement techniques.138 Most of them did not have a control group, and goiter rates depend on a number of factors unrelated to the program that can change over time (e.g. goitrogens in the diet).139 Other potential biases for some studies include non-representative sampling, a lack of blinding of outcome assessment, and large and unexplained differences in sample size between baseline and follow-up.140

Keeping these caveats in mind, we take this as reasonably strong evidence that large-scale iodization programs have successfully reduced goiter rates.

### Household surveys of iodized salt

Several large household surveys include a module testing the iodine content of table salt used in the household.141 As of 2012, 52 countries have a survey measuring household iodized salt coverage.142 The median coverage was 82% (IQR: 55% to 95%).143 We are not sure how to interpret these figures. For example, data shows that 80.2% of households in Angola have access to iodized salt, but Michael Zimmerman of ICCIDD has told us that Angola's iodization efforts require significant attention.144

A number of these surveys measure iodine content in household salt using rapid testing kits.145 But rapid testing kits (RTKs) may not reliably assess iodine content.146 WHO Assessment of Iodine Deficiency Disorders and Monitoring their Elimination 2007 contends that RTKs can reliably differentiate between iodized and non-iodized salt.147 We have not yet vetted that claim.148 In either case, these surveys do not capture iodine intake from other sources, such as processed foods and dairy products.149

We have very little information about the costs of salt iodization. Mannar and Dunn 1994 estimates that the cost of setting up and operating an iodization plant ranges from $0.02 to$0.10 per person per year.150 The manual first estimates the cost per ton of iodized salt produced based on "costs of materials and services in several Asian and African countries."151 It then assumes that the plant produces enough iodized salt to serve a population consuming 3-5 kg of salt per person per year to arrive at a cost per person per year.152 A program that works with a large number of smaller salt producers probably costs more, because "capital costs are spread over a smaller volume."153 We also got the views of Venkatesh Mannar, the former president of the Micronutrient Initiative, who estimated the cost at $0.06 per person per year.154 We assume that a high-quality salt iodization program that increases the median UIC of a population from one that indicates mild-to-moderate deficiency to one that indicates adequate iodine intakes will produce an average increase in IQ in children in that population of around 4 IQ points.155 But salt iodization programs may fail to increase iodine intakes for a variety of reasons.156 Assuming a treatment effect along the lines of the RCTs for children in households with salt that contains iodine, neglible negative effects and a neglible prevalence of cretinism, we find a cost-effectiveness in line with our priority programs. We combine these in a cost-effectiveness model using inputs provided by GiveWell staff members. We estimate that salt iodization programs are in the same range of cost-effectiveness as our other priority programs. See this spreadsheet for details. ## Our process For assessing the evidence on cretinism, we rely on Mahomed and Gulmezoglu 1997 and Zhou et al 2013. For assessing the evidence on mental function, we rely on Angermayr and Clar 2004 and Taylor et al 2013. We also had a conversation with Dr. Jelte Wicherts and a conversation with Dr. James Flynn about cognitive testing and IQ. See What are the benefits of salt iodization? for a fuller discussion of these studies. The section on the track record of salt iodization relies on Abudou et al 2014. We rely on Achieving Universal Salt Iodization and WHO Assessment of Iodine Deficiency Disorders and Monitoring their Elimination 2007 for their discussions of national salt iodization programs, and Zimmermann and Andersson 2012 and Andersson, Karumbunathan and Zimmermann 2012 for their discussions of iodine deficiency surveillance. We've also had several conversations with Michael Zimmermann, Executive Director of ICCIDD and Head of the Human Nutrition Laboratory, Institute of Food, Nutrition and Health, Department of Health Sciences and Technology, ETH Zurich, Switzerland, and Jonathan Gorstein, Clinical Associate Professor of Global Health at the University of Washington and Senior Advisor at ICCIDD.157 ## Questions for further investigation We discuss what we might like to learn more about from the existing literature on iodine deficiency. • What is the evidence for the effect of salt iodization on other health outcomes related to cognition? Children born to iodine deficient women may also suffer from impaired mental function even if they do not develop cretinism, but neither of the 2 quasi-RCTs identified by Zhou et al 2013 reporting on mental function found an effect on intelligence.158 We've also found little evidence for the effect of iodine supplementation of adults on their cognition or for the effect of iodine supplementation of lactating women on their infant's cognition (soon after supplementation or later in life). Zimmermann 2009, a narrative review, does not mention any randomized controlled trials (RCTs) on either of these populations.159 We could still spend more time investigating the evidence on these other potential cognitive benefits. • What is the evidence for the effect of salt iodization on other health outcomes besides cognition, such as infant mortality? • Is the relationship between iodine deficiency and health outcomes better understood as a threshold effect or a continuum of risk?160 • Does the connection between iodine deficiency and goiter and hypothyroidism strengthen the evidence for the connection between mild-to-moderate iodine deficiency and cognitive development? • Does the connection between severe iodine deficiency and cretinism strengthen the evidence for the connection between mild-to-moderate iodine deficiency and cognitive development? We may want to spend more time investigating the biological mechanism leading to cretinism and the postulated mechanism for general improvements in cognitive development. • To what extent are cognitive disabilities (both cretinism and sub-clinical cognitive impairment) caused by in utero iodine deficiency reversible later in life? We currently believe that cretinism is irreversible but that sub-clinical cognitive impairment is partially or fully reversible. • What is the role of other indicators of iodine deficiency, such as TSH, T3 and T4 levels?161 • Does economic development have less of an impact on iodine deficiency than other micronutrient deficiences?162 • If salt iodization is inconsistent over time in a country, causing people's iodine levels to fluctuate, could this cause temporary increases in hyperthyroidism each time iodine levels go back up?163 • How does the number of years lived with iodine deficiency and the age at which iodine deficiency is experienced relate to impairments in mental function? • To what extent do advocates of salt iodization promote increases in salt intake that might increase the risk of hypertension?164 ## Sources  Abudou et al 2014 Forthcoming Achieving Universal Salt Iodization Source (archive) Andersson et al 2005 Source (archive) Andersson et al 2007 Source (archive) Andersson, Karumbunathan and Zimmermann 2012 - Supplemental Table 1 Source (archive) Andersson, Karumbunathan and Zimmermann 2012 Source (archive) Angermayr and Clar 2004 Source (archive) Ascoli and Arroyave 1970 Source (archive) Baczyk et al 2007 Source (archive) Barnett and Hustedt 2005 Source (archive) Bautista et al 1982 Source (archive) Benoist et al 2007 Source (archive) Bhutta et al 2008 Source (archive) Bimenya et al 2002 Source (archive) Blinding, Cochrane Glossary Source (archive) Bourdoux et al 1996 Source (archive) Burgi, Kohler and Morselli 1998 Source (archive) Cerqueira et al 2009 Source (archive) Combining groups, Cochrane Handbook for Systematic Reviews of Interventions 2011 Source (archive) Formulae for combining groups, Cochrane Handbook for Systematic Reviews of Interventions 2011 Source (archive) The standardized mean difference, Cochrane Handbook for Systematic Reviews of Interventions 2011 Source (archive) Childinfo Iodized salt consumption 2012 Source (archive) Cobra et al 1997 Source (archive) Costs of salt iodization, estimates from Venkatesh Mannar, September 2, 2014 Source CPI Inflation Calculator Source (archive) Dary 2011 Source (archive) Delange 1998 Source (archive) DeLong et al 1997 Source (archive) Dietary Reference Intakes for Vitamin A, Vitamin K, Arsenic, Boron, Chromium, Copper, Iodine, Iron, Manganese, Molybdenum, Nickel, Silicon, Vanadium, and Zinc Source (archive) Dr. Jelte Wicherts, Curriculum Vitae Source (archive) Dr. Sheila Skeaff, Associate Professor, Department of Nutrition, University of Otago, email to GiveWell, November 14, 2014 Unpublished Dr. Michael Zimmermann, Executive Director of ICCIDD, email to GiveWell, November 20, 2014 Unpublished Dunn, Semigran and Delange 1998 Source (archive) Eltom et al 1985 Source (archive) Ethiopia, WHO Global Database on Iodine Deficiency Source (archive) Field, Robles and Torero 2009 Source (archive) Furnee et al 1994 Source (archive) Garber et al 2012 Source (archive) GiveWell's non-verbatim summary of a conversation with James Flynn, Emeritus Professor of Political Studies and Psychology, University of Otago, New Zealand, October 17, 2014 Source GiveWell's non-verbatim summary of a conversation with Jelte Wicherts, Associate Professor, Department of Methodology and Statistics, Tilburg University, August 28, 2014 Source GiveWell’s non-verbatim summary of a conversation with Michael Zimmermann, International Council for the Control of Iodine Deficiency Disorders, May 9 2012 Source Goiter Symptoms, Mayo Clinic Source (archive) Golkowski et al 2003 Source (archive) Golkowski et al 2007 Source (archive) Gordon et al 2009 Source (archive) Heinisch et al 2002 Source (archive) Hintze et al 1988 Source (archive) Horton 2006 Source (archive) Horton, Mannar and Wesley 2008 Source (archive) Huda, Grantham-McGregor and Tomkins 2001 Source (archive) Hyperthyroidism, MedlinePlus Source (archive) Hypothyroidism, MedlinePlus Source (archive) Hyperthyroidism Complications, Mayo Clinic Source (archive) Hyperthyroidism Symptoms, Mayo Clinic Source (archive) Hypothyroidism Symptoms, Mayo Clinic Source (archive) Hypothyroidism, PubMed Health Source (archive) Ibrahim, Sinn and McGuire 2006 Source (archive) ICCIDD conversations with GiveWell in Zurich, April/May 2014 Unpublished ICCIDD Global Efforts Source (archive) ICCIDD Our Leadership Source (archive) ICCIDD Our Mandate Source (archive) ICCIDD Newsletter Feb 2013 Source (archive) ICCIDD Protecting Children Source (archive) ICCIDD Scorecard Source (archive) ICCIDD Where is the problem? Source (archive) Iodine Deficiency Disorders, UpToDate Source (archive) Iodine, FAO Human mineral and vitamin requirements Source (archive) Isa et al 2000 Source (archive) Jooste, Weight and Lombard 2000 Source (archive) Kassim et al 2014 Source (archive) Kevany et al 1969 Source (archive) Kohn 1976 Source (archive) Kotwal, Priya and Qadeer 2007 Source (archive) Leung, Braverman and Pearce 2012 Source (archive) Laurberg et al 2010 Source (archive) Lim et al 2012 Source (archive) Lv et al 2009 Source (archive) Mahomed and Gulmezoglu 1997 Source (archive) Malone et al 1996 Source (archive) Mannar and Dunn 1994 Source (archive) MathWorks Coefficient Standard Errors and Confidence Intervals Source (archive) Mattsson and Lindstrom 1995 Source (archive) Neisser 2011 Source (archive) Nyumuah et al 2012 Source (archive) Palpation, MedlinePlus Source (archive) Pandav et al 2000 Source (archive) Pearce 2006 Source (archive) Pharoah and Connolly 1987 Source (archive) Pharoah and Connolly 1991 Source (archive) Pharoah, Buttfield and Hetzel 1971 Source (archive) Powles et al 2013 Source (archive) Pretell et al 1974 Source (archive) Rastogi and Mathers 2000 Source (archive) Recommended iodine levels in salt and guidelines for monitoring their adequacy and effectiveness Source (archive) Salt as a Vehicle for Fortification Source (archive) Salt Iodization, WHO Source (archive) Santiago et al 2013 Source (archive) Schweinhart et al 2005 Source (archive) Shrestha et al 1994 Source (archive) SI Unit Conversion Calculator, Society for Biomedical Diabetes Research Source (archive) Sooch and Ramalingaswami 1965 Source (archive) Speeckaert et al 2011 Source (archive) Stanbury et al 1998 Source (archive) Taylor et al 2013 Source (archive) Tazhibayev et al 2008 Source (archive) The High/Scope Perry Preschool Study Through Age 40 Source (archive) Thilly et al 1994 Source (archive) Thyroid Disease Manager Source (archive) Thyrotoxicosis, Virginia Mason Source (archive) Todd et al 1995 Source (archive) Todd and Dunn 1998 Source (archive) Toxic nodular goiter, MedlinePlus Source (archive) Untoro et al 1999 Source (archive) Vanderpump 2011 Source (archive) Vitamin and Mineral Requirements in Human Nutrition Source (archive) Wang et al 2009 Source (archive) Wang 2012 Source (archive) Wechsler Intelligence Scale for Children Source (archive) WHO Assessment of Iodine Deficiency Disorders and Monitoring their Elimination 2007 Source (archive) WHO Global Database on Iodine Deficiency Source (archive) WHO Global Database on Iodine Deficiency, Ghana, 2006 Source (archive) WHO Iodine status worldwide 2004 Source (archive) Wicherts et al 2010 Source (archive) Sum of correlated variables, Wikipedia Source (archive) Wong et al 2011 Source (archive) Zhao et al 1999 Source (archive) Zhao et al 2013 Source (archive) Zhou et al 2013 Supplemental Data 1 Source (archive) Zhou et al 2013 Supplemental Data 2 Source (archive) Zhou et al 2013 Supplemental Data 3 Source (archive) Zimmermann et al 2004 Source (archive) Zimmermann 2009 Source (archive) Zimmermann and Andersson 2012 Source (archive) Zimmermann et al 2006 Source (archive) Zimmermann et al 2005 Source (archive) Zimmermann et al 2003 Source (archive) Zhou et al 2013 Source (archive) • 1. "Your thyroid is a butterfly-shaped gland in your neck, just above your collarbone. It is one of your endocrine glands, which make hormones. Thyroid hormones control the rate of many activities in your body. These include how fast you burn calories and how fast your heart beats. All of these activities are your body's metabolism." Hypothyroidism, MedlinePlus • 2. "IODINE (atomic mass, 126.9 amu) is an essential component of the hormones produced by the thyroid gland." Zimmermann 2009, Pg. 377 • 3. • "Iodine-deficient soils are common in mountainous areas (e.g., the Alps, Andes, Atlas, and Himalayan ranges) and areas of frequent flooding, especially in South and Southeast Asia (for example, the Ganges River plain of northeastern India). Although many inland areas, including central Asia and Africa and central and eastern Europe are iodine deficient, iodine deficiency may also affect coastal and island populations. Iodine deficiency in populations residing in these areas will persist until iodine enters the food chain through addition of iodine to foods (e.g., iodization of salt) or dietary diversification introduces foods produced outside the iodine-deficient area." Zimmermann 2009, Pg. 377 • Iodine intake may not always be deficient without salt iodization. It may even be excessive: "The median UICs for nonpregnant women and children were 329 and 416 micrograms/L, respectively, indicating excessive iodine intake (>300 micrograms/L). The prevalence of visible goiter was <4%. The coverage of salt iodization was low, with a national average of 7.7% (95% CI: 3.2%, 17.4%). Spatial analysis revealed localized areas of relatively high and low iodine status. Variations could not be explained by food consumption or salt iodization but were associated with the main source of household drinking water, with consumers of borehole water having a higher UIC (569 vs. 385 micrograms/L;P<0.001)." Kassim et al 2014, Pg. 1 • 4. "Both insufficient and excessive iodine intake can result in thyroid disease. The term 'iodine deficiency disorders' refers to the several consequences that iodine deficiency imposes on individuals." Iodine Deficiency Disorders, UpToDate • 5. • "Goiter is the most obvious manifestation of iodine deficiency. Low iodine intake leads to reduced thyroxine (T4) and triiodothyronine (T3) production, which results in increased thyrotropin (TSH) secretion in an attempt to restore T4 and T3 production to normal. TSH also stimulates thyroid growth; thus, goiter occurs as part of the compensatory response to iodine deficiency." Iodine Deficiency Disorders, UpToDate • "Early Chinese medical writings in approximately 3600 B.C. were the first to record the decreases in goiter size upon ingestion of seaweed and burnt sea sponge [1]." Leung, Braverman and Pearce 2012, Pg. 1740 • "Iodine in organic form occurs in high amounts in certain seaweeds. Inhabitants of the coastal regions of Japan, whose diets contain large amounts of seaweed, have remarkably high iodine intakes amounting to 50 to 80 mg/d." Zimmermann 2009, Pg. 377 • "In 1852, Adolphe Chatin, a French chemist, was the first to publish the hypothesis of population iodine deficiency associated with endemic goiter [3]." Leung, Braverman and Pearce 2012, Pg. 1741 • 6. • We discuss these surveys in more detail below. • "By the mid-1990s, the median UIC <100 micrograms/L in SAC [school-aged children] had become the most widely used biochemical measure of ID in a population. For national, school-based surveys of iodine nutrition, the median UIC from a representative sample of spot urine collections from approximately 1,200 children (30 sampling clusters x 40 children per cluster) could be used to derive the median UIC used to classify a country’s iodine status (Table 1)." Zimmermann and Andersson 2012, Pg. 6 • Daily iodine intake can be estimated from measured UIC and estimated daily urine volume: "Daily iodine intake can be extrapolated from urinary concentration as follows. The median 24-hour urine volume for ages 7 through 15 years is approximately 0.9 mL/hr/kg (or 0.0009 L/hr/kg) (Mattsson and Lindstrom, 1995). The 24-hour urine volume for adults is approximately 1.5 L (Larsson and Victor, 1988), a value in general agreement with an extrapolation of the calculation for children and adolescents. Urine volume among individuals and over time can vary considerably, but these numbers for daily volume appear reasonable for population estimates. From the above information and assuming an average bioavailability of 92 percent, the daily iodine intake is calculated from urinary iodine concentration by the following formula: Urinary iodine (μg/L) ÷ 0.92 × (0.0009 L/h/kg × 24 h/d) × wt (kg) = daily iodine intake;" Dietary Reference Intakes for Vitamin A, Vitamin K, Arsenic, Boron, Chromium, Copper, Iodine, Iron, Manganese, Molybdenum, Nickel, Silicon, Vanadium, and Zinc, Pg. 264 • The World Health Organization (WHO) provides the following classification for the iodine status of a population based on its median UIC: Median UIC in school-aged children (µg/L) Iodine status <20 Severe 20-49 Moderate 50-99 Mild 100-199 Adequate 200-299 Above requirements ≥300 Excessive • See How reliable and predictive is measuring urinary iodine content? for more information on how the WHO determines these thresholds. • For pregnant women, whose iodine requirements differ from children and the general adult population, the WHO classifies a median UIC of <150 as insufficient, 150-249 as adequate, 250-499 as above requirements and ≥500 as excessive. Table 5, WHO Assessment of Iodine Deficiency Disorders and Monitoring their Elimination 2007, Pg. 33 • 7. "WHO recommends universal salt iodization - the fortification with iodine of all salt used for human and animal consumption - as the main strategy for eliminating iodine deficiency." Salt Iodization, WHO • 8. "Salt is the vehicle of choice for the following reasons: 1. It is consumed by nearly everyone at roughly equal amounts throughout the year, 2. Salt production is limited to a few centres, facilitating quality control, 3. Addition of iodate or iodide does not affect the taste or smell of the salt, and 4. Iodisation is cheap (less than 0.01 US Dollar a day for the estimated amount of salt intake). " Angermayr and Clar 2004, Pg. 3 • 9. For a list of iodine deficiency disorders that might be prevented by correcting iodine deficiency, see Table 1, WHO Assessment of Iodine Deficiency Disorders and Monitoring their Elimination 2007, Pg. 7 • 10. • 11. "The three characteristic features of neurological cretinism in its fully developed form are severe mental retardation with squint, deaf mutism, and motor spasticity (Fig. 2A). The mental deficiency is characterized by a marked impairment of abstract thought, whereas autonomic and vegetative functions and memory are relatively well preserved, except in the most severe cases. Vision is unaffected, whereas deafness is characteristic. This may be complete in as many as 50% of cretins, as confirmed by studies of auditory brainstem-evoked potentials. The motor disorder shows proximal rigidity of both lower and upper extremities and the trunk, and corresponding proximal spasticity with exaggerated deep tendon reflexes at the knees, adductors, and biceps (100). Spastic involvement of the feet and hands is unusual, and their function is characteristically preserved so that most cretins can walk." Zimmermann 2009, Pg. 385 • 12. See the section on Improving mental function below • 13. • See Table 1: Summary of results from 6 RCTs evaluating the effect of iodine on children’s mental function for a description of the studies’ qualities. • We also put some small weight on the fact that Taylor et al 2013 only includes Gordon et al 2009 and Zimmermann et al 2006: “Only two RCTs which determined cognitive performance in children living in areas with mild-to-moderate iodine deficiency were suitable for meta-analysis. In a double blind trial, Zimmermann et al. (25) measured cognitive and motor performance in 310 Albanian children aged 10–12 years (median UIC 44µg/l), who were randomly assigned to receive 400mg of intramuscular iodine or placebo (25)...Gordon et al. (24) in a marginally iodine-deficient New Zealand population randomly assigned 166 children aged 10–13 years (median UIC 63µg/l) to receive a daily tablet of 150 mg of iodine or placebo for 28 weeks (24).” Taylor et al 2013, Pgs. R7-R9 • "Several randomized, controlled trials in school-aged chil- dren have tried to measure the effect of iodized oil on cognition (158 –161 [158: Bautista et al 1982; 159: Shrestha et al 1994; 160: Huda, Grantham-McGregor, and Tomkins 2001; 161: Isa et al 2000 (Isa et al 2000 is not actually a RCT)]). Three of the studies found no effect (158, 160, 161), whereas one found that cognition im- proved with treatment (159). However, methodological problems limit their interpretation because two of the studies were confounded by a significant improvement in iodine status in the control group (158, 161), whereas in the other two, the treated group remained iodine deficient at retesting (159, 160). In a recent placebo-controlled, double-blind, 6-month intervention trial, moderately iodine-deficient 10- to 12-yr-old children (n = 310) in Albania were randomized to receive either 400 mg of iodine as oral iodized oil or placebo. The children were given a battery of seven cognitive and motor tests that included measures of information processing, working memory, vi- sual problem solving, visual search, and fine motor skills. Treatment with iodine markedly improved iodine and thyroid status: at 24 wk, median UI in the treated group was 172 µg/liter, and mean circulating T4 increased approximately 40%. Compared with placebo, iodine treatment significantly improved performance on tests of informa- tion processing, fine motor skills, and visual problem solving. These findings need to be confirmed in other populations, but it appears that in children born and raised in areas of iodine deficiency, cognitive impairment is at least partially reversible by iodine repletion (162 [Zimmermann et al 2006])." Zimmermann 2009, Pg. 389 • 14. • The 6 RCTs: Gordon et al 2009, Zimmermann et al 2006, Huda, Grantham-McGregor and Tomkins 2001, Untoro et al 1999, Shrestha et al 1994 and Bautista et al 1982. • "Only few studies investigated physical or mental development in the children... mental development [was assessed] in four studies (Bautista 1982; Shrestha 1994; Untoro 1999; Huda 2001)." Angermayr and Clar 2004, Pg. 9 • "Only two RCTs which determined cognitive performance in children living in areas with mild-to-moderate iodine deficiency were suitable for meta-analysis. In a double blind trial, Zimmermann et al. (25) measured cognitive and motor performance in 310 Albanian children aged 10-12 years (median UIC 44 µg/l), who were randomly assigned to receive 400mg of intramuscular iodine or placebo (25)...Gordon et al. (24) in a marginally iodine-deficient New Zealand population randomly assigned 166 children aged 10-13 years (median UIC 63 µg/l) to receive a daily tablet of 150 µg of iodine or placebo for 28 weeks (24)." Taylor et al 2013, Pg. R7-R9 • Angermayr and Clar 2004 is a Cochrane review. It includes all of the six trials except for Gordon et al 2009 and Zimmermann et al 2006, both published after it was released. • Taylor et al 2013 examined the effect of iodine supplementation of mild-to-moderately deficient children on their mental function. • "Objectives: We undertook a systematic review of the impact of iodine supplementation in populations with mild-to-moderate iodine deficiency. Methods: We searched Medline and the Cochrane library for relevant articles published between January 1966 and April 2013, which investigated the effect of iodine supplementation on maternal and newborn thyroid function, infant neurodevelopment and cognitive performance in school-age children" Taylor et al 2013, Pg. R1. • Taylor et al 2013 only included the two most recent RCTs: Gordon et al 2009 and Zimmermann et al 2006. • "Only two RCTs which determined cognitive performance in children living in areas with mild-to-moderate iodine deficiency were suitable for meta-analysis. In a double blind trial, Zimmermann et al. (25) measured cognitive and motor performance in 310 Albanian children aged 1012 years (median UIC 44 µg/l), who were randomly assigned to receive 400mg of intramuscular iodine or placebo (25)...Gordon et al. (24) in a marginally iodine-deficient New Zealand population randomly assigned 166 children aged 10-13 years (median UIC 63µg/l) to receive a daily tablet of 150 µg of iodine or placebo for 28 weeks (24)." Taylor et al 2013, Pg. R7-R9 • Taylor et al 2013 conducted a systematic search and described their inclusion criteria in the paper. • "We searched Medline and the Cochrane library for relevant articles published in the English language between January 1966 and April 2013. We used various combinations of the following terms: iodine, deficiency, supplementation, mild, moderate, pregnancy, goitre, IQ, childhood, development, neurological, thyroid, TSH, hypothyroidism, thyroiditis, hyperthyroidism, side effects, cost, salt, fortification, maternal and foetus. Additional publications were sourced from references in individual articles. Relevant articles were selected after reading through all titles and abstracts and full texts were obtained if the information contained in the title or abstract was insufficient to exclude the study." Taylor et al 2013, Pg. R2 • "We selected studies for review if they were randomised controlled trials (RCTs), quasi-randomised trials, or prospective cohort or case–control studies which investigated: i) the effects of maternal iodine supplementation in pregnancy on a) maternal thyroid function, b) foetal thyroid function and c) child neurodevelopment; and ii) the effects of childhood iodine supplementation on child cognitive performance. Studies were chosen if: i) participants received iodine supplements; ii) an appropriate control group was included which comprised participants who either received no supplements or received a significantly lower dose of supplements; iii) thyroid function, thyroid volume or cognitive performance were determined as outcomes; and iv) the study was limited to populations with mild-to-moderate iodine deficiency as determined from the median population urinary iodine." Taylor et al 2013, Pg. R2 • "The study selection flow process is shown in Fig. 1. Nine RCTs (24, 25, 26, 27, 28, 29, 30, 31, 32) and eight observational studies (33, 34, 35, 36, 37, 38, 39, 40) were included in the review. Seven RCTs were reported on the effects of maternal supplementation on maternal thyroid function (26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40). Of these, four RCTs also contained data on neonatal thyroid function (27, 29, 30, 32). One RCT (32) and four observational studies (33, 36, 39, 40) addressed the impact of gestational iodine supplementation on infant neuropsychological function, while two RCTs investigated the impact of iodine supplementation on cognitive performance in school-age children (24, 25)." Taylor et al 2013, Pg. R3 • But Taylor et al 2013 did not include a list of publications excluded from the review, so it is unclear why the authors chose to exclude Untoro et al 1999 and Huda, Grantham-McGregor and Tomkins 2001. • "We selected studies for review if they were randomised controlled trials (RCTs), quasi-randomised trials, or prospective cohort or case–control studies which investigated: i) the effects of maternal iodine supplementation in pregnancy on a) maternal thyroid function, b) foetal thyroid function and c) child neurodevelopment; and ii) the effects of childhood iodine supplementation on child cognitive performance. Studies were chosen if: i) participants received iodine supplements; ii) an appropriate control group was included which comprised participants who either received no supplements or received a significantly lower dose of supplements; iii) thyroid function, thyroid volume or cognitive performance were determined as outcomes; and iv) the study was limited to populations with mild-to-moderate iodine deficiency as determined from the median population urinary iodine." Taylor et al 2013, Pg. R2 • Participants in both Untoro et al 1999 and Huda, Grantham-McGregor and Tomkins 2001 received iodine supplements. Both studies were randomized controlled trials, "an appropriate control group was included which comprised participants who either received no supplements or received a significantly lower dose of supplements," and measured cognitive performance (see Table 1: Summary of results from 6 RCTs evaluating the effect of iodine on children’s mental function) • Presumably, Taylor et al 2013 did not include Shrestha et al 1994 and Bautista et al 1982, because those studies examined populations with severe, rather than mild-to-moderate iodine deficiency (see Table 1: Summary of results from 6 RCTs evaluating the effect of iodine on children’s mental function) ). The control group's iodine status also improved almost as much as the treatment's group in Bautista et al 1982 and the treatment group's iodine status improved only temporarily in Shrestha et al 1994 (see “Table 1: Summary of results from 6 RCTs evaluating the effect of iodine on children’s mental function”; "The group to be treated with iodine had a slightly lower initial mean level of urinary iodine than the controls, and the difference was significant at p = 0.04...The increases in the control group and the large SD's in both groups suggest the introduction of iodine into the community from other sources, or perhaps a recycling of the administered iodine within the population." Bautista et al 1982, Pg. 129). We would guess that they excluded Untoro et al 1999 because iodine status improved somewhat in the control group, and Huda, Grantham-McGregor and Tomkins 2001 because the treatment group remained mildly iodine deficient (see “Table 1: Summary of results from 6 RCTs evaluating the effect of iodine on children’s mental function”). • Still, we don't see a strong reason to completely ignore any one study. Angermayr and Clar 2004 classified the quality of Huda, Grantham-McGregor and Tomkins 2001, Untoro et al 1999, Shrestha et al 1994 and Bautista et al 1982 as moderate and the two more recent RCTs seem of similar quality or higher. • "Most studies were of poor methodological quality ('weak'). None of the studies were classified as ’high quality’, and only six studies were classified as 'moderate quality' (Bautista 1982; Shrestha 1994; Cobra 1997; Untoro 1999; Zhao 1999; Huda 2001) and described methodological issues in some detail (for example, power calculation, randomisation method, flow of participants, blinding of outcome assessment, allocation concealment)." Angermayr and Clar 2004, Pg. 9 • "The group assignments were done by using Excel software (version XP 2002; Microsoft, Redmond, WA) with the random-number-generator function and by using a Bernoulli distribution (P = 0.5), in which random variables have the value 0 or 1. The capsules were swallowed with water under direct supervision. The study design was double-blind. Children were tested at baseline in September 2004 and retested in March 2005, or 23–24 wk after receiving the treatment or placebo. After the second round of testing was complete, all children in the study received 400 mg oral I (26)." Zimmermann et al 2006, Pg. 110 • "This study was a randomized, placebo-controlled, double-blind intervention trial undertaken in schoolchildren living in Dunedin, New Zealand, between August 2007 and October 2008...children were block-randomized (block size = 20) according to sex and method of recruitment (ie, school or advertisement) to receive a daily supplement containing either 150 µg I or an identical placebo." Gordon et al 2009, Pg. 1265 • "At the end of the baseline testing session, children were provided with 4 wk worth of supplements in 28-d compliance packaging blister packs (Medico Pak; Douglas Pharmaceuticals, Auckland, New Zealand), and an information sheet on how to take their supplements. Every 4 wk a new blister pack of supplements was posted to each child, and a return envelope was provided so that the previous month’s supplements could be collected and counted to assess compliance. If a pack of supplements was not returned, the compliance was assumed to be zero for that month. Movie vouchers, small stationary items, or shopping vouchers were sent out with supplements during the study to aid with compliance." Gordon et al 2009, Pg. 1266 • 15. • "Mild iodine deficiency is defined as a median urinary iodine concentration of 50 to 99 mcg/L, moderate deficiency as 20 to 49 mcg/L,and severe deficiency as [less than] 20 mcg/L [6]." Iodine Deficiency Disorders, UpToDate • See “Table 1: Summary of results from 6 RCTs evaluating the effect of iodine on children’s mental function” • 16. • 17. • 18. Shrestha et al 1994: "BLINDING: participants blinded, investigators blinded, outcome assessors unclear", Angermayr and Clar 2004, Pg. 39 • 19. • "In a double-blind intervention trial, 10 12-y-old children (n = 310) in primary schools in rural southeastern Albania were randomly assigned to receive 400 mg I (as oral iodized oil) or placebo." Zimmermann et al 2006, Pg. 108 • "A randomized, placebo-controlled, double-blind trial was conducted in 184 children aged 10-13 y in Dunedin, New Zealand" Gordon et al 2009, Pg. 1264 • 287 participants. Table 2, Huda, Grantham-McGregor and Tomkins 2001, Pg. 75 • 164 participants. Table 2, Untoro et al 1999, Pg. 77 • 144 participants in the group receiving iodine and the group receiving a placebo (Table 3, Shrestha et al 1994, Pg. 49). Shrestha et al 1994 also had a treatment arm, which received both iodine and iron supplements, and a treatment arm, which received just iron (156 participants in these groups). Comparing these two arms yields an average standardized treatment effect of 0.93 (similar to the average standardized treatment effect comparing the group receiving iodine supplementation to the group receiving a placebo). When studies have multiple relevant comparisons, the Cochrane handbook suggests combining the treatment arms together and combining the control arms together and then comparing the combined groups (Combining groups, Cochrane Handbook for Systematic Reviews of Interventions 2011). We have not undertaken the analysis that Cochrane suggests; in the context of the various limitations of the RCTs, we do not think that this analysis would change our conclusions. • "From these we chose at random for further study 100 boys and 100 girls, all between ages 5½ and 12 yr and all with some degree of thyroid enlargement" Bautista et al 1982, Pg. 128 • 20. • 21. • 22. • Gordon et al 2009: "Children were randomly assigned to receive a daily tablet containing either 150 lg I or placebo for 28 wk." Gordon et al 2009, Pg. 1264 • Zimmermann et al 2006: "In a double-blind intervention trial, 10 –12-y-old children (n = 310) in primary schools in rural southeastern Albania were randomly assigned to receive 400 mg I (as oral iodized oil) or placebo." Zimmermann et al 2006, Pg. 108 • Bautista et al 1982: "INTERVENTION: single dose of oral iodised oil (Ethiodol, 1ml, 475 mg iodine/ml)(N= 95 CONTROL: single dose of 1 ml of non-iodine mineral oil, same colour (N=94)" Angermayr and Clar 2004, Pg. 25 • Huda, Grantham-McGregor and Tomkins 2001: "INTERVENTION: single oral dose of 400 mg iodised poppyseed oil (Lipiodol, iodine content not stated) (N=156) CONTROL: single oral dose of poppyseed oil without iodine, similar appearance and taste (N=149)" Angermayr and Clar 2004, Pg. 31 • Shrestha et al 1994: "INTERVENTION: single oral dose of iodised oil, (Lipiodol Ultrafluid, 1 ml, 490 mg/ml) (N=79) CONTROL: single oral dose of poppyseed oil, 1ml (N=82)" Angermayr and Clar 2004, Pg. 39 • Untoro et al 1999: "INTERVENTION A: single oral dose of 1ml poppyseed oil, (Lipiodol Ultrafluide, 400 mg iodine/ml, total dose 400 mg iodine) (N=49) INTERVENTION B: single oral dose of iodised peanut oil (Yodiol, 0.5 ml, 400 mg iodine/ml, total dose 200 mg iodine) (N=50) INTERVENTION C: single oral dose of iodised peanut oil (Yodiol, 1.0 ml, 400 mg iodine/ml, total dose 400 mg iodine) (N=51) INTERVENTION D: single oral dose of iodised peanut oil (Yodiol, 2.0 ml, 400 mg iodine/ml, total dose 800 mg iodine) (N=50) CONTROL: single oral dose of placebo peanut oil , 1 ml (N=51)" Angermayr and Clar 2004, Pg. 42 • 23. • See the complete list of mental tests used here: Analysis of studies on the effect of iodine supplementation of children on their mental function (xlsx) • We note that tests may have been chosen in order to maximize the estimated effect of iodine treatment on cognitive performance. For example, in Gordon et al 2009, one criterion for choosing tests was the outcome of cognitive tests in Zimmermann et al 2006, and Zimmermann et al 2006 partly chose its tests based on "sensitivity to iodine deficiency or hypothyroidism (or both) in previous studies." • "Each test was chosen on the basis of 3 criteria: 1) the likelihood of the test to respond to iodine supplementation in mildly iodine-deﬁcient children based on previous research (11 [Zimmermann et al. 2006]), 2) the aspect of cognition assessed by the test, and 3) the relative ease of administration of the test." Gordon et al 2009, Pg. 1265 • "Tests were chosen for their simplicity and ease of administration, as well as their sensitivity to iodine deficiency or hypothyroidism (or both) in previous studies (see Discussion)." Zimmermann et al 2006, Pg. 109 • Additionally, 4 of the 11 cognitive tests used in Gordon et al 2009 and Zimmermann et al 2006 had only modest test-retest reliability. • "Test-retest reliability was estimated by testing 30 schoolchildren, boys and girls, on each of the tests and then testing them again 14 d later. The children were not retested by the same tester. Children used in the estimation of test reliabilities were the same age as the study population, but they were not participants in the subsequent intervention study. Correlations between the 2 scores for each test were: Raven’s Coloured Progressive Matrices, 0.80; digit span, 0.78; coding, 0.74; symbol search, 0.54; rapid object naming, 0.71; bead threading, 0.52; and rapid target marking, 0.68." Zimmermann et al 2006, Pg. 110 • "A further limitation was the modest test-retest reliabilities on 2 of the 7 tests (bead threading and symbol search), which may have reduced our ability to distinguish a treatment effect on these tests." Zimmermann et al 2006, Pg. 113 • "The 4 cognitive tests chosen were picture concepts, letter-number sequencing, matrix reasoning, and symbol search, which have a test-retest reliability of 0.74, 0.59, 0.73, and 0.55, respectively (17)." Gordon et al 2009, Pg. 1265 • 24. • "Mixed model regression was performed on the cognitive test scores to obtain an estimate of the overall effect of iodine supplementation on cognition, with control for baseline scores (standardized by using baseline SD), sex, method of recruitment, cohort, household income, and ethnicity and by using a random-subjects effect to control for pairs of cognitive scores (ie, at baseline and 28 wk) from each participant." Gordon et al 2009, Pg. 1266 • "After baseline scores, sex, method of recruitment, cohort, ethnicity, and household income were controlled for, the overall cognitive score was 0.19 (95% CI: 0.04, 0.34) SD higher in iodine-supplemented children than in children who received the placebo at 28 wk (P = 0.011)." Gordon et al 2009, Pg. 1268 • 25. • See Analysis of studies on the effect of iodine supplementation of children on their mental function (xlsx) • Our standardized treatment effect is calculated using the standardized mean difference (The standardized mean difference, Cochrane Handbook for Systematic Reviews of Interventions 2011). When available, we've used the estimate of the treatment effect from a regression instead of the "Difference in mean outcome between groups," but we also report the standardized treatment effects calculated only using the difference in mean outcome between the treatment group and control group in the spreadsheet. • We use the standard deviation of the control group at endline, because more studies report this quantity than the standard deviation of the treatment group or control group at baseline. We don’t use the standard deviation of the treatment group at endline, because the treatment may have affected the standard deviation, introducing more potential variability between studies. • Note that the standard deviation of the treatment effect assessed by a single subtest will be larger than the standard deviation of the treatment effect assessed by a battery of subtests, so calculating a standardized treatment effect for each subtest individually and then averaging the results may tend to underestimate the overall treatment effect. • "The first step in determining whether the results from the subtests used in the iodine supplementation RCTs are likely to correlate with cognitive abilities tapped by IQ tests is to see how strongly the results of the subtests correlate with each other. If the results of the subtests do not correlate very highly with each other, they are likely not measuring what IQ tests measure. Even simple measures of short-term memory, like reciting a list of animal names, will probably have some correlation with IQ. However, estimates of cognitive gains through iodine supplementation that only relied on a few very simple tests might not correlate very highly with the results of an official IQ test. The greater the variety and number of subtests used, the more likely that the results of the tests would correlate with a standard measure of IQ. A motor coordination test, such as threading beads, which was used in some of the iodine supplementation RCTs, is likely to be at least somewhat predictive of IQ. John B. Carroll’s research is a good source for more information on the connection between motor coordination, cognitive subtests, and IQ." GiveWell's non-verbatim summary of a conversation with Jelte Wicherts, Associate Professor, Department of Methodology and Statistics, Tilburg University, August 28, 2014, Pg. 1 • "For each of the other studies, GiveWell obtains an overall treatment effect by standardizing the treatment effects for the individual tests and averaging them. GiveWell then multiples this standardized treatment effect by 15 to convert it to an IQ scale. Dr. Wicherts thought this approach for estimating the average treatment effect across the subtests in the iodine supplementation RCTs and converting them to an IQ scale seemed reasonable. However, if one of the subtests in the iodine supplementation RCTs used for GiveWell’s estimates did not show an effect, it is difficult to determine if the lack of effect is because the subtest was not reflective of general intelligence, or because the iodine supplementation did not cause cognitive gains. If the problem was an unreliable subtest not reflective of general intelligence, the average treatment effect would be biased downwards." GiveWell's non-verbatim summary of a conversation with Jelte Wicherts, Associate Professor, Department of Methodology and Statistics, Tilburg University, August 28, 2014, Pg. 2 • 26. • Control change in median UIC (urinary iodine concentration) (µg/L) • "Median; interquartile range (25th–75th percentile) in parentheses (all such values)." Gordon et al 2009, Pg. 1267 • "Median; range in parentheses (all such values)." Zimmermann et al 2006, Pg. 111 • "Geometric mean", Huda, Grantham-McGregor and Tomkins 2001, Pg. 74 • "Median urinary iodine concentration of population studied improved from 0.38 nmol/L to 0.70 (µmol/L, but this level is still below 0.79 µmol/L which is regarded as adequate iodine status." Untoro et al 1999, Pg. 88 • "After 3 mo, urinary iodine concentration increased in the iodine and iodine plus iron groups from a median of 0.15 (0.09, 0.25; 25th and 75th percentiles) µg/L and 0.15 (0.08, 0.26) µg/L to 0.42 (0.01, 1.19) µg/L and 0.32 (0.11, 0.84) µg/L respectively. After 11 mo, the values had fallen to 0.18 (0.08, 0.50) and 0.16 (0.06, 0.34) µg/L respectively. There was no change in the other two groups which did not receive iodized oil." Shrestha et al 1994, Pg. 47 • "Table 3 shows the mean levels of urinary iodine at several times during the study." Bautista et al 1982, Pg. 130. Presumbly, Table 2 also reports the mean. • Treatment change in median UIC (µg/L) • "Median; interquartile range (25th–75th percentile) in parentheses (all such values)." Gordon et al 2009, Pg. 1267 • "Median; range in parentheses (all such values)." Zimmermann et al 2006, Pg. 111 • "Geometric mean", Huda, Grantham-McGregor and Tomkins 2001, Pg. 74 • "Median urinary iodine concentration of population studied improved from 0.38 nmol/L to 0.70 (µmol/L, but this level is still below 0.79 µmol/L which is regarded as adequate iodine status." Untoro et al 1999, Pg. 88 • "After 3 mo, urinary iodine concentration increased in the iodine and iodine plus iron groups from a median of 0.15 (0.09, 0.25; 25th and 75th percentiles) µg/L and 0.15 (0.08, 0.26) µg/L to 0.42 (0.01, 1.19) µg/L and 0.32 (0.11, 0.84) µg/L respectively. After 11 mo, the values had fallen to 0.18 (0.08, 0.50) and 0.16 (0.06, 0.34) µg/L respectively. There was no change in the other two groups which did not receive iodized oil." Shrestha et al 1994, Pg. 47 • "Table 3 shows the mean levels of urinary iodine at several times during the study." Bautista et al 1982, Pg. 130. Presumbly, Table 2 also reports the mean. • 27. Table 1, Gordon et al 2009, Pg. 1267 • 28. Table 1, Gordon et al 2009, Pg. 1267 • 29. Table 1, Gordon et al 2009, Pg. 1267 • 30. Analysis of studies on the effect of iodine supplementation of children on their mental function (xlsx) • 31. Table 1, Zimmermann et al 2006, Pg. 111 • 32. Table 1, Zimmermann et al 2006, Pg. 111 • 33. Table 1, Zimmermann et al 2006, Pg. 111 • 34. Analysis of studies on the effect of iodine supplementation of children on their mental function (xlsx) • 35. • Table 2, Shrestha et al 1994, Pg. 46 • "After 3 mo, urinary iodine concentration increased in the iodine and iodine plus iron groups from a median of 0.15 (0.09, 0.25; 25th and 75th percentiles) µg/L and 0.15 (0.08, 0.26) µg/L to 0.42 (0.01, 1.19) µg/L and 0.32 (0.11, 0.84) µg/L respectively. After 11 mo, the values had fallen to 0.18 (0.08, 0.50) and 0.16 (0.06, 0.34) µg/L respectively. There was no change in the other two groups which did not receive iodized oil." Shrestha et al 1994, Pg. 47 • 36. "After 3 mo, urinary iodine concentration increased in the iodine and iodine plus iron groups from a median of 0.15 (0.09, 0.25; 25th and 75th percentiles) µg/L and 0.15 (0.08, 0.26) µg/L to 0.42 (0.01, 1.19) µg/L and 0.32 (0.11, 0.84) µg/L respectively. After 11 mo, the values had fallen to 0.18 (0.08, 0.50) and 0.16 (0.06, 0.34) µg/L respectively. There was no change in the other two groups which did not receive iodized oil." Shrestha et al 1994, Pg. 47 • 37. • "Anthropometric measurements were made and seven mental performance tests were administered before and 10 mo after supplementation began." Shrestha et al 1994, Pg. 41 • We mean “follow up” to represent the time from first administration of iodine supplements to the tests of mental function. In Shrestha et al 1994, the children’s UICs were measured about a month after they were given the tests. • 38. Analysis of studies on the effect of iodine supplementation of children on their mental function (xlsx) • 39. • 40. • 41. Table 2, Huda, Grantham-McGregor and Tomkins 2001, Pg. 75 • 42. Analysis of studies on the effect of iodine supplementation of children on their mental function (xlsx) • 43. • 44. • 45. Table 2, Untoro et al 1999, Pg. 77 • 46. • 47. • Table 2 Bautista et al 1982, Pg. 130 • "If the daily volume of urine produced by a group approximates 1 L/day, as it does in healthy primary SAC [school-aged children], then the UIC (mg/L) is interchangeable with the 24 h UIE (mg/24 h)." Zimmermann and Andersson 2012, Pg. 557 • Using the mean to measure UIC may be unreliable: "He [Pierre Bourdoux] showed that UIC distributions in populations generally do not follow a normal distribution and the limitations of the use of the mean UIC to describe iodine status of a population." Zimmermann and Andersson 2012, Pg. 557 • 48. • Table 2, Bautista et al 1982, Pg. 130 • "If the daily volume of urine produced by a group approximates 1 L/day, as it does in healthy primary SAC [school-aged children], then the UIC (mg/L) is interchangeable with the 24 h UIE (mg/24 h)." Zimmermann and Andersson 2012, Pg. 557 • "He showed that UIC distributions in populations generally do not follow a normal distribution and the limitations of the use of the mean UIC to describe iodine status of a population." Zimmermann and Andersson 2012, Pg. 557 • 49. "On follow-up 22 months later the urinary iodine had increased and goiter size had decreased in both groups, more strikingly in the iodine-treated children. There were no consistent differences between the two treatment groups in rate of somatic growth or performance on the Stanford-Binet and Bender tests." Bautista et al 1982, Pg. 127 • 50. • 51. • See the table above (Table 1: Summary of results from 6 RCTs evaluating the effect of iodine on children’s mental function) for data on urinary iodine in the studies • Gordon et al 2009 • "After 28 wk, the median UIC in both the placebo and iodine groups had increased significantly (P , 0.001); however, the median UIC of 81 ug/L in the placebo group still categorized these children with mild iodine de- ficiency, whereas the UIC of 145 ug/L in the iodine group was indicative of adequate iodine status." Gordon et al 2009, Pg. 1267 • "There was a small improvement in the iodine status of the placebo group, but these children were still categorized as mildly iodine deficient at the end of the study. The iodine content of the diet, as estimated by the FFQ, did not differ between the 2 groups nor did the use of iodized salt at the table; however, a higher pro- portion of children in the placebo group than in the supple- mented group reported using iodized salt in cooking (68% compared with 59%)." Gordon et al 2009, Pg. 1268 • "A limitation of many previous trials was the concurrent improvement in iodine status in the control group that necessitated a post hoc change in data analysis; this was not required in our study because a clear distinction in iodine status between the supplemented and control groups was maintained after 28 wk." Gordon et al 2009, Pg. 1268 • Bautista et al 1982 • "Urinary iodine excretion increased approximately 4-fold in the iodine-treated group and doubled in the controls. As shown in Table 3, the mean urinary iodine levels remained fairly constant, around 50 µg/day, during the first 18 months after iodine administration, with a slight subsequent decrease at 22 months. The group to be treated with iodine had a slightly lower initial mean level of urinary iodine than the controls, and the difference was significant at p = 0.04. This was much less than the highly significant difference (p = 0.0001) after treatment. The increases in the control group and the large SD’s in both groups suggest the introduction of iodine into the community from other sources, or perhaps a recycling of the administered iodine within the population." Bautista et al 1982, Pg. 129 • "The group comparisons were complicated by an apparent increase in iodine ingestion by the controls." Bautista et al 1982, Pg. 130 • "The apparent introduction of some exogenous iodine into the community during our study complicated interpretation of the results. Fortunately, this could be monitored by the urinary iodine levels." Bautista et al 1982, Pg. 132 • "We did not identify the source of the increased iodine in the controls. Iodized salt was available in nearby Cochabamba, but not widely used there because of its cost, and was not sold in Tiquipaya. No educational programs promoting iodination were being conducted during the study. Recycling of excreted iodine within a community has been a significant source of iodine conservation in other endemias (23). We cannot say whether it was a likely factor in this study." Bautista et al 1982, Pg. 132 • Untoro et al 1999: "Thirdly, the effect observed had been attenuated by exposure of the population, both the iodized oil supplemented group and the placebo group, to iodine derived from iodized salt which has become increasingly available in the area during the study period (43)." Untoro et al 1999, Pg. 82 • 52. • See the table above (Table 1: Summary of results from 6 RCTs evaluating the effect of iodine on children’s mental function) for data on urinary iodine in the studies • Table 4, WHO Assessment of Iodine Deficiency Disorders and Monitoring their Elimination 2007, Pg. 33 • "A significant treatment effect was found on urinary iodine levels with a clear increase in the treated group. Although the urinary iodine levels in the treated group at 4 mo were more than double initial levels, they were not within the range of normal values. Recent studies have shown that rapeseed oil (35), peanut oil (36) and Oriodol (37) may produce greater and longer lasting increases in iodine levels...Another hypothesis is that the iodine preparation (Lipoidol) was no sufficiently effective, and it is possible that other preparations that produce normal levels of iodine excretion may have had benefits." Huda, Grantham-McGregor and Tomkins 2001, Pg. 76 • The study participants in Bautista et al 1982 and Shrestha et al 1994 also remained mildly deficient, but from a baseline level of severe deficiency. • 53. • For instance, researchers might only report a subset of the tests they administered in a study. • "Preregistration for trials would somewhat alleviate concerns about the choice of subtests, since it would be easier to see if it seemed like specific subtests were chosen simply for their likelihood of observing the desired effect. Publication bias is a potential issue with the iodine supplementation RCTs, since it is much more interesting to find an effect from iodine supplementation than it would be to not find an effect. Studies with very small sample sizes are also of concern, since they may inflate the estimated treatment effect. If the studies with small sample sizes show the biggest treatment effects, there is reason to suspect that the combined results are not reliable." GiveWell's non-verbatim summary of a conversation with Jelte Wicherts, Associate Professor, Department of Methodology and Statistics, Tilburg University, August 28, 2014, Pg. 2 • 54. Perhaps children eventually catch up at older ages despite slowed cognitive development at younger ages, or perhaps iodine deficiency actively impedes mental function throughout childhood. • 55. • "Some past reviewers (Haskins, 2004; McKey et al., 1985; White & Casto, 1985) have found that positive impacts of Head Start and early childhood programs for disadvantaged children decrease over time and eventually fade altogether. However, recent meta-analyses of longitudinal studies (Gorey, 2001; Nelson et al., 2003) suggest that effects persist over time although there may be some diminution of effects over the long term. These findings are consistent with the work done by Barnett, Young, and Schweinhart (1998), who used causal modeling to show that long-term effects of early childhood education are built upon short-term effects. Reviews focused on long-term studies of early education programs serving economically disadvantaged children (eg, Barnett, 1998, 2004) find that the evidence regarding Head Start’s long-term outcomes is mixed. In a recent examination of Head Start’s long- term cognitive effects, Barnett (2004) identified only 39 studies in which educational programs included treatment and control groups, served children from low-income families, began during or before the preschool years, and were followed up with cognitive or academic measures at least through third grade, of which 15 were studies of “model”programs and 24 were studies of large-scale public programs. Twelve of the public program studies focused on Head Start, and an additional 4 included both Head Start and public school programs. Several of the model program studies, but none of the large-scale public program studies, employed random assignment. Studies of model programs typically show initial gains in children’s IQ scores that fade out over time (Barnett, 2004). Studies of large-scale programs have less often measured IQ, although the Peabody Picture Vocabulary Test has sometimes been used as a proxy for verbal IQ, making it more difficult to evaluate whether Head Start produces persistent IQ gains. However, it is likely that initial increases in IQ scores by Head Start children also fade out over time. Findings regarding other types of benefits are more promising." Barnett and Hustedt 2005, Pg. 18 • "Although the program had a strong effect on children’s intellectual performance, didn’t it fade out over time? It is true that the High/Scope Perry Preschool program had a statistically significant effect on children’s IQs during and up to a year after the program, but not after that. This pattern has been found in numerous other studies, such as those in the Consortium for Longitudinal Studies (1983). The pattern raises two questions: How far does it generalize, and what does it mean? For some time, the pattern of children’s intellectual performance found in this study was taken to represent all outcomes of this and similar programs. It was concluded that the program had strong effects that faded out over time. However, all of the subsequent findings of program effects in this study (effects on school achievement, high school graduation, adult earnings, and crime prevention) disprove this conclusion. Indeed, so many studies have now found evidence of long-term effects of high-quality preschool programs that the opposite conclusion is practically indisputable: High-quality preschool programs for young children living in poverty do have long-term effects. So what is the meaning of the fadeout of program effect on children’s intellectual performance? More than anything else, it teaches us about the nature of multiage intelligence tests. Unlike most achievement tests that are age-specific, most intelligence tests, like the Stanford-Binet (Terman & Merrill, 1960), are designed to be used with individuals of a wide range of ages, from early childhood to adulthood. Also unlike achievement tests, intelligence tests were not designed to assess program effects, and so the way they function in this role was not, and is not, well understood. Multiage intelligence tests actually consist of a series of age-specific test batteries (the Stanford-Binet has 6 items per battery) designed to function with a specific age level, such as children 4 years old or children 4 years and 6 months of age. The preschool studies found effects at the ages during and a year or two after the program, but not subsequently. Children with preschool program experience got more items right on those age- specific batteries, but did not get more right on age-specific batteries designed for older children. It seems reasonable to conclude that, when used to assess preschool program outcomes, intelligence tests functioned more like achievement tests than intelligence tests, and indeed that is precisely the use to which they were put. Imagine if achievement tests for grades 4–8 were all combined into one grand multiage test of achievement. It would not be at all surprising if a really good grade 4 classroom improved children’s achievement test scores on this test at grades 4 and 5, but not at grades 6, 7, and 8. That is precisely what happened in the temporary effects of high-quality preschool programs on children’s intellectual performance. To take this thinking to a theoretical level regarding children’s intellectual performance, we might simply say that the preschool studies showed intellectual performance to be environmentally sensitive—it went up in intellectually stimulating preschool settings and down in less intellectually stimulating elementary school settings. Or, to put it in terms of program and no-program groups, it went up when the program group’s experience was more intellectually stimulating than that of the no-program group and returned to the same level as that of the no-program group when both found themselves in the same elementary school settings." Schweinhart et al 2005, Pg. 16 • 56. For example, other benefits of preschool programs appear to persist, despite the fact that IQ increases from preschool programs fade out: "For some time, the pattern of children’s intellectual performance found in this study was taken to represent all outcomes of this and similar programs. It was concluded that the program had strong effects that faded out over time. However, all of the subsequent findings of program effects in this study (effects on school achievement, high school graduation, adult earnings, and crime prevention) disprove this conclusion. Indeed, so many studies have now found evidence of long-term effects of high-quality preschool programs that the opposite conclusion is practically indisputable: High-quality preschool programs for young children living in poverty do have long-term effects." Schweinhart et al 2005, Pg. 16 • 57. • 58. • Analysis of studies on the effect of iodine supplementation of children on their mental function (xlsx) • If we assume the two studies for which we could not calculate a standardized treatment effect had a standardized treatment of effect of 0, then the unweighted average of all the studies would be 0.23. • Shrestha et al 1994 also had a treatment arm, which received both iodine and iron supplements, and a treatment arm, which received just iron. Comparing these two arms yields an average standardized treatment effect of 0.93 (similar to the average standardized treatment effect comparing the group receiving iodine supplementation to the group receiving a placebo). When studies have multiple relevant comparisons, the Cochrane handbook suggests combining the treatment arms together and combining the control arms together and then comparing the combined groups (Combining groups, Cochrane Handbook for Systematic Reviews of Interventions 2011). We have not undertaken the analysis that Cochrane suggests; in the context of the various limitations of the RCTs, we do not think that this analysis would change our conclusions. • 59. • 60. "That distribution itself—the crucial reference for assigning IQ scores—is simply the empirical result that was obtained when the test was initially standardized. By convention, the mean of each age group in the standardization sample defines an IQ score of 100; by further convention, the standard deviation of the sample defines 15 IQ points. Given appropriate sampling and a normal distribution, this implies that about two-thirds of the population in any given age group will have IQs between 85 and 115." Neisser 2011 • 61. "An IQ score is an averaged summary statistic of typically ten to twenty subtests for fluid intelligence, short-term memory, spatial reasoning, and other measures of intelligence." GiveWell's non-verbatim summary of a conversation with Jelte Wicherts, Associate Professor, Department of Methodology and Statistics, Tilburg University, August 28, 2014, Pg. 1 • 62. "The measurements from these subtests tend to correlate highly with each other in standard IQ tests. In contexts where children may not have the same levels of formal education as children in developed countries, researchers should choose specific subtests to get the most accurate and fair scores for a group. For example, presenting questions with multiple-choice answers to a group that is not used to the format would not result in a fair comparison to groups that answer multiple-choice questions frequently. Instead, researchers in these contexts usually focus on subtests like Raven’s Progressive Matrices, which can be understood without scholastic training." GiveWell's non-verbatim summary of a conversation with Jelte Wicherts, Associate Professor, Department of Methodology and Statistics, Tilburg University, August 28, 2014, Pg. 1 • 63. • If we only look at the tests of matrix reasoning ("The Raven's tests are generally considered to be good non-verbal indicators of general intelligence or g (Carroll, 1993; Jensen, 1998), at least in western samples, and have often been administered in sub-Saharan Africa." Wicherts et al 2010, Pg. 1), we get an increase of 5.7 IQ points. See Analysis of studies on the effect of iodine supplementation of children on their mental function (xlsx) • "The first step in determining whether the results from the subtests used in the iodine supplementation RCTs are likely to correlate with cognitive abilities tapped by IQ tests is to see how strongly the results of the subtests correlate with each other. If the results of the subtests do not correlate very highly with each other, they are likely not measuring what IQ tests measure. Even simple measures of short-term memory, like reciting a list of animal names, will probably have some correlation with IQ. However, estimates of cognitive gains through iodine supplementation that only relied on a few very simple tests might not correlate very highly with the results of an official IQ test. The greater the variety and number of subtests used, the more likely that the results of the tests would correlate with a standard measure of IQ. A motor coordination test, such as threading beads, which was used in some of the iodine supplementation RCTs, is likely to be at least somewhat predictive of IQ. John B. Carroll’s research is a good source for more information on the connection between motor coordination, cognitive subtests, and IQ." GiveWell's non-verbatim summary of a conversation with Jelte Wicherts, Associate Professor, Department of Methodology and Statistics, Tilburg University, August 28, 2014, Pg. 1 • 64. "No one in the academic community studying IQ would claim that a gain of four IQ points would not have a significant impact on life outcomes. Academic literature on the relationship between IQ and life outcomes examines impacts on the following categories: • Education: Most of the literature focuses on performance in formal education, including how difficult it is for an individual to graduate from high school. Whereas someone with an IQ of 76 may have difficulty in a formal educational setting without special assistance, someone with an IQ of 80 would likely be able to function independently. • Job performance: An increase in IQ of four points would impact job performance, even for most menial jobs. As long as a job requires learning new skills, a four-point gain in IQ would be an advantage, since it would enable a person to learn skills more quickly. Teaching an employee with an IQ of 80 to be a file clerk would be easier than teaching an employee with an IQ of 76. For jobs that do not require many learned skills, like sweeping a floor, the difference in job performance between two people with a four point IQ difference would likely be less pronounced. However, the ability to learn new skills quickly not only applies to job skills, but also to everyday activities learned through socialization.” • 65. "There is overwhelming evidence from a wide variety of experimental and observational studies that IQ has an independent effect on a wide range of life outcomes (an experimental study, for example, might randomly assign subjects to tasks with varying demands on cognition and then examine the differences in performance between low IQ and high IQ groups on the different tasks). Some researchers have tried to think of variables confounded with IQ that would explain different life outcomes, but there is always variation in outcomes that can only be explained by IQ differences. IQ is also mostly stable throughout one's life; possible confounders then have to both predict life outcomes and remain relatively stable throughout one's life." GiveWell's non-verbatim summary of a conversation with James Flynn, Emeritus Professor of Political Studies and Psychology, University of Otago, New Zealand, October 17, 2014, Pg. 2 • 66. • 67. "People with an IQ less than 70 are usually considered to have an intellectual disability and may have difficulty functioning independently. If the IQ of a population were normally distributed with a mean of 100 and a standard deviation of 15, then a four-point increase in the IQ of the population would reduce the prevalence of intellectual disability from about 2.3% to 1.2%." GiveWell's non-verbatim summary of a conversation with James Flynn, Emeritus Professor of Political Studies and Psychology, University of Otago, New Zealand, October 17, 2014, Pg. 1 • 68. • Shifting the IQ distribution (xlsx) • Limited evidence on IQ in Sub-Saharan Africa: "This paper presents a systematic review of published data on the performance of sub-Saharan Africans on Raven's Progressive Matrices. The specific goals were to estimate the average level of performance, to study the Flynn Effect in African samples, and to examine the psychometric meaning of Raven's test scores as measures of general intelligence. Convergent validity of the Raven's tests is found to be relatively poor, although reliability and predictive validity are comparable to western samples. Factor analyses indicate that the Raven's tests are relatively weak indicators of general intelligence among Africans, and often measure additional factors, besides general intelligence. The degree to which Raven's scores of Africans reflect levels of general intelligence is unknown. Average IQ of Africans is approximately 80 when compared to US norms. Raven's scores among African adults have shown secular increases over the years. It is concluded that the Flynn Effect has yet to take hold in sub-Saharan Africa." Wicherts et al 2010, Pg. 1 • 69. "The three characteristic features of neurological cretinism in its fully developed form are severe mental retardation with squint, deaf mutism, and motor spasticity (Fig. 2A). The mental deficiency is characterized by a marked impairment of abstract thought, whereas autonomic and vegetative functions and memory are relatively well preserved, except in the most severe cases. Vision is unaffected, whereas deafness is characteristic. This may be complete in as many as 50% of cretins, as confirmed by studies of auditory brainstem-evoked potentials. The motor disorder shows proximal rigidity of both lower and upper extremities and the trunk, and corresponding proximal spasticity with exaggerated deep tendon reflexes at the knees, adductors, and biceps (100). Spastic involvement of the feet and hands is unusual, and their function is characteristically preserved so that most cretins can walk." Zimmermann 2009, Pg. 385 • 70. ICCIDD conversations with GiveWell in Zurich, April/May 2014 • 71. • Mahomed and Gulmezoglu 1997 • Authors' conclusions: "Iodine supplementation in a population with high levels of endemic cretinism results in an important reduction in the incidence of the condition with no apparent adverse effects." Mahomed and Gulmezoglu 1997, Pg. 1 • Only Pharoah, Buttfield and Hetzel 1971 reported on cretinism. "Characteristics of included studies", Mahomed and Gulmezoglu 1997 Pgs. 4-5 • "Types of Participants Women, before or during pregnancy living in areas with low iodine intake (iodine deficiency)." Mahomed and Gulmezoglu 1997, Pg. 2 • "The finding that 6 cretinous children were born to mothers who had had iodised oil but were pregnant at the time of treatment is of some significance. The uncertainty about the date of birth of the 7th case makes it possible that this mother also had conceived before treatment. These data indicate that iodine deficiency in the mother during the first trimester is probably the main factor in the causation of endemic cretinism." Pharoah, Buttfield and Hetzel 1971, Pgs. 590-591 • Note that "The 'Maternal iodine supplements in areas of deficiency' review has been withdrawn from Issue 3, 2006 of The Cochrane Library because it is out-of-date. A new review team have taken responsibily for this review and are currently preparing an update for publication. " Mahomed and Gulmezoglu 1997. We would guess that its discussion of the effect of iodine supplementation of mild-to-moderately iodine deficient women on the mental function of their children, which is the focus of ongoing research, is the section considered outdated rather than its discussion of cretinism. • Zhou et al 2013 • "Despite the relatively broad inclusion criteria, only 2 of the 8 included RCTs reported limited clinical outcome data including the birth rate, birth anthropometric measures, childhood growth rate, 15-y survival, and development, and both trials were conducted in regions of severe iodine deficiency with a high incidence of endemic goiter and cretinism (17, 21)." Zhou et al 2013, Pg. 11 • "17. Kevany J, Fierro-Benitez R, Pretell EA, Stanbury JB. Prophylaxis and treatment of endemic goiter with iodized oil in rural ecuador and peru. Am J Clin Nutr 1969;22:1597–607." Zhou et al 2013, Pg. 13 • "21. Pharoah PO, Buttfield IH, Hetzel BS. Neurological damage to the fetus resulting from severe iodine deficiency during pregnancy. Lancet 1971; 1:308–10." Zhou et al 2013, Pg. 13 • However, Kevany 1969 did not find any cretins to study: "10) Cretinism: no obvious cretins detected in the series", Table 1, Zhou et al 2013, Pg. 1243 • "We searched the Cochrane Central Register of Controlled Trials (http://onlinelibrary.wiley.com/cochranelibrary/search?sear- chRow.searchOptions.searchProducts=clinicalTrialsDoi), EMBASE (http://www.embase.com/), MEDLINE (http://www.nlm.nih.gov/ medlineplus/), Cumulative Index to Nursing and Allied Health Literature (http://www.ebscohost.com/academic/cinahl-plus- with-full-text), PubMed (http://www.ncbi.nlm.nih.gov/pubmed/), and PsycINFO (http://www.apa.org/pubs/databases/psycinfo/index.aspx) databases from inception through December 2012 for relevant articles according to the search strategy (see Supplement Appendix 1 under “Supplemental data” in the online issue) specified in the predefined protocol. All RCTs (including those with a quasirandom design) that compared the effect of iodine supplementation with a parallel control group who received no iodine supplementation (or placebo) during pregnancy or the preconceptional period on clinical or biochemical outcomes were eligible for inclusion in the review. Pregnant women or those of childbearing age regardless of iodine status or gestation at trial entry were included. Trials were eligible for inclusion in which women received any form of iodine supplementation, with or without other nutrients, in which the only difference between the treatment and comparison group was the presence or absence of iodine. The primary outcome was the cognitive development of children. Secondary outcomes included pregnancy and birth outcomes, childhood growth and mortality, iodine status, and thyroid function of mothers and infants. The search was restricted to human studies without language restrictions. Reference lists of relevant retrieved publications identified by the search and recent review articles were also checked for additional studies. Titles and abstracts of all articles retrieved by the search were used to assess eligibility by 2 independent reviewers. If there was insufficient information to warrant the exclusion of an article from the abstract, the full text of the article was retrieved to determine eligibility. Any discrepancies were resolved by consensus through discussion or by a third re- viewer." Zhou et al 2013, Pgs. 1-2 • Zhou et al 2013 Supplemental Data 1, Zhou et al 2013 Supplemental Data 2, Zhou et al 2013 Supplemental Data 3 • Zimmermann 2009 • We also examined Zimmermann 2009, a narrative review of iodine deficiency and supplementation for studies on cretinism. Zimmermann 2009 describes 5 controlled interventions in "Controlled interventions in severe deficiency" Zimmermann 2009, Pg. 386. Of these studies, 3 were quasi-randomized or randomized. Only Pharoah, Buttfield and Hetzel 1971 reported treatment effects on cretinism. • "In a landmark trial in an area of severe iodine deficiency in Papua New Guinea (114, 115), alternate families received saline (control) or iodized oil injection. The primary outcome was the prevalence of cretinism at 4- and 10-yr follow-up, with more sensitive diagnostic tests applied at the 10-yr follow-up." Zimmermann 2009, Pg. 386 • "In a study in Zaire, participants were pregnant women attending antenatal clinics in an area of severe iodine deficiency with a 4% cretinism rate (117-119). Pregnant women were randomly allocated to two groups: one received iodized oil injection, the other an injection of vitamins." Zimmermann 2009, Pg. 386 • "In a study in western China, an area of severe iodine deficiency and endemic cretinism, participants were groups of children from birth to 3 yr and women at each trimester of pregnancy (120). Untreated children 13 yr of age, who were studied when first seen, served as controls." Zimmermann 2009, Pg. 386 • "In a randomized Peruvian trial (122, 123), women of childbearing age from three Andean villages in an area of severe iodine deficiency with a 13% cretinism rate were studied." Zimmermann 2009, Pg. 387 • "In two villages in Ecuador with severe iodine deficiency and a cretinism rate of up to 8%, one village received iodine treatment, and one did not and served as an iodine-deficient control (124)." Zimmermann 2009, Pg. 387 • 72. • Table 1, Pharoah, Buttfield and Hetzel 1971, Pg. 591 • "Iodine supplementation was associated with a significant reduction in the prevalence of endemic cretinism: at 4 yr of age, the relative risk (95% CI) was 0.27 (0.12–0.60), and at 10 yr of age, the relative risk (95% CI) was 0.17 (0.05–0.58)." Zimmermann 2009, Pg. 386 • The investigators diagnosing cretinism did know whether the child's mother had received iodine, but it is unclear whether the assessors of squint, deafness and mental function were also blinded. The outcome of cretinism appears relatively objective, so we worry less about blinding. "The parents of any child who showed motor retardation were questioned as to whether they thought the child was deaf or mentally normal. Deafness was formally assessed by noting any response of the child to a tuning-fork, or, if this was negative, the response to a handclap. Formal assessment of mental development was not attempted. A squint, if present, was also noted. The clinical diagnosis of endemic cretinism was accepted if motor retardation was present together with deafness and/or a squint. The diagnosis was made without knowledge of whether a mother had received iodised oil or saline." Pharoah, Buttfield and Hetzel 1971, Pg. 590 • 73. • Analysis 01.03, Mahomed and Gulmezoglu 1997, Pg. 8 • "This paper presents the final follow-up results on the children born between 1966 and 1972. The children were last examined in 1982 when they were aged between 10 and 16 years." Pharoah and Connolly 1987, Pg. 68 • 74. "The controlled trial was started in August to September, 1996, in collaboration with the local ad- ministration, who were due to conduct a census. Twenty-seven villages with a population of 16500 were seen during a walking patrol lasting six weeks. At each village all the people were gathered in order to have a census taken, and the name of every man, woman, and child together with an estimate of their age was noted. Each woman of child- bearing age was asked if she was pregnant; if she answered affirmatively, this was recorded. However, no formal examination to confirm or deny this was carried out. Alternate families were injected with either iodised oil or saline solution, each member receiving 4ml. if aged 12 years or over and 2ml. if under 12 years of age." Pharoah, Buttfield and Hetzel 1971, Pg. 590 • 75. • The first follow-up included only 16 of the original twenty seven villages. The study did not include all of the original villages in the follow-up, nor did it report on how it selected which villages to include. We would guess it examined villages with the highest prevalence of cretinism like in another follow-up study of the population in 1982. • "Subsequent follow-up patrols were carried out in July, 1967, November, 1969, and January, March, and November 1970. Sixteen of the original twenty-seven villages were visited, all mothers and children in each village were assembled and checked against the census sheets." Pharoah, Buttfield and Hetzel 1971, Pg. 590 • This quote from the second follow-up seems to suggest 13 villages were selected instead of 16: "Since 1972 the follow-up of the children in the trial has been limited to five of the original 13 villages which had the highest prevalence of cretinism ranging from 29.3 to 33.3 per 1000 total population. The decision to limit further follow-up to five of the 13 villages was made for logistical reasons. The area in which the villages lie is rugged, mountainous country with no roads and the only means of visiting villages is on foot. In order to spend more time in examining and assessing each child it was necessary to reduce the area covered." Pharoah and Connolly 1987, Pg. 69 • 76. • "With rapid global progress in correcting iodine deficiency, examples of iodine excess are being recognized, particularly when salt iodization is excessive and poorly monitored (20). Tolerance to high doses of iodine is quite variable, and many individuals ingest amounts of several milligrams or more per day without apparent problems. The major epidemiological consequence of iodine excess is iodine-induced hyperthyroidism (IIH) (24,31). This occurs more commonly in older subjects with pre-existing nodular goitres, and may occur even when iodine intake is within the normal range. Iodine intakes above 300 µg/l per day should generally be discouraged, particularly in areas where iodine deficiency has previously existed. In these situations, more individuals may be vulnerable to adverse health consequences, including iodine-induced hyperthyroidism and autoimmune thyroid diseases. In populations characterized by long-standing iodine deficiency and a rapid increase in iodine intake, median value(s) for urinary iodine above 200 µg/l (and in pregnant women, above 250 µg/l) are not recommended because of the possible risk of iodine-induced hyperthyroidism. This adverse condition can occur during the 5 to 10 years following the introduction of iodized salt (24,31). Beyond this period of time, median values up to 300 µg/l have not demonstrated side-effects, at least not in populations with adequately iodized salt." WHO Assessment of Iodine Deficiency Disorders and Monitoring their Elimination 2007, Pg. 34 • "If your thyroid is too active, it makes more thyroid hormones than your body needs. This is called hyperthyroidism." Hyperthyroidism, MedlinePlus • 77. • In addressing the risks of salt iodization, the WHO Assessment of Iodine Deficiency Disorders and Monitoring their Elimination 2007 cites Stanbury et al 1998, a narrative review of the evidence (mostly hospital admission records or from very small-scale, uncontrolled, observational studies) on iodine-induced hyperthyroidism from several countries. • "With rapid global progress in correcting iodine deficiency, examples of iodine excess are being recognized, particularly when salt iodization is excessive and poorly monitored (20). Tolerance to high doses of iodine is quite variable, and many individuals ingest amounts of several milligrams or more per day without apparent problems. The major epidemiological consequence of iodine excess is iodine-induced hyperthyroidism (IIH) (24,31). This occurs more commonly in older subjects with pre-existing nodular goitres, and may occur even when iodine intake is within the normal range. Iodine intakes above 300 µg/l per day should generally be discouraged, particularly in areas where iodine deficiency has previously existed. In these situations, more individuals may be vulnerable to adverse health consequences, including iodine-induced hyperthyroidism and autoimmune thyroid diseases. In populations characterized by long-standing iodine deficiency and a rapid increase in iodine intake, median value(s) for urinary iodine above 200 µg/l (and in pregnant women, above 250 µg/l) are not recommended because of the possible risk of iodine-induced hyperthyroidism. This adverse condition can occur during the 5 to 10 years following the introduction of iodized salt (24,31). Beyond this period of time, median values up to 300 µg/l have not demonstrated side-effects, at least not in populations with adequately iodized salt." WHO Assessment of Iodine Deficiency Disorders and Monitoring their Elimination 2007, Pg. 34 • Stanbury et al 1998 appears to be the primary source upon which the WHO relies, because the other source cited is not as comprehensive: The WHO Assessment of Iodine Deficiency Disorders and Monitoring their Elimination 2007 also cites Todd et al 1995, a letter to the Lancet describing an increase in the incidence of IIH associated with the introduction of iodine supplementation in Zimbabwe: "Increase in thyrotoxicosis associated with iodine supplements in Zimbabwe" Todd et al 1995, Pg. 1563. • "We have critically reviewed the available information on iodine-induced hyperthyroidism (IIH) from published sources and other reports as well as the experience of the authors in Tasmania, Zaire, Zimbabwe, and Brazil." Stanbury et al 1998, Pg. 83 • "24. Todd CH et al. Increase in thyrotoxicosis associated with iodine supplements in Zimbabwe. Lancet, 1995, 346:1563–1564." WHO Assessment of Iodine Deficiency Disorders and Monitoring their Elimination 2007, Pg. 96 • "31. Stanbury JB et al. Iodine-induced hyperthyroidism: occurrence and epidemiology. Thyroid, 1998, 8:83–100." WHO Assessment of Iodine Deficiency Disorders and Monitoring their Elimination 2007, Pg. 96 • "Thyrotoxicosis means an excess of thyroid hormone in the body...Hyperthyroidism, also referred to as an overactive thyroid is the most common cause of thyrotoxicosis and, occurs when your thyroid gland produces too much thyroid hormone." Thyrotoxicosis, Virginia Mason • "Between 1924 and 1928, a sharp rise in hospital admissions and surgical thyroidectomies was observed in several major clinics in the United States. This has been admirably reviewed by Kohn (44,45)." Stanbury et al 1998, Pg. 84 • In Tasmania: "Within 6 months of the introduction of iodized bread, a rise in admissions for thyrotoxicosis was noted independently in the two principal hospitals in the state, the Launceston General Hospital in the northeast and the Royal Hobart Hospital in the south." Stanbury et al 1998, Pg. 86 • In Zaire: "The data showed a sharp rise in patients with thyrotoxicosis seen at the Pararenyatwa Central Hospital in Harare, the principal referral hospital in the country, from 1992 onward." Stanbury et al 1998, Pg. 87 • In Switzerland: "Review of admission records of the Bergerspital, Solothurn, from 1978 to 1990 (catchment area 109,000 persons) disclosed that there was a 27% rise in admissions for thyrotoxicosis in the first year after the iodine content of salt was raised to 15 ppm which rapidly declined to below prior levels (90)." Stanbury et al 1998, Pg. 88 • In Austria: "In 1990 the mean incidence of toxic adenomas rose from approximately 3.3 per hundred to 4.7 in 1992 in the Innsbruck Clinic, and has slowly declined thereafter." Stanbury et al 1998, Pg. 89 • "Endemic goiter is severe in many regions of Zaire, and has been exhaustively studied, both in the Ubangi region of the northwest, and especially in the Kivu region of the northeast. Iodinated oil was successfully used in the Kivu region at doses of only 0.1 and 0.25 mL (480 mg/mL) and IIH was not observed. When iodized salt (1 to 840 ppm, with 25% of samples > 50 ppm), was recently introduced into the Kivu region and was produced and iodized erratically, a striking increase in urinary iodine from 1.6 µg/dL to 24 µ/dL was recorded in 191 subjects who were shown by ultrasonography to have thyroids larger than 60 mL." Stanbury et al 1998, Pg. 87 • "Studies on IIH from Brazil were published in 1984 and 1989. Twenty-three subjects from a region of western Brazil were studied 3 to 8 weeks after arriving in Sao Paulo, where the daily iodine intake was estimated to be between 150 and 200 µg/d. Eight developed mild thyrotoxicosis and had no response of TSH to TRH." Stanbury et al 1998, Pg. 88 • "Four instances of IIH have been described from Argentina. A 46-year-old woman with a large multinodular goiter but without signs or symptoms of abnormal activity of the thyroid was given 1500 µ/g iodine by mouth daily. By the 33rd day, an elevated plasma protein-bound iodine concentration was detected, and by day 78 she was clinically thyrotoxic." Stanbury et al 1998, Pg. 88 • In Nepal: "In a remote area of the Himalayas 14 subjects with goiter were injected with iodinated oil containing 400 mg iodine (96). Blood samples were drawn 4 to 10 days later. Five of this group showed a sharp reduction in serum TSH concentrations and of these, 4 had rises of T4 into the toxic range." Stanbury et al 1998, Pg. 88 • In Ecuador: "In the course of a follow-up survey of approximately 900 persons who had received iodinated oil intramuscularly 6 months earlier, several older women with nodular thyroids were found to have warm skin, in contrast to the usual finding in that cold Andean climate. Only on close questioning did they admit to weight loss and other symptoms suggestive of thyrotoxicosis (97,98)." Stanbury et al 1998, Pg. 88 • 78. • "IIH is most commonly encountered in older persons with long standing nodular goiter and in regions of chronic iodine deficiency, but instances in the young have been recorded...The risks of IIH are principally to the elderly who may have heart disease, and to those who live in regions where there is limited access to medical care." Stanbury et al 1998, Pg. 83 • "Toxic nodular goiter involves an enlarged thyroid gland. The gland contains areas that have increased in size and formed nodules. One or more of these nodules produce too much thyroid hormone." Toxic nodular goiter, MedlinePlus • 79. • 80. • "Within 6 months of the introduction of iodized bread, a rise in admissions for thyrotoxicosis was noted independently in the two principal hospitals in the state, the Launceston General Hospital in the northeast and the Royal Hobart Hospital in the south. The incidence rose from a baseline of 24 patients per 100,000 in 1963 to 40 in 1964 to 1965, to peak at 125 in 1967, in a total population of approximately 370,000. There were minor peaks in 1971, 1977 to 1978 and 1980 coinciding with rises in ambient iodine levels (Fig. 1). The graph represents all cases treated by either Dr. Vidor or Dr. Stewart in their thyroid clinics, private patients, and other patients treated surgically or who received radioiodine. The diagnosis was clinically supported by isotopic or biochemical tests, or both. For inclusion the following criteria were met: clinical diagnosis of thyrotoxicosis supported by positive isotopic or biochemical tests and improvement after treatment. Not included in the graph are cases with a clinical diagnosis but with high normal tests who improved after conservative treatment and patients with 'toxic' tests who were clinically (by history and examination) euthyroid and therefore not treated." Stanbury et al 1998, Pg. 86 • Figure 1, Stanbury et al 1998, Pg. 86 • 81. • "The mortality associated with IIH in the Tasmanian experience is uncertain. The Bureau of Statistics figures based on death certificates suggested that there were 49 deaths that had thyrotoxicosis as one of the diagnoses between 1966 and 1987. Study of 44 case notes retrieved from this period at the Launcheston General Hospital and the Royal Hobart Hospital showed that 21 were thyrotoxic at the time of death." Stanbury et al 1998, Pg. 86 • "The incidence rose from a baseline of 24 patients per 100,000 in 1963 to 40 in 1964 to 1965, to peak at 125 in 1967, in a total population of approximately 370,000." Stanbury et al 1998, Pg. 86 • 82. "The frequency with which IIH occurs depends on a number factors, among them the severity of iodine deficiency that existed before prophylactic iodine was introduced, the magnitude of the incremental rise in iodine intake, the frequency of autonomous elements in the thyroid, the age groups examined, the skills and instruments used in ascertainment, and when or at what intervals the disorder is investigated." Stanbury et al 1998, Pg. 92 • 83. "We estimated incidence rates from 1990 to 1996 for Graves’ disease and toxic nodular goitre among the 109 000 people living in the catchment area of the Bürgerspital, Solothurn. It is the only institution in the area that has a thyroid specialist and facilities for radioiodine studies and treatment. We assumed, therefore, that most cases of hyperthyroidism were seen at least once in this hospital. In 1980, after full correction of iodine deficiency, the yearly incidence of toxic nodular goitre and Graves’ disease per 100 000 rose overall by 27% (figure). After 1980, the incidence of toxic nodular goitre decreased substantially per 100 000 population to 6·9 whereas that of Graves’ disease decreased to only 20·6 (mean of 2 last years). The decrease of Graves’ disease might have been owing to a classification artefact, because of multifocal disseminated autonomy sometimes being mistaken for Graves’ disease.4 Our data suggest that full correction of iodine deficiency leads to a greatly decreased incidence of thyrotoxicosis, in addition to the disappearance of goitre, cretinism, and minor deficiencies of intellect." Burgi, Kohler and Morselli 1998, Pg. 1034 • 84. • ICCIDD conversations with GiveWell in Zurich, April/May 2014 • "In children, excess dietary iodine has been associated with goiter and thyroid dysfunction. In the reports of 'endemic coast goiter' in Hokkaido, Japan (253), the traditional local diet was high in iodine-rich seaweed. UI excretion in children consuming the local diet was approximately 23,000 µg/d. The overall prevalence of visible goiter in children was 3–9%, but in several villages approximately 25% had visible goiter." Zimmermann 2009, Pg. 396 • "Two mechanisms may be responsible for the increase in hypothyroidism in a population where the iodine intake is chronically high. One mechanism is the inhibitory effect of iodine on thyroid hormone synthesis and secretion. This autoregulatory process is thought to protect against thyroid hormone hypersecretion in the face of high iodine intake. However, this autoregulation is not perfect and commonly induces some degree of thyroid hypofunction. In Japanese adults with chronic excess iodine intakes, many with overt hypothyroidism will become euthyroid if their iodine intakes are normalized (277, 278). The other proposed mechanism is induction of thyroid hypofunction due to iodine-induced autoimmune thyroiditis (279), although not all studies agree (82, 83). The frequency of histological thyroiditis in surgical thyroid specimens is increased in an area of endemic goiter after iodine prophylaxis (280)." Zimmermann 2009, Pg. 396 • So, excess iodine intake seems to sometimes lead to hypothyroidism, just as iodine deficiency also leads to hypothyroidism. We do not fully understand the mechanism that causes mental deficiencies in children who are iodine-deficient, but one possible mechanism could be the thyroid not producing enough hormones. Thus, if iodine excess causes hypothyroidism, (where the thyroid is not producing enough hormones) or other effects similar to those that iodine deficiency causes, it may be that iodine excess could also cause mental deficiencies in children. • 85. "In an international sample of 6–12-y-old children (n = 3319) from 5 continents with iodine intakes ranging from adequate to excessive, Tvol [thyroid volume] was measured by ultrasound, and the urinary iodine (UI) concentration was measured. Regressions were done on Tvol and goiter including age, body surface area, sex, and UI concentration as covariates....Chronic iodine intakes approximately twice those recommended—indicated by UI concentrations in the range of 300 – 500 µg/L— do not increase Tvol in children. However, UI concentrations ≥500 µg/L are associated with increasing Tvol." Zimmermann et al 2005, Pg. 840 • 86. • 11 countries out of 148 countries with data have a median UIC classified as excessive (≥300). Table 4, Andersson, Karumbunathan and Zimmermann 2012, Pg. 749 • "Chronic iodine intakes approximately twice those recommended—indicated by UI concentrations in the range of 300 – 500 µg/L— do not increase Tvol in children. However, UI concentrations ≥500 µg/L are associated with increasing Tvol." Zimmermann et al 2005, Pg. 840 • 87. • 88. "The results of early studies were the basis for advocating iodine supplementation to decrease goiter…Iodine supplementation, primarily through the fortification of table salt, did not begin until the early 1920s and occurred initially in Switzerland and the U.S." Leung, Braverman and Pearce 2012, Pg. 1742 • 89. "Recognizing the importance of preventing IDD, the World Health Assembly adopted in 1991 the goal of eliminating iodine deficiency as a public health problem." WHO Assessment of Iodine Deficiency Disorders and Monitoring their Elimination 2007, Pg. 1 • 90. • "Only a few countries, Switzerland, some of the Scandinavian countries, Australia, the US, and Canada, were completely iodine sufficient before 1990. Since then, there has been a major effort to introduce salt iodization as a safe, cost-effective, and sustainable strategy to ensure sufficient intake of iodine in deficient areas. Iodized salt programs are now implemented in many countries worldwide, and two-thirds of the world’s population (71%) is estimated to be covered by iodized salt (3)." Andersson, Karumbunathan and Zimmermann 2012, Pg. 744 • "Large scale salt iodisation is a preferred strategy for control of iodine deficiency disorders and has been implemented in more than 120 countries around the world (UNICEF 2008)." Abudou et al 2014, Pg. 6 • 91. Table 4, Andersson, Karumbunathan and Zimmermann 2012, Pg. 749 • 92. • "Assessment of thyroid size by palpation is the time-honoured method of assessing IDD prevalence. However, because of the lack of sensitivity to acute changes in iodine intake, this method is of limited usefulness in assessing the impact of programmes once salt iodization has commenced. In this case, urinary iodine is the most useful indicator because it is reflective of the current intake of iodine in the diet (23). Since most countries have now started to implement IDD control programmes, urinary iodine rather than thyroid size is emphasized in this manual as the principal indicator of impact. Thyroid size is more useful in baseline assessments of the severity of IDD, and also has a role in the assessment of the long-term impact of control programmes." WHO Assessment of Iodine Deficiency Disorders and Monitoring their Elimination 2007, Pg. 28 • "The specificity and sensitivity of palpation are low in grades 0 and 1 due to a high inter-observer variation. As demonstrated by studies of experienced examiners, misclassification can be high." WHO Assessment of Iodine Deficiency Disorders and Monitoring their Elimination 2007, Pg. 36 • "In areas of mild to moderate IDD, the sensitivity and specificity of palpation are poor and measurement of thyroid size using ultrasound is preferable." WHO Assessment of Iodine Deficiency Disorders and Monitoring their Elimination 2007, Pg. 37 • 93. "Thyroid size is more useful in baseline assessments of the severity of IDD, and also has a role in the assessment of the long-term impact of control programmes." WHO Assessment of Iodine Deficiency Disorders and Monitoring their Elimination 2007, Pg. 28 • 94. • 95. • Note that Abudou et al 2014 did not include studies without a baseline before salt iodization: "Multiple cross-sectional studies without a baseline before iodisation were not included in the current review. There are publications of data from monitoring of iodisation programs that report change in health outcomes with changes in the level of iodine in salt. These studies, if included, would provide more information on the effectiveness of varying levels of iodine in salt. This information would be particularly important as countries strengthen efforts to decrease population overall salt intake." Abudou et al 2014, Pg. 46 • 38 studies reporting on urinary iodine: Aghini-Lombardi 1993, Aziz 2002, Baczyk 2007, Bauch 1990, Bimenya 2002, Chen 1999, Chen 2001, Chen 2002a, Dai 2008, Fei 1996, Foo 1996, Gatti 1980, Guo 1984, Han 2006, Heydarian 2007, Hintze 1998, Hou 2003, Jia 2004, Jooste 2000, Kimiagar 1990, l'Ons 2000, Lv 2009, Pongpaew 1998, Regalbuto 2010, Romano 1991, Saowakhontha 1994a, Szybinski 2001, Tang 1992, Tazhibayev 2008, Wang 2000, Wang 2009, Weber 1987, Yang 1984a, Yang 1984b, Yang 2011, Zhou 2004, Zimmermann 2003, Zimmermann 2004b • "Urinary iodine excretion was reported as population mean value or population median value. Some studies reported ug Iodine/ L urine (ug/L), other studies reported ug Iodine / g Creatinine (ug/gCr), and still others reported ug Iodine / 24 hours (ug/24hr). Meta-analyses were conducted when studies reported mean and a measure of variance and were done separately for studies reporting ug/L and those reporting ug/gCr. The results of other studies were compiled in summary tables (Table 3; Table 4; Table 5). Urinary iodine excretion was also used to report population severity of iodine deficiency indicated by percentage of the population with a urinary iodine excrtion below 100ug/L." Abudou et al 2014, Pg. 35 • Figures 13-17, Abudou et al 2014, Pgs. 301-302 • Hintze 1998, l'Ons 2000, Pongpaew 1998, Romano 1991, Saowakhontha 1994a. "3 Summary Table: Urinary iodine excretion in RCT and non RCTs", Abudou et al 2014, Pg. 215 • Foo 1996, Guo 1984, Weber 1987, Yang 1984a, Yang 1984b, Zimmermann 2004b. "4 Summary Table: Urinary iodine excretion in observational cohort studies", Abudou et al 2014, Pg. 216 • Bacyzk 2007, Chen 1999, Chen 2001, Chen 2002a, Dai 2008, Han 2006, Hou 2003, Jooste 2000, Lv 2009, Regalbuto 2010, Wang 2000, Wang 2009, Yang 1984a, Yang 2011, Zhou 2004, Zimmermann 2003. "5 Summary Table: Urinary iodine excretion in cross-sectional studies", Abudou et al 2014, Pgs. 217-218 • Kimiagar 1990, Tang 1992, Tazhibayev 2008. Figure 13, "Cohort: Urinary iodine excretion (ALL µg/L)", Abudou et al 2014, Pg. 301 • Tazhibayev 2008, Zimmermann 2004b. Figure 14, "Cohort: Iodine deficiency measured by low UIE (ALL)", Abudou et al 2014, Pg. 301 • Aziz 2002, Fei 1996, Heydarian 2007, Jia 2004, Szybinski 2001. Figure 15, "Multiple cross-sectional: Urinary iodine (µg/L ALL)", Abudou et al 2014, Pg. 301 • Aghini-Lombardi 1993, Bauch 1990, Gatti 1980. Figure 16, "Multiple cross-sectional: Urinary iodine (µg/gCr ALL)", Abudou et al 2014, Pg. 302 • Baczyk 2007, Bimenya 2002, Chen 2001, Fei 1996, Jooste 2000, Lv 2009, Szybinski 2001, Wang 2009, Yang 2011, Zimmermann 2003. Figure 17, "Multiple cross-sectional: Iodine deficiency measured by low UIE (ALL)", Abudou et al 2014, Pg. 302 • 96. • "After goitre, urinary iodine excretion was the outcome measured in the most number of studies. This outcome was measured in different units and was reported in many different ways, making it impossible to do one overall meta-analysis with this outcome. Nonetheless, the results of the different studies were almost entirely consistent with an increase in urinary iodine excretion with the introduction of iodised salt. This result was found in all study types and was consistent across age groups, concentration of iodine in salt, or baseline risk of IDD based on goitre prevalence of urinary iodine excretion." Abudou et al 2014, Pg. 41 • "The urinary iodine excretion body of evidence was of moderate quality. All data were from cohort and multiple cross-sectional studies; however, all studies reported large effects of salt iodisation on mean urinary iodine excretion or prevalence of low urinary iodine excreation (UIE < 100 ug/L) in the population. Additionally there were more than 15 studies that could not be combined in the meta-analyses which had results consistent with a positive effect of iodised salt on urinary iodine excretion." Abudou et al 2014, Pg. 45 • Figures 13-17, Abudou et al 2014, Pgs. 301-302 • "3 Summary Table: Urinary iodine excretion in RCT and non RCTs", Abudou et al 2014, Pg. 215 • "4 Summary Table: Urinary iodine excretion in observational cohort studies", Abudou et al 2014, Pg. 216 • "5 Summary Table: Urinary iodine excretion in cross-sectional studies", Abudou et al 2014, Pgs. 217-218 • 97. • "Assessment of thyroid size by palpation is the time-honoured method of assessing IDD prevalence. However, because of the lack of sensitivity to acute changes in iodine intake, this method is of limited usefulness in assessing the impact of programmes once salt iodization has commenced. In this case, urinary iodine is the most useful indicator because it is reflective of the current intake of iodine in the diet (23). Since most countries have now started to implement IDD control programmes, urinary iodine rather than thyroid size is emphasized in this manual as the principal indicator of impact. Thyroid size is more useful in baseline assessments of the severity of IDD, and also has a role in the assessment of the long-term impact of control programmes." WHO Assessment of Iodine Deficiency Disorders and Monitoring their Elimination 2007, Pg. 28 • "Studies have convincingly demonstrated that a profile of iodine concentrations in morning or other casual urine specimens (child or adult) provides an adequate assessment of a population’s iodine nutrition, provided a sufficient number of specimens are collected. Round the clock urine samples are difficult to obtain and are not necessary. Relating urinary iodine to creatinine, as has been done in the past, is cumbersome, expensive, and unnecessary. Indeed, urinary iodine/ creatinine ratios are unreliable, particularly when protein intake – and consequently creatinine excretion – is low." WHO Assessment of Iodine Deficiency Disorders and Monitoring their Elimination 2007, Pg. 29 • "The median value for the sampled population is the most commonly assessed indicator. Urinary iodine values from populations are usually not normally distributed. Therefore, the median rather than the mean should be used as the measure of central tendency." WHO Assessment of Iodine Deficiency Disorders and Monitoring their Elimination 2007, Pg. 32 • "UI is an excellent indicator of recent iodine intake because ≥92% of dietary iodine is absorbed and, in healthy, iodine-replete adults, >90% is excreted in the urine within 24–48 h.27,28 UI can be expressed as a 24-h excretion (UIE; mg/day), as a concentration (UIC; mg/L), or in relationship to creatinine excretion (mg iodine/g creatinine). But these are not interchangeable, as discussed below. Because it is impractical to collect 24-h samples in field studies, UICs are usually measured in spot urine collections. In healthy well-nourished adults, daily creatinine excretion is fairly constant at about one gram, so expressing the UIE from spot samples in adults as mg iodine/g creatinine approximates the value in a 24-h collection and reduces variation due to hydration status.29 But in malnourished populations with poor protein intakes, daily creatinine excretion is more variable and often lower than 1 g.30 In these settings, expressing the UIE as mg iodine/g creatinine may introduce greater variation. Due to these limitations and the additional expense of measuring creatinine, the routine co-measurement of creatinine fell out of favor and was replaced by the expression of UIC in mg/L. If a large number of samples are collected, variations in hydration among individuals and day-to-day variation in iodine intake generally even out, so that the median UIC in spot samples correlates well with the median from 24-h samples and with the estimated UIE from creatinine- corrected UICs.31" Zimmermann and Andersson 2012, Pg. 556 • The WHO uses median UIC expressed in µg/L to classify the iodine status of a population: Tables 4-5, WHO Assessment of Iodine Deficiency Disorders and Monitoring their Elimination 2007, Pg. 33 • 98. • 99. • 4 studies: Tazhibayev et al 2008, Baczyk et al 2007, Jooste, Weight and Lombard 2000, Zimmermann et al 2003 • 38 studies reporting on urinary iodine: Aghini-Lombardi 1993, Aziz 2002, Baczyk 2007, Bauch 1990, Bimenya 2002, Chen 1999, Chen 2001, Chen 2002a, Dai 2008, Fei 1996, Foo 1996, Gatti 1980, Guo 1984, Han 2006, Heydarian 2007, Hintze 1998, Hou 2003, Jia 2004, Jooste 2000, Kimiagar 1990, l'Ons 2000, Lv 2009, Pongpaew 1998, Regalbuto 2010, Romano 1991, Saowakhontha 1994a, Szybinski 2001, Tang 1992, Tazhibayev 2008, Wang 2000, Wang 2009, Weber 1987, Yang 1984a, Yang 1984b, Yang 2011, Zhou 2004, Zimmermann 2003, Zimmermann 2004b • "Urinary iodine excretion was reported as population mean value or population median value. Some studies reported ug Iodine/ L urine (ug/L), other studies reported ug Iodine / g Creatinine (ug/gCr), and still others reported ug Iodine / 24 hours (ug/24hr). Meta-analyses were conducted when studies reported mean and a measure of variance and were done separately for studies reporting ug/L and those reporting ug/gCr. The results of other studies were compiled in summary tables (Table 3; Table 4; Table 5). Urinary iodine excretion was also used to report population severity of iodine deficiency indicated by percentage of the population with a urinary iodine excrtion below 100ug/L." Abudou et al 2014, Pg. 35 • Figures 13-17, Abudou et al 2014, Pgs. 301-302 • Hintze 1998, l'Ons 2000, Pongpaew 1998, Romano 1991, Saowakhontha 1994a. "3 Summary Table: Urinary iodine excretion in RCT and non RCTs", Abudou et al 2014, Pg. 215 • Foo 1996, Guo 1984, Weber 1987, Yang 1984a, Yang 1984b, Zimmermann 2004b. "4 Summary Table: Urinary iodine excretion in observational cohort studies", Abudou et al 2014, Pg. 216 • Bacyzk 2007, Chen 1999, Chen 2001, Chen 2002a, Dai 2008, Han 2006, Hou 2003, Jooste 2000, Lv 2009, Regalbuto 2010, Wang 2000, Wang 2009, Yang 1984a, Yang 2011, Zhou 2004, Zimmermann 2003. "5 Summary Table: Urinary iodine excretion in cross-sectional studies", Abudou et al 2014, Pgs. 217-218 • Kimiagar 1990, Tang 1992, Tazhibayev 2008. Figure 13, "Cohort: Urinary iodine excretion (ALL µg/L)", Abudou et al 2014, Pg. 301 • Tazhibayev 2008, Zimmermann 2004b. Figure 14, "Cohort: Iodine deficiency measured by low UIE (ALL)", Abudou et al 2014, Pg. 301 • Aziz 2002, Fei 1996, Heydarian 2007, Jia 2004, Szybinski 2001. Figure 15, "Multiple cross-sectional: Urinary iodine (µg/L ALL)", Abudou et al 2014, Pg. 301 • Aghini-Lombardi 1993, Bauch 1990, Gatti 1980. Figure 16, "Multiple cross-sectional: Urinary iodine (µg/gCr ALL)", Abudou et al 2014, Pg. 302 • Baczyk 2007, Bimenya 2002, Chen 2001, Fei 1996, Jooste 2000, Lv 2009, Szybinski 2001, Wang 2009, Yang 2011, Zimmermann 2003. Figure 17, "Multiple cross-sectional: Iodine deficiency measured by low UIE (ALL)", Abudou et al 2014, Pg. 302 • Type of study for studies reporting on urinary iodine • Cross-sectional studies reporting median UIC in units of µg/L: Baczyk 2007, Chen 1999, Chen 2001, Chen 2002a, Dai 2008, Han 2006, Hou 2003, Jooste 2000, Lv 2009, Wang 2000, Wang 2009, Yang 2011, Zhou 2004, Zimmermann 2003 • Excluded Yang 1984a: reported in units of µg iodine / g creatinine • Excluded Regalbuto 2010: reported mean, not median • "There were 14 multiple cross-sectional studies that reported urinary iodine excretion as median value before and after iodised salt consumption and one study that reported mean and standard deviation without a sample size. The results of the 15 [16 studies are in the table] studies compiled in a summary table (Table 5). In all comparisons in all studies, the urinary iodine excretion increased with iodised salt intake." Abudou et al 2014, Pg. 37 • "5 Summary Table: Urinary iodine excretion in cross-sectional studies", Abudou et al 2014, Pg. 217 • Cross-sectional studies that did not directly provide iodized salt and reported median UIC in units of µg/L: Baczyk 2007, Chen 1999, Chen 2002a , Dai 2008, Han 2006, Hou 2003, Jooste 2000, Lv 2009, Wang 2000, Wang 2009, Yang 2011 (not sure based on the description, but we'd guess that this study did not directly provide iodized salt), Zhou 2004 (not sure based on the description, but we'd guess that this study did not directly provide iodized salt), Zimmermann 2003. Chen 2001 directly provided iodized salt. • Baczyk 2007: "Multiple cross-sectional observational study conducted in Wielkopolska Region, Poland. Cross-sectional surveys of goitre, urinary iodine excretion, and adverse effects were conducted before obligatory salt iodisation in Poland and 9 years after obligatory salt iodisation in Poland." Abudou et al 2014, Pg. 53 • Chen 1999: "A survey was conducted in 1995 before the availability of iodized salt. Salt then became iodized throughout the province at a concentration of 33ppm. There was then another survey conducted with the same methods in 1997" Abudou et al 2014, Pg. 66 • Chen 2001: "A survey in the school children was conducted in 1995 before the availability of iodized salt. In 1996 the government started providing families iodizated salt on a monthly basis. A second survey was conducted in the same schools in 1997. The concentration of Iodine in the salt was 33ppm (1/30000)." Abudou et al 2014, Pg. 67 • Chen 2002a: "Primary school children aged 8 - 10 years were randomly selected for inclusion in the baseline survey of goitre before the introduction of iodised salt. A follow-up survey was conducted three years after the introduction of iodised salt and goitre was again measured in randomly sampled school children...Two methods were used to introduce Iodised salt in two separate provinces. In Hui An County (HAC) the government distriburted iodized salt on the basis of 6kg/person/year and non-iodized salt was taken completely off the market. In Quan Gang Regioin (QGR) iodized salt was introduced in the market and non- iodized was still allowed to be in the market as well." Abudou et al 2014, Pg. 68 • Dai 2008: "Multiple cross sectional study conducted in Xiamen City, Fujian province, China. School children aged 8-10 years were randomly sampled from 90 schools before supplementation of iodised salt in 1995. Goitre and urinary iodine excretion were measured in these children. Another similar cross- sectional survey was conducted 12 years after introduction of iodised salt....Comprehensive intervention used in local salt mines and manufactoring facilities to all salt on the market. Coverage was reported at 90% at follow-up survey." Abudou et al 2014, Pg. 71 • Han 2006: "Multiple Cross-Sectional study conducted in Tianjin, China. The participants were sampled by Probability Proportionate to Size Sampling (PPS) according to the population. Samples were taken from 30 schools. One hundred and fifty housewives were also sampled though selection process was not described...A survey was conducted in 1995, then in 1996 unversial salt iodization was implemented. A follow-up study was conducted in 2005." Abudou et al 2014, Pg. 87 • Hou 2003: "The first survey was conducted in 1986, iodized salt was introduced in 1993 and USI achieved in 1996. The follow-up survey was conducted in 1999" Abudou et al 2014, Pg. 94 • Jooste 2000: "The baseline survey measured goitre and UIE during the month before USI was begun in S. Africa. The follow-up survey occurred one year later using the same sampling technique." Abudou et al 2014, Pg. 102 • Lv 2009: "Multiple cross-sectional observational study conducted in Hebei province, China. Probability proportionate to size sampling (PPS) technique wass used to select the participants for a survey on Urinary iodine excretion and goitre. 1200 school children aged 8-10 years were selected from 30 schools (40 children in each school). In the same time, salt samples were collected from those children's households. 360 urinary specimens were randomly selected from the 1200 children. The same sampling occurred at the baseline survey conducted before the provision of iodised salt and at the follow-up survey 10 years after the provision of iodised salt through mandatory means in the province. Baseline survey conducted in 1995 then USI implemented. A follow-up survey was conducted in 2005." Abudou et al 2014, Pg. 111 • Wang 2000: "In two localities, a survey was conducted in 1995 before implementation of universal salt iodization and USI was implemented in 1996 with a follow-up survey in 1997." Abudou et al 2014, Pg. 153 • Wang 2009: "Baseline survey conducted before USI implemented and follow-up survey conducted 10 years after implementation of USI." Abudou et al 2014, Pg. 157 • Yang 2011: "Multiple cross-sectional observational study conducted in China. Randomly selected a primary school, 40~100 schoolchildren, 8~10 years of age. The first survey was conducted before the provision of iodised salt in the community and the follow-up surveys, conducted yearly, were conducted up to 15 years after the initiation of iodisation of salt in the community. The data includes every year from 1995 to 2009, we compared 1995 and 2009 data only in this review...Iodised salt provided at a concentration of 35 ppm through provision at the population level." Abudou et al 2014, Pgs. 167-168 • Zhou 2004: "Randomly sampled students from one primary school. A survey of students 8-10 years of age was conducted before iodisation of salt and then another sample of students were surveyed 8 years after introduction of iodised salt." Abudou et al 2014, Pg. 176 • Zimmermann 2003: "Multiple cross sectional observatoinal study, conducted in 6 remote villages in the Danane Health District of western Cote d'Ivoire. Surveys of goitre and other indicators of iodine status were conducted in schools of the villages. All children, 5 - 14 years of age, who were present at school on the day of measurement were included in the sample. The first survey was conducted in 1997 before iodised salt was available in the villages and children from two schools were included. In 1998, universal salt iodisation was initiated. Other surveys using similar methods were conducted in school children in the same villages yearly until 2001." Abudou et al 2014, Pg. 180 • Cross-sectional studies that did not directly provide iodized salt and reported median UIC in units of µg/L and were published in English and had baseline median UIC less than 100 µg/L: Baczyk 2007, Jooste 2000, Zimmermann 2003. Excluded because published in Chinese (we haven't obtained translations; we may revisit these studies in the future): Chen 1999, Chen 2002a, Dai 2008, Han 2006, Hou 2003, Wang 2000, Yang 2011, Zhou 2004. Excluded because baseline median UIC greater than or equal to 100 µg/L: Lv 2009, Wang 2009 • Table 5, Abudou et al 2014, Pg. 217 • Baczyk 2007: "The report was published in peer reviewed journal in details English." Abudou et al 2014, Pg. 54 • Chen 1999: "Published in Chines language journal" Abudou et al 2014, Pg. 66 • Chen 2002a: "The report was published in Chinese language journal." Abudou et al 2014, Pg. 69 • Dai 2008: "The report was published in a peer reviewed journal in Chinese." Abudou et al 2014, Pg. 72 • Han 2006: "Published in Chinese language journal." Abudou et al 2014, Pg. 87 • Hou 2003: "Published in Chinese language journal" Abudou et al 2014, Pg. 94 • Jooste 2000: "The report was published in a peer reviewed journal in English." Abudou et al 2014, Pg. 102 • Lv 2009: "The report was published in a peer reviewed journal in English." Abudou et al 2014, Pg. 112 • Wang 2000: "Published in Chinese language journal" Abudou et al 2014, Pg. 153 • Wang 2009: "Published in peer-reviewed journal in English." Abudou et al 2014, Pg. 158 • Yang 2011: "The report was published in a peer reviewed journal in Chinese." Abudou et al 2014, Pg. 168 • Zhou 2004: "The report was published in a peer reviewed journal in Chinese." Abudou et al 2014, Pg. 176 • Zimmermann 2003: "The report was published in a peer reviewed journal in English." Abudou et al 2014, Pg. 181 • Cohort studies that did not directly provide iodized salt and reported median UIC in units of µg/L: Tazhibayev 2008. Tang 1992 and Jia 2004 also did not directly provide iodized salt. Jia 2004 appears not to have reported a median UIC. We are not sure if Tang 1992 reported median UIC. Both studies are in Chinese and we haven't translated them. • All cohort studies reporting urinary iodine: Foo 1996, Guo 1984, Jia 2004, Kimiagar 1990, Tang 1992, Tazhibayev 2008, Weber 1987, Yang 1984a, Yang 1984b, Zimmermann 2004b. (See bullet point above on "Type of study for studies reporting on urinary iodine") • Guo 1984, Yang 1984a, Yang 1984b report median UIC in units of µg iodine / g creatinine (we are assuming that Abudou et al 2014 would have reported the median UIC (µg/L) in the table if the studies had reported it). "4 Summary Table: Urinary iodine excretion in observational cohort studies", Abudou et al 2014, Pg. 216 • At least Foo 1996, Kimiagar 1990, Weber 1987, Zimmermann 2004b provided iodized salt: "The cohort observational studies followed the same individuals from before salt iodisation to after salt was iodised and consumed by those individuals. In one study, all women and young children living in selected longhouse villages in Malaysia were followed over time (Foo 1996) from before the provision of iodised salt and as they were provided iodised salt. In four studies, participants were randomly sampled from the population and followed over time (Golkowski 2007; Ibanez Gonzalez 1956; Tazhibayev 2008; Wang 1985b) from pre-availability of iodised salt in the population to post availability of iodised salt in the community. In two studies, households were selected to receive iodised salt for household consumption and all members of the household were followed over time (Kimiagar 1990; Weber 1987). In another study, school students were included in the study and provided iodised salt for their homes and followed over time (Zimmermann 2004b). In seven studies all inhabitants of villages, towns or some other geographic area were sampled before USI and followed over time as USI was implemented in the area (Chen 1984; Guo 1984; Pedersen 2002; Tang 1992; Xu 1984; YANG 1984b; Zhang 1988a), and in one study all students in a village were measured before USI and followed as USI was introduced in the community (Wang 2001)." Abudou et al 2014, Pgs. 20-21 • "Median UIE (µg/L)" Table 5, Tazhibayev et al 2008, Pg. 262 • Tazhibayev 2008 did not directly provide iodized salt: "Background. A project for universal salt iodation with potassium iodate and wheat flour fortification with a vitamin-mineral premix was implemented in Azerbaijan, Kazakhstan, Kyrgyzstan, Mongolia, Tajikistan, and Uzbekistan between 2002 and 2007.Objective. To determine the potential effectiveness of the food fortification programs in improving the micronutrient status of selected families in a sentinel population in each country." Tazhibayev et al 2008, Pg. 255 • Tang 1992: "Published in Chinese language journal" Abudou et al 2014, Pg. 136 • Jia 2004 was also a cross-sectional study. Because Table 5 did not include it, we assume it did not report median UIC: "There were 14 multiple cross-sectional studies that reported urinary iodine excretion as median value before and after iodised salt consumption and one study that reported mean and standard deviation without a sample size. The results of the 15 [16 studies are in the table] studies compiled in a summary table (Table 5). In all comparisons in all studies, the urinary iodine excretion increased with iodised salt intake." Abudou et al 2014, Pg. 37; "5 Summary Table: Urinary iodine excretion in cross-sectional studies", Abudou et al 2014, Pg. 217 • Jia 2004 did not directly provide iodized salt: "Cohort observational study and quasi-experimental, multiple cross-sectional study conducted in Heshui county, Gansu province, China. A survey was conducted of all residents of Heshui county in 1990. In that year USI was implemented. Another study of all residents was conducted in 2000. The results of all persons over age of 10 in 2000 are presented as cohort study results because they were measured at both time points. Individuals aged 0 - 10 in 1990 are comparison group in quasi-experiment as they had no exposure to iodized salt and the intervention group are those children 0-10 at follow-up because they did have exposure to iodized salt. The measure of urinary iodine excretion was only reported for 7-14 year olds and is a multiple-cross sectional study...Iodised salt provided through USI at an unknown concentration." Abudou et al 2014, Pg. 101 • Jia 2004: "The report was published in a peer reviewed journal in Chinese." Abudou et al 2014, Pg. 101 • No studies besides cohort and cross-sectional studies that did not directly provide iodized salt • Remaining studies: Hintze 1998, Romano 1991, l'Ons 2000, Pongpaew 1998, Saowakhontha 1994a (See bullet point above on "Type of study for studies reporting on urinary iodine") • Hintze 1988: "All school children were invited to participate. Of the 1160 parents contacted, nearly 50% did not respond, 227 did not wish to participate and the parents of a total of 334 boys and girls agreed to the study and allowed their children to participate. Measurements were taken in the schools of body weight and height, goitre classified by WHO standards, and neck circumference. After baseline measurements students were randomly assigned to group A or group B. Randomization method not explained. Group A was assigned to buy from the market and consume only iodised salt while Group B was assigned to buy from the market and consume only non-iodised salt. Groups were not blinded to treatment but assessors were blinded. Follow-up measurements of goitre were undertaken at 2 and 4 years and UIE was measured yearly. The final 4 year follow-up measure is used in this review." Abudou et al 2014, Pg. 92 • Romano 1991: "Randomised controlled trial conducted in Italy. Pregnant women were randomly selected from a health care facility. Women were first stratified by smoking status and then randomly assigned to either group A (iodised salt) or group B (non-iodised salt). Each trimester, measurements were taken at the clinic. Units of comparison: Individuals...1. Iodised salt provided at 20ppm (N=17) 2. Control - plain salt, but not specifically distributed (N=18)" Abudou et al 2014, Pg. 122 • l'Ons 2000: "Non-randomised controlled study conducted in South Africa. Non-random controlled trial where two schools were selected for convenience and because there was no iodized salt available in the region where the schools were located. All children in both schools participated. Goitre, measured by palpation and defined using WHO criteria, and UIE were measured at baseline and after a 4-month follow-up period. Children in the experiment school were all given iodised salt for use in the household and children in the control school were given non-iodized salt for use in the household." Abudou et al 2014, Pg. 108 • Pongpaew 1998: "Non-randomized controlled trial conducted in Thailand. Concurrent controlled study conducted in four schools in Thailand. Four schools were randomly selected to participate in a baseline study of iodine status and status of other nutritional indicators. After the baseline study, each of the schools was assigned to one of three interventions (iodinated salt, iodised water, iodised fish sauce) or control (no intervention). In the iodised salt intervention arm, iodised salt was given to the students on a regular basis for use at all times. The participants were measured again after one year of intervention. The data from the iodised salt intervention school and the control school were compared in this review" Abudou et al 2014, Pg. 119 • Saowakhontha 1994a: "Non-randomized, controlled study conducted in four villages of Thailand. The four villages sampled in the study were randomly selected from 12 villages. The participants were randomly selected from the inhabitants of the villages. One of three interventions (fish sauce, iodised salt, iodised water) or control were applied to each village. All participants in each village received the intervention assigned to his/her village. The iodised salt group received the salt at the household on a regular basis. The control group received no intervention. The data used in this analysis was that from the iodised salt and the control groups." Abudou et al 2014, Pgs. 126-127 • 100. "In late 1997, Côte d’Ivoire legislated mandatory USI at a production level of 30–50 ppm. In February–March 1998, iodized salt was introduced into the Danané region. By 1999, it was estimated that > 80% of Ivorian households had access to iodized salt at a market level of 20–30 ppm (P Adou, unpublished data, 2000). The present study was done from 1997 through 2001." Zimmermann et al 2003, Pg. 664 • 101. • "The study was done in 6 remote villages in the Danané Health District, a mountainous region of western Côte d’Ivoire. The villages are located within an ~10-km radius in dense forest and have no electricity or running water. Most families are engaged in small-scale subsistence farming. The staple foods are rice and cassava. During the 5-y study period, the quantity and quality of local harvests were stable. The villages are similar ethnically and socioeconomically. Before the introduction of USI, the GR by pal- pation in western and northern Côte d’Ivoire was 40–60% (13)." Zimmermann et al 2003, Pg. 664 • "The subjects were schoolchildren recruited from 6 primary schools. The study visits were done in the same month (November) in the midst of the dry season for 5 consecutive years. All children aged 5–14 y attending school on days when the fieldwork was done were measured. School attendance is only sporadic in this region, so samples from the 5 y varied in size. Children were recruited from 2 schools in 1997 and 1998 and all 6 schools in 1999–2001." Zimmermann et al 2003, Pg. 664 • 102. "The subjects were schoolchildren recruited from 6 primary schools. The study visits were done in the same month (November) in the midst of the dry season for 5 consecutive years. All children aged 5–14 y attending school on days when the fieldwork was done were measured. School attendance is only sporadic in this region, so samples from the 5 y varied in size. Children were recruited from 2 schools in 1997 and 1998 and all 6 schools in 1999–2001." Zimmermann et al 2003, Pg. 664 • 103. • "In late 1997, Côte d’Ivoire legislated mandatory USI at a production level of 30–50 ppm. In February–March 1998, iodized salt was introduced into the Danané region. By 1999, it was estimated that > 80% of Ivorian households had access to iodized salt at a market level of 20–30 ppm (P Adou, unpublished data, 2000). The present study was done from 1997 through 2001." Zimmermann et al 2003, Pg. 664 • "The subjects were schoolchildren recruited from 6 primary schools. The study visits were done in the same month (November) in the midst of the dry season for 5 consecutive years. All children aged 5–14 y attending school on days when the fieldwork was done were measured. School attendance is only sporadic in this region, so samples from the 5 y varied in size. Children were recruited from 2 schools in 1997 and 1998 and all 6 schools in 1999–2001." Zimmermann et al 2003, Pg. 664 • 104. "The subjects were schoolchildren recruited from 6 primary schools. The study visits were done in the same month (November) in the midst of the dry season for 5 consecutive years. All children aged 5–14 y attending school on days when the fieldwork was done were measured. School attendance is only sporadic in this region, so samples from the 5 y varied in size. Children were recruited from 2 schools in 1997 and 1998 and all 6 schools in 1999–2001." Zimmermann et al 2003, Pg. 664 • 105. Table 1, Zimmermann et al 2003, Pg. 665 • 106. "Mandatory iodization of household salt was introduced in South Africa through revised legislation in December 1995. The salt-related regulations of the Foodstuffs, Cosmetics and Disinfectants Act No. 54 of 1972 were revised to make iodization mandatory rather than optional and to increase the concentration of iodine in the form of potassium iodate from 10–20 to 40–60 µg/g. As a result of this new regulation, the availability of iodized salt in food shops was expected to increase from ~30% (7) to > 90% within 6 mo." Jooste, Weight and Lombard 2000, Pg. 75 • 107. "The study was carried out in the primary schools of 4 communities [Haarlem (school 1), Louterwater (school 2), Krakeel (school 3), and Joubertina (school 4)] in the Langkloof area, a 150-km-long fruit-producing valley ~70 km inland from the southeastern coast of South Africa. These 4 communities were chosen because the Langkloof area had been a focal point of studies on endemic goiter in the past (10). In addition, the socioeconomic status of the 4 communities ranged from low to high and there were sufficient numbers of schoolchildren. Children in grades 4–7 (ie, with 4–7 years of schooling) attending primary schools in the 4 communities, situated over a distance of ~100 km in the Langkloof area, were used as subjects in both the baseline and follow-up studies." Jooste, Weight and Lombard 2000, Pgs. 75-76 • 108. • 109. "An intensified iodine prophylaxis project based on an obligatory model of salt iodization (30 mg KI/kg) was introduced in Poland - a country with mild/moderate iodine deficiency - in 1996." Baczyk et al 2007, Pg. 511 • 110. "This study was carried out in the Wielkopolska Region (the lowlands of the western part of Poland) in a group of children between 8 and 12 years of age. The study covered a total of 1,215 children, who came not only from urban areas but also from rural areas, and were represented by a proportional division into sex and age groups. Four hundred and two children were examined in 1992, before salt iodization, and 408 in 2000 and 405 in 2005, after salt iodization was implemented in 1996. Almost all children were prepubertal. The population was stable, and there was no significant migration." Baczyk et al 2007, Pg. 511 • 111. Table 1, Baczyk et al 2007, Pg. 513 • 112. "A high-level policy agreement of the Commonwealth of Independent States in Minsk, Byelorussia, 31 May 2001, established a uniform level of salt iodination at 40 parts per million (ppm) of potassium iodate. The vitamin-mineral premix KAP-complex #1, which consists of six micronutrients (electrolytic iron – 34.7%, zinc oxide – 18.7%, niacinamide – 6.7%, riboflavin – 2%, thiamin mononitrate – 1.3%, folic acid – 1%, and stuff [silicon dioxide 0.5 to 1.5% and calcium sulfate q.s. to make 100%] – 35.6%), was developed by the Kazakh Academy of Nutrition (KAN), and approved by all participating governments. The premium grade wheat flour (extraction rate 55–60%) was fortified with 150 g and the first grade wheat flour (extraction rate up to 72%) with 120 g of KAP Complex #1 per metric ton (MT) and supplied with the following ppm of micronutrients: 50 – iron, 22 – zinc, 10 – niacin, 3.0 – riboflavin, 2.0 – thiamin, and 1.5 – folic acid in premium grade; and 40 – iron, 17.6 – zinc, 8 – niacin, 2.4 – riboflavin, 1.6 – thiamin, and 1.2 – folic acid in first grade flour." Tazhibayev et al 2008, Pg. 256 • 113. "To determine the potential effectiveness of the food fortification programs in improving the micronutrient status of selected families in a sentinel population in each country…A project for universal salt iodation with potassium iodate and wheat flour fortification with a vitamin-mineral premix was implemented in Azerbaijan, Kazakhstan, Kyrgyzstan, Mongolia, Tajikistan, and Uzbekistan between 2002 and 2007...An area was selected in each country in a sentinel population expected to have early access to iodated salt and fortified wheat flour." Tazhibayev et al 2008, Pg. 255 • 114. "The study was designed to evaluate the effectiveness of the JFPR Program implementation in a favorable pilot region in each country. It was limited to 40 households and 120 individuals (80 children and 40 women) in each country. The baseline round of data collection was carried out from December 2002 to April 2003 before implementation of the food fortification programs. Within each area, 40 families with at least one woman of reproductive age and two children between 2 and 15 years of age were selected. A list of families meeting these criteria was prepared in a child polyclinic in a pilot area of each country. Forty families were randomly sampled from the list. In the second round, from May to August 2004, women of reproductive age living in these 40 households were also selected. This coincided with more extensive food fortification. The third round was carried out in the second and third quarters of 2007...The same data were collected in all three rounds from the same individuals..." Tazhibayev et al 2008, Pg. 256 • 115. Table 5, Tazhibayev et al 2008, Pg. 262 • 116. Table 5, Tazhibayev et al 2008, Pg. 262 • 117. • "The urinary iodine excretion body of evidence was of moderate quality. All data were from cohort and multiple cross-sectional studies; however, all studies reported large effects of salt iodisation on mean urinary iodine excretion or prevalence of low urinary iodine excreation (UIE < 100 ug/L) in the population. Additionally there were more than 15 studies that could not be combined in the meta-analyses which had results consistent with a positive effect of iodised salt on urinary iodine excretion." Abudou et al 2014, Pg. 45 • Abudou et al 2014 classified all of the 14 studies as having a high risk of selection bias (random sequence generation and allocation concealment, i.e. not a randomized controlled trial), performance bias (blinding of participants and personnel) and detection bias (blinding of outcome assessment). Figure 25, Abudou et al 2014, Pg. 305; Figure 26, Abudou et al 2014, Pgs. 306-307. Because urinary iodine is objectively assessed outcome, we don't worry much about blinding. • No additional biases • Jooste 2000 • Zimmermann 2003 • Lv 2009 • Wang 2009 • Wang 2000 • Chen 1999 (Abudou et al 2014, Pgs. 66-67) • Unclear risk of attrition bias (Figure 25, Abudou et al 2014, Pg. 305; Figure 26, Abudou et al 2014, Pgs. 306-307 unless otherwise noted) • Tazhibayev 2008: attrition bias. "Sample size at follow-up not reported so same size assumed to be same as baseline" Abudou et al 2014, Pg. 139 • Dai 2008: attrition bias (Abudou et al 2014, Pg. 72) "Multiple cross-sectional survey with fewer participants at follow-up than baseline" Abudou et al 2014, Pg. 72. "The report was published in a peer reviewed journal in Chinese." Abudou et al 2014, Pg. 72 • At least 2 of attrition bias, reporting bias and other bias (Figure 25, Abudou et al 2014, Pg. 305; Figure 26, Abudou et al 2014, Pgs. 306-307 unless otherwise noted) • Hou 2003: attrition bias, reporting bias. "two cross sectional studies but number of individuals measured was not reported" Abudou et al 2014, Pg. 95; "prevalence reported but number of individuals not reported" Abudou et al 2014, Pg. 95 • Chen 2002a: reporting bias, other bias (Abudou et al 2014, Pgs. 69-70) "Multiple cross-sectional survey with similar number of participants at each time point" Abudou et al 2014, Pg. 69; "There was no baseline data provided for % of children with urinary iodine excretion < 100 ug/L" Abudou et al 2014, Pg. 69; "Methodology for follow-up survey not described though assumed to be the same as baseline as sample size similar" Abudou et al 2014, Pg. 70 • Han 2006: attrition bias, other bias. "Multiple cross-sectional study but sample sizes much smaller at follow- up than at baseline" Abudou et al 2014, Pg. 88; "PPS sampling described and reported that same at all timepoints but no reason given for very different sample sizes" Abudou et al 2014, Pg. 88 • Baczyk 2007: other bias (Abudou et al 2014, Pgs. 54-55). "Sampling methodology in two surveys not described in sufficient detail to discern potential bias" Abudou et al 2014, Pg. 55 • Yang 2011: attrition bias, other bias. "multiple cross sectional survey with much lower sample size at follow-up" Abudou et al 2014, Pg. 168; "sampling methodolgy was not detailed and very different sample sizes suggest high risk of bias" Abudou et al 2014, Pg. 169 • Zhou 2004: attrition bias, other bias. "Only 40 subjects measured at follow-up" Abudou et al 2014, Pg. 177; "Sampling methodology not detailed" Abudou et al 2014, Pg. 177 • 118. • Jooste, Weight and Lombard 2000: "Schools in four communities in South Africa were surveyed for goitre and urinary iodine excretion. The schools were not representative of any larger population but were selected to have a variety of socio-economic status participants." Abudou et al 2014, Pg. 102 • Zimmermann et al 2003: "Multiple cross sectional observatoinal study, conducted in 6 remote villages in the Danane Health District of western Cote d'Ivoire. Surveys of goitre and other indicators of iodine status were conducted in schools of the villages. All children, 5 - 14 years of age, who were present at school on the day of measurement were included in the sample." Abudou et al 2014, Pg. 180 • Tazhibayev et al 2008: "Children 2 -15 years of age in households randomly selected from health clinic registries in the sentinel sites in six countries." Abudou et al 2014, Pg. 138; "An area was selected in each country in a sentinel population expected to have early access to iodated salt and fortified wheat flour." Tazhibayev et al 2008, Pg. 255; "Sentinel studies should not be confused with effectiveness studies that are based on a national sample. However, when positive, sentinel studies such as this can indicate whether an intervention has the potential to be effective on a national scale if implemented similarly in all of a country." Tazhibayev et al 2008, Pg. 263; "The main potential confounding factor in comparing the results from the different countries is the variable sustainability of fortified wheat flour production nationally and for the sentinel sites. This can differentially influence the availability and consumption of fortified wheat flour in sentinel households." Tazhibayev et al 2008, Pg. 264 • Baczyk et al 2007: "Sampling methodology in two surveys not described in sufficient detail to discern potential bias" Abudou et al 2014, Pg. 55 • 119. See Ghana for example: WHO Global Database on Iodine Deficiency, Ghana, 2006 • 120. See for instance our blog post on Rethinking SCI's evidence of impact, where we discuss how sentinel sites for a national deworming campaign received additional treatment in studies. • 121. Abudou et al 2014 would have classified them as quasi-experimental, randomized controlled trials or non-randomized controlled trials: "This review included randomised controlled trials (both individual and cluster randomization), non-randomized tirals, quasi-experimental studies, cohort observational studies, and multiple cross-sectional observations studies which compared results before and after introduction of iodised salt." Abudou et al 2014, Pg. 8 • 122. "Multiple cross-sectional observationalstudy conducted in Wielkopolska Region, Poland. Cross-sectional surveys of goitre, urinary iodine excretion, and adverse effects were conducted before obligatory salt iodisation in Poland and 9 years after obligatory salt iodisation in Poland." Abudou et al 2014, Pg. 53 • 123. • "After goitre, urinary iodine excretion was the outcome measured in the most number of studies. This outcome was measured in different units and was reported in many different ways, making it impossible to do one overall meta-analysis with this outcome. Nonetheless, the results of the different studies were almost entirely consistent with an increase in urinary iodine excretion with the introduction of iodised salt. This result was found in all study types and was consistent across age groups, concentration of iodine in salt, or baseline risk of IDD based on goitre prevalence of urinary iodine excretion." Abudou et al 2014, Pg. 41 • "The urinary iodine excretion body of evidence was of moderate quality. All data were from cohort and multiple cross-sectional studies; however, all studies reported large effects of salt iodisation on mean urinary iodine excretion or prevalence of low urinary iodine excreation (UIE < 100 ug/L) in the population. Additionally there were more than 15 studies that could not be combined in the meta-analyses which had results consistent with a positive effect of iodised salt on urinary iodine excretion." Abudou et al 2014, Pg. 45 • Figures 13-17, Abudou et al 2014, Pgs. 301-302 • "3 Summary Table: Urinary iodine excretion in RCT and non RCTs", Abudou et al 2014, Pg. 215 • "4 Summary Table: Urinary iodine excretion in observational cohort studies", Abudou et al 2014, Pg. 216 • "5 Summary Table: Urinary iodine excretion in cross-sectional studies", Abudou et al 2014, Pgs. 217-218 • 124. • 125. • 126. "Since the establishment of the standards, salt iodization in Ghana has been ongoing for well over 16 years, with the target of covering at least 90% of the population. Although the percentage of households with adequately iodized salt has increased from only 0.7% in 1996 to 32% in 2006 [3, 17], a survey carried out in 2007 [18] showed that the prevalence of goiter dropped by 20% at two sentinel sites (Bongo and Jirapa districts) since the implementation of salt iodization legislation in 1996. Two major factors were identified as the causes for the low coverage of iodized salt in Ghana. The main challenge was that small salt producers were experiencing difficulties in accessing reputable international suppliers for their individual quantities and minimum order quantities, and pack sizes were not suitable for the production requirements, making it difficult and very expensive to source their potassium iodate...Another major reason identified by the government for lack of access to iodized salt was the inability of the numerous small-scale salt producers in Ghana to adequately iodize their production of salt. " Nyumuah et al 2012, Pg. S295 • 127. "Most iodine absorbed in the body eventually appears in the urine. Therefore, urinary iodine excretion is a good marker of very recent dietary iodine intake." WHO Assessment of Iodine Deficiency Disorders and Monitoring their Elimination 2007, Pgs. 38-39 • 128. • "Meanwhile, taking a more rigorous approach to defining UIC cut-offs, experts examined the limited data available on the association between goiter and UIC in populations before iodine prophylaxis. A key study was that of Ascoli and Arroyave (1970) that included data from 186 regions of Central America with goiter palpa- tion in 21,000 people (children and adults) and measurements of UIC and creatinine in over 3,000, with calculation of 24 h UIE based on urinary creatinine.35 This study found that endemic goiter (defined at that time as a GR >10%) was found in the following areas: 1) all areas in which the mean UIE was <25 mg/day; 2) most areas in which the mean UIE was 25–49 mg/day; 3) about one-third of areas in which the mean UIE was 50–99 mg/ day; and 4) virtually none of the areas in which UIE was >100 mg/day (Figure 1). Based largely on these data suggesting the GR was <10% when the mean UIE was >100 mg per day, WHO endorsed a median UIC of >100 mg per liter as an indicator of iodine sufficiency in a population (1993).[6] But clearly the UIC in mg/L is not necessarily interchangeable with the 24 h UIE (mg/24 h), as it depends on daily urine volume.36 If the daily volume of urine produced by a group approximates 1 L/day, as it does in healthy primary SAC, then the UIC (mg/L) is interchangeable with the 24 h UIE (mg/24 h). But these two indices are not comparable in older adolescents and adults, in whom the mean daily urine volume approximates 1.5 L/day37; thus, the UIC (mg/L) in spot samples is usually about 60–65% of the amount excreted in 24 h." Zimmermann and Andersson 2012, Pg. 557 • "Average urinary iodine excretion and prevalence of goitre by clinical examination in 186 localities of Central America 1965–1967. In each locality members of around 20 randomly selected families were investigated. A total of 21,611 people from 3,712 families were investigated for goitre, and iodine and creatinine were measured in late morning spot urine in a random sample of 3,181 participants. Daily iodine excretion was estimated from iodine and creatinine concentrations using an equation correcting for body weight, age and sex depending differences in 24 hour urinary creatinine excretion.20 Data from Ascoli and Arroyave.19" Laurberg et al 2010, Pg. 16 • "UI is an excellent indicator of recent iodine intake because ≥92% of dietary iodine is absorbed and, in healthy, iodine-replete adults, >90% is excreted in the urine within 24–48 h.27,28 UI can be expressed as a 24-h excretion (UIE; mg/day), as a concentration (UIC; mg/L), or in relationship to creatinine excretion (mg iodine/g creatinine). But these are not interchangeable, as discussed below. Because it is impractical to collect 24-h samples in field studies, UICs are usually measured in spot urine collections. In healthy well-nourished adults, daily creatinine excretion is fairly constant at about one gram, so expressing the UIE from spot samples in adults as mg iodine/g creatinine approximates the value in a 24-h collection and reduces variation due to hydration status.29 But in malnourished populations with poor protein intakes, daily creatinine excretion is more variable and often lower than 1 g.30 In these settings, expressing the UIE as mg iodine/g creatinine may introduce greater variation. Due to these limitations and the additional expense of measuring creatinine, the routine co-measurement of creatinine fell out of favor and was replaced by the expression of UIC in mg/L. If a large number of samples are collected, variations in hydration among individuals and day-to-day variation in iodine intake generally even out, so that the median UIC in spot samples correlates well with the median from 24-h samples and with the estimated UIE from creatinine-corrected UICs.31" Zimmermann and Andersson 2012, Pg. 556 • "In 1992, at a joint WHO/UNICEF/ICCIDD consultation, it was first proposed that the goal of monitoring progress toward eliminating IDD as a public health problem was to achieve a median UIC of 100 mg/L in SAC [school-aged children], with less than 20% of subjects having UICs below 50 mg/L.6 The UIC cut-off of 100 mg/L in SAC was sup- ported by a multicenter study in which UIC and thyroid volume were measured in 5,709 European school chil- dren.39 Although this study may have overestimated the goiter prevalence,19 the authors concluded that goiter begins to appear in SAC below the critical UIC threshold of 100 mg/L, and this agrees well with the earlier data from SAC in the report of Ascoli and Arroyave.35" Zimmermann and Andersson 2012, Pg. 558 • 129. • "Based largely on these data suggesting the GR was <10% when the mean UIE was >100 mg per day, WHO endorsed a median UIC of >100 mg per liter as an indicator of iodine sufficiency in a population (1993).[6] But clearly the UIC in mg/L is not necessarily interchangeable with the 24 h UIE (mg/24 h), as it depends on daily urine volume.36 If the daily volume of urine produced by a group approximates 1 L/day, as it does in healthy primary SAC [school-aged children], then the UIC (mg/L) is interchangeable with the 24 h UIE (mg/24 h). But these two indices are not comparable in older adolescents and adults, in whom the mean daily urine volume approximates 1.5 L/day37; thus, the UIC (mg/L) in spot samples is usually about 60–65% of the amount excreted in 24 h." Zimmermann and Andersson 2012, Pg. 557 • "In 1992, at a joint WHO/UNICEF/ICCIDD consultation, it was first proposed that the goal of monitoring progress toward eliminating IDD as a public health problem was to achieve a median UIC of 100 mg/L in SAC [school-aged children], with less than 20% of subjects having UICs below 50 mg/L.6 The UIC cut-off of 100 mg/L in SAC was supported by a multicenter study in which UIC and thyroid volume were measured in 5,709 European school chil- dren.39 Although this study may have overestimated the goiter prevalence,19 the authors concluded that goiter begins to appear in SAC below the critical UIC threshold of 100 mg/L, and this agrees well with the earlier data from SAC in the report of Ascoli and Arroyave.35" Zimmermann and Andersson 2012, Pg. 558 • 130. • 131. "Although the median UIC is a good population indicator of iodine status, the distribution of UIC around the median in iodine surveys is often misinterpreted in an attempt to define the number of individuals who are deficient. A common mistake is to assume that all subjects with a spot UIC <100 mg/L are iodine deficient. But dietary iodine intake and therefore UIC are highly variable from day to day. In iodine-sufficient countries where most iodine intake comes from iodized salt, UIC (both spot and 24-h urine collections) show an individual day-to-day variation of 30–40% (Figure 3a,b).29,31,42 Therefore, in an individual whose average daily iodine intake is adequate to maintain normal thyroidal iodine stores, iodine intake will show wide daily variation that will result in many individual days when a UIC value will be less than adequate. Thus, even in populations in which iodized salt ensures adequate thyroid stores, there will nearly always be individuals with a UIC<100mg/L on the day of the survey, but they are not truly iodine deficient. This common error in UIC interpretation was unfortunately indirectly endorsed by WHO. When pushed to define the number of individuals with low iodine intakes in order to give prevalence estimates, WHO made the decision to classify all children in iodine surveys with a spot UIC <100 mg/L as having low iodine intakes.43 This allowed WHO to generate regional and global prevalence data, but the approach also led to the apparent paradox that a country like Switzerland, with a model iodized salt program, a national median UIC of 120 mg/L and a GR of <3% in SAC [school-aged children],44 is classified as having 'optimal' country iodine status, but at the same time 36% of the population is classified as having inadequate iodine intake.9" Zimmermann and Andersson 2012, Pgs. 558-559 • 132. ICCIDD Scorecard • 133. • 54 studies reporting goiter prevalence: Aghini-Lombardi 1993, Azizi 2002, Baczyk 2007, Bimenya 2002, Charania 1988, Chen 1976, Chen 1984, Chen 1999, Chen 2001, Chen 2002a, Dai 2008, Fei 1996, Foo 1996, Gongora 1952, Guo 1984, Han 2006, Heydarian 2007, Hintze 1988, Hou 2009, Hu 1998, Ibanez-Gonzalez 1956, Jia 2004, Jooste 2000, Kimball 1931, Kimball 1946, l'Ons 2000, Lv 2009, Mostafavi 2005, Nicod 1953, Regalbuto 2010, Romano 1991, Rueda Williamson 1966, Salvaneschi 1991, Scrimshaw 1966, Sooch 1965, Szybinski 2001, Tang 1992, Wang 1981, Wang 1985a, Wang 2000, Wang 2001, Wang 2009, Wei 1985, Xu 1984, Xue 1993, Yang 1984a, Yang 1984b, Yang 2011, Yuan 1993, Yusuf 2008, Zhang 1988a, Zhou 2004, Zimmermann 2003, Zimmermann 2004b • 2 RCTs: Romano 1991, Hintze 1988 • "One RCT reported goitre as an outcome (Hintze 1988)." Abudou et al 2014, Pg. 27 • "The second RCT (Romano 1991) reported change in thyroid volume in pregnant women during the duration of pregnancy and qualitatively noted that thyroid volume did not change in the group of women consuming iodised salt whereas in the control group thyroid volume increased significantly (p<0.001)." Abudou et al 2014, Pg. 28 • 4 non-randomized controlled trials: Gongora 1952, l'Ons 2000, Sooch 1965, Kimball 1931 • "Four non-RCTs reported goitre as an outcome. Three studies with four comparisons could be combined in the meta-analysis. Iodised salt reduced the risk of goitre relative to control (RR=0.59 [0.36, 0.95]) (Figure 2). One study only reported prevalence without a sample size and therefore could not be combined in the meta-analysis (Kimball 1931)." Abudou et al 2014, Pg. 28 • Gongora 1952, l'Ons 2000, Sooch 1965. Figure 2, Abudou et al 2014, Pg. 294 • 1 quasi-experimental study: Jia 2004. "Only one study with a quasi-experiment component reported goitre (Jia 2004). In that study children 0-10 years of age who had been exposed to iodised salt their entire lives (experiment) were compared to children 0-10 years who had never been exposed to iodised salt (comparison). Only one child in the experimental group out of 23,685 participants had goitre, whereas 349 of 18,682 (1.8%) in the comparison group had goitre (Peto OR = 0.10 [0.08, 0.13])." Abudou et al 2014, Pg. 28 • 12 cohort studies: Chen 1984, Foo 1996, Guo 1984, Jia 2004, Tang 1992, Wang 2001 Xu 1984, Yang 1984a, Yang 1984b, Zhang 1988a, Zimmermann 2004b, Ibanez-Gonzalez 1956 • "Eleven cohort studies with 13 comparisons contributed to the meta-analysis of the effect of iodised salt on goitre. Consumption of iodised salt was associated with decreased risk of goitre (RR=0.30 [0.23, 0.41]) (Figure 3). One study with four comparisons reported prevalence of goitre without sample sizes and therefore could not be included in the meta-analysis (Ibanez Gonzalez 1956)." Abudou et al 2014, Pg. 28 • Figure 3, Abudou et al 2014, Pg. 294 • 38 multiple cross-sectional studies: Aghini-Lombardi 1993, Aziz 2002, Baczyk 2007, Charania 1988, Chen 1976, Chen 1999, Chen 2001, Chen 2002a, Dai 2008, Fei 1996, Han 2006, Heydarian 2007, Hou 2009, Hu 1998, Jia 2004, Jooste 2000, Kimball 1946, Lv 2009, Mostafavi 2005, Regalbuto 2010, Rueda Williamson 1966, Salvaneschi 1991, Scrimshaw 1966, Sooch 1965, Szybinski 2001, Wang 1981, Wang 1985a, Wang 2009, Wei 1985, Xue 1993, Yang 2011, Yuan 1993, Zhou 2004, Zimmermann 2003, Bimenya 2002, Nicod 1953, Wang 2000, Yusuf 2008 • "Thirty-four multiple cross-sectional studies with 44 comparisons contributed to the meta-analysis of the effect of iodised salt on goitre. Consumption of iodised salt was associated with decreased risk of goitre (RR=0.18 [0.14, 0.22]) (Figure 4). Four studies with 10 comparisons did not provide the minimum information necessary to be included in the meta-analysis. The results of these studies were consistent with a beneficial effect of iodised salt consumption on goitre prevalence (Table 1)." Abudou et al 2014, Pg. 28 • Aghini-Lombardi 1993, Aziz 2002, Baczyk 2007, Charania 1988, Chen 1976, Chen 1999, Chen 2001, Chen 2002a, Dai 2008, Fei 1996, Han 2006, Heydarian 2007, Hou 2009, Hu 1998, Jia 2004, Jooste 2000, Kimball 1946, Lv 2009, Mostafavi 2005, Regalbuto 2010, Rueda Williamson 1966, Salvaneschi 1991, Scrimshaw 1966, Sooch 1965, Szybinski 2001, Wang 1981, Wang 1985a, Wang 2009, Wei 1985, Xue 1993, Yang 2011, Yuan 1993, Zhou 2004, Zimmermann 2003. Figure 4, Abudou et al 2014, Pg. 295 • Bimenya 2002, Hou 2003, Nicod 1953, Wang 2000, Yusuf 2008. Table 1, Abudou et al 2014, Pg. 214 • 134. • "The outcome of goitre was reported by the most number of studies.The data across all studies designs very consistently showed that iodised salt was effective at reducing goitre risk. This result was true of studies conducted across 8 decades and across countries and continents including Africa, Asia, Europe, North America, and South America. Results were also consistent regardless age group, concentration of iodine in salt, or baseline risk of IDD based on goitre prevalence or urinary iodine excretion. In cohort and multiple cross-sectional studies, the longer the duration of the salt iodisation, the lower the risk of goitre." Abudou et al 2014, Pg. 41 • "1 Summary Table: Goitre prevelance with and without iodisation of salt", Abudou et al 2014, Pg. 214 • "Non-randomized controlled - Goitre (ALL)." Figure 2, Abudou et al 2014, Pg. 294 • "Cohort studies - Goitre (ALL)." Figure 3, Abudou et al 2014, Pg. 294 • "Multiple cross-sectional - Goitre (ALL)." Figure 4, Abudou et al 2014, Pg. 295 • 135. • 26 studies outside of China reporting goiter prevalence: Aghini-Lombardi 1993, Azizi 2002, Baczyk 2007, Bimenya 2002, Charania 1988, Foo 1996, Gongora 1952, Heydarian 2007, Hintze 1998, Ibanez-Gonzalez 1956, Jooste 2000, Kimball 1931, Kimball 1946, l'Ons 2000, Mostafavi 2005, Nicod 1953, Regalbuto 2010, Romano 1991, Rueda Williamson 1966, Salvaneschi 1991, Scrimshaw 1966, Sooch 1965, Szybinski 2001, Yusuf 2008, Zimmermann 2003, Zimmermann 2004b • 28 studies in China reporting goiter prevalence: Chen 1976, Chen 1984, Chen 1999, Chen 2001, Chen 2002a, Dai 2008, Fei 1996, Guo 1984, Han 2006, Hou 2009, Hu 1998, Jia 2004, Lv 2009, Tang 1992, Wang 1981, Wang 1985a, Wang 2000, Wang 2001, Wang 2009, Wei 1985, Xu 1984, Xue 1993, Yang 1984a, Yang 1984b, Yang 2011, Yuan 1993, Zhang 1988a, Zhou 2004 • Aghini-Lombardi 1993: "Multiple cross-sectional observational study conducted in Toscana, Italy." Abudou et al 2014, Pg. 50 • Azizi 2002: "Multiple cross-sectional observational study conducted in Shahriar, Iran." Abudou et al 2014, Pg. 52 • Baczyk 2007: "Multiple cross-sectional observationalstudy conducted in Wielkopolska Region, Poland." Abudou et al 2014, Pg. 53 • Bimenya 2002: "Multiple cross sectional observational study conducted in Uganda." Bimenya et al 2002, Pg. 56 • Charania 1988: "Multiple cross sectional study conducted in Pakistan." Abudou et al 2014, Pg. 59 • Chen 1976: "Multiple cross-sectional study conducted during 1964-5 and June 1971 in Taiwan, China." Abudou et al 2014, Pg. 61 • Chen 1984: "Observational cohort study conducted in China." Abudou et al 2014, Pg. 63 • Chen 1999: "Multiple cross sectional study, before-after design carried out in the Hui-an county, Fujian province, China." Abudou et al 2014, Pg. 65 • Chen 2001: "Multiple cross sectional study, before-after design, conducted in Nonghai city, Fuanjian province, China." Abudou et al 2014, Pg. 67 • Chen 2002a: "Multiple cross sectional study conducted in Hui-an, Fujian, China." Abudou et al 2014, Pg. 68 • Dai 2008: "Multiple cross sectional study conducted in Xiamen City, Fujian province, China." Abudou et al 2014, Pg. 71 • Fei 1996: "Multiple cross sectional study conducted in Rizhao City, Shandong province, China." Abudou et al 2014, Pg. 74 • Foo 1996: "Observational Cohort in Malaysia." Abudou et al 2014, Pg. 75 • Gongora 1952: "Non-randomized controlled trial conducted in Colombia." Abudou et al 2014, Pg. 84 • Guo 1984: "Observational cohort study conducted in Houchang community, Ziyun county, Guizhou province, China." Abudou et al 2014, Pg. 85 • Han 2006: "Multiple Cross-Sectional study conducted in Tianjin, China." Abudou et al 2014, Pg. 87 • Heydarian 2007: "Multiple cross sectional observational study conducted in Tehran province, Iran." Abudou et al 2014, Pg. 90 • Hintze 1998: "Randomised controlled trial conducted in Gottingen, Germany..." Abudou et al 2014, Pg. 91 • Hou 2009: "Multiple cross sectional observational study conducted in Lantian county, Shanxi province, China." Abudou et al 2014, Pg. 95 • Hu 1998: "Multiple cross sectional observational study conducted in Taiping village, Hanyin county, Shanxi province" Abudou et al 2014, Pg. 96 • Ibanez-Gonzalez 1956: "Cohort observational study conducted in Guejar Sierra, Granada, Spain." Abudou et al 2014, Pg. 99 • Jia 2004: "Cohort observational study and quasi-experimental, multiple cross-sectional study conducted in Heshui county, Gansu province, China." Abudou et al 2014, Pg. 100 • Jooste 2000: "Multiple cross sectional observational study conducted in South Africa." Abudou et al 2014, Pg. 102 • Kimball 1931: "Non-randomized controlled trial conducted in the United States." Abudou et al 2014, Pg. 103 • Kimball 1946: "Multiple Cross Sectional Study conducted in Michigan, USA." Abudou et al 2014, Pg. 105 • l'Ons 2000: "Non-randomised controlled study conducted in South Africa." Abudou et al 2014, Pg. 108 • Lv 2009: "Multiple cross-sectional observational study conducted in Hebei province, China." Abudou et al 2014, Pg. 111 • Mostafavi 2005: "Multiple cross sectional study conducted in Shiraz, Iran." Abudou et al 2014, Pg. 113 • Nicod 1953: "Multiple cross-sectional study conducted in Switzerland." Abudou et al 2014, Pg. 114 • Regalbuto 2010: "Multiple cross-sectional study conducted in Sicily, Italy." Abudou et al 2014, Pg. 120 • Romano 1991: "Randomised controlled trial conducted in Italy." Abudou et al 2014, Pg. 122 • Rueda Williamson 1966: "Multiple cross-sectional study conducted in Colombia." Abudou et al 2014, Pg. 124 • Salvaneschi 1991: "Multiple cross-sectional study conducted in Argentina." Abudou et al 2014, Pg. 125 • Scrimshaw 1966: "Multiple cross-sectional study conducted in Guatemala." Abudou et al 2014, Pg. 128 • Sooch 1965: "Non-randomized controlled trial: conducted in Punjab, India.(Sooch 1965)" Abudou et al 2014, Pg. 132 • Szybinski 2001: "Multiple cross-sectional study conducted in Poland." Abudou et al 2014, Pg. 135 • Tang 1992: "Observation cohort study conducted in villages near Kunmin, China." Abudou et al 2014, Pg. 136 • Wang 1981: "Multiple cross-sectional study conducted in Anqiu county, Shandong province, China." Abudou et al 2014, Pg. 143 • Wang 1985a: "Multiple cross-sectional study, before-after data conducted in Chengde City, China." Abudou et al 2014, Pg. 144 • Wang 2000: "Multiple cross-sectional study, before-after design, carried out in Hefei and Caohu area, China." Abudou et al 2014, Pg. 153 • Wang 2001: "Observational cohort study, before-after design, conducted in Qingyuan county, Hebei province." Abudou et al 2014, Pg. 155 • Wang 2009: "Multiple cross-sectional survey conducted in Gansu province,China." Abudou et al 2014, Pg. 157 • Wei 1985: "Multiple cross-sectional observational study conducted in Daxin county, Guangxi, China." Abudou et al 2014, Pg. 160 • Xu 1984: "Prospective observational cohort study conducted in Yushugou village, Dajiagou community, China." Abudou et al 2014, Pg. 161 • Xue 1993: "Multiple cross-sectional study conducted in Hami, Xinjiang, China." Abudou et al 2014, Pg. 163 • Yang 1984a: "Multiple cross-sectional study (outcome of cretinism) and cohort observational study (all other outcomes) conducted in Pingliang community, Guizhou, China." Abudou et al 2014, Pg. 164 • Yang 1984b: "Cohort observational study conducted in Qiandongnan Autonomous Prefecture, China." Abudou et al 2014, Pg. 166 • Yang 2011: "Multiple cross-sectional observational study conducted in China." Abudou et al 2014, Pg. 167 • Yuan 1993: "Multiple cross-sectional study conducted in Nanhai and Beihe communities of Qianjiang county, Sichuan, China." Abudou et al 2014, Pg. 169 • Yusuf 2008: "Multiple cross-sectional study conducted in Bangladesh." Abudou et al 2014, Pg. 170 • Zhang 1988a: "Cohort observational study conducted in Xiguanying community, Beipiao city, Liaoning province, China." Abudou et al 2014, Pg. 173 • Zhou 2004: "Multiple cross-sectional study conducted in Zhaotong City, Yuannan province, China." Abudou et al 2014, Pg. 176 • Zimmermann 2003: "Multiple cross sectional observatoinal study, conducted in 6 remote villages in the Danane Health District of western Cote d'Ivoire." Abudou et al 2014, Pg. 180 • Zimmermann 2004b: "Cohort observational study conducted in the northern Morocco." Abudou et al 2014, Pg. 182 • 136. • 20 studies of programs broadly similar to those that ICCIDD promotes and assists (i.e., programs that work with salt producers to iodize salt, rather than provide it directly to study participants or encourage study participants to buy iodized salt): Aghini-Lombardi 1993, Azizi 2002, Baczyk 2007, Bimenya 2002, Charania 1988, Gongora 1952, Heydarian 2007, Ibanez-Gonzalez 1956, Jooste 2000, Kimball 1931, Kimball 1946, Mostafavi 2005, Nicod 1953, Regalbuto 2010, Rueda Williamson 1966, Salvaneschi 1991, Scrimshaw 1966, Szybinski 2001, Yusuf 2008, Zimmermann 2003 • 6 studies of programs that provide iodized salt directly to study participants or encourage study participants to buy iodized salt: Foo 1996, Hintze 1988, l'Ons 2000, Romano 1991, Sooch 1965, Zimmermann 2004b • Aghini-Lombardi 1993: "Multiple cross-sectional observational study conducted in Toscana, Italy...All schoolchildren (6-14 years) from two villages participated...Iodized salt at 15ppm became available in the two villages. Before the intervention, no iodized salt was available for purchase." Abudou et al 2014, Pg. 50 • Azizi 2002: "Multiple cross-sectional observational study conducted in Shahriar, Iran. The sampling technique was stratified random sampling resulting in 48 urban and rural centers as primary sampling units. The same 48 PSUs selected for the first survey which was undertaken before salt iodisation in Iran were sampled during the follow surveys. The first survey was conducted before iodised salt was available and the second survey was conducted 12 years after the introduction of iodised salt in the country...Iodised salt at a concentration of 40 ppm in the form of potassium iodide was made available in the area." Abudou et al 2014, Pg. 52 • Baczyk 2007: "Multiple cross-sectional observationalstudy conducted in Wielkopolska Region, Poland. Cross-sectional surveys of goitre, urinary iodine excretion, and adverse effects were conducted before obligatory salt iodisation in Poland and 9 years after obligatory salt iodisation in Poland." Abudou et al 2014, Pg. 53 • Bimenya 2002: "Multiple cross sectional observational study conducted in Uganda. A sample of 2880 school children aged 6-12 years from 72 primary schools in 6 districts of Uganda were selected for evaluation of goitre in 1999, 5 years after the provision of iodised salt in the form of Universal Salt Iodisation in the country of Uganda. The results of goitre were compared to the results of a similar study conducted in 1991, 3 years before Universal Salt Iodization in Uganda. Study districts were randomly selected but geographic representation ensured through stratification of country. PPS sampling used to select the sample of schools and random sampling was used to select the students from each school. 40 students attending selected schools were included." Abudou et al 2014, Pg. 56 • Charania 1988: "Multiple cross sectional study conducted in Pakistan. Villages in the Gilgit and Hunza regions of Pakistan were stratified by accessibility to main road and then randomly selected. Households within villages were then randomly selected. Villagers in randomly selected household had goitre and urinary iodine concentration measured. The first survey was conducted in 1978 and described as baseline (pre-iodization) though manuscript says salt fortified with iron became available in Pakistanin 1977. iodised salt then became available in this region of Pakistan." Abudou et al 2014, Pg. 59 • Foo 1996: "Observational Cohort in Malaysia. The original study design was a randomised controlled trial which was conducted in Malaysia. The groups were randomly allocted to receive iodised salt or iodised water." Abudou et al 2014, Pg. 75 • Gongora 1952: "Non-randomized controlled trial conducted in Colombia. Salt iodisation was introduced into one community (Caldas, Colombia) and not into another community (Bogota, Colombia). The investigators selected participants randomly from schools to measure goitre before iodised salt was introduced. After 2 years of use of iodised salt in the intervention community, another cross-sectional study where the sample was said to have been selected in the same fashion as the previous study was conducted in both communities. The community of Bogota acted as a control because iodised salt was not used in that community. Methods of selection of participants was not described. Methods of measurement of goiter were not described but it was noted that the same methods were used in both surveys." Abudou et al 2014, Pg. 84 • Heydarian 2007: "Multiple cross sectional observational study conducted in Tehran province, Iran. Cluster random sampling was used to select adults for each survey. The first survey was conducted in 1983- 84 and the second survey was conducted in 1999-2000. National salt iodization became law in 1994." Abudou et al 2014, Pg. 90 • Hintze 1988: "All school children were invited to participate. Of the 1160 parents contacted, nearly 50% did not respond, 227 did not wish to participate and the parents of a total of 334 boys and girls agreed to the study and allowed their children to participate. Measurements were taken in the schools of body weight and height, goitre classified by WHO standards, and neck circumference. After baseline measurements students were randomly assigned to group A or group B. Randomization method not explained. Group A was assigned to buy from the market and consume only iodised salt while Group B was assigned to buy from the market and consume only non-iodised salt. Groups were not blinded to treatment but assessors were blinded. Follow-up measurements of goitre were undertaken at 2 and 4 years and UIE was measured yearly. The final 4 year follow-up measure is used in this review." Abudou et al 2014, Pg. 92 • Ibanez-Gonzalez 1956: "A random sample of the population living in the town of Guejar Sierra, Granada, Spain was evaluated for Goitre before and after the introduction of iodised salt in the community." Abudou et al 2014, Pg. 99 • Jooste 2000: "The baseline survey measured goitre and UIE during the month before USI was begun in S. Africa. The follow-up survey occurred one year later using the same sampling technique." Abudou et al 2014, Pg. 102 • Kimball 1931: "Two cross-sectional surveys were conducted in a community which started a salt iodisation programme (Detroit, Michigan, USA) and in a community without salt iodization which acted as a control (Cleveland, Ohio,USA). the first survey was before iodised salt was introduced into the Detroit community and the second survey was 7 years after the introduction of iodised salt" Abudou et al 2014, Pg. 103 • Kimball 1946: "Multiple Cross Sectional Study conducted in Michigan, USA. All school children in three counties in Michigan, USA were measured for goitre in 1924 before the introduction of iodised salt to that state. In 1924, iodised salt was introduced to the state of Michigan at a concentration of 20ppm. In 1928 and in 1936 all the school children in the same schools in the same four counties were again measured for goiter." Abudou et al 2014, Pg. 105 • l'Ons 2000: "Non-randomised controlled study conducted in South Africa. Non-random controlled trial where two schools were selected for convenience and because there was no iodized salt available in the region where the schools were located. All children in both schools participated. Goitre, measured by palpation and defined using WHO criteria, and UIE were measured at baseline and after a 4-month follow-up period. Children in the experiment school were all given iodised salt for use in the household and children in the control school were given non-iodized salt for use in the household." Abudou et al 2014, Pg. 108 • Mostafavi 2005: "Iodised salt at an unknown concentration through universal salt iodization in the country." Abudou et al 2014, Pg. 113 • Nicod 1953: "Article provides an overview of the prophylaxis of goitre through the use of iodised salt in various cantons of Switzerland. The study includes data for discharged military adults, national prevalence of goitre, and level of fortification in each Canton and when iodisation programmes were initiated. The data included in the review comes from surveys of school-aged children in two Cantons where complete data from pre-iodisation and post-iodisation of salt is available. No data available on how students were selected to participate in the evaluations. Numerous indicators of the thyroid size were reported based on palpation of the neck. The indicators were normal thyroid, large neck, and advanced goitre." Abudou et al 2014, Pg. 114 • Regalbuto 2010: "This study compared the goitre prevalence in 1977, in 1994 during which the use of iodised salt was less than 1% of the total salt consumed, and 2007, ten years after the promotion of iodised salt which began in 1996. The use of iodised salt reached 44% (28% to 55%). School children from different towns were surveyed for goitre, thyroid volumen and urinary iodine excretion. The selection of subjects was not described in detail." Abudou et al 2014, Pgs. 120-121 • Romano 1991: "Randomised controlled trial conducted in Italy. Pregnant women were randomly selected from a health care facility. Women were first stratified by smoking status and then randomly assigned to either group A (iodised salt) or group B (non-iodised salt). Each trimester, measurements were taken at the clinic.Units of comparison: Individuals...1. Iodised salt provided at 20ppm (N=17) 2. Control - plain salt, but not specifically distributed (N=18)" Abudou et al 2014, Pg. 122 • Rueda Williamson 1966: "A representative sample of school-aged children from one department in Colombia were selected for measurement of goitre status before iodisation of salt in Colombia. The first survey was conducted in 1945. Salt was iodised at 50ppm starting as pilot is 1950, then law implemented in 1955 for country. Large-scale iodization began in 1959 and in 1964 salt in country varied from 60-75ppm in iodine level based on monitoring activities. Another survey sample was measured for goitre 10 years after the implementation of a law requiring iodisation of salt. Goitre was defined using WHO standards." Abudou et al 2014, Pg. 124 • Salvaneschi 1991: "Before/after study design with cross sectional representative surveys of goitre in city of Buenos Aires, Argentina. Goitre was measured using WHO methods and definition. The first survey occurred before the mandatory fortification of salt with iodine and the second survey occured 18 years after the implementation of mandatory fortification of salt with iodine in the city." Abudou et al 2014, Pg. 125 • Scrimshaw 1966: "Before and after design to monitor effect of fortification of salt with iodine on goitre rates in school children in Guatemala. Two cross-sectional studies were conducted, one before and one after the implementation of mandatory salt iodisation in the country. The surveys were representative of 19 of the 22 departments of the country. Coverage of salt iodization at sufficient concentration in the country was estimated at 80% at the year of of the follow-up survey." Abudou et al 2014, Pg. 128 • Sooch 1965: "This study was set up as a prospective controlled study of the effect of iosidation of salt on goitre rates. The study was undertaken in the Himalayan endemic goitre belt of northern India. Over 20,000 villagers were surveyed and the analysis of results was divided between children 5-16 years attending school, children not attending school and the general population of adults and children combined. The area of the study was divided into three zones. One zone received salt with potassium iodide, one received salt with potassium iodate, and the third was a control zone that did not receive fortified salt. Randomisation and blinding were not mentioined in the study reports. All school children in each of the zones was surveyed at the beginning, in the middle and at the end of the study period. In each zone 2/3rds of the villages were samples and all of the villagers present on the day of survey were included in the surveys. The first, baseline, survey was in 1956, another survey was conducted five years later (1962), and a final survey was conducted in 1968. From 1956 to 1962, two zones received the intervention and one zone was the control. In 1956, the control zone was also provided salt fortified with salt and all three villages were followed over time until the final follow-up. Additional data was reported in Sooch 1973 which was of multiple cross sectional design describing the results after twelve years of salt iodisation...1. Iodised salt provided to one zone in the concentration of 20 ppm KI (i.e. 15.27 ppm iodine) and as 25 ppm KIO3 (i.e. 14.83 ppm iodine) to another zone (the results from the two intervention groups were combined and used as intervention and compared to control) 2. Plain salt" Abudou et al 2014, Pgs. 132-133 • Szybinski 2001: "Multiple cross-sectional study conducted in Poland. Children attending selected schools were surveyed before and 5 years after the initiation of a national programme for the iodisation of salt." Abudou et al 2014, Pg. 135 • Yusuf 2008: "Nationally representative surveys on nutritional status in children and women of reproductive age in Bangladesh were conducted in numerous years and the data provide national estimates of Iodine Deficiency Disorders before and after the introduction on iodised salt in that country. Probably proportionate to size (PPS) sampling was used to select households for inclusion to select a nationally representative sample. The first sample used in this review was from the year 1993. The Salt-Bye law was passed in Bangladesh in 1994 and at that time substantial investments were made in the infrastructure for salt iodisation. By 1995 all salt factories were equipped for iodisation (original law passed in 1989 but no effort made to enforce and there was little iodisation in the country until 1994). The data from a second nationally representative survey undertaken 10 years later (2005) was used to compare the iodine status of children and women pre and post fortification of salt with iodine." Abudou et al 2014, Pg. 170 • Zimmermann 2003: "Multiple cross sectional observatoinal study, conducted in 6 remote villages in the Danane Health District of western Cote d'Ivoire. Surveys of goitre and other indicators of iodine status were conducted in schools of the villages. All children, 5 - 14 years of age, who were present at school on the day of measurement were included in the sample. The first survey was conducted in 1997 before iodised salt was available in the villages and children from two schools were included. In 1998, universal salt iodisation was initiated. Other surveys using similar methods were conducted in school children in the same villages yearly until 2001." Abudou et al 2014, Pg. 180 • Zimmermann 2004b: "Iodised salt provided at 25 ppm through provision to households." Abudou et al 2014, Pg. 182 • 137. • Developing: Bimenya 2002 (Uganda), Gongora 1952 (Colombia), Rueda Williamson 1966 (Columbia), Salvaneschi 1991 (Argentina), Scrimshaw 1966 (Guatemala), Yusuf 2008 (Bangladesh), Zimmermann 2003 (Cote d'Ivore), Charania 1988 (Pakistan), Heydarian 2007 (Iran), Mostafavi 2005 (Iran), Azizi 2002 (Iran) • Non-developing: Aghini-Lombardi 1993 (Italy), Baczyk 2007 (Poland), Ibanez-Gonzalez 1956 (Spain), Jooste 2000 (South Africa), Kimball 1931 (USA), Kimball 1946 (USA), Nicod 1953 (Switzerland), Szybinski 2001 (Poland), Regalbuto 2010 (Italy) • Aghini-Lombardi 1993: "Multiple cross-sectional observational study conducted in Toscana, Italy." Abudou et al 2014, Pg. 50 • Azizi 2002: "Multiple cross-sectional observational study conducted in Shahriar, Iran." Abudou et al 2014, Pg. 52 • Baczyk 2007: "Multiple cross-sectional observational study conducted in Wielkopolska Region, Poland." Abudou et al 2014, Pg. 53 • Bimenya 2002: Uganda, Table 1, Abudou et al 2014, Pg. 214 • Charania 1988: "Multiple cross sectional study conducted in Pakistan." Abudou et al 2014, Pg. 59 • Gongora 1952: "Non-randomized controlled trial conducted in Colombia." Abudou et al 2014, Pg. 84 • Heydarian 2007: "Multiple cross sectional observational study conducted in Tehran province, Iran." Abudou et al 2014, Pg. 90 • Ibanez-Gonzalez 1956: Spain, Table 1, Abudou et al 2014, Pg. 214 • Jooste 2000: "Multiple cross sectional observational study conducted in South Africa." Abudou et al 2014, Pg. 102 • Kimball 1931: USA, Table 1, Abudou et al 2014, Pg. 214 • Kimball 1946: "Multiple Cross Sectional Study conducted in Michigan, USA." Abudou et al 2014, Pg. 105 • Mostafavi 2005: "Multiple cross sectional study conducted in Shiraz, Iran." Abudou et al 2014, Pg. 113 • Nicod 1953: Switzerland, Table 1, Abudou et al 2014, Pg. 214 • Regalbuto 2010: "Multiple cross-sectional study conducted in Sicily, Italy." Abudou et al 2014, Pg. 120 • Rueda Williamson 1966: "Multiple cross-sectional study conducted in Colombia." Abudou et al 2014, Pg. 124 • Salvaneschi 1991: "Multiple cross-sectional study conducted in Argentina." Abudou et al 2014, Pg. 125 • Scrimshaw 1966: "Multiple cross-sectional study conducted in Guatemala." Abudou et al 2014, Pg. 128 • Szybinski 2001: "Multiple cross-sectional study conducted in Poland." Abudou et al 2014, Pg. 135 • Yusuf 2008: "Multiple cross-sectional study conducted in Bangladesh." Abudou et al 2014, Pg. 170 • Zimmermann 2003: "Multiple cross sectional observatoinal study, conducted in 6 remote villages in the Danane Health District of western Cote d'Ivoire." Abudou et al 2014, Pg. 180 • 138. "Many of the analyses suffered from heterogeneity stemming from known (age group, region of world, decade of study implementation, sampling methodology, measurement techniques, iodine status at baseline, iodine concentration in salt, salt consumption patterns, other dietary patterns) and unknown factors. This limitation notwithstanding, the results were very consistent for most outcomes measured. Iodised salt reduced prevalence of goitre in children, adults and the entire population; in populations consuming iodised salt at <20ppm, 20-40ppm, or >40ppm; and in populations with severe, moderate, or mild iodine deficiency at baseline. The other outcomes with sufficient data to make such comparisons (i.e. cognitive function, urinary iodine excretion) showed the same results. Additionally, studies that could not be combined in the meta-analyses for these outcomes were also consistent with a benefit of consuming iodised salt. Therefore, the heterogeneity of the studies did not proclude the ability to draw the conculsion that iodised salt is effective at reducing risk of many adverse effects of iodine deficiency." Abudou et al 2014, Pg. 43 • 139. • "The response to supplemental iodine may be reduced by other factors in the environment, such as consumption of specific goitrogenic foods (for example, cassava (manioc)), and deficiency of other trace elements in the diet, such as selenium (Yang 1997; Khrle 1999) or iron (Zimmermann 2000)." Angermayr and Clar 2004, Pg. 3 • "The size of the thyroid gland changes inversely in response to alterations in iodine intake, with a lag interval that varies from a few months to several years, depending on many factors. These include the severity and duration of iodine deficiency, the type and effectiveness of iodine supplementation, age, sex, and possible additional goitrogenic factors." WHO Assessment of Iodine Deficiency Disorders and Monitoring their Elimination 2007, Pg. 35 • 140. • "For goitre, the quality of evidence ranged from low to moderate. The non-RCTs had serious risk of bias because of lack of randomisation of treatment allocation and therefore were moderate in quality. The observational cohort studies were also moderate in quality. Because of study design they started as low and were upgraded due to the strong association between iodised salt and reduced risk of goitre. The mutliple cross sectional studies were of low quality. They were downgraded due to serious risk of bais because of the incomplete reporting of study methodology in many studies; however the body of evidence was upgraded due to the strong association between iodised salt and reduced risk of goitre. Most of the other studies faced risk of bias and methodological weaknesses, such as differences in baseline goitre rates and other comparability issues, sampling differences between measurement time periods, etc. In some studies, including one of the RCTS (Hintze 1988), the level of iodisation was likely not high enough to supply dialy recommended levels of iodine intake. Multiple cross-sectional studies were considered to have a high risk of bias if they had large sample size differences between surveys but lacked sufficient detail of methodology to understand those differences. Results would be biased if individuals with goitre at baseline, before introduction of iodised salt, were not included in the second or follow-up survey after introduction of iodised salt." Abudou et al 2014, Pgs. 44-45 • "Neither of the RCTs reported blinding of participants; however, both studies reported blinding of outcome assessors. One non-RCT (l'Ons 2000) mentioned blinding of participants and no non-RCTs mentioned the blinding of outcome assessors. No quasi-experimental studies reported blinding of participants and personnel. One cohort observational study (Kimiagar 1990) and one multiple cross-sectional study (Yuan 1993) were unclear on blinding of participants and personnel, while all ofther cohort observational and multiple cross-sectional studies reported that participants and personnel were not blinded. Most quasi-experimental, cohort, and multiple cross-sectional studies were unclear on the blinding of outcome assessment. Those that were clear, reported no blinding of outcome assessment (Bauch 1990; Bimenya 2002; Cerqueira 2009; Charania 1988; Chen 1999; Dong 1991; Heydarian 2007; Mostafavi 2005; Nicod 1953; Salvaneschi 1991; Wang 2009; Yusuf 2008; Zhang 1988a; Zhou 2004)." Abudou et al 2014, Pgs. 25-26 • 141. "From the mid-1990s, UNICEF started assisting countries in collecting health data through its international household survey initiative, the Multiple Indicator Cluster Surveys (MICS). Countries may also participate in other standardized household survey programs, such as the Demographic and Health Surveys (DHS) sponsored by the US Agency for International Development (USAID). These surveys have large sample sizes (usually between 5,000 and 30,000 households) and typically are conducted about every 5 years, to allow comparisons over time." ICCIDD Where is the problem? • 142. Countries with a value in the "> 0 ppm (Any iodine)" column: Childinfo Iodized salt consumption 2012 • 143. "> 0 ppm (Any iodine)", Childinfo Iodized salt consumption 2012 • 144. This casts some doubt on how accurately the surveys’ data reflects a need (or lack of need) for iodine fortification. Because we are unsure of what the surveys imply, they do not significantly influence our views. • 145. • “The assessment of iodine content has generally been accomplished through the use of rapid test kits. While the test kit cannot provide data that are as precise and accurate as titration, it remains the test method associated with most data points in UNICEF’s global database. Based on recommendations from WHO to only use rapid test kits to indicate the presence of iodine, and to use another method – titration – to quantify and report on salt with a particular amount of iodine, a number of countries have stopped reporting on the cut-off of ≥15 ppm and only report on salt with any iodine (>0 ppm) in many recent surveys.” Childinfo Iodized salt consumption 2012 • "From the mid-1990s, UNICEF started assisting countries in collecting health data through its international household survey initiative, the Multiple Indicator Cluster Surveys (MICS). Countries may also participate in other standardized household survey programs, such as the Demographic and Health Surveys (DHS) sponsored by the US Agency for International Development (USAID). These surveys have large sample sizes (usually between 5,000 and 30,000 households) and typically are conducted about every 5 years, to allow comparisons over time. One of the nutrition indicators in these surveys is the presence of iodized salt in the households. Traditionally, an adequate iodine level in household salt has been defined as salt containing at least 15 ppm of iodine. Testing of the iodine content in household salt samples is usually done with Rapid Testing Kits (RTKs). Data on household coverage with iodized salt is updated and summarized yearly by UNICEF in their State of the World’s Children (SOWC) Reports and in the Iodine Scorecard on this website. The data compiled by UNICEF are mainly from developing countries and countries in transition; household coverage data from industrialized countries are limited." ICCIDD Where is the problem? • 146. • "A limitation of this data is that, although in an increasing number of surveys iodometric titration is used to quantitatively assess salt iodine content, much of the data has been generated using the RTK to assess salt iodine content. These kits have been shown not to be highly reliable for quantifying salt iodine content and thus, the data have to be interpreted with some caution." ICCIDD Where is the problem? • "Recent evaluations of these kits showed that the colour reaction cannot be used as a quantitative indication of the iodine content (18). These kits should therefore be regarded as qualitative rather than quantitative and are most appropriate to indicate the presence or absence of iodine, but not of the concentration. An advantage of rapid test kits is that they can be used in the field to give an immediate result. They are therefore useful to health inspectors and others who are involved in carrying out spot checks on food quality or household surveys. They may also play a valuable educational role, in that they provide a visible indication that salt actually is iodized. Accordingly, they can be used for demonstration purposes in schools and other institutions. However, because rapid test kits do not give a reliable estimate of iodine content (19,20), results must be backed up by titration." WHO Assessment of Iodine Deficiency Disorders and Monitoring their Elimination 2007, Pg. 21 • "Due to the low specificity and resulting high numbers of false-positives for the kit when used by multiple observers (''real-life situations''), kits were likely to consistently overestimate the availability of iodized salt. This overestimation could result in complacency. Therefore, we conclude that until a valid alternative is available, the titration method should be used for monitoring the iodine content of salt at all levels, from producer to consumer, to ensure effectiveness of the programme." Pandav et al 2000, Pg. 975 • 147. "These kits should therefore be regarded as qualitative rather than quantitative and are most appropriate to indicate the presence or absence of iodine, but not of the concentration." WHO Assessment of Iodine Deficiency Disorders and Monitoring their Elimination 2007, Pg. 21 • 148. "Comparison of spot-testing kits versus the iodometric titration method for the determination of iodine content in salt: multiple-observer data", Table 3, Pandav et al 2000, Pg. 978. The table does not show strong evidence of that claim. • 149. "In addition, household salt iodization coverage does not take all dietary iodine sources into account, particularly the contribution of salt in processed foods and dairy products that are important sources in many industrialized countries, as well as iodine-rich groundwater found in some regions (although often confined to smaller geographical areas)." ICCIDD Where is the problem? • 150. • "Thus, the total cost of salt iodization ranges from$3.95 to $14.80 per ton, with an average of$7.40, using existing packaging, or $11.40 with new packing material. This amounts to 0.4 to 1.2 cents per kg or 1-6 cents per person per year, based on an estimated annual consumption of 3 kg to 5 kg. The retail price of crystalline salt for domestic use ranges from$0.25 to $1.00 per kg. From these estimates, iodization increases the retail price of salt by 1 to 24%." Mannar and Dunn 1994, Pg. 76 • Converted USD 1994 to USD 2014 using the CPI Inflation Calculator • 151. • "Table 12.1 summarizes the range of costs for several components of the iodization process. Table 12.2 presents the capital and operating cost for a continuous spray mixing plant, with a capacity of 20,000 tons per year built at a new location. The estimates are based on costs of materials and services in several Asian and African countries." Mannar and Dunn 1994, Pg. 75 • "Thus, the total cost of salt iodization ranges from$3.95 to $14.80 per ton, with an average of$7.40, using existing packaging, or $11.40 with new packing material. This amounts to 0.4 to 1.2 cents per kg or 1-6 cents per person per year, based on an estimated annual consumption of 3 kg to 5 kg. The retail price of crystalline salt for domestic use ranges from$0.25 to $1.00 per kg. From these estimates, iodization increases the retail price of salt by 1 to 24%." Mannar and Dunn 1994, Pg. 76 • 152. "Thus, the total cost of salt iodization ranges from$3.95 to $14.80 per ton, with an average of$7.40, using existing packaging, or $11.40 with new packing material. This amounts to 0.4 to 1.2 cents per kg or 1-6 cents per person per year, based on an estimated annual consumption of 3 kg to 5 kg. The retail price of crystalline salt for domestic use ranges from$0.25 to $1.00 per kg. From these estimates, iodization increases the retail price of salt by 1 to 24%." Mannar and Dunn 1994, Pg. 76 • 153. "Table 1 summarizes best estimates of relative costs of different methods of fortification, along with potential population coverage of each. These figures are not exact, and meant more to indicate orders of magnitude. It is estimated that up to 70% coverage with USI can be achieved within a country by fortifying salt from large producers. The coverage may be higher in some countries, for example most countries in sub‐Saharan Africa tend to have fewer small salt producers, whereas countries with archipelagos such as Indonesia or the Philippines may have many small, remote producers. The cost of$0.05 per person per year represents the cost of the actual ingredients (only about 20% of this total), plus the cost of amortizing the capital equipment and processing costs. If salt from small producers if fortified, the capital costs are spread over a smaller volume and this may double the cost of fortification as compared to fortification by larger producers, but may permit another 10% of the population to be reached. Double‐fortified salt has not been produced on a large scale, and is about 4 times as expensive as iodized salt and likely more expensive than using iodized salt and iron fortified flour. However, it may be a good option where flour fortification does not reach a majority of the population." Horton, Mannar and Wesley 2008, Pg. 22

• 154.

Costs of salt iodization, estimates from Venkatesh Mannar, September 2, 2014

• 155.
• 156.

"The actual availability of iodine from iodized salt at the consumer level can vary over a wide range as a result of:

• Variability in the amount of iodine added during the iodization process;
• Uneven distribution of iodine in the iodized salt, within batches and individual bags, due to insufficient mixing of salt after the salt iodization process and/or variation in particle size of salt crystals in a batch or bag;
• The extent of loss of iodine due to salt impurities, packaging (for instance, 1kg versus 20 or 50kg), and environmental conditions during storage and distribution;
• Loss of iodine due to food processing, and washing and cooking processes in the household;
• The availability of non-iodized salt from unconventional marketing sources.
• 157.

ICCIDD conversations with GiveWell in Zurich, April/May 2014

• 158.

"Only 2 included trials reported the growth and development of children and clinical outcomes. Iodine supplementation during pregnancy or the periconceptional period in regions of severe iodine deficiency reduced risk of cretinism, but there were no improvements in childhood intelligence, gross development, growth, or pregnancy outcomes, although there was an improvement in some motor functions." Zhou et al 2013, Pg. 1

• 159.
• "C. Adulthood In adults, mild-to-moderate iodine deficiency appears to be associated with higher rates of more aggressive subtypes of thyroid cancer, increases risk for diffuse goiter, and increases risk of nontoxic and toxic nodular goiter and associated hyperthyroidism (4, 30) (see detailed discussion in Section XI). Observational studies also suggest subtle but widespread adverse effects in adults secondary to hypothyroidism, including impaired mental function with decreased educability, apathy, and reduced work productivity (3)." Zimmermann 2009, Pg. 390
• "VII. Effects of Deficiency through the Life Cycle", Zimmermann 2009, Pgs. 385-390
• 160.
• A few studies have demonstrated correlations between goiter rates and UICs, which the WHO has used to help determine population-level goals for iodine status. Iodine deficiency as a continuum of risk is not clear from these correlational studies.
• "Meanwhile, taking a more rigorous approach to defining UIC cut-offs, experts examined the limited data available on the association between goiter and UIC in populations before iodine prophylaxis. A key study was that of Ascoli and Arroyave (1970) that included data from 186 regions of Central America with goiter palpation in 21,000 people (children and adults) and measurements of UIC and creatinine in over 3,000, with calculation of 24 h UIE based on urinary creatinine.35 This study found that endemic goiter (defined at that time as a GR >10%) was found in the following areas: 1) all areas in which the mean UIE was <25 mg/day; 2) most areas in which the mean UIE was 25–49 mg/day; 3) about one-third of areas in which the mean UIE was 50–99 mg/ day; and 4) virtually none of the areas in which UIE was >100 mg/day (Figure 1). Based largely on these data suggesting the GR was <10% when the mean UIE was >100 mg per day, WHO endorsed a median UIC of >100 mg per liter as an indicator of iodine sufficiency in a population (1993).6 But clearly the UIC in mg/L is not necessarily interchangeable with the 24 h UIE (mg/24 h), as it depends on daily urine volume.36 If the daily volume of urine produced by a group approximates 1 L/day, as it does in healthy primary SAC [school-aged children], then the UIC (mg/L) is interchangeable with the 24 h UIE (mg/24 h). But these two indices are not comparable in older adolescents and adults, in whom the mean daily urine volume approximates 1.5 L/day37; thus, the UIC (mg/L) in spot samples is usually about 60–65% of the amount excreted in 24 h...The proper interpretation of these data would have produced a cut-off for ID using the median UI in SAC of 100 mg/L and a cut-off of about 60–70 mg/L for adults. This adult cut-off is supported by a large study of young women in Slovakia conducted before the introduction of iodine prophylaxis; in this study, GRs were estimated by palpation in approximately 10,000 women and UICs in spot samples (expressed as mg/L) were measured in approximately 2,50038 (Figure 2)." Zimmermann and Andersson 2012, Pg. 557
• Figure 1, Zimmermann and Andersson 2012, Pg. 557
• Figure 2, Zimmermann and Andersson 2012, Pg. 557
• 161.
• "Measurement of TSH levels, which are expected to be elevated in iodine deficiency, is not a reliable indicator in school children and adults as differences to normal levels are small and there is a large overlap between values in people who are iodine sufficient and those who are iodine deficient. However, TSH levels are a good indicator of iodine deficiency in neonates. Increased thyroglobulin levels are a good indicator of thyroid hyperplasia resulting from iodine deficiency. Thyroglobulin levels reflect iodine nutrition over months and years, whereas urinary iodine levels measure the more immediate effects of increased iodine intake on iodine status. Measurement of thyroid hormones (T3 and T4) is not recommended, as the tests are difficult and expensive and the measurements are not very reliable indicators of iodine deficiency." Angermayr and Clar 2004, Pg. 4
• Table 9, WHO Assessment of Iodine Deficiency Disorders and Monitoring their Elimination 2007, Pg. 43
• 162.

"Though dietary patterns vary geographically, with respect to a variety of micronutrients important for human development, iodine availability is likely to exert a stronger independent influence on economic outcomes than dietary prevalence of other micronutrients and many climatic conditions due to the fact that it has little correlation with local food availability. Hence, while other micronutrient deficiencies are likely to be resolved with economic development by way of rising caloric intake, iodine deficiency is more likely to exert a persistent influence on economic outcomes." Field, Robles and Torero 2009, Pg. 141

• 163.

"The epidemic of thyrotoxicosis reported from Denmark in the years between 1941 and 1945 (40,41) and in Sweden (42) could not be related to fluctuations in iodine intake, but a correlation between seasonal fluctuations in iodine content of milk in England and in incidence of thyrotoxicosis was inferred (43)." Stanbury et al 1998, Pg. 84

• 164.

"Reliance on salt as a vehicle for the delivery of dietary iodine should not be used to justify promotion of salt intake to the public, and additional vehicles to salt for micronutrient fortification should continue to be explored." Salt as a Vehicle for Fortification, Pg. 15