Effects of exposures to carcinogens during infancy
by DP Meulenberg
Section 1: There were dramatic increases of toxins in the environment during the 20th century: For all of human history until relatively recently, human milk was the ideal food for infants. But, beginning in about the middle of the 20th century, exposures to toxins in the environment greatly increased, and that affected the contents of human milk.
Preview of Figures 1 and 2
Increases of dioxins and PCBs, both carcinogens, were especially strong. Despite declines in levels of both in recent decades, both are still present in human milk far in excess of established safe levels. Exposures to both at common background levels have been found in various studies to be associated with greatly increased risk of cancer. The U.S. National Toxicology Program has stated that there is no known safe dose or threshold below which dioxins would not cause cancer. By excellent authority, many other chemicals in breast milk are also known carcinogens.
Summary of Section 2: PCBs, dioxins and similar chemicals from the environment are attracted to fat, which means they become greatly concentrated in breast milk, as recognized by undisputed authorities. Children who have grown up in developed and many developing countries during the more recent decades would have been likely, during their vulnerable developmental periods, to have been exposed to large amounts of these chemicals via breastfeeding. While levels of dioxins and PCBs have declined, other chemicals with similar properties (such as PBDE flame retardants) have greatly increased in human milk. Even at age 4, concentrations of PBDEs in breastfed children were found still to be almost three times as high as in formula-fed children.
Summary of Section 3: See preview charts below.
Studies that have avoided the confounded comparisons that are normal in typical studies, by means of sibling comparison and a randomized controlled trial, have both found that obesity is increased in association with increased breastfeeding.
Summary of Section 4: Childhood cancer.
Section 4.a: Why typical studies have arrived at false conclusions about breastfeeding and childhood cancer:
Section 4.a.1: Women in developed countries who breastfeed, or who breastfeed for longer durations, are normally of higher socio-economic status (SES) and are less exposed to carcinogenic pollution (including tobacco smoke) than are women who don't breastfeed. This biases study results in favor of breastfeeding. Typical studies report lower breastfeeding among those with cancer, but they do not mention the greater environmental causes of cancer among the low-SES families in which breastfeeding rates are well below average.
Summary of Section 4.a.2: When a study is carried out in a country where, contrary to the pattern in developed countries, the more-educated may not be more likely to breastfeed, more breastfeeding is associated with increased leukemia.
Summary of Section 4.a.3: When a study is carried out by a means (with siblings for the comparison group) that avoids the usual confounding, breastfeeding is positively correlated with leukemia.
Summary of Section 4.a.4: Control (comparison) groups are a source of bias and misinterpretation of many studies. Health consciousness tends to be much greater in control groups, due to difficulty of getting average people to participate in them. Therefore healthful lives and following health recommendations (including breastfeeding) are more prevalent among the people with whom the diseased group is compared. But only the greater breastfeeding receives attention, in typical studies.
Summary of Section 4.b: There have been increases of cancer among U.S. children but not among adults, raising the question about any increasing exposures that children have had that adults have not had. Breastfeeding has been increasing substantially.
Summary of Section 4.c: In many countries, there have been very low rates of both breastfeeding and childhood cancer.
The UK's incidence for all cancers combined is reported to be higher than those of 90% of the countries of the world. But its incidence of childhood cancer is unusually low.
Preview of Fig. 8
Could the UK's unexpectedly low childhood cancer incidence be related to its breastfeeding rates? (see arrow for UK data)
Summary of Section 5.a: Transfer of PCBs as of birth is minimal, but it increases greatly after birth via lactation.
Summary of Section 5.b: Transfer of dioxins to the fetus is moderate, but that changes dramatically after birth. The major, sharp increases and declines in dioxins in infants during and after the period of breastfeeding are very similar to major, sharp increases and declines in leukemia incidence in children during the years after birth.
Summary of Section 6 through 6.a: There have been substantial international variations in childhood cancer incidence, with levels and changes that are notably parallel with levels and changes in breastfeeding. This section draws heavily on publicly-available data, including from international cancer registries. Reporting that includes estimates of cancer patterns based on such data has received authoritative recognition over several decades as having been basically accurate.
In almost all of the European countries of substantial size, there have been parallel relationships between breastfeeding rates and childhood cancer incidence -- highs, lows, mid-levels, increases, stable periods, and declines.
Summary of Section 6.b: In 1986 the Nordic countries received considerable radioactivity as a direct result of the Chernobyl nuclear accident, which was well known in those countries. Ample scientific evidence as of that time indicated that radioactive toxins would pass through human milk to infants, and governmental advice reflected that knowledge. There were almost certainly temporary declines in breastfeeding taking place in those countries at that time. Three years later there was a simultaneous major temporary drop in childhood leukemia in all Nordic countries, for which there was no known cause.
Summary of Section 8: In some important respects, the case linking breastfeeding with childhood cancer is more substantial than was the case against cigarette smoking when a U.S. Surgeon General first came out in strong opposition to that practice.
Summary of Section 9: As explained, there are many good reasons to see toxins in human milk to be a cause of childhood cancer, which has been increasing along with increases in breastfeeding. The highly respected Sir Austin Bradford Hill posed the question that ought to be asked, "Is there any other way of explaining the set of facts before us, is there any other answer equally, or more, likely than cause and effect?" He did not say that lack of proof means we should do nothing about what appears to be a cause of adverse outcomes.
Professionals who ought to be knowledgeable on this topic do not dispute the findings indicating that breastfeeding transfers both dioxins and PCBs in amounts that far exceed established safe doses. And both of those chemicals are authoritatively considered to be carcinogens. When asked, those professionals do not suggest any other possible pathway that might exist for widespread exposures of infants to toxins in unsafe doses. Hill's basic question, above, deserves serious consideration by anyone concerned about reducing childhood cancer.
Chemical exposures that far exceed established safe levels
Appendix B: Chemical exposures determined to exceed safe levels even more than previously thought
Appendix C: Ongoing sources of PCBs, dioxins and PBDEs, including in traffic emissions
Appendix D: More about PBDEs
Appendix E: Low quality of evidence from observational studies
Appendix F: In the Nordic countries, there were especially high early increases in both breastfeeding and childhood cancer, followed by parallel plateaus in breastfeeding and childhood cancer
Appendix G: Statement by the American Academy of Pediatrics on absence of toxins in infant formula
Appendix H: Evidence of substantial under-reporting of maternal smoking during pregnancy, leading to inaccurate findings in studies
Appendix I: Sources of serious distortion of studies' findings probably resulting from extra participation in control groups by the health conscious
Appendix J: The Doll and Peto 1981 report in retrospect
Appendix K: A possibly unique pathway of widespread infant exposure to toxins in doses exceeding established safe levels
Appendix L: Exposures of mothers to toxins and times of exposures of the developing infant to those toxins
Appendix M: Basis for recommending against breastfeeding in case of nuclear accident, which would have predated the Chernobyl meltdown
Appendix N: The normal relationship between higher education and breastfeeding in China, unlike that in developed countries
Appendix O: “Recall bias” as a problem in observational studies
Appendix P: Long-term accumulation of body burden of dioxins
Appendix Q: Other reasons to be skeptical about findings of studies in a matter, such as breastfeeding, about which strong opinions are widely held
Appendix R: Very misleading statement by the U.S. CDC on toxins in breast milk
Appendix S: Additional indications of inaccuracy in typical studies such as those dealing with breastfeeding
Appendix T: Incidence trends of the five most common types of cancer in U.S.
Appendix U: Breastfeeding and childhood cancer in France
Appendix V: Other parallels between breastfeeding and childhood cancer in Eastern Europe
There appears to be no reason to doubt the generalization that, for all of human history until well into the 20th century, human milk was the ideal food for infants. But, beginning in about the middle of the 20th century, exposures to developmental toxins in modern environments greatly increased. A basic element of much of the economic development that took place since then, and an important contributor to environmental toxins, has been production of chemicals, which rose dramatically during the last 70 years or so.
Beginning in the years following World War II, there were major environmental increases of chemicals to which humans became exposed.1is especially strong about increases in exposures to dioxins and PCBs (the two toxins of principal concern in this article) beginning in the 1940's.2 PCBs are ranked by the U.S. Agency for Toxic Substances and Disease Registry (ATSDR) as being among the top 5 priority hazardous substances.2a Dioxin is "often called the most toxic man-made chemical," as stated in a document of the EPA, with no question expressed about the correctness of that assessment.4 Dioxins and PCBs are both considered by the International Agency for Research in Cancer (IARC) to cause cancer in humans.5
Chemicals enter the human environment not only via the well-known routes (smokestacks, etc.) but also via off-gassing from materials in older buildings,8 from pollutants in landfills and on land, and via emissions from many widespread forms of combustion; included in the latter category are both diesel- and gasoline-powered vehicles,8a,8b,9d and wildfires.9 Dioxin releases generally increase with chlorine content of the combustibles,9b,9d so the rapid increase in uses of plastics (which typically contain chlorine) beginning after 19459c would have greatly increased dioxin emissions from the previously fairly harmless practice of backyard burning, beginning not long after the increase of plastics.9 Similarly, the huge increases in vehicular traffic over the course of the 20th century, with accompanying scores-to-100-fold increases in dioxin levels within a few hundred feet of major traffic locations,8b would have had a substantial impact on toxic emissions close to people.
PCBs: A 2017 study (Donat-Vargas et al., 20176) found that, in the population studied, top-third PCB exposures resulting from normal food consumption were associated with a 4-fold increased risk of the only kind of cancer assessed, compared with women in the bottom third. (The main sources of dietary exposure to PCBs are fish, meat and dairy products.61a)
Notice that the dramatically increased risk of malignant melanoma associated with dietary exposures to PCBs appeared to be in a dose-response relationship; dose-response is considered to be good evidence of causation.
This appears to have been an unusually strong study, both in its type (prospective, with 4.5 years of virtually complete follow-up) and in other important respects.6 The authors referred to the only previous study of cancer risk from common PCB exposures in a general population, which found a 5-6-fold increased risk of melanoma associated with being in the highest versus the lowest quartile of PCB exposure. That was in close agreement with the Donat-Vargas study, given the greater difference between top-and-bottom quartile exposures versus the difference in a similar comparison by tertiles. Again, the higher the (typical) PCB exposures, compared with lower exposures, the greater the increased risk of cancer.
Another study found that upper-quartile exposures of children to PCBs as measured in their residential carpet dust was associated with a nearly-three-fold increase in risk of ALL leukemia.72c The risk of leukemia increased with increases in concentrations of PCBs in the dust. The tripled risk of leukemia was associated with rather ordinary differences in environmental exposures to PCBs. The authors explained various ways in which PCBs are thought to cause cancer, including promotion of cell proliferation and toxicity to immune cells; immunotoxicity of PCBs was said to have been especially well demonstrated, and with specific reference to ALL leukemia.
Dioxins: Multiple studies concur in showing an appreciable excess risk of cancer due to relatively small increases in dioxin above background levels, such as increases that could result from dioxin in normal foods.61 (Animal fats are the main sources of major dioxin contents in foods.) In a 1998 Dutch study of 1031 workers in a chemical plant, those who received merely “medium” exposures to dioxins were found to have 5-fold increases in cancer mortality compared with workers at the same plant with low exposure.6a The U.S. National Toxicology Program has stated that there is no known safe dose or threshold below which dioxins would not cause cancer.7 In the IARC’s 1997 evaluation, dioxin was unprecedented in that it was judged to cause an increase in cancers at all sites rather than at only a few specific sites.7
Given the evidence of carcinogenicity of both PCBs and dioxins separately, when exposures are at common levels, consider the extent to which infants are typically exposed to both of these toxins, and in concentrations far exceeding established relatively safe levels, as will be seen in Figure 2 just below.
Discussing passage of pollutants from the environment to young children, the American and Danish experts, P. Grandjean and A. A. Jensen (who are authors or coauthors of 481 and 127 scientific studies and articles respectively), pointed out in 2004 the following: "these substances have caused contamination of human milk only during the last half century, and long-term health impacts are now being discovered." Specific chemicals to which they referred were PCBs, dioxins, and brominated flame retardants (PBDEs).9a
Dioxins and PCBs in human milk
For original source of the above, go to van den Berg et al., WHO/UNEP Global Surveys of PCDDs, PCDFs and DDTs in human milk and benefit-risk evaluation of breastfeeding, at https://www.ncbi.nlm.nih.gov/pubmed/27438348
There is considerable other evidence that supports the findings indicated in these charts (see Appendix A). And there are other carcinogens of concern here in addition to PCBs and dioxins: according to a team of authors, six of whom were with the EPA, writing in 2018 in a peer-reviewed journal that receives support from the U.S. National Institute of Environmental Health Sciences, "many environmental chemicals commonly found in breast milk are also known carcinogens."22 (emphasis added)
Note that infant formula has been found to be essentially free of these toxins. According to a major 2012 document of the American Academy of Pediatrics (which strongly promotes breastfeeding), referring specifically to PCBs, PBDEs, and major types of pesticides, "Infant formula is free of these residues...." (then going on to explain the various factors that contribute to that outcome). (see Appendix G) Health Canada's Bureau of Chemical Safety (on the basis of 120 samples, mostly imported) found dioxin to be present in infant formula in concentrations less than one percent of the EU limit for dioxin equivalency in milk and dairy products.32
We should also note that the toxicity standards that were in effect during the 2000-2010 surveys, results of which are shown in the charts above, have been subsequently replaced by far stricter standards (see Appendix B); so the updated exceedances of established safe intakes of these toxins, according to the best currently-available evidence and research methods, are many times the already major exceedances shown in the charts above. There is also substantial other evidence, from the U.S. and other developed countries, indicating that exposures of breastfed infants to both PCBs and dioxins during the 2000's have exceeded established safe levels by scores to hundreds of times.10
Section 2: How PCBs and dioxins become concentrated before exposure to infants:
PCBs and dioxins are chemically related: both are lipophilic (attracted to fat, which is why they become attracted to human milk); both are considered to be “persistent” in the environment and in the body; and both are considered by the International Agency for Research on Cancer to cause cancer in humans.5
The U.S. ATSDR, in support of its statement that "the amount of PCBs transferred to offspring is expected to be higher during lactation than during gestation," refers to a laboratory experiment as an example: "In female rats administered PCBs before gestation, an average of 0.003% of the administered dose was transferred to the fetus, whereas 5% was transferred to sucklings."18 That works out to lactational transfer of PCBs over 1600 times greater than gestational transfer, resulting from the same original prenatal exposure; this is a ratio that the ATSDR obviously considers to be relevant to how human lactational exposure compares with human gestational exposure to PCBs. Studies of humans have found lactational transfers of toxins to be hundreds of times greater than gestational transfers.19 It may be worth reviewing the dramatic extent to which exposures of breastfed infants to PCBs and dioxins are considered to exceed established safe levels. (Figure 2 and accompanying text)
Declining levels of these toxins? PCBs and dioxins have been declining in many countries' overall environments since the 1970s; but they are called persistent for good reason, and they continue to be emitted by many widely-distributed processes and structures to which people are closely exposed, including vehicular combustion, smoking, backyard burning, and older buildings -- see Appendix C. As shown in surveys from the 2000s in Figure 2 and in the accompanying text, those toxins are still present in typical human milk in concentrations far exceeding established safe levels, decades after their emissions started to decline. A German study found that, in 2003 after dioxins in breast milk had substantially declined and then practically leveled out over preceding decades, a breastfed infant would still take in dioxins in the range of 100 times the intake of an adult.19a A study published in 2014, over three decades after PCBs were mainly outlawed, found the following remarkable result: even at 45 months of age, PCB levels of children who had been breastfed for 7 months were still 9 times as high as those of children who had not been breastfed.19d
And, while PCBs and dioxins declined, other chemicals with similar toxic properties, such as PBDEs, greatly increased in environments and in human milk. (See Figure 3 below and Appendix D.) According to the U.S. EPA, studies with rats and mice have shown that PBDEs cause cancer.19b
A study by a team of seven scientists estimated PBDE intake from food to be 307 ng/kg/day for nursing infants compared with 0.9 ng/kg/day in adult females.20 The EPA quotes estimates from two other studies that were almost that extreme.20a
Concentrations in the body are especially of concern during the highly vulnerable period of infancy, a period of rapid growth and a time when major increases in leukemia are already strongly beginning (see Figure 10). A German study found that the body burden of dioxin-like compounds was more than ten times higher for breast-fed infants than for formula-fed infants tested during each of two assessment periods.21a This was compatible with data from Karger et al. (2007) that is incorporated into Figure 9, later.
a recognized risk factor for cancer:
The PROBIT study,
with over 17,000 mother-infant pairs, was a randomized trial on the health
effects of breastfeeding, publishe
d in the Journal of the American
Medical Association. Randomized trials are widely considered to be the
gold standard of study types, and the far more common observational studies are
recognized by leading authorities in medical evidence to be highly subject to
false conclusions. (Appendix E)
And large size, such as in this case, is also a recognized important
characteristic in assessing quality of findings.
The researchers gathered data concerning an experimental group whose breastfeeding rates were greatly increased by major promotion of breastfeeding, comparing that with data from a control group whose breastfeeding was relatively minimal. The study found that risk of overweight/ obesity at age 16 was increased among children who had received more breastfeeding.52 A recent major study (Colen et al.) used siblings as subjects, which is another means of avoiding the confounding that is normal in observational studies; and that study, also, found that obesity increased with greater breastfeeding.53
See (below) what happened in obesity of American children following major increases in breastfeeding:
The more-than-tripling of child obesity during these years is compatible with the dramatic increases in breastfeeding (especially long-term breastfeeding) that took place after the 1960s. (See below.)
The increases in breastfeeding in the U.S. were especially notable after 1970, but they actually started gradually in the late 1960s, following a low period in the early 1960s.48 Breastfeeding in most countries of Western Europe also increased greatly, apparently beginning in the 1970s,48a,49 although data for before 1980 is scarce.
Non-increasing obesity among those born before the late 1960s, but a sharp increase later:
Looking at the first two assessment periods shown in the chart below, notice that 18-to-29-year-olds in those years would have been born before the (1970s) period of breastfeeding increases.
Notice that obesity was not increasing among those people who were born before breastfeeding increased. (This is seen in the first two periods shown above). Then, in the first assessment period in which a large part of this age group would have been born in the late 1960s and later (when breastfeeding was increasing greatly), obesity was increasing rapidly.
In explanation of how breastfeeding could lead to obesity, note the findings of the Jacobson et al. 1990 study,54 as follows: "reduced activity (of children) was associated with both 4-year PCB body burden and its principal determinant, exposure to PCB-contaminated breast milk." (There is ample evidence indicating that major PCB contamination is still a normal characteristic of human milk in the 21st century; see Figure 2 and the text following it.) The most-exposed children were found in this study to be ”about 2-3 times more likely to be rated 'usually quiet and inactive.' " The children who were observed were mostly from Great Lakes fish-eating populations, with merely slightly-above-average PCB exposures.55 Three animal studies have also found reduced -- often greatly reduced -- activity to be an effect of postnatal PCB exposure, including at upper-background levels.56 The causal connection between reduced activity and increases in obesity probably needs no explanation. But the recognition of breastfeeding history as the principal determinant of (activity-reducing) PCB levels in children over a period of years is worthy of emphasis. Both inactivity and obesity are independent risk factors for cancer;56v inactivity has been authoritatively described as a "major" risk factor for various types of cancer;56w overweight and obesity have been referred to as "strong risk factors for multiple malignancies." 56x
The American Academy of Pediatrics points to observational studies that have found breastfeeding to be associated with reduced leukemia in children.60f Aside from the matter of the generally low quality of evidence from observational studies (see Appendix E), there are problems that apply especially to studies of effects of breastfeeding:
Section 4.a.1: Women in developed countries who don’t breastfeed, or who breastfeed for shorter durations, are disproportionately of lower socioeconomic status and are typically exposed to more carcinogenic pollution (including tobacco smoke) than are women who breastfeed more.
In developed countries, where the vast majority of studies on breastfeeding and childhood cancer are carried out,72k less-educated, lower-income mothers are unusually likely to smoke and to live in worse air (see below), and they also breastfeed at unusually low rates; the lower-education/lower-breastfeeding relationship is verified in data provided by the U.S. Surgeon General.71 And it is applied to "rich" countries in general in a WHO publication, specifically in relation to people with both lower incomes and less education within those countries.71 A study published in the British Medical Journal discusses that same kind of relationship with reference to "developed" countries.71k
Smoking may be the most important contributor of carcinogenic exposures to mothers and infants. According to a report of the International Consultation on Environmental Tobacco Smoke and Child Health, in consultation with a WHO group, "tobacco smoke, whether voluntarily or involuntarily inhaled, includes numerous carcinogens." 71f The chart below shows newborns' concentrations of one of those carcinogens (PCBs, see Section 1), varying substantially according to higher or lower exposures of the mother to smoking.
When seeing (above) how much difference smoking makes to an infant's carcinogenic PCB levels at birth, bear in mind that PCBs are then ingested by typical infants in concentrations far exceeding established safe doses via breastfeeding (Figures 2 and 8.a); any exposures by inhalation of sidestream smoke would be additional. And PCBs are just one of numerous carcinogens found in tobacco smoke.
When considering the significance of higher or lower levels of PCBs in infants (due to smoking exposure), remember the evidence in the paragraphs in Section 1.a dealing with PCBs, indicating major increases in various kinds of cancer (including leukemia) associated with common increased levels of PCBs, in dose-response relationships, as found in several studies.
Substantial levels of dioxins, also, have been found in cigarette smoke.72d And dioxins, also, are known carcinogens.
See below some of the evidence indicating that relative risk (RR) of childhood cancer increases with increasing intensity of exposure to parental smoking.
(Multiple studies have not found increases in childhood leukemia to be linked with maternal smoking exposure,72o which at first seems to conflict with the major presence of carcinogens in tobacco smoke, and also with the studies that have found very significant positive relationships between parental smoking and childhood cancer.71z,72f,72g,72q,72a But strong evidence indicates that smoking mothers would have under-reported continuation of that habit to researchers because of smoking's recognized harmfulness to the fetus; this would have led to incorrect findings. The evidence in that regard is compelling but too extensive to include here, so it has been placed in Appendix H.)
According to the U.S. Surgeon General, secondhand smoke is a known human carcinogen and contains more than 50 chemicals that can cause cancer; and scientific evidence indicates that there is no risk-free level of exposure to secondhand smoke.73p The CDC also points out that (a) sidestream smoke contains higher concentrations of many of the toxins found in inhaled cigarette smoke, since (being emitted by the tip of a cigarette) it is generated at lower temperatures than mainstream smoke;73q and (b) because their bodies are developing, infants and young children are especially vulnerable to the poisons in secondhand smoke.73r
Considering the good reasons to see a causal link between parental smoking and childhood cancer, the following should be of interest: Mothers who breastfeed, especially for medium or longer durations, are very predominantly non-smokers. It was found in a 2016 study that 97.6% of U.S. women who exclusively breastfed for 3 months or more were non-smokers, prenatally.72b
On the other side of that coin, non-breastfeeders are disproportionately smokers,71s,71v, with a nearly four times greater likelihood of early weaning among daily smokers, as found in a review of studies.71t An Australian study pointed out that women who smoke, being aware of possible harmful effects on the infant but having difficulty quitting, are likely to stop breastfeeding instead.71u A dose response effect has been found, with the heaviest smokers having the least likelihood of establishing exclusive breastfeeding.71v
Therefore, with smoking being so closely linked with non-breastfeeding, and smoke being so carcinogenic, an association observed between cancer and lack of breastfeeding is very likely to have had its actual origin as an outcome of smoking. But the lack of breastfeeding is what is observed and reported, while the (very likely causal) smoking is normally not reported.
Health consciousness is central to this discussion. An authoritative assessment of the weaknesses of the observational studies about hormone replacement therapy (HRT) indicated that those studies' misleading findings were mainly attributed to three principal confounders, including socioeconomic status; but the first confounder to be mentioned was health consciousness, which differed between the groups being compared.73w An article provided on the website of the U.S. CDC also points to the primary importance of differences in health consciousness between the groups that were compared, in explaining how the studies that supported HRT could have been so subject to later reversal. Better health outcomes were seen in women who received the hormone therapy, and it was concluded that the hormone therapy caused the better outcomes. But subsequent, higher-quality studies were able to create more equal health consciousness in the two groups being compared, and that resulted in the reversal of the earlier (confounded) findings.74f
The CDC also pointed out how heavily influenced the medical establishment can be by studies that reach such erroneous conclusions that they will later be reversed by higher-quality studies: in the 1990s, about half of the postmenopausal women in the U.S. were taking HRT.74f
It is well established that health-conscious people are normally very over-represented in studies' control groups (comparison groups, made up of people free of the disease in question). Health-conscious women are unusually likely to be willing to participate in studies that are unrelated to problems in their families; and, paying attention to doctors' recommendations, they are also unusually likely to both (a) breastfeed, and (b) to provide their children with environments that have minimal exposures to carcinogens. An association is found between these women's high breastfeeding and their children's lower incidence of cancer; and the lower exposures to carcinogens among these health-conscious women's families are overlooked, in the studies' findings. These women's high numbers in control groups lead to distortions when their children's health data is used for comparisons. Supporting evidence for this, and more information about the confounding effects of health consciousness in studies, will be discussed in Section 4.a.4.
Less-breastfed children are also exposed to worse air in general. In addition to exposures to carcinogens in tobacco smoke, there is also considerable evidence indicating that, in developed countries (where most studies are carried out), children who are breastfed less are also exposed to more carcinogenic air pollution in general. It is very much a matter of socio-economic status: as mentioned earlier, in developed countries, less-educated, low-income mothers typically breastfeed less than average.71 They also live and/or work in places where toxic air pollution is greater, due to industry, vehicular traffic, heating combustion, and other sources of toxic air that are concentrated where populations are dense.72z It is common that, due to high housing costs, low-income neighborhoods are clustered around ports, industrial sites, truck routes, and other air pollution hotspots.73k Benzene and diesel engine exhaust are specifically recognized to be carcinogenic,71h,71b and leukemia has been found to be four-fold higher where exposures to benzene from road traffic were found to be higher;71y automobile exhaust (more concentrated in the urban areas discussed) is considered by WHO to be the largest source of benzene in the environment.73b
It has been found that people in low-SES groups are also more likely to consume unhealthy foods and to be physically inactive and overweight or obese -- all of which are linked with cancer.73g Many pesticides, also, are recognized to be carcinogenic;71w,71q,71r and exposures to pesticides are higher in densely-populated, low-income neighborhoods and among farm workers. In a 2016 study in Iran, it was found that children of mothers in the least-educated of four groupings were 390 times as likely to be diagnosed with leukemia and lymphoma as children of mothers with university degrees; children whose family incomes were in the lowest of three groups had 286 times the risk of those diseases compared with children in the highest group; but no significant relationship was found between breastfeeding duration and those diseases.72r
All of the above support the generalization that children of less-educated, low-income mothers (who breastfeed less) are unusually subject to cancer, for reasons other than the breastfeeding.
But the focus of most studies is directed toward what fits in well with prevailing beliefs, as represented in the simplified picture as shown on the left below:
The comparisons are not at all complete or valid unless they include the accompanying characteristics on the right, to give due consideration to differences in environmental exposures between breastfeeders and non-breastfeeders. There is considerable evidence indicating that the factors on the above right
(a) are normally found in combination with the respective breastfeeding levels on the left, and
(b) are likely to be the real causes of the differences in leukemia incidence. (see earlier)
But it is almost certainly easier to complete a study if just part of what matters (such as that on the left, above) is the focus of attention. The completed studies then help in getting publications to one's credit, for career advancement. Findings that support prevailing beliefs can be expected to facilitate getting approval by peer reviewers and publishers as well as to encourage future funding.74h (There can be no doubt that there is a strong prevailing belief that breastfeeding is beneficial for children.)
Researchers typically attempt to arrive at valid findings by statistically compensating ("adjusting") for various confounding factors. But, by excellent authority, that is unlikely to lead to valid results. (See "Adjusting the data" later, in Section 4.a.4.) There are good reasons why leading authorities on medical evidence consider evidence found in observational studies (such as those that have associated child leukemia with less breastfeeding) to be of generally low quality -- see Appendix E.
As pointed out by one research group, "the upsurge in breast‐feeding after the mid‐1970s" (see Figure 5) was followed by increases in childhood leukemia.121 Aside from that, and in addition to the undisputed knowledge about presence of known carcinogens in human milk in concentrations far exceeding established safe levels (Figure 2 and accompanying text), there is also very considerable epidemiological evidence indicating that breastfeeding has been leading to increased cancer (see just ahead and Section 6, later).
Section 4.a.2: When a study is carried out in a country where, contrary to the pattern in developed countries, the more-educated are commonly not more likely to breastfeed, we see that more breastfeeding is associated with increased child leukemia.
As mentioned, in published reviews of studies of breastfeeding in relation to childhood cancer, almost all studies reviewed were carried out in developed countries. In a 2004 meta-analysis, 13 of the 14 studies analyzed by Kwan et al.71m were carried out in developed countries. The 14th study (Shu et al., 1995) was carried out in China, where the evidence was very mixed but showed a tendency toward highly-educated mothers' being less likely to breastfeed exclusively than average (see Appendix N); and, contrary to the pattern normally found in developed countries, more breastfeeding was found to be associated with slightly increased childhood leukemia in China, in all 6 of the 6 categories assessed. The numbers in the individual associations were not sufficiently large to reach the typical standard for statistical significance, but their complete consistency in positively associating breastfeeding with childhood leukemia (as presented in the meta-analysis) was noteworthy.
Section 4.a.3: When a study is carried out with selection of a comparison group that avoids the usual confounding, the relationship between breastfeeding and leukemia is again very different from what has typically been found in studies.
A 2018 study (Lingappa et al.), carried out in India, was especially free of the normal confounding because it used a control (comparison) group consisting of older siblings, thereby achieving unusually good similarity in socio-economic status and environmental exposures between disease cases and controls.72s Based on substantial online research, this is the only published study of significant size on the relationship between breastfeeding and childhood cancer that has had a characteristic that sets it clearly above the quality level of the typical observational studies. So it should be of interest to see the findings of this study that is unusually unaffected by the typical confounding (below):
Notice in the bottom 3 rows above that, for those who were breastfed for more than a year, the odds (OR) of Acute leukemia were reduced (1.0 is the no-effect level); however, just above those lines notice that, for those who received continued breastfeeding for a year or less, the risk of leukemia was greatly increased.
So continued breastfeeding for a (rather typical) period of up to a year after birth was associated with a greatly increased likelihood of the child's being diagnosed with leukemia.
On the other hand, breastfeeding well into the second year was associated with reduced leukemia, and breastfeeding exclusively for more than six months was also associated with reduced leukemia. So, on the basis of this study that is unusually free of the typical confounding, one might suggest that what a mother should do in order to gain beneficial cancer-related effects of breastfeeding is to continue breastfeeding well into the second year after birth; and she should also breastfeed exclusively for most of the first year. But such goals would be difficult or impossible to achieve for a very high percentage of mothers; according to CDC data accessed in mid-2020, only 36% of U.S. mothers breastfeed for as long as 12 months, and even fewer breastfeed exclusively for six months; and most mothers do not breastfeed for as long as they intended to.72w Reasons for discontinuing include, "sore, cracked, or bleeding nipples," "too painful," "breasts infected or abscessed," sickness, "infant began to bite," low milk supply, inverted nipples, and trouble sucking72x (most of these meaning that the infant could not take in sufficient nutrition). Additional problems would be the mother's normal need to return to work full time, and her physical and mental needs for uninterrupted sleep. All of those are understandable, compelling reasons for not breastfeeding into the later periods when childhood-cancer-related benefits of breastfeeding seem to become attainable, as indicated in this unusually good study.
Therefore mothers should be cautioned that, even though they may intend to reach a duration of breastfeeding at which cancer risk is reduced, there is a high risk that they will discontinue sooner; they would very likely only reach a duration at which the outcome of the breastfeeding appears to be greatly increased cancer. See the 5-fold and 10-fold increased odds of cancer in the latter part of the first year of continued breastfeeding, in the chart above.
It is worth carefully considering why the outcomes associated with medium durations of breastfeeding are much greater risk of leukemia, whereas the outcomes go over to reduced leukemia with longer durations. That is actually as should be expected, given the decline in transfer of toxins in breast milk that takes place over the course of many months of lactation. Decreasing infant intakes of dioxins and PCBs (per kg of body weight) during lactation are indicated in data from three UK cities, below.
Decline of dioxins and PCBs taken in over the course of 10 months of breastfeeding, Part 1:
(Above chart accessed at https://cot.food.gov.uk/sites/default/files/cot/cotsuremilk.pdf)
Most of the declines in intakes shown above are results merely of increases in the infant's body weight, into which non-increasing toxins in the milk would be distributed. However, the UK's Committee on Toxicity points out that "all data (shown above) are derived from pooled milk and do not allow for the decreasing concentrations of dioxins and PCBs with duration of lactation." So the well-recognized declines in concentrations of dioxins and PCBs in human milk that take place during lactation, as the mother's long-term accumulations of toxins are depleted,72y, 67,124 would be in addition to the declines shown above.
Therefore it is very plausible that the effects of the toxins on cancer risk would be substantial during the earlier period after birth when the amounts of toxins taken in have been larger (and in high concentrations -- Figure 2)); on the other hand, the beneficial nutrients in human milk would have a predominant effect after the intakes of toxins in the milk have declined greatly. That transition would lead to the high odds ratios (ORs) for leukemia associated with breastfeeding and weaning within a somewhat limited period after birth, followed by reduced ORs later, as seen in Figure 6.d above.
Notice that the transition from increased odds to reduced odds of leukemia was found to take place sooner with exclusive than with continued breastfeeding. That is in line with the more rapid depletion of the mother's stores of toxins that would take place when breastfeeding is exclusive.
The above study, carried out by a means that avoided the normal confounding that distorts the findings of typical breastfeeding-leukemia studies, provides evidence that suggests an important relationship: Childhood leukemia appears to be greatly increased when children are breastfed for durations that are normal and achievable for most women in many countries. There is considerable other evidence indicating that this is what has actually been happening all over Europe, to be presented in Section 6. Aiming to breastfeed for the longer durations that are associated with reduced cancer risk would be goals that are likely to leave many mothers falling short -- into a period of greatly increased cancer risk for the child.
An Iranian study (Mohammadi et al.72i) found a pattern somewhat similar to that of the Lingappa study:
Similarly to the LIngappa study (Figure 6.d), the ORs for the different periods in this chart show (a) during earlier months of lactation, a positive association of exclusive breastfeeding with childhood leukemia, (b) a greatly reduced, actually negative, association in later months (at 4-6 months in the above chart); (c) at the final stage, evidence of an increase of leukemia in relation to continued breastfeeding.
A proposed explanation has already been offered for the (a) and (b) stages presented above: as indicated in Figure 6.e and accompanying text, exposures to carcinogens in human milk are highest initially, but the concentrations decline as the infant's weight increases and as the mother's burden of the toxins is depleted. In explaining the last stage (c), the upturn in risk of leukemia with continued exclusive breastfeeding, the authors of the Mohamaddi study suggested the following: after 6 months of age, a child would need sources of nutrients other than breast milk, especially iron, to prevent "anemia and malnutrition which may be associated with an increased risk of developing leukemia."72i
This study did not have the high-quality features that were characteristics of the Lingappa study (siblings for comparison group, and breastfeeding exposures and ORs broken down into numerous duration periods); but it did indicate ORs for four separate breastfeeding exposure periods, and it was carried out in a country where the usual confounding by health consciousness among breastfeeders may have been reduced or absent. And it did arrive at findings that support those of the Lingappa study, reinforcing a pattern of increased leukemia associated with more breastfeeding during early months, transitioning to a negative relationship later, and returning to a positive relationship at the final stage.
Confounding by higher socioeconomic status and health consciousness among women who breastfeed more might be absent in a developing country, but that is not always the case. So the few studies on this subject that have been conducted in developing countries are inconsistent about the correlation between breastfeeding and childhood cancer. Another problem that is apparently present in many studies is lumping data for no-breastfeeding together with data for breastfeeding for up to 6 months, resulting in combined figures that are reported but which are likely to have very little rational meaning.72l
Section 4.a.4: Control groups, a source of bias and misinterpretation in many studies:
When researchers carry out studies, trying to make valid comparisons between a group of subjects with a disease and a sample of people in the same population without the disease, they typically try to take into account confounding factors by attempting to match cases and controls and by "adjusting" the primary data. But that holds little promise for effectively preventing confounding, at least in the studies on the topic discussed here, as will be discussed below.
Possibly the most serious weakness that distorts the findings of essentially all of the typical studies in this area is that many (sometimes most) people who are asked to participate in the studies are not willing to do so; this can greatly affect the types of people who end up providing the data in the studies. People who are willing to participate in medical research studies are unrepresentative of the general population in that they are typically more health conscious, better educated, and of higher-than-average socioeconomic status (SES), according to many studies.73c
The SES difference has consistently been found to apply particularly to people who are willing to take the time to participate in cancer studies' control (comparison) groups.71r,73c,73y According to one research team, providing data from studies carried out in the US, UK, Germany and Canada, controls have included an average of over 20% fewer individuals in the lowest SES or income category and over 30% more in the highest SES category, compared with the disease cases.73h People of higher SES are often more health conscious and more willing to take the time to participate in studies intended to advance medical knowledge, compared with people of lower SES; and that would disproportionately affect the makeup of the control group, whose members would not be personally affected by the disease being studied.73u Higher levels of health consciousness in control groups are known to occur often when people declining to participate in studies are replaced by people who are willing and health conscious.73y,73z,74a,74b,74c
When discussing case-control studies on nutrition and cancer carried out previously, one research team noted that "participation among controls was often low and the control participants represented a more health-conscious group than the study population."74d A Spanish research team put it succinctly in a 2011 publication: "Control subjects who agree to participate are typically very health-conscious persons…." 74e
The resulting differences in health consciousness between case and control groups could clearly have very distorting effects on comparisons between those two groups in studies of disease risk. Those who are health conscious would be very unlikely to smoke, would tend to have better diets and exercise more,74e would minimize their families' exposures to pesticides, and would typically reside in places more removed from carcinogenic exposures such as traffic pollution. Children of health-conscious mothers (in control groups) would therefore have less cancer risk because of reduced carcinogenic exposures. But researchers would be likely to only point out that those (health-conscious) mothers breastfeed more; and the reduced childhood cancer would be attributed to greater breastfeeding.
There is substantial evidence of the serious extent of the probable distortion in studies' findings resulting from the elevated health consciousness in control groups, and some of that evidence can be found in Appendix I.
"Adjusting" the data: Researchers often attempt to compensate or statistically "adjust" for SES differences between study groups being compared. But in the Kwan et al. meta-analysis71m discussed earlier, many of the studies analyzed did not even try to match or adjust for differences in maternal education or socio-economic status; only two considered differences in smoking; and none attempted to adjust or control for differences in health consciousness. (An authoritative article on sources of bias in studies mentions health consciousness distinctly separately from SES.73w). Of the studies that did try to match appropriately or make relevant adjustments, their likely success in such efforts is very much in doubt, considering
-- the crudeness of the criteria used, such as telephone exchange or degree of urbanization, when trying to match cases with controls,
-- problems of adjusting, with even minimal accuracy, for differences in carcinogenic exposures between cases and controls,73f and
-- the very negative comments that a committee of the U.S. Institute of Medicine and a separate, authoritative research team have made about statistical adjustment of that kind and about the impossibility of correcting for unknown factors (see Appendix E).
Researchers would observe that breastfeeding is higher among the control (cancer-free) group and would focus on that. The fact that the control group is very disproportionately made up of people with healthy lifestyles and minimal exposures to carcinogens is normally not given suitable attention.
Returning to the topic of why the Lingappa study was able to arrive at findings about breastfeeding's effects that (a) stood out from those of typical studies and (b) were almost certainly more accurate: as mentioned, the control group in that study was made up of older siblings of the children with leukemia, providing reasonable assurance that comparisons would be made between cancerous and non-cancerous children with similar exposures to environmental carcinogens. This would allow differences in effects of breastfeeding to be seen more correctly.
Section 4.a.5: Brief review of important points:
Professionals who ought to be knowledgeable on this topic do not dispute the very high levels of known carcinogens in human milk, nor do they suggest any other possible pathway for widespread exposures of infants to toxins in unsafe doses. (Summary of Section 9)
Ordinary dietary exposures to PCBs have been strongly linked to increased cancer incidence. The U.S. National Toxicology Program has stated that there is no known safe dose or threshold below which dioxins would not cause cancer. (Section 1)
In the only published study of significant size on the relationship between breastfeeding and childhood cancer that has had a characteristic that sets it clearly above the quality level of the typical observational studies (Lingappa et al.), the risk of leukemia was greatly increased among children who received breastfeeding for a year or less. That would apply to the vast majority in developed countries.
Even though mothers may intend to breastfeed for long enough to reach a duration at which childhood cancer is reduced, most mothers do not breastfeed for as long as they intended to; there is a high risk that they will only reach a duration at which the outcome of the breastfeeding appears to be greatly increased cancer.
Higher levels of health consciousness in control groups are known to occur often when normal people, declining to participate in studies, are replaced by volunteers who are health conscious. Children of health-conscious mothers (in those control groups) would therefore have less cancer risk because of reduced carcinogenic exposures. But researchers would be likely to only point out that those (health-conscious) mothers breastfeed more; and the reduced childhood cancer would be attributed to greater breastfeeding, as reported in studies.
Above chart found at https://seer.cancer.gov/csr/1975_2016/browse_csr.php , selecting Childhood Cancer and Figure 28.1, for ages<20
For comparison with what has been happening in the population in general, total cancer statistics are not useful since they are distorted by the effects of the major changes that have taken place in smoking. But Cancer.gov shows data for cancer incidence trends by type of cancer, and Appendix T shows those trends for the most common types of cancer individually, from which we can exclude lung and bronchus cancer because of their link with smoking; a glance there will make it obvious that cancer in general has not been increasing significantly, if at all. Given the increases that have been taking place specifically in childhood cancer, it is therefore extremely important to carefully consider any and all sources of carcinogens to which children specifically, and not adults, may have been increasingly exposed in recent decades. The high levels of carcinogens in human milk (Figure 2), consumption of which has been greatly increasing (Figure 5), would be a logical exposure to think about.
As pointed out in a systematic review with meta-analysis on this subject, the continued increase in childhood leukemia in the U.S. while breastfeeding has been rapidly increasing is in conflict with the claims of breastfeeding's being protective against leukemia.121 But a reasonable interpretation of these temporal trends should go beyond merely seeing lack of positive effect of breastfeeding in this area, if one takes into account the undisputed presence of carcinogens in human milk in high concentrations.
In what may be the only other well-supported hypothesis that has been proposed to explain the increase of cancer distinctly among children, decline of immunity-stimulating infections in young children in developed countries has been suggested, in relation to leukemia. In case-control studies, reduced ALL leukemia has been consistently associated with greater exposure to infections, via daycare and later birth order.64e Readers may want to consider how development of the child's own immune system, which should be stimulated by early infections, would fit in with the (temporary) suppression of infections by the immune cells that are present in human milk. Breastfeeding advocates often point out the reduction in infections during infancy that results from the maternal immune cells transferred via breastfeeding.
Section 4.c: Possibly surprising relationship between breastfeeding and childhood cancer in the UK and elsewhere:
The UK's incidence for all cancers combined is reported to be higher than those of 90% of the countries of the world;61 this may be related to the UK's problems with air pollution.62 However, in stark contrast with adult cancer rates, childhood cancer incidence in the UK is unusually low.89 One might wonder whether that difference could be related to the UK's breastfeeding rates.
Above OECD chart can be accessed at https://www.oecd.org/els/family/43136964.pdf
When looking at the above and relating it to the UK's remarkably low childhood cancer rates, bear in mind the major exceedances of carcinogens in human milk in relation to established safe levels -- see Figure 2 and accompanying text.
The only European countries with lower childhood cancer incidence rates than the UK, according to the International Agency for Research in Cancer, are in Eastern Europe: Bulgaria, Lithuania, Poland, and Ukraine.89 And all four of those countries have breastfeeding rates that are either near-bottom or at the bottom of rates shown in international charts that permit comparisons.89f, 89j, see Figure 14 re Lithuania and Ukraine
Major, matching ethnic differences in both breastfeeding and childhood cancer: Indian children in Leicester, England were found to have a statistically significant 82% increased risk of all cancers combined compared with white children, with evidence of more than doubled risks of leukemia and CNS tumors.89i In that regard, it may be relevant that Indian mothers in the UK have been found to breastfeed at over twice the rate of white mothers at three months after birth.89i
The above is just a preview of substantial evidence indicating a link between breastfeeding and childhood cancer. There is considerable other such evidence to be presented just ahead.
By way of introduction to more details about childhood cancer in recent decades, a generalization stated in a publication of the U.S. Institute of Medicine should be quoted, as follows: "The fact that childhood cancer occurs in the context of rapid and dramatic growth and development… distinguishes it from adult cancers." 64d Childhood cancer latencies can be very short.
The 2015 study that was the source of the charts below analyzed data from seven European birth cohorts. Notice how the mother's accumulated levels of PCBs decline as the chemicals are transferred to the infant, and observe that the infant's PCB levels continue rising rapidly while breastfeeding continues.
Notice (in the right-hand chart) how much higher the typical breastfed child's PCB levels are than the mother's, even years after nine months of breastfeeding. Then remember (from Section 1) the studies that found that differences in a person's PCB levels, varying according to different ordinary exposures, were associated with major differences in risk of malignancies.
The lines showing decline of the mother's PCB levels during breastfeeding provide visualization of the process that helps explain an often-mentioned benefit of breastfeeding: a mother's levels of carcinogens (therefore her risk of cancer) are reduced by lactation. The longer the breastfeeding continues, the greater the reduction of her levels of carcinogens.
But some people might be concerned about the opposite side of that coin, which is seen in the upper lines: as the mother excretes PCBs, levels of those chemicals in the breastfed infant increase rapidly, farther into the range in which exposure has been shown to be a risk factor for cancer. (Figure 1.a and text above and below it)
Remember the statement in a major 2012 document of the American Academy of Pediatrics that infant formula is essentially free of these toxins, referring specifically to PCBs, PBDEs, and major types of pesticides.31
Perfluorinated chemicals, or PFASs: A maternal-excreting-and-infant-ingesting process also takes place with regard to the perfluorinated chemicals found in emissions resulting from stain- and liquid-repellant and non-stick coatings. Unlike PCBs and dioxins, these chemicals are not confirmed as being causes of cancer in humans, but there is substantial evidence indicating that they are carcinogenic, including "clear evidence of carcinogenic activity" in male rats.31a PFASs have also been found to very significantly reduce immune function, especially as a result of exposures during the first six months after birth, while the immune system is developing.31b
Major transfers of PFASs via breastfeeding: A 2018 study found that exclusive breastfeeding was associated with increases of about 17% per month in PFOS and PFOA levels in an infant. (Table 3 of Grandjean et al., ref. 31b) A 2017 study found that infant serum PFAS concentrations increased about 30% per month of exclusive breastfeeding, while concentrations declined slightly for non-breastfed children31c (see below). (Differences in the above studies were probably attributable to variations in mothers' exposures in different local environments.) In both studies, the findings quoted had very high levels of statistical significance. (p-values of <0.001)
The original from which the above was taken is at http://nrs.harvard.edu/urn-3:HUL.InstRepos:37221737
The authors of a 2014 study (Mondal et al.31d) referred to a longitudinal study that found that PFOA concentrations in breastfeeding mothers fell by about 94% during one year of nursing; the Mondal authors considered this to be compatible with their own study's findings of 34% and 60% declines for two different groups, taking into account the differences in current exposures of the mothers during the breastfeeding year, which would replenish levels of chemicals being excreted.
Even though there is substantial variation in how far the mother's concentrations of the toxins are reduced by breastfeeding, it appears that the excretion of toxins (also including PCBs, PBDEs and dioxins) is probably sufficiently large to explain the reported reduction of cancer in women who have breastfed. On the other hand, the effects of those excreted toxins in the small child who ingests them ought to be of at least equal interest; this will be considered in Section 6.
These chemicals, like PCBs and dioxins, are persistent in the body; in a 2018 study, it was determined that, at three years after birth, children who had been breastfed for over one year still had PFOA and PFOS serum concentrations about 2.5 times their concentrations at birth; during that same interval, concentrations in non-breastfed children had declined.31e Remember what was said early in this subsection about evidence of carcinogenicity of such chemicals and of their leading to reduction of immune function while the immune system is developing after birth.
Section 5.b: Transfer of dioxins to the fetus is also limited, but that, too, typically changes rapidly after birth:
Compared with dioxin levels in maternal blood, dioxins in cord blood of fetuses have been found to be much lower;65 the ATSDR concludes a substantial paragraph by stating, "these results suggest that the transfer of CDDs (dioxins) to the fetus may be somewhat limited."66 The minimal mother-to-fetus transfer of dioxins apparently results from the filtering, protective effect of the placenta. But at birth the low prenatal dioxin level rapidly turns sharply upward for breastfed children (Figure 9, just below), since human milk contains concentrations of dioxins that come not only from current exposures but also from the mother's long-term, accumulated burden of dioxins (see Appendix P). The ATSDR refers to the “efficient” removal of dioxins from the woman's body burden during lactation.67 If we were to consider the likely effects of those long-term adult accumulations in the small, rapidly growing body that ingests them, we should bear in mind (from Section 1) that,
-- there is no known safe dose or threshold below which dioxins would not cause cancer, according to the U.S. National Toxicology Program7 (although it clearly causes cancer in only a small percentage of cases), and
-- dioxin has been authoritatively found to be present in typical human milk in concentrations far exceeding established relatively safe levels (Figure 2), and in infant formula in concentrations less than one percent of the EU limit for dioxins in milk.32
Early childhood levels of dioxin toxic equivalency
This chart and study accessible at https://www.ncbi.nlm.nih.gov/pubmed/17682006
Note (above) the outstanding peaks of exposures to dioxins during the early months after birth. Then, in the disease chart below, observe the early-childhood peaks of ALL leukemia, in a distinctive pattern very similar to the earlier peaks of dioxin exposures. Childhood leukemia latencies have been observed to be as short as 0.4 year68 and some scientists consider that the latencies may be much shorter than that,68a but they typically vary between 18 months and 4 years.69,70
ALL Leukemia incidence rates for 3 U.S. ethnic groups
Above chart accessed at https://www.acco.org/wp-content/uploads/2014/11/ACS-Special-Report-2014.pdf, p 30
The bottom line in the chart above is for a different type of leukemia. The second line from the bottom, with a very subdued similarity to the upper lines, is for ALL leukemia among African Americans, who breastfeed at unusually low rates.71
It should be noted that the odd, distinctive shape of ALL leukemia incidence is not a natural phenomenon. It apparently was first observed in a minor form in the 1920s in the UK, following no observed peak during the preceding decade, with moderate peaks forming in the 1920s and 1930s, and higher peaks appearing in later decades (see chart below).71e
It has been suggested that electromagnetic fields from power lines, newly installed in the 20th century, could have been the source of increasing leukemia; but that does not provide a good explanation for the sharp decline from the high peak (at about age 3) that has been observed in recent decades. It is of interest that the time of a strong beginning of smoking among women (i.e., current and future mothers of infants) in the UK was during World War I;71x that was several years before a peak in childhood leukemia at age 3 was first recorded in the UK. Related to that, we should remember the text dealing with carcinogenic effects of tobacco smoke in Section 4.a.1, including the quotation from an authoritative body, "tobacco smoke, whether voluntarily or involuntarily inhaled, includes numerous carcinogens." 71f
Note above, as of 1916-20, the absence of the peak at age 3 that was to form in later years.
Whether resulting from power lines or maternal smoking or both, or from something else in the environment, the peak in ALL leukemia was started moderately in the early 20th century, almost certainly caused by something in the environment. But the very high peaks (as shown in Figures 10 and 21) apparently only occurred in recent decades.
The important message is that the outstanding peak in ALL leukemia should not be regarded as an inherent characteristic of the disease, something that we have little possibility of controlling. It appears to have resulted from environmental exposures that (a) first started early in the 20th century and increased greatly during succeeding decades, and (b) rapidly become potent in early life and then decline sharply within a very few years. We should consider what a good fit in those regards would be provided by
-- maternal smoking beginning in the UK during World War I (1914-1918),71x
-- subsequent major increases in carcinogenic infant exposures as a result of
-- environmental developments beginning in mid-century (Figures 1 and 2) and
This section will draw heavily on publicly-available data about both childhood cancer incidence and breastfeeding rates in many European countries, noting parallels where appropriate. That sort of reporting based on publicly-available information is not often done, compared with carrying out typical studies; but it has been done very successfully, and there are good reasons to do it that way. Information from national breastfeeding surveys is normally collected during children's infancies, typically based on recall of feeding within the previous 24 hours;60n therefore data about breastfeeding in relation to childhood cancer is likely to be relatively free of incorrect recollections about breastfeeding in earlier years. National surveys also normally have much higher response rates compared with participation in control groups of typical studies, resulting in more valid data. (see Section 4.a.4)
As evidence that analysis of publicly-available data can result in conclusions generally recognized to be valid, note the following: Doll and Peto published an article in 198158 that over later years has been widely referred to as a "landmark" report,60, 60a,60b including in a 2005 editorial in the Journal of the American Medical Association.59 The authors arrived at estimates of the avoidability of specific cancers based on publicly-available information about differences in cancer mortality rates in different populations and geographic areas, and on information about suggested causes of cancers.60 Note that this 1981 report was largely based on national cancer registry data and publicly-available dietary information, as is Section 6, here; and it has over the succeeding decades been authoritatively recognized to have been substantially accurate and an important contribution to knowledge about origins of cancer. (see Appendix J)
As a brief summary of considerable text that is to follow, some graphic illustration is provided below:
With very few exceptions, higher and lower levels in breastfeeding rates in European countries have been in accord with higher and lower levels in childhood cancer incidence in those countries.
Exceptions to the overall pattern can be related to factors of relevant interest: Belgium had a major contamination of its food supply with carcinogens (PCBs and dioxins) in 1999; human milk in Czech Republic has been found to be extraordinarily high in PCBs (Figure 2); and Ireland is reported to have a very exceptionally high prevalence of smoking during pregnancy,72e which exposes the fetus to considerable carcinogens. (Section 4.a.1)
It is also of interest to see what may cause countries to have abnormally low childhood cancer incidence. PM (particulate matter), referred to in the caption of the pollution map below, is said by WHO to be "a common proxy indicator for air pollution;" and it is the "component of air pollution most closely associated with increased cancer incidence…."127 PM2.5 (the specific toxin, exposures to which are represented below) is considered to be the more harmful of the two major types of PM. So it is of interest to see which European countries have lower (as well as higher) levels of this kind of pollution.
Above chart can be accessed at https://link.springer.com/chapter/10.1007/978-3-319-39745-0_16
Notice the unusually low pollution indicated above in Norway, Sweden, Spain, Portugal, and Switzerland; that very likely explains why those countries' childhood cancer incidences are all in the lower half of a childhood cancer incidence group in the chart below (circled, right). Also in the chart below, notice the three countries in the upper left that have exceptionally high rates of childhood cancer, which is in accord with the high exposures of infants to carcinogens in those countries. (see "Exceptions…" below Figure 11, above)
(See individual country paragraphs in Section 6.a for sources of data incorporated above.)
So a careful look at the central, normal pattern above will reveal a basically straight line from the lower left to the upper right; the only deviations from a steadily increasing line are ones that are explainable by known, exceptional exposures or by lack of common exposures.
The normal pattern appears to be a strong association between more breastfeeding and more childhood cancer.
That relationship is as should be expected, considering how far the concentrations of carcinogenic PCBs and dioxins in human milk have been found to exceed established safe levels. (Figure 2 and accompanying text)
Section 6.a: Variations in childhood cancer incidence among European countries in recent decades:
In a 2004 study that drew high-quality data on childhood cancer from 63 European population-based cancer registries, it was found (and confirmed by a 2010 study) that overall incidence in children aged 0-14 increased by about 1.0% per year during the period 1970–99;72 and increases continued into the 2000s.73a Improvements in diagnosis and in registration of cancers over time were seen as possible partial explanations for those increases.73
But of greater interest are the numerous individual variations in childhood cancer incidence, by time period and among different countries, which have been in close parallel with variations in breastfeeding rates by time period and country. These will be discussed below.
By themselves, a few such parallels would not constitute strong evidence of causation. But in this case, causation seems to be especially likely because
-- dioxins and PCBs are officially recognized to be carcinogenic, including at low doses (see Section 1);
-- there have been numerous authoritative findings that PCBs and dioxins are present in human milk in doses far exceeding established safe levels. (Figure 2 and accompanying text),
-- scientists and government sources who ought to be knowledgeable on the subject don't know of any other likely source of widespread exposure of infants to carcinogens beyond safe levels, aside from human milk. (Appendix K), and
-- the large number of individual national parallels between breastfeeding rates and childhood cancer, and their consistency, are remarkable, as will be apparent in the upcoming text.
NOTE: When going through this section, the reader should bear in mind the statement from a publication of the U.S. National Academy of Sciences that states, "Positive human epidemiologic data are the strongest evidence in evaluating any human-health risk." 42a (epidemiologic = related to incidence, distribution and control of disease in a population42b)
Of relevance to the steady increase of childhood cancer in Europe in recent decades: breastfeeding, also, has been mostly increasing in Europe in recent decades, as can be seen below.
Generally increasing breastfeeding trend in Europe in recent decades
(Similar increases in Eastern Europe, also, can be seen farther down in Figure 14)
(Above chart can be accessed at https://gateway.euro.who.int/en, then following above instructions)
The Nordic region after the mid-1990s has been an exception to the increase of breastfeeding in Europe, as seen below:
(Above chart can be accessed at https://gateway.euro.who.int/en, then following above instructions)
Mixed increase followed by flatness in the Nordic countries: Note above that breastfeeding was increasing in only two countries of this region as of the earlier years shown above, and those remaining increases appeared to be coming to an end by the middle of the 1990s; after that, rates appeared to be uniformly stable, in contrast with the general increases that were continuing in the rest of Europe during those years (Figures 12 and 14). It is therefore of interest to see (below) that incidence of the principal type of childhood cancer (ALL leukemia) appears to have followed a pattern similar to that of breastfeeding: moderate average increase through 1996, followed by years of no increase. (Bear in mind that latencies of childhood cancer can be as little as 0.4 year.68)
Before the moderate increases and then plateaus of both breastfeeding and leukemia in the Nordic countries, there were especially high earlier increases in both breastfeeding and childhood cancer in the Nordic region. The data to support this statement is somewhat lengthy and dense with numbers, so it has been placed in Appendix F, where readers are encouraged to find it.
Parallels between breastfeeding and childhood cancer also in other Western European countries:
-- United Kingdom: The UK has been low in breastfeeding (see Figures 8 and 12); and its childhood cancer incidence has also been unusually low for Western Europe (140.5 per million in data for 2000-2012 from the IARC).89
-- Spain and Portugal are shown above (in Figure 12) to have had unusually high breastfeeding rates, and they also had unusually high incidence of childhood cancer. (165 and 164 (median) age 0-14 age-specific rates per million).88
-- Germany: Germany overall showed average breastfeeding in Figure 12 and overall rather average incidence of childhood cancer.89c But separation of the data by regions is of interest. An article published in 2018 reported that, four months after birth, half of the mothers from the western section but only a third from the eastern part were still breastfeeding.89d In line with that, the IARC has reported the recent childhood cancer data for Germany at two significantly different levels: 157.5 per million in Western Germany and 141.2 per million in Eastern Germany.89e
-- Italy: Notice in Figure 12 that the breastfeeding rate for Italy more than doubled between 1987 and 1996, followed by many years of minor fluctuations, including one somewhat noticeable decrease. In relation to that, note that an extensive report on childhood cancer has pointed out that malignant cancer in Italian children increased at 3.2% per year until 1997, followed by a plateau and by decreases in several types of cancer during the following decade.83
-- Switzerland: Although not shown in Figure 12, Switzerland's breastfeeding rates are unusually high.83a And, averaging childhood cancer data for the various registries provided for Switzerland yields an incidence of over 165 cases per million, well into the high range.89
-- Belgium: At first, Belgium appears to have been an exception in the present discussion, in having had a very high rate of childhood cancer (192 cases per million)89a despite having had medium-to-low breastfeeding rates in recent years. That is partly explainable by the outstandingly high concentrations of dioxins in human milk in Belgium. (see below)
Also, in 1999 considerable waste transformer oil was accidentally added to Belgian farm animal feed, which was not brought to light until four months later.89b It has been variously estimated that the additional cancer deaths that would have been expected to result from the additional dioxins and PCBs in the human food supply following that contamination ranged between 44 and over 8000. Thirdly, if one were to look at the map of particulate air pollution (which contains many carcinogens) in Europe in Figure 11.b and focus on the blackest, central part of the large nearly-black area, that is Belgium. So it should only be expected that -- as ended up being the case -- childhood cancer incidence in Belgium in the 2000s would turn out to be high even in the absence of high breastfeeding rates.
-- Ireland is another case similar to that of Belgium, with low breastfeeding rates but a medium rate of childhood cancer (153 cases per million).89 Calculations have led to the estimate that, among 174 countries considered, Ireland has by far the highest rate of smoking during pregnancy,72e which exposes the fetus to substantial carcinogens. (See Section 4.a.1)
-- Norway has long had high rates of breastfeeding (Figure 13), and its only childhood cancer incidence shown by the IARC is 165.2, in the high range despite the country's unusually low pollution (Figure 11.b)..
-- Sweden, also, has long had high rates of breastfeeding (Figure 13), and its only childhood cancer incidence shown by the IARC is 155.3, in the upper part of the middle range despite the country's unusually low pollution (Figure 11.b)..
-- Finland, like its sister Nordic countries above, has long had high rates of breastfeeding (Figure 13), and its childhood cancer incidence is not shown by the IARC but is available from another source that shows a rate of 173.3 per million,89z again a high rate despite the country's unusually low pollution (Figure 11.b)..
-- Denmark in Figure 13 shows very limited data, but it does show a major, rapid increase between 1985 and 1992. In relation to that, it is of interest that Denmark's childhood cancer incidence increased 40% between 1985 and 1995.89y Since then, its incidence has averaged about 180 per million,89y compatible with its high 6-month breastfeeding rate shown in Figure 13.
Published data about childhood cancer in Luxembourg does not appear to be available online. For the more complicated case of France, see Appendix U.
Similar data for Eastern Europe:
-- Bulgaria: Data from WHO shows Bulgaria's breastfeeding rates for recent decades to be the lowest among all 25 of the WHO European Region countries for which comparative data is provided.89f And Bulgaria's childhood cancer rate for recent decades (120.8 per million) was the lowest among all 28 European countries for which data was provided by the IARC in 2017.88
-- Croatia: Summarizing data from UNICEF and other sources, one data source ranked Croatia number one in its list of 141 countries in rate of exclusive breastfeeding at six months.89n And Croatia's childhood cancer incidence is shown by the IARC to be the highest among all of the European countries for which data is provided (193.9 per million).88 (Only a few parts of countries were higher.).
-- Poland can be seen in Figure 14 at the bottom of breastfeeding rates as of the 1980s -- no more recent data for Poland is provided by this WHO source. And, as found in a major 1988 WHO/IARC publication on childhood cancer, the only data for Poland (for Warsaw City) showed that "incidence rates overall and for most major diagnostic groups were low compared with those for nearly all other European registries…." 89h A 2016 study showed Poland's exclusive breastfeeding rate at 6 months to be second lowest among 25 reporting countries.89j And data from the IARC for the 2000s shows childhood cancer incidence for Poland to be low (132.8/million) in comparison with other European countries.89k .
-- Lithuania is near the bottom of breastfeeding rates in the early 2000s in Figure 14 below, and it was also found to have had near-bottom childhood cancer incidence in the early 2000s. (133.5 cases per million)89l
-- Slovenia and Slovakia: Those two countries had recent incidence rates of 144.2 and 144.6, mid-range for Eastern Europe,89o compatible with their middle-range positions in breastfeeding rates as seen just below.
(Above chart can be accessed at https://gateway.euro.who.int/en, then following above instructions)
Hungary can be seen to have undergone a major, rapid increase in breastfeeding. (The major increase was actually spread out over a number of years, as can be seen at the above source but selecting breastfeeding for six months.) And childhood cancer was reported to have increased at an average rate of 3.3% per year in Hungary during the period 1988 through 1997.89r
When reading about the 3.3% annual childhood cancer increase in Hungary over a 10-year period, remember that the increases that have been normal for Europe in recent decades have been about 1% per year (see earlier), commensurate with the moderate increases in breastfeeding that have been normal in Europe (see Figures 12 and 14).
Austria's incidence of childhood cancer was at a high level in 2011, at 182 cases per million;89s this was compatible with other data for Austria that showed breastfeeding in that country to be very unusually high well into the 2000s. (same WHO HFA source as for Figure 14 but selecting 6-month data)
Romania's general breastfeeding rates were mid-level as of the latest reports (for year 2000) indicated by WHO, but its exclusive breastfeeding rate in recent years has been reported to be very unusually low.91a And Romania's childhood cancer incidence as of 2010 was reported to be so low (mean 62 per million91b) that it seems not to be credible; the National Registry for Pediatric Cancer had only been established in Romania in 2009, so incompleteness of registration in the first full year of the system very likely contributed to such a low figure. But, even if it were to be doubled, the figure for childhood cancer in Romania would still be among the lowest in Europe. That may be related to especially great significance of exclusive breastfeeding, which was reported to be very low in Romania.
Ukraine's breastfeeding rate has in recent years reached a very low level, as shown above; and its most recent childhood cancer incidence has been near the bottom among European countries.88 Ukraine has also seen an exceptional decline in its breastfeeding in recent decades (see above); in relation to that, note that Ukraine has also had an exceptional decline of its childhood cancer incidence. See Appendix V for details and sources about breastfeeding and childhood cancer in Ukraine.
See Appendix V for other parallels between breastfeeding and childhood cancer in Eastern Europe
Why the focus on Europe?
When making international comparisons to try to ascertain effects of environmental risk factors, it is relevant that the countries of Europe whose data are discussed here comprise a relatively homogeneous grouping in various important respects:
-- their citizens are overwhelmingly Caucasian (ethnicity is considered to be a risk factor for childhood cancer);89v
-- essentially all of these countries are considered by the World Bank to be either high-income or upper-middle-income,89w implying some similarity in sanitation, medical resources, and nutrition; (Republic of Moldova, which is lower middle income, is found in one chart, but its data is not used in any comparisons);
-- climatic variations within the region are relatively moderate, especially among that large majority of countries located within the central latitudes of the region.
The similarities among countries of Europe mean that comparisons of exposures to a varying factor such as breastfeeding within that region are potentially meaningful, with a minimum of differences in other factors that could affect the outcomes. So there are good reasons to focus comparisons mainly within this one region.
Review of Figure 8
Why all of the above goes beyond coincidence:
In addition to the large number and the strong consistency of the associations between breastfeeding rates and childhood cancer incidence, we need to remember that
-- there is no dispute that typical human milk contains carcinogens in concentrations far exceeding established safe levels (Figure 1.a), and also,
-- "Positive human epidemiologic data are the strongest evidence in evaluating any human-health risk,"42a as stated in a publication of the U.S. National Academy of Sciences.
Section 6.b: Three years after the Chernobyl nuclear disaster, radioactivity from which drifted to the Nordic countries, there was an extraordinary, unexplained decline in childhood cancer among those countries.
Preview of Fig. 20
Arrival of serious amounts of radioactivity via air currents was well known in the Nordic countries. Substantial evidence had previously been published indicating that radioactive chemicals are readily transferred via breastfeeding, and there was governmental advice to breastfeeding mothers reflecting that knowledge. That would have brought a temporary decline in breastfeeding in 1986, preceding what turned out to be a remarkable temporary decline in childhood cancer incidence in 1989. (Average latency of childhood leukemia is about 2-1/2 years.)
The details and sources on this topic are rather lengthy and have been placed in Appendix M.
Looking at the above WHO charts as well as childhood cancer data from European countries, all of which can be readily verified, we have seen the following as discussed above:
-- In every case in which countries were found to have had unusually high or rapidly increasing rates of breastfeeding, their childhood cancer incidence rates were also high or rapidly increasing at about the same times: Norway, Denmark, Sweden, Spain, Portugal, Italy, Switzerland, Hungary, Austria, Belarus and Croatia. (Bear in mind that latencies of childhood cancer can be as short as 0.4 year.)
-- The Nordic region stood out from most of the rest of Europe in having a stable incidence of childhood cancer after 1996 (Figure 13a), and that region also had stable breastfeeding rates after 1996 (Figure 13); these countries' histories contrasted with most of Europe, which in those same years had increasing rates of childhood cancer72,73a as well as increasing breastfeeding rates. (Figures 12 and 14)
-- European countries with unusually low breastfeeding rates have been found to have had unusually low childhood cancer rates: the UK (in Section 4.c), Poland, Lithuania, and Bulgaria, and Romania and Ukraine in recent years.
-- Countries that have had mid-range breastfeeding rates (Germany, the Netherlands, Slovenia, and Slovakia) have had mid-range incidences of childhood cancer. Estonia, with mid-range breastfeeding, has had cancer incidence not far below average. Germany, with higher and lower breastfeeding durations in its western and eastern regions, has had corresponding higher and lower incidences of childhood cancer in the two regions. The minimal information that is available about breastfeeding in France is compatible with what is known about its incidence of childhood cancer.
-- In Ukraine, childhood cancer has shown an exceptional decline in relation to the general trends in Europe (See Ukraine in Appendix V.), and breastfeeding in Ukraine has also had an exceptional decline during recent decades (Figure 14).
There has been a pattern of positive associations between breastfeeding and childhood cancer incidence that was found to apply in most of the Western European countries of substantial size as well as in the major part of Eastern Europe discussed here. The few exceptions were:
-- countries for which not all of the relevant information is available: France, Latvia and Republic of Moldova;
Changes in breastfeeding or in breastfeeding-related activity have come before similar changes in childhood cancer, where sufficiently specific information is available:
a) A significant, temporary decrease almost certainly occurred in breastfeeding in the Nordic countries after the Chernobyl accident in 1986, and a significant, temporary decrease in childhood cancer in the Nordic countries followed in 1989;
b) In Norway,
-- incidence of childhood CNS tumors (a major type of childhood cancer) in Norway nearly doubled between 1970 and 1999, but leveled off during the 1990s;78
c) distinctive, sharp peaks in infants' levels of (carcinogenic) dioxins occur during the first six months of breastfeeding, followed by rapid declines (Figure 9); and similarly-shaped peaks in incidence of ALL leukemia occur later, at 2-3 years after birth (Figure 10).
Why all of the above goes beyond random coincidence:
In addition to the large number and the strong consistency of the positive associations between breastfeeding rates and childhood cancer incidence, we need to remember that
-- there appears to be no dispute that typical human milk contains many known carcinogens,22 at least two of which are typically in concentrations far exceeding established safe levels (Figure 2 and Appendix K), and also,
-- "Positive human epidemiologic data are the strongest evidence in evaluating any human-health risk,"42a as stated in a publication of the U.S. National Academy of Sciences.
Section 8: Comparison with the U.S. Surgeon General's major statement on smoking:
When a U.S. Surgeon General first made a major public statement saying that cigarette smoking causes cancer, the causal connection between smoking and cancer was in important respects not as strong as the case linking breastfeeding with childhood cancer is now. In that first report of the Surgeon General’s Advisory Committee on Smoking and Health, there was some uncertainty about risk of smoke exposure, as indicated by the Committee's statement, "The amount of known carcinogens in tobacco smoke is too small to account for their carcinogenic activity" (p. 146); it was also acknowledged that studies had (with one possible exception) been unsuccessful in producing lung cancer in animals by means of exposure to tobacco smoke, extracts, or condensates (p. 165).91 On the other hand, carcinogenicity of both dioxins and PCBs in humans (as well as animals) has been established,12b and infants' ingestion of both of those toxins by way of human milk in doses far exceeding established safe levels is well recognized (Figure 2 and accompanying text). So the chemical-effects evidence linking childhood cancer with breastfeeding may be significantly stronger than was the equivalent case linking lung cancer with cigarette smoking.
And the international epidemiological evidence, also, appears to be stronger in the case of the link between childhood cancer and breastfeeding. In the case of smoking, the number of countries identified where exposures to toxins of concern aligned well with adverse outcomes was relatively small (see chart below); that is, small as compared with the large number of countries with close associations between breastfeeding rates and childhood cancer incidence, as presented at length in Section 6.
Above U.S. HEW chart at p. 176, https://www.govinfo.gov/content/pkg/GPO-SMOKINGANDHEALTH/pdf/GPO-SMOKINGANDHEALTH.pdf
The Surgeon General's Committee clearly paid close attention to the epidemiological evidence showing associations between cigarette smoking and lung cancer deaths in different countries, as represented in the chart above. (pp. 175-177.) That attention was as should have been expected, especially given the conflicting nature of the other evidence: the Committee reviewed more than 7,000 articles related to smoking and disease in the biomedical literature;7f if there had been that many articles on this topic without there ever before having been a decision by a Surgeon General to strongly oppose smoking, there must have been numerous scientific studies and articles disagreeing with the proposition that smoking has harmful effects. The epidemiological relationship between smoking and lung cancer deaths among nations -- as can be seen in Figure 18 above -- correctly pointed toward the exposure of interest as causing cancer, probably more clearly than the numerous studies on the subject. The same would probably be true with regard to the consistent relationship between breastfeeding and childhood cancer among nations, as presented in Section 6. But in the latter case the international epidemiological evidence appears to be much stronger than was the case with smoking and cancer, because of
(a) the much larger number of country examples in the childhood cancer case, and
(b) the provision of data for all contiguous countries of substantial size within a very large area of consideration (most of Europe).
The chart of countries by smoking history drew its examples from three separate continents; countries within the areas of consideration were omitted that may have yielded data that conflicted with the perception of risk from smoking. Since no major countries were omitted from the contiguous group of nations providing data on breastfeeding in relation to childhood cancer, this epidemiological evidence should receive even greater respect than was the case with the chart showing lung cancer as related to smoking.
We should also remember the statement from a publication of the U.S. National Academy of Sciences that states, "Positive human epidemiologic data are the strongest evidence in evaluating any human-health risk." 42a
Section 9: An overall assessment of the link between breastfeeding and childhood cancer:
The highly respected Sir Austin Bradford Hill (once president of the Royal Statistical Society) provided a set of "viewpoints" (normally referred to as criteria) to aid in judging causality, which have been widely used since then, including by the EPA.68a Hill said that none of his viewpoints can lead to indisputable evidence of cause-and-effect, but that they can help with "the fundamental question -- is there any other way of explaining the set of facts before us, is there any other answer equally, or more, likely than cause and effect?"68b He did not say that lack of proof means we should do nothing about what appears to be a cause of adverse outcomes. As explained above, there are many good reasons to see toxins in human milk to be a cause of childhood cancer. It would seem that it should be up to the promoters of breastfeeding to come up with any other way(s) of explaining the set of facts before us, which could equally or better explain the numerous, consistent associations between breastfeeding and childhood cancer, as discussed in this article; that is especially true considering
-- the authoritatively recognized carcinogenicity of at least two different toxins each of which is known to normally be present in human milk in concentrations far exceeding established safe levels (Figure 2 and accompanying text), and
-- the lack of other known widespread pathways for hazardous doses of carcinogens to children. (Appendix K)
Judging by considerable past correspondence with the leading medical organizations specializing in children's health, no explanations will be forthcoming from them.
As the author of the above, my role has not been to carry out original research, but instead it has been to read through very large amounts of scientific research that has already been completed on the subjects of environmental toxins and infant development, and then to summarize the relevant findings; my aim has been to put this information into a form that enables readers to make better-informed decisions related to these matters. The original research articles and government reports on this subject (my sources) are extremely numerous, often very lengthy, and are usually written in a form and stored in locations such that the general public is normally unable to learn from them.
My main qualification for writing these publications is ability to find and pull together large amounts of scientific evidence from authoritative sources and to condense the most significant parts into a form that is reasonably understandable to the general public and also sufficiently accurate as to be useful to interested professionals. My educational background included challenging courses in biology and chemistry in which I did very well, but at least as important has been an ability to correctly summarize in plain English large amounts of scientific material. I scored in the top one percent in standardized tests in high school, graduated cum laude from Oberlin College, and stood in the top third of my class at Harvard Business School.
There were important aspects of the business school case-study method that have been helpful in making my work more useful than much or most of what has been written on this subject, as follows: After carefully studying large amounts of printed matter on a subject, one is expected to come up with well-considered recommendations that can be defended against criticisms from all directions. The expected criticisms ingrain the habits of (a) maintaining accuracy in what one says, and (b) not making recommendations unless one can support them with good evidence and logical reasoning. Established policies receive little respect if they can't be well supported as part of a free give-and-take of conflicting evidence and reasoning. That approach is especially relevant to the position statements on breastfeeding of the American Academy of Pediatrics60f and the American Academy of Family Physicians60h, which statements cite only evidence that has been
(a) selected, while in no way acknowledging the considerable contrary evidence,92 and
(b) of a kind that has been authoritatively determined to be of low quality. (See about observational studies, in Appendix E.)
When a brief summary of material that conflicts with their breastfeeding positions is repeatedly presented to the physicians' associations, along with a question or two about the basis for their breastfeeding recommendations, those associations never respond. That says a great deal about how well their positions on breastfeeding can stand up to scrutiny.
The credibility of the contents of the above article is based on the authoritative sources that are referred to in the references: The sources are mainly U.S. government health-related agencies and reputable academic researchers (typically highly-published authors) writing in peer-reviewed journals; those sources are essentially always referred to in footnotes that follow anything that is said in the text that is not common knowledge. In most cases a link is provided that allows easy referral to the original source(s) of the information. If there is not a working link, you can normally use your cursor to select a non-working link or the title of the document, then copy it (control - c usually does that), then paste it (control - v) into an open slot at the top of your browser, for taking you to the website where the original, authoritative source of the information can be found.
The reader is strongly encouraged to check the source(s) regarding anything he or she reads here that seems to be questionable, and to notify me of anything said in the text that does not seem to accurately represent what was said by the original source. Write to firstname.lastname@example.org. I will quickly correct anything found to be inaccurate.
Comments or questions on the above are invited, including criticisms if they are specific, and will usually receive a response. At the next link are past comments and questions from a number of readers, including eight doctors, followed by our responses. Some of the doctors have been critical but others have been substantially in agreement with us (including one with children with asthma, one who says she has delivered thousands of babies, and one with a son with autism); they put into briefer, everyday language and personal terms some important points that tend to be immersed in detail when presented in our own publications. Topics discussed in that section include about having breast milk tested for toxins and about means of trying to achieve milk that is relatively free of toxins, including the "pump and dump" option. To read the above, go to www.pollutionaction.org/comments.htm
In criticisms, please point out any specific passages that you feel are not accurately based on authoritative sources (as cited) or that do not logically follow from the evidence presented. Note that the author of this article feels no obligation to present the pro-breastfeeding case as long as the medical associations and other promoters of breastfeeding fail to inform parents about the developmental toxins that are, without dispute, present in high concentrations in human milk. Please e-mail criticisms or other comments to email@example.com. Most will receive a reply.
For a more complete statement about the author and Pollution Action, please go to www.pollutionaction.org.
Shenandoah, VA, USA
A: Exposures that
far exceed established safe levels:
The American Academy of Pediatrics and the American Academy of Family Physicians, both of which strongly advocate breastfeeding, do not deny that PCBs and dioxins are transferred to infants via breastfeeding in quantities that far exceed established safe levels. In an August, 2018 letter to the AAP, the author of this article wrote the following, based on considerable evidence from authoritative sources, which were cited: "PCBs have been found to be present in human milk in doses 63 to 270 times the minimal risk level established by the U.S. Agency for Toxic Substances and Disease Registry.4,6 Dioxins have been found to be present in typical U.S. human milk in concentrations exceeding the EPA's RfD (estimated reasonably safe dose) by scores to hundreds of times. PCBs have been found to have long-term effects including cancer and damage to the blood-brain barrier; dioxins have been found to have long-term effects including cancer, heart disease, reproductive abnormalities, and immunity deficits." (Footnotes were included from authoritative sources supporting all of the above)
The letter requested a response from the AAP that would contradict the above based on scientific evidence, and offered $5,000 to compensate for the time required to write a response. (The financial offer was due to the lack of response to several earlier, similar letters to both the AAP and the American Academy of Family Physicians.) Return receipt for the letter was requested and received from the U.S. Postal Service, and (despite offer to place the funds in escrow to assure payment upon receipt of reply) as of over six months later, no response has been received. Any organization that promotes breastfeeding (such as the AAP) should have responded if the statements quoted just above were not well-substantiated or if there existed scientific evidence that could contradict the substance of the letter that was received.
Likewise, there has
been no reply to a similar letter to the AAP (dated Oct. 5, 2018)
stating, "Of toxins with recognized long-term effects, none are known to
be ingested or inhaled by U.S. infants in doses beyond established safe levels,
except for toxins that are transferred via breast milk." Again,
there was an ample financial offer to compensate for a science-based,
contradictory response, return receipt for the letter was received, and again
(after over six months) there has been no response and no sign of disagreement
with the statement in quotes.
For additional sources about authoritatively-established safe doses of these chemicals and actual exposures in the U.S. and other countries, see reference no. 10 below.
Appendix B: Chemical exposures determined to exceed safe levels even more than previously thought:
In 2001, the European Commission's Scientific Committee on Food set a safe level for dioxins and dioxin-like PCBs that was similar to the level determined by WHO in 2001, which was the standard that was in effect during the WHO surveys results of which are shown in Figure 1.a.93 However, in November of 2018, the European Food Safety Authority (EFSA) issued a revised statement of tolerable intake that was seven times lower than the one established in 2001 by its predecessor Committee, based on new data and on improved assessment methods.94
Appendix C: Ongoing sources of PCBs, dioxins and PBDEs, including in traffic emissions and older buildings, to which people are closely exposed: In a 2013 U.S. study of PCB concentrations in carpet dust in homes at various distances from possible pollution sources, PCB concentrations were found to be about five times as high in homes within 600 yards from major roads in Los Angeles as they were in homes that were over 1100 yards from those roads;104 a study in the Czech Republic found even more extreme differences in traffic-affected soils, compared with soils in reference areas.8b In a study in Colombia, measurements of PCBs and dioxins taken near vehicular sources were found to be almost six times as high as measurements made over a mile farther away.105 Similar results concerning PCBs in traffic emissions were found in a study in Stockholm, Sweden.106 PCBs were a component of caulks used in construction of buildings between the 1960s and 1970s; PCBs continue to be emitted from those sources and have been found at elevated levels in people who spend considerable time in some of those buildings. See Figure 6.a and accompanying text about PCBs in cigarette smoke.
Dioxins: A 2009 Czech study found "dramatic decreases" in concentrations of pollutants (including dioxins) in soil according to distances from roadways. Concentrations next to roadways were found to be as high as in industrially-polluted areas; and the concentrations of dioxins were found to decrease, stepwise, with progressive distances from roadways up to the last distance measured.107
PBDEs are apparently produced in vehicle emissions in even greater amounts than PCBs. A Taiwanese study found that concentrations of PBDEs in vehicle exhausts were 17 to 140 times as high as concentrations of PCBs, in the different vehicles tested.108 (When reading that, remember from above how high PCBs have been found to be in vehicle emissions.) In a 2010 study, the PBDE emission rate from tailpipe exhaust of just one unleaded-gas-fueled vehicle was found to be about four times higher than that from one U.S. house and garage.109 (Bear in mind that U.S. houses, usually containing many flame-retardant-treated furnishings and electronics, are themselves substantial sources of PBDE dust.110
PBDEs are chemical relatives of PCBs and best known as flame retardants. They are similar to PCBs in that they are toxins that can cause impairment, being listed in the TENDR group of the six toxins of greatest concern.111 In a discussion of PBDEs, the EPA refers to those toxins' adverse neurobehavioral effects following exposure during the postnatal period." 112 (The italics were added here to reinforce the point that these chemicals are known to have adverse neurodevelopmental effects resulting from exposures occurring after birth; that emphasis was added because prenatal exposures seem to draw inappropriately predominant attention.)
The leading authorities on medical evidence have determined that evidence from observational (non-randomized) studies is predominantly of low quality, with evidence from only exceptional ones reaching a medium level of quality. One such determination has been provided by Dr. Gordon Guyatt and an international team of 14 associates;113 Dr. Guyatt is chief editor of the American Medical Association's Manual for Evidence-based Clinical Practice, in which 26 pages are devoted to examples of studies (almost all of which were observational) that were later refuted by high-quality studies.114
A similar assessment of the generally low quality of evidence from observational studies has been provided by the other chief authority on medical evidence, Dr. David Sackett,115 writing about "the disastrous inadequacy of lesser evidence," in reference to findings from observational studies.116 Dr. Sackett was said by the British Medical Journal to have been "widely regarded as 'the father of evidence-based medicine.'"117
A team of MD's, in an article in the AMA Journal of Ethics, studied the matter brought up by John P.A. Ioannidis (MD, PhD) in the latter's 2005 article entitled, "Why most published research findings are false." The authors were in basic agreement with what Ioannidis wrote. Observational studies of health effects of breastfeeding are especially subject to false conclusions because of
a) confounding by socio-economic status,
b) inaccurate, biased data, which is especially likely to occur in retrospective (that is, most) studies of breastfeeding (see Section 4.a.2);
d) selective publication;1a,1b aside from selective publication's being recognized in peer-reviewed literature as a problem in research on effects of breastfeeding,1a it appears that screening out of contrary information regarding breastfeeding is encouraged and practiced at the highest levels of medicine; this can be seen in the policy statement on breastfeeding of the American Academy of Pediatrics,60f which cites numerous studies that are favorable to breastfeeding but refers to none of the many negative studies60g and does not mention the word "toxin" even once, despite ample undisputed evidence about presence of developmental toxins in high concentrations in human milk. (Figure 2 and accompanying text)
On the basis of a large and consistent body of evidence from observational studies showing that use of hormone replacement therapy (HRT) was associated with a 40–60% lower incidence of coronary heart disease, several medical associations in the mid-1990's officially recommended HRT for post-menopausal women;2b that is, they did so until about 2002 when results came in from long-term randomized studies; those studies found that HRT either had no benefit for women with cardiovascular disease or actually led to worse outcomes, including a small increase in breast cancer.2c Until that time, it had not yet been generally recognized that observational studies were so greatly subject to erroneous findings, as compared with randomized trials, which are recognized to be the highest-quality type.3 The erroneous results favoring HRT had been arrived at even after adjustments had been made in the observational studies (as is normal) in attempting to correct for numerous potential confounders.3a
A publication of the Institute of Medicine (of the U.S. National Academy of Sciences) discussed the dramatic discrepancies that were found when results of observational studies were compared with results of better-quality studies on the same topic.3c The authors devoted a large paragraph to the problem of selection bias (failure to achieve initially comparable groups) that is inherent in typical observational studies, ending with, "Although a variety of statistical methods can be used to attempt to reduce the impact of selection bias, there is no way that analysis can be used to correct for unknown factors that may be associated with prognosis. Thus, it is generally acknowledged that 'adjustment' in the analysis cannot be viewed as a substitute for a study design that minimizes this bias." In an article funded by Harvard Medical School and the U.S. CDC and published on the CDC's website, the authors indicate their negative opinion of the usual means of trying to deal with unintended differences between groups being compared in typical studies, saying, "strenuous statistical machinations are then needed to 'adjust for' irreconcilable differences between study and control groups."2c
The best and most established way of minimizing distortions resulting from differences between groups being compared is by randomization of selection, as in randomized controlled trials (RCTs), like the PROBIT study in Section 3. Use of siblings as controls (as in the Lingappa et al. study discussed in Section 4.a.1 and the Colen et al. study discussed in Section 3) has in recent years become increasingly established as another means of minimizing such bias.
There is a question as to whether evidence from typical observational studies might be thought to be of some value if better-quality evidence is not available in a particular area, on the assumption that low- or-sometimes-medium-quality evidence might be better than no evidence. The committee writing for the Institute of Medicine took up that question in their 2011 publication quoted earlier (page 113),3c and its answer was as follows: " The committee did not find evidence to support a recommendation about substituting observational data in the absence of data from RCTs." In other words, the committee of the Institute of Medicine concluded that, if the evidence is only from typical observational studies, there is no reason to consider that evidence to be of value. Given the above, bear in mind that the studies that have found childhood cancer to be reduced by breastfeeding were entirely observational, according to the U.S. Surgeon General's 2011 breastfeeding-promoting document.51
Apparently, as of the writing of the above document by the Institute's committee, sibling studies had not yet become a major alternative type of study. A 2011 article in the International Journal of Epidemiology referred in the present tense to "the advent of sibling designs" in epidemiology, and said (while clearly pointing to the future) that sibling designs "have great potential" and "offer an avenue for significant advance in epidemiology."73e So the lack of mention of sibling studies by the Institute's committee (in a 2011 publication) as an alternative means of gaining high-quality evidence only meant that they had not yet seen sibling studies to be a substantial source of high-quality evidence. The Colen and Ramey sibling study53 (2014, discussed in Section 3), which found a positive association between breastfeeding and an important risk factor for cancer, had not yet been published; nor had the Lingappa sibling study72s (2018, discussed in Section 4.a.1) which found strong positive associations between childhood leukemia and breastfeeding for typical durations.
For more about quality problems in evidence from observational studies, see Section 1.a of http://www.pollutionaction.org/breastfed.htm.
Appendix F: In the Nordic countries, there were especially high early increases in both breastfeeding and childhood cancer, followed by parallel plateaus in breastfeeding and childhood cancer:
-- In Norway, there were parallel, especially rapid increases as follows:
a) incidence of childhood CNS tumors -- a major type of childhood cancer -- nearly doubled between 1970 and 1999, but with a leveling off during the 1990s;78 and
-- In Denmark, there were extended, major increases in both
a) CNS tumors, which increased at the rate of 2.9% per year during the period 1980-1996,79) and
b) breastfeeding, which increased greatly in Denmark during the 1980s (Figure 13)
(When reading about increases of 2.9% or so per year, and about non-increases, bear in mind that those are all substantial deviations from the average annual increases of about 1% that were normal in European countries during these decades; see earlier.)
-- In Sweden, there were extended, major increases in both
a) CNS tumors, which increased an average 3.7% per year among children aged 0-4 from 1973 to 1992,80 and
b) breastfeeding, which more than doubled between 1970 and 1987.81
The increases discussed above came at about the same time as a major increase in CNS tumors in U.S. children, with an average annual increase of 2.7% during the period 1970-1989;82 note how compatible that was with the rapid increases in breastfeeding in the U.S. during that period, especially breastfeeding at 6 months, as seen in Figure 5.
Appendix G: Statement by the American Academy of Pediatrics on absence of toxins in infant formula:
(The following is a picture of page 200 in: American Academy of Pediatrics Council on Environmental Health, Ch. 3, Food and Water, in Etzel, R.A. ed., Pediatric Environmental Health, 3rd Edition, American Academy of Pediatrics, 2012.)
Since the AAP has long been an active promoter of breastfeeding, the above acknowledgement of formula's freedom from typical toxins is clearly not based on bias in favor of formula. It is simply based on established scientific facts. The AAP, however, repeated its long-standing position in favor a breastfeeding later in this chapter, alleging that toxins such as PCBs in human milk are merely "at or near the upper regulatory allowances for formula or infant foods." Although the study that was the source of Figure 2 had not yet been published as of this AAP book's publication, substantial other evidence had already been published indicating that PCBs and dioxins were present in human milk far in excess of established safe levels.10 The AAP apparently does not deny the validity of that evidence. (See Appendix K below)
Appendix H: Evidence of substantial under-reporting of maternal smoking during pregnancy, leading to inaccurate findings in studies:
Parents in recent decades, especially mothers, are likely to have been well-informed about dangers of smoking to the fetus and infant; so, if they had been unwilling or unable to quit smoking for the baby, mothers were likely to under-report their smoking when questioned about it later, as part of medical research. It is well established that responses to surveys are heavily affected by what respondents feel would be the desirable response for a well-behaved person -- see Section 4.a.2.
However, trustworthy data related to effects of maternal smoking can more consistently be found by searching back to the period before it was well known that smoking is harmful to the child. It was apparently not until the late 1960s and the 1970s that awareness in that regard increased greatly.72j The Canadian/British data shown below from 1960-61 provides a credible representation of the effects of maternal smoking, compared with the (probably) often-distorted findings from recent decades.
Look at the figures for the first two age groups, showing rates of childhood cancer diagnosed at the ages most indicative of likely effects of smoking exposures during the especially sensitive stages of gestation and infancy. The cancer rates per 100,000 were 10.4 and 8.7 for non-smoking mothers in those two groups, compared with 17.5 and 15.6 for smokers, representing dramatic increases in childhood cancer that could be logically linked with maternal smoking.
A study of data from the 1950s provides similar results in the data for maternal and paternal smoking combined, showing a highly-significant 70% increased risk of child leukemia in association with moderate or heavy parental smoking.72m
So research results for the earlier period, before information about harmfulness of smoking to developing fetuses and infants had been publicized, apparently provided more consistent findings linking parental smoking and childhood cancer. The results became inconsistent in later decades, by which time mothers who smoked would have had reason not to want to report that.
If the question about a mother's exposure to smoking is asked such that the source of the smoke is considered to be someone other than the mother, one can find recent verification of increase of childhood cancer linked with smoking; a 2016 study in Iran found that childhood leukemia and lymphoma were more than twice as high as normal among children of mothers who were reported to be exposed to smoke from others.72r
Also, if researchers gather maternal smoking data prospectively, during the breastfeeding period, rather than relying on (likely inaccurate) recall in later years, that can improve accuracy of findings; a 2017 prospective study found more than doubled risk of one common type of childhood cancer and 9-fold increased risk of another type, associated with smoking during pregnancy.72p
Appendix I: Sources of serious distortion of studies' findings, probably resulting from above-average participation in control groups by the health conscious:
Data from the Kwan meta-analysis71m provides evidence of the distortion in studies' findings that probably results from elevated health consciousness in control groups. "Participation" in a study, representing the proportion of those invited who agreed to participate, indicates the potential for distortion; the lower the participation, the greater the likely distortion. Of the 14 studies analyzed by Kwan et al., 12 showed their recognition of the (obvious) importance of participation by providing the appropriate data for cases, but only 6 of those provided the data for their controls. (Table 1 in Kwan et al.) A logical reason for that selective provision of relevant data would be that the participation rates among controls were often so low that the authors realized that the data would reflect poorly on the perceived validity of their studies' findings. Among those studies in which the authors did provide participation data for controls, the following characteristics were found:
-- the non-participation rates were over twice as high among controls as among cases, and
-- the median participation among controls was 71%. That should be compared with 75%, which is the lowest figure that has been authoritatively stated as being within the "acceptable" participation range for surveys utilizing interviews.73x Bear in mind that these very minimal participation rates applied to the 6 (out of 14) studies that probably had the best participation rates among controls in the Kwan meta-analysis, the ones whose authors were willing to provide this important information.
So dissimilarities between groups being compared, in important traits that could distort comparisons between the groups, would have been substantial. Health consciousness in the control groups would have been disproportionately high, resulting from health-conscious people joining the study when the originally-invited people declined. (Section 4.a.4) Their children would have reduced cancer because of the healthy practices and environments, but studies would only point out the higher breastfeeding rates of those people.
Appendix J: The Doll and Peto 1981 report in retrospect:
This publication became a reference standard when estimates were made of the relative roles of lifestyle, environmental, and host factors as causes of cancer in the US.60b
A 2015 article in the Journal of the National Cancer Institute summarized that Doll and Peto's article could be seen as "holding generally true for 35 years."60b The authors pointed to only one of the estimates made by Doll and Peto that they felt had subsequently turned out to have been significantly inaccurate; but other authors (with Harvard Medical School and Harvard School of Public Health) wrote later in 2015 that the reportedly inaccurate estimate by Doll and Peto should actually be seen to have been reasonable, after all.60c
A 2008 article by an author with the CDC in the journal, Epidemiology, stated that there had been "serious criticisms" of Doll and Peto's methods, but that respect for their 1981 estimates still continued within the larger cancer control community.60a
Appendix K: Very likely a unique, undisputed pathway of widespread infant exposure to developmental toxins in doses exceeding established safe levels:
As can be seen in Figure 2 and accompanying text, with considerable authoritative supporting evidence, breastfeeding is a pathway for ingestion of multiple developmental toxins by infants, with PCBs and dioxins at especially high levels. But it is more than just a pathway; it is almost certainly the only pathway by which infants are widely exposed to developmental toxins in doses exceeding established safe levels. The author of this article has written relevant letters of inquiry to appropriate officials at:
a) the American Academy of Pediatrics,
b) the American Academy of Family Physicians,
c) the entire science team at the major autism-advocacy organization, Autism Speaks,
d) the U.S. Agency for Toxic Substances and Disease Registry (ATSDR), and
e) the U.S. Centers for Disease Control and Prevention (CDC).
Those letters asked whether the addressees or their organizations were aware of
a) any toxins that are believed to widely reach infants in doses well in excess of a recognized safe level (e.g., EPA's RfD), aside from the several such toxins that are ingested by means of breast milk, or
b) any other pathway of infant exposure to toxins in concentrations exceeding established safe levels.
As of six or more months later, none of the five replies that were received suggested any other such toxins or other pathway.
Related to this are the very large differences between concentrations of toxins in breast milk and those in the main alternative infant feeding. The toxins being discussed here are present in infant formula in concentrations usually less than 1% as high as their concentrations in human milk. (See below Figure 2 above and also near the beginning of each of the subsections of Section 3 of www.pollution-effects.info)
Appendix L: Exposures of mothers to toxins and times of exposures of the developing infant to those toxins:
The period of potentially adverse exposures of developing children to toxins is often said to be prenatal, with no mention of concerns about postnatal exposures. However, prenatal exposure of the mother typically merely adds to a mother’s accumulated body burden of toxins, contributing to a much greater postnatal exposure that will occur later for breastfed infants; remember the ATSDR's illustrative example of the laboratory experiment in which the following was observed: after a maternal dose of PCB administered before gestation, postnatal exposure of rat sucklings to those PCBs via lactation was over 1600 times greater than fetal exposure (see beginning of Section 2). Also remember that studies of humans have found lactational transfers of toxins to be hundreds of times greater than gestational transfers.19
Also note in Figure 9 that dioxin toxic equivalency levels in infants as of time of birth are very low, indicating low exposures during gestation; but the dioxin levels very rapidly increase among breastfed children right after birth, indicating that the postnatal exposures of breastfed children are far greater than prenatal.
Appendix M: A striking decline from the normal level of childhood cancer among all Nordic countries in the same year, three years after Chernobyl: coincidence?
The authors of the article from which the above chart was taken remarked (p.1543) about the occurrence of a "very low" incidence rate of acute lymphoblastic leukemia in the Nordic countries in 1989, and they were unable to come up with an explanation for it other than coincidence.90 It would be extremely unlikely for there to have been random, coincidental decline in leukemia of such magnitude in all of the Nordic countries in the same year, followed by rapid return to normal in all of those countries, without some common underlying cause. It is of interest that the 1986 nuclear disaster at Chernobyl, to the south of the Nordic countries, resulted in considerable transfer of radioactivity to those countries via air currents.90a. ALL Leukemia is known to have latencies of 0.4 to 4 years,68,69,70 and the latency period for childhood leukemia after radiation exposure is estimated to peak at about 2.5 to 3 years;90n so it would have been reasonable to foresee an increase in leukemia incidence in about 1989, the opposite of what actually happened.
There was most likely an underlying cause for the decline of leukemia in the Nordic countries, following the nuclear meltdown. The level of concern and caution in the public was high: after it was recommended to wash leafy vegetables before eating them, the public response was instead to essentially completely stop buying leafy vegetables.90c When Norwegian authorities set an upper threshold for radiation intake via food for breastfeeding mothers that was one-eighth as high as the standard for the general public,90d that very likely led to many or most mothers simply going over to other types of infant feeding. Responsible authorities in the Nordic countries met repeatedly to inform each other and to harmonize their countermeasures,90d so similar advisories were most likely in place in the other Nordic countries. In addition, ample scientific basis for medical recommendations not to breastfeed when radiation exposure was high had been published before the Chernobyl accident; a description of some of that evidence can be found at the end of this appendix. Favorable alternative infant feedings would have been available to mothers in the form of various products such as infant formula and baby foods that were manufactured before the contamination period and/or imported from less-affected regions, where there were many sources.90f
Therefore it is almost certain that there was significant, temporary reduction of breastfeeding in the Nordic countries during 1986. That would have meant reduction in 1986 of the (apparently undisputed) typical ingestion by infants of carcinogens, including dioxins and PCBs at very high levels (Figure 2 and accompanying text). Note (below) that ALL leukemia comes to a fairly sharp peak in the Nordic countries at 3 years after birth; considering how greatly that 3-year peak contributes to total childhood leukemia, higher or lower incidence at 3 years of age is a potentially useful indicator of variations in children's early exposures to carcinogens.
A child's first two years after birth are recognized by the EPA to be a period of especially great vulnerability to effects of an important type of carcinogens -- carcinogens that are mutagenic,90o which includes dioxin90q,90t as well as many types of PCBs90p,90r. (Remember from Figure 2 that dioxins and PCBs are both heavily present in human milk.) That EPA publication goes on to say that risk of cancer from exposures to mutagenic carcinogens between birth and age 2 should be multiplied by a factor of 10, when comparing those chemicals' risks for infants as opposed to adults.90o Given such exceptionally increased vulnerability of infants and 1-year-olds to important types of carcinogens, and the recognized 0.4- 4-year latencies of this disease, as well as the very high incidence of ALL leukemia at about age 3 (above), a large drop in leukemia in 1989 should prompt a careful consideration of what was happening to infants in that region within a very few years before 1989.
The Chernobyl accident was the outstanding occurrence in the environment at about that time, and that would obviously not have been directly beneficial to young children. It may have caused caregivers to keep them indoors more; but avoiding the minimal air pollution of the Nordic region (see below, mostly above the red area) in 1986, compared with earlier and later years, would have been unlikely to cause a substantial reduction of leukemia among children.
Above chart can be accessed at https://link.springer.com/chapter/10.1007/978-3-319-39745-0_16
But, as discussed earlier in this appendix, a significant, temporary reduction of breastfeeding probably did occur. And it appears to be undisputed that such a reduction would have significantly reduced infants' exposures to large doses of known carcinogens. (Figure 2 and accompanying text)
Seeing the substantial decline in childhood leukemia that occurred in the Nordic countries three years after the Chernobyl accident, people ought to consider all possible changes in child care that may have occurred in response to the accident, which may have been causes of the reduction in leukemia. Reduction of breastfeeding, as a practical response to governmental (and probably other) advice, would have substantially reduced infants' exposures to many known carcinogens.22
Evidence basis for recommending against breastfeeding in case of nuclear accident, which would have predated the Chernobyl meltdown:
Among other pre-Chernobyl studies that would have discouraged breastfeeding when radiation is high (see references at 90h) was another study, in the internationally-known journal, Pediatrics, which found that fully half of certain radioactive material administered to a mother (as part of medical treatment) was excreted in her breast milk within the first 24 hours.90j Another study, on the topic of nuclear medicine and the nursing mother and published in the British Medical Journal in 1985, pointed out the finding in various studies that "selective uptake may result in iodine concentrations in milk being 40 times those in plasma."90m (The iodine discussed here was considered to be a vehicle for radioactivity, and radioactive iodine was known to be important in the incoming substances from Chernobyl.) This remarkable summary finding, published in a prestigious journal just a year before the Chernobyl accident, may have been fresh in the minds of some medical professionals at the time of the accident, leading to medical advice not to breastfeed for a significant period after the surge in radioactivity.
Appendix N: The normal relationship between higher education and breastfeeding in China, unlike that in developed countries:
Two studies have found that the greater likelihood of breastfeeding among more educated mothers did not apply in China. One "nationally representative" study of over 10,000 mothers found that higher education was inversely associated with exclusive breastfeeding, although positively associated with early introduction of breastfeeding; the authors said that their finding of inverse association was consistent with earlier studies in China, including a systematic review.71n Another "nationally representative" study, surveying about 2900 children, found that socioeconomic status was not a significant predictor of breastfeeding initiation, but mothers with higher education were likely to breastfeed for longer duration.71o Overall, breastfeeding in China in relation to socio-economic status is a mixed picture.
Appendix O: “Recall bias” as a problem in observational studies: In a 2005 meta-analysis of 26 publications on the association between breastfeeding and childhood cancer, 85% of the studies relied on the mothers’ long-term recall of feeding history, and those studies overwhelmingly found that breastfeeding was associated with reduced cancer.121 But, of the four studies in that meta-analysis that were based on feeding data gathered during the children’s infancies, not one found any statistically significant association of breastfeeding with reduced cancer. One of those study teams had access to an especially complete record of the breastfeeding histories (those for every child born in Sweden), on which to base its finding of a substantial increase in non-Hodgkins lymphoma in relation to breastfeeding; and there was a dose-response relationship for that finding.122
Appendix P: Long-term accumulation of body burden of dioxins:
Exposures of breastfed infants to toxins are much more than merely a pass-through of current exposures of the mother. According to most studies that have been conducted on this topic, dioxins that are excreted in human milk result from mobilization of accumulations that have built up in a woman's adipose tissue over many years.123 A publication of the National Academies Press, from the Institute of Medicine, refers to a woman’s “body burden” of dioxins, which is reduced during breastfeeding.124 (Also See Appendix L about this process.) And, according to the ATSDR, "lactation provides an efficient mechanism for decreasing the body burden of these compounds," since lipophilic compounds (such as dioxins) concentrate in maternal milk. (Toxicological profile for chlorinated dibenzo-p-dioxins, ATSDR, U.S. Dept. of Health and Human Services, 1998, Section 220.127.116.11, at https://www.atsdr.cdc.gov/toxprofiles/tp104.pdf)
Appendix Q: Other reasons to be skeptical about findings of studies in a matter, such as breastfeeding, about which strong opinions are widely held: According to U.K. authorities on scientific research, in addition to the biases that affect observational studies as referred to in Appendix E, other common kinds of bias (which could favor breastfeeding) include "cherry-picking of findings that support a researcher's own position," publication bias, and pressure to attract funding and to publish regularly, for career progression.7g There are strong pro-breastfeeding biases in organizations whose senior members are likely to influence awarding of research grants as well as selecting which articles to publish, which would be quite apparent to researchers who are seeking funding and trying to publish; these biases are indicated by
-- public statements on breastfeeding by the medical establishment that distort the truth as part of an effort to promote breastfeeding (see Appendix S just ahead), and
-- pro-breastfeeding statements by the U.S. CDC in support of which no evidence is offered despite good reasons to believe that they are not true. (See Appendix R below)
Appendix R: Very misleading statement by the U.S. CDC on toxins in breast milk:
A web page of the CDC states, "While some women may have detectable levels of chemical agents in their breast milk, no established “normal” or “abnormal” levels exist to aid in clinical interpretation." (CDC web page at
https://www.cdc.gov/breastfeeding/breastfeeding-special-circumstances/environmental-exposures/index.html ) The reader is invited to review the chart below from a high-quality scientific study, and the text below it, and possibly also the large amount of authoritative information about toxins in human milk in studies cited at reference no.10, and then count the several different ways in which the above CDC statement is basically incompatible with evidence.
Fig. 2 reviewed
Dioxins and PCBs in human milk
For original source of the above, go to van den Berg et al., WHO/UNEP Global Surveys of PCDDs, PCDFs and DDTs in human milk and benefit-risk evaluation of breastfeeding, at https://www.ncbi.nlm.nih.gov/pubmed/27438348
There is considerable other evidence that supports the findings indicated in these charts (see Appendix A). Also note that the toxicity standards that were in effect during the 2000-2010 surveys, results of which are shown above, have been subsequently replaced by much stricter standards (see Appendix B); so the updated exceedances of established safe intakes of these toxins, according to the best currently-available evidence and research methods, are many times the already major exceedances shown in the charts above. There is also substantial other evidence, from the U.S. and other developed countries, indicating that exposures of breastfed infants to both PCBs and dioxins during the 2000's have exceeded established safe levels by scores to hundreds of times.10
The CDC also states, "To date, effects on the nursing infant have been seen only where the mother herself was clinically ill from a toxic exposure." The phrase, "have been seen only where…," in order to deserve respect, should only be made after a thorough consideration of long-term effects, some evidence of which consideration ought to be given; but none is provided by the CDC. The reader is invited to review Sections 3 to 6 in http://www.pollutionaction.org/breastfed.htm to see what has actually occurred in important disease prevalences while breastfeeding has increased. Then, on the (fairly safe) assumption that long-term effects were not adequately investigated by the CDC, consider whether the above CDC statement constitutes anything other than promotion of an opinion in the absence of actual evidence.
Appendix S: Additional indications of low quality in typical studies such as those dealing with breastfeeding, and non-rational bias in the medical establishment:
Contents of Section 1 are only part of the reason why the many studies cited by medical associations as evidence for beneficial effects of breastfeeding merit only minimal credibility. A respected physician and scientist, John P.A. Ioannidis, Professor of Medicine at Stanford University and member of the U.S. National Academy of Medicine,64a authored an article entitled, "Why Most Published Research Findings Are False," in which he states that "claimed research findings may often be simply accurate measures of the prevailing bias;" he provides considerable evidence for that conclusion.64b There can be little doubt about the prevalence of fundamental, non-rational bias among key medical associations on the subject of breastfeeding when, for instance, the American Academy of Pediatrics, in its policy statement on breastfeeding, does not mention the word "toxin" even once;60f that is despite the very large amount of authoritative evidence indicating considerable toxins in human milk in concentrations far exceeding established safe levels;60m much less does the AAP mention any of the numerous studies that have found adverse effects of breastfeeding.60g Deep-seated bias is also apparent when the American Academy of Family Physicians makes the false statement (in its position statement on breastfeeding) as follows: "By using formula, they (mothers) do not reduce exposure to environmental toxins;" 60h the AAFP makes no attempt to support that statement with evidence either together with the statement or when challenged on it. (Appendix A)
The reader is also encouraged to
-- review all of the many diseases and disorders among children and young adults that have been greatly increasing since the 1970's (Sections 5 and later in http://www.pollutionaction.org/breastfed.htm) and also
-- remember the major increases in breastfeeding that have occurred since the late 1960's (Figure 5), and then
-- reflect on the fact that the medical establishment publicly states that almost all of those diseases and disorders are reduced by breastfeeding.60k
It should be apparent from the above (including the preceding Appendix R) that bias in favor of breastfeeding, intense enough to lead to suppression of contrary views and facts where they are able to do so, is pervasive in the medical establishment and in the U.S. government's health establishment.
Above cancer trend info is found at https://seer.cancer.gov/statfacts/.
The next five most common types of cancer (Non-Hodgkin lymphoma, kidney and renal pelvis, uterus, leukemia, and pancreas) were either stable or showed minor increases.
Appendix U: Breastfeeding and childhood cancer in France:
Although very little is shown for France in Figure 12, the short line that does show indicates a minor decline of breastfeeding in recent years; there is also other evidence of a minor decline in breastfeeding in France in recent years (exclusive breastfeeding went from 60% to 52%, 2010-2016).89x This is compatible with what a study team observed about childhood cancer incidence in France for the period 2000-2014: no overall increase, and decreasing trends in many subtypes of childhood cancer.87 And, since reported incidence in recent decades has been artificially driven up by increased ascertainment (advanced diagnostic equipment, etc.),73a the reported average flatness of childhood cancer in France in the last two decades has probably in reality represented a minor decrease of overall actual childhood cancer.
Appendix V: Other parallels between breastfeeding and childhood cancer in Eastern Europe:
-- Estonia -- low-to-medium, with moderate increases in both breastfeeding and childhood cancer: Childhood cancer incidence figures for Estonia were 122.2 per million for 1970-1994 and 138.1 for 1995-2016,89p low and low-medium, while that country's breastfeeding rates were low-medium to average (judging by what we can see in Figure 14). And the moderate increase in childhood cancer in Estonia was in line with the moderate increase in breastfeeding that has occurred in Estonia.
-- Czechia -- high childhood cancer and high PCBs: Czechia is seen in Figure 14 to have had a low breastfeeding rate in recent decades; that is another example of environmental contamination sometimes being even more important than breastfeeding rates and duration. Czechia's childhood cancer rate is relatively high,89m which is compatible with its exceptionally high level of environmental PCBs (see top right country in Figure 2).
-- Ukraine has in recent years been at the bottom position in the breastfeeding rates shown in Figure 14; and its most recent childhood cancer incidence has been just above the bottom among European countries, at 133.7 per million (mean for 2013-2017).89t.
Ukraine has also seen an exceptional decline in its breastfeeding since 1997 (Figure 14); and Ukraine has had a decline of its childhood cancer incidence (down to 133.7 per million from 139.6 per million for the years 2002-2012 89u); a decline such as that has been exceptional compared with the average increases that have been the norm in Europe in recent decades. (see earlier)
This recent decline in Ukraine's childhood cancer incidence was not merely an example of a random, year-to-year fluctuation; the earlier, higher rate was a summary figure for a decade, and the later, lower figure was an average for the following half-decade. It may be something other than random coincidence that this exceptional decline of childhood cancer in recent years occurred in a country that has also been exceptional in having gone through a large decline in its breastfeeding rates during the most recent decades.
1) Sperm Counts Continue to Fall, Ashley Fetters, The Atlantic, Oct. 12, 2018, athttps://www.theatlantic.com/family/archive/2018/10/sperm-counts-continue-to-fall/572794/
1a) Martin et al., Effects of Promoting Longer-Term and Exclusive Breastfeeding on Cardiometabolic Risk Factors at Age 11.5 Years, A Cluster-Randomized, Controlled Trial, Circulation (journal), at https://www.ahajournals.org/doi/full/10.1161/CIRCULATIONAHA.113.005160
1b) Owen et al., Effect of breast feeding in infancy on blood pressure in later life: systematic review and meta-analysis. The BMJ. 2003, at https://www.bmj.com/content/327/7425/1189.short
1c) Children with autism at significant risk for feeding problems and nutritional deficits, Woodruff Health Sciences Center | Feb. 4, 2013 at http://www.news.emory.edu/stories/2013/02/autism_nutritional_deficits/
Lucas et al., Dysregulated Breastfeeding Behaviors in Children Later Diagnosed With Autism, J Perinat Educ. 2015; 24(3): 171–180 at http://www.ncbi.nlm.nih.gov/pmc/articles/PMC4720870/
Also see Section 6.c, "Early weaning in relation to autism" at http://www.pollution-effects.info/.
2) ATSDR Toxic Substances Portal: Addendum to the Profile, 1998: Toxicological Profile for Chlorinated Dibenzo-p-dioxins (CDDs), pp. 378 ff, at https://www.atsdr.cdc.gov/ToxProfiles/tp.asp?id=366&tid=63
Also ATSDR: Toxicological Profile for Chlorinated Dibenzo-P-Dioxins, at https://www.atsdr.cdc.gov/toxprofiles/tp104.pdf, p. 2
Also see Section 2.a in http://www.pollutionaction.org/shorter-lives.htm.
2b) Gary Taubes, Do We Really Know What Makes Us Healthy? The New York Times Magazine, Sept. 16, 2007, at http://www.nytimes.com/2007/09/16/magazine/16epidemiology-t.html?_r=0
2c) Soumerai et al., How Do You Know Which Health Care Effectiveness Research You Can Trust? A Guide to Study Design for the Perplexed, Preventing Chronic Disease, 2015, at https://www.cdc.gov/pcd/issues/2015/15_0187.htm
3) Lawlor et al, Commentary: The hormone replacement-coronary heart disease conundrum: is this the death of observational epidemiology? International Journal of Epidemiology, v. 33, Issue 3, June 2004, Oxford University Press, at https://academic.oup.com/ije/article/33/3/464/716652
3a) Maki et al., Limitations of Observational Evidence: Implications for Evidence-Based Dietary Recommendations, Advances in Nutrition, 2014, at https://academic.oup.com/advances/article/5/1/7/4558018 In the section, "Implications for Evidence-Based Reviews…", the authors refer to "the frequency with which results from RCTs have failed to confirm hypotheses suggested by observational evidence…."
3b) Huded et al., When Research Evidence is Misleading, AMA Journal of Ethics, 2013, at https://journalofethics.ama-assn.org/article/when-research-evidence-misleading/2013-01
3c) Institute of Medicine: Standards for Finding and Assessing Individual Studies, Chapter 3, pp. 134-5, in Finding What Works in Health Care: Standards for Systematic Reviews. 2011. Washington, DC: The National Academies Press. At https://www.nap.edu/read/13059/chapter/5#134
3d) Perez et al., A comprehensive review of randomized clinical trials in three medical journals reveals 396 medical reversals, eLife, 2019, at https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6559784/
3e) Haslam et al., Medical Reversals in Family Practice: A Review, Curr Ther Res Clin Exp, 2020, at https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7063107/
3f) Prasad et al., A Decade of Reversal: An Analysis of 146 Contradicted Medical Practices, Mayo Clinic Proceedings, 2013, at https://www.mayoclinicproceedings.org/article/S0025-6196(13)00405-9/fulltext
3g) Solo-Rivera and Agos, Tight Glycemic Control in the Pediatric Intensive Care Unit, J Pediatr Intensive Care. 2016, at https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6512400/
4) EPA: Health Consequences of Dioxin Exposure, at https://cfpub.epa.gov/si/si_public_record_Report.cfm?Lab=NHEERL&dirEntryId=171623
5) IARC Monographs Volume 107: Polychlorinated Biphenyls and Polybrominated Biphenyls at https://www.iarc.fr/news-events/iarc-monographs-volume-107-polychlorinated-biphenyls-and-polybrominated-biphenyls/
Also IARC web page on dioxin at https://monographs.iarc.fr/wp-content/uploads/2018/06/mono100F-27.pdf
6) Donat-Vargas et al., Dietary polychlorinated biphenyls, long-chain n-3 polyunsaturated fatty acids and incidence of malignant melanoma, European Journal of Cancer, 2017, at https://www.sciencedirect.com/science/article/abs/pii/S0959804916325941
6a) Hooiveld et al., Second Follow-up of a Dutch Cohort Occupationally Exposed to Phenoxy Herbicides, Chlorophenols, and Contaminants, American Journal of Epidemiology, 1998, Table 7, at https://academic.oup.com/aje/article/147/9/891/68169
7) Steenland et al., Dioxin Revisited: Developments Since the 1997 IARC Classification of Dioxin as a Human Carcinogen, Environ Health Perspect, 2004, at https://ehp.niehs.nih.gov/doi/full/10.1289/ehp.7219
8) MacIntosh et al., Mitigation of building-related polychlorinated biphenyls in indoor air of a school, Environ Health. 2012; 11: 24, at https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3353159/
European Commission: Science for Environment Policy: Building materials used between 1950 and 1980 in Europe may contribute to PCB air pollution, 15 Apr. 2016, at http://ec.europa.eu/environment/integration/research/newsalert/pdf/building_materials_used_1950_1980_in_europe_may_contribute_to_pcb_air_pollution_453na2_en.pdf
8a) 2,3,7,8-Tetrachlorodibenzo-P-dioxin, U.S. National Library of Medicine, at https://pubchem.ncbi.nlm.nih.gov/compound/15625#section=Use-and-Manufacturing
8b) Sidlova et al., Dioxin-Like and Endocrine Disruptive Activity of Traffic-Contaminated Soil Samples, Archives of Environmental Contamination and Toxicology, 200, at https://link.springer.com/article/10.1007%2Fs00244-009-9345-4
9) EPA: An Inventory of Sources and Environmental Releases of Dioxin-Like Compounds in the United States for the Years 1987, 1995, and 2000, EPA/600/P-03/002F November 2006, Table 1-12, at https://cfpub.epa.gov/ncea/dioxin/recordisplay.cfm?deid=159286 ; under Downloads, choose to download the Dioxin Inventory of Sources (Optimized Report).
At the EPA "An Inventory of Sources...." listed above, see p. xliv and Section 6.5, "Backyard Barrel Burning."
Tox Town: Polyvinyl Chloride (PVC), at https://toxtown.nlm.nih.gov/text_version/chemicals.php?id=84
New York Times: In Rural Areas, the Heat Is on Over Practice of Trash Burning, By Lisa W. Foderaro, MARCH 7, 2005 , at https://www.nytimes.com/2005/03/07/nyregion/in-rural-areas-the-heat-is-on-over-practice-of-trash-burning.html?pagewanted=all
9a) P Grandjean and AA Jensen, Breastfeeding and the Weanling's Dilemma Am J Public Health. 2004 July; 94(7): 1075. PMCID: PMC1448391 at http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1448391/
9b)Thomas and Spiro, An estimation of dioxin emissions in the United States, Toxicological and Environmental Chemistry, 1995, at https://www.tandfonline.com/doi/abs/10.1080/02772249509358202
9c) Freinkel: A Brief History of Plastic's Conquest of the World, Scientific American, at https://www.scientificamerican.com/article/a-brief-history-of-plastic-world-conquest/
9d) Dopico and Gomez, Review of the current state and main sources of dioxins around the world, Journal of the Air & Waste Management Assn., 2015, at https://www.tandfonline.com/doi/full/10.1080/10962247.2015.1058869?src=recsy
10) Re: EPA's RfD for dioxin: At www.epa.gov/iris/supdocs/dioxinv1sup.pdf in Section 4.3.5, at end of that section, "...the resulting RfD in standard units is 7 10−10 mg/kg-day." (that is, O.7 pg of TEQ/kg-d)
Re: breastfed infants' exposures to dioxins and PCBs in U.S. and internationally:
- Arisawa et al., Background exposure to PCDDs/PCDFs/PCBs and its potential health effects : a review of epidemiologic studies, The Journal of Medical Investigation Vol. 52 2005, at https://www.jstage.jst.go.jp/article/jmi/52/1%2C2/52_1%2C2_10/_pdf
- Lorber et al., Infant Exposure to Dioxin-like Compounds in Breast Milk, Vol. 110 No. 6, June 2002, Environmental Health Perspectives at http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=54708#Download, indicating 242 pg of TEQ/kg-d at initiation of breastfeeding.
Also see Patandin et al. in ref 84c for other U.S. information.
- Focant et al., Levels of polychlorinated dibenzo-p-dioxins, polychlorinated dibenzofurans and polychlorinated biphenyls in human milk from different regions of France, Science of The Total Environment, Volumes 452-453, 1 May 2013, Pages 155-162 abstract at http://www.sciencedirect.com/science/article/pii/S0048969713002404
- Yang J, et al., PCDDs, PCDFs, and PCBs concentrations in breast milk from two areas in Korea: body burden of mothers and implications for feeding infants. Chemosphere. 2002 Jan;46(3):419-28. At www.ncbi.nlm.nih.gov/pubmed/11829398
- Bencko V et al., Exposure of breast-fed children in the Czech Republic to PCDDs, PCDFs, and dioxin-like PCBs. Environ Toxicol Pharmacol. 2004 Nov;18(2):83-90. Abstract at http://www.ncbi.nlm.nih.gov/pubmed/21782737/
- Nakatani T, et al., Polychlorinated dibenzo-p-dioxins, polychlorinated dibenzofurans, and coplanar polychlorinated biphenyls in human milk in Osaka City, Japan Arch Environ Contam Toxicol. 2005 Jul;49(1):131-40. Epub 2005 Jun 22. Found at http://www.ncbi.nlm.nih.gov/pubmed/15983863
- Deng B, et al., Levels and profiles of PCDD/Fs, PCBs in mothers' milk in Shenzhen of China: estimation of breast-fed infants' intakes.Environ Int. 2012 Jul;42:47-52.. At www.ncbi.nlm.nih.gov/pubmed/21531025
- Chovancov J, et al., PCDD, PCDF, PCB and PBDE concentrations in breast milk of mothers residing in selected areas of Slovakia Chemosphere. 2011 May;83(10):1383-90. doi: 10.1016/j. At www.ncbi.nlm.nih.gov/pubmed/21474162
- J Grigg, Environmental toxins; their impact on children's health, Arch Dis Child 2004;89:244-250 doi:10.1136/adc.2002.022202 at http://adc.bmj.com/content/89/3/244.full
-- Oregon Department of Environmental Quality Environmental Cleanup Program, Oct. 2010, 10-LQ-023, p. D2-4 (near very end) at http://www.deq.state.or.us/lq/pubs/docs/cu/HumanHealthRiskAssessmentGuidance.pdf The doses of PCBs that a breastfeeding infant may be expected to receive, given breast milk PCB concentrations measured in the literature, are presented in table 1. These doses range from 0.0019 to 0.0081 mg/kg/day and are 63-270 times higher than ATSDR's minimal risk level (0.00003 mg/kg/day) for PCB exposures that last between 15 and 364 days.
14e) Qin et al., Does physical activity modify the risk of obesity for type 2 diabetes: a review of epidemiological data, Eur J Epidemiol, 2010, at https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2807936/
14f) Of the first 10 studies provided in the American Academy of Pediatrics' Policy Statement on Breastfeeding to indicate better outcomes for breastfed children, all 10 measured the outcomes in infancy or early childhood. American Academy of Pediatrics, Policy Statement, Breastfeeding and the Use of Human Milk, at https://pediatrics.aappublications.org/content/129/3/e827, as accessed May 4, 2020. Starting with footnote number 13.
15) Grandjean et al., The Faroes Statement: Human Health Effects of Developmental Exposure to Chemicals in Our Environment, Basic and Clinical Pharmacology and Toxicology, Volume 102, Issue 2, February 2008 at http://onlinelibrary.wiley.com/doi/10.1111/j.1742-7843.2007.00114.x/full
15a) Committee on Developmental Toxicology, Board on Environmental Studies and Toxicology, in Scientific Frontiers in Developmental Toxicology and Risk Assessment (2000) , Commission on Life Sciences, The National Academies Press, p. 56
15b) Early Experiences Can Alter Gene Expression and Affect Long-Term Development, Working Paper 10 of The National Scientific Council on the Developing Child, Center on the Developing Child at Harvard University, 2010, at https://developingchild.harvard.edu/wp-content/uploads/2010/05/Early-Experiences-Can-Alter-Gene-Expression-and-Affect-Long-Term-Development.pdf
16) Endocrine disrupters and child health, Possible developmental early effects of endocrine disrupters on child health, at https://apps.who.int/iris/bitstream/handle/10665/75342/9789241503761_eng.pdf;jsessionid=6C174B18182AABB475365AA1BC52E5AF?sequence=1 p. 51
WHO: Endocrine disrupters and child health, Possible developmental early effects of endocrine disrupters on child health, p. 17, at https://apps.who.int/iris/bitstream/handle/10665/75342/9789241503761_eng.pdf;jsessionid=6C174B18182AABB475365AA1BC52E5AF?sequence=1
ATSDR: Toxicological Profile for Chlorinated Dibenzo-P-Dioxins, at https://www.atsdr.cdc.gov/toxprofiles/tp104.pdf, p. 279
17a) Nagayama et al., Postnatal exposure to chlorinated dioxins and related chemicals on thyroid hormone status in Japanese breast-fed infants, Chemosphere. 1998 Oct-Nov;37(9-12):1789-93.at http://www.ncbi.nlm.nih.gov/pubmed/9828307
18) U.S. ATSDR, Persistent chemicals found in breast milk, Appendix A, p. 180, at https://www.atsdr.cdc.gov/interactionprofiles/ip-breastmilk/ip03-a.pdf
19a) Furst: Dioxins, polychlorinated biphenyls and other organohalogen compounds in human milk, Mol Nutr Food Res, 2006
19b) p. 3, top, of EPA: Technical Fact Sheet -- Polybrominated Diphenyl Ethers (PBDEs) 2017, at https://www.epa.gov/sites/production/files/2014-03/documents/ffrrofactsheet_contaminant_perchlorate_january2014_final_0.pdf
19d) Jusko et al., Prenatal and Postnatal Serum PCB Concentrations and Cochlear Function in Children at 45 Months of Age, Environmental Health Perspectives, 22 July 2014 (Advance Pub.) at http://ehp.niehs.nih.gov/wp-content/uploads/advpub/2014/7/ehp.1307473.pdf
See also U.S. ATSDR: Toxicological Profile for Polychlorinated Biphenyls (PCBs), Nov. 2000, at
See also paragraph 39 in UK Committee on Toxicity of Chemicals in Food, Consumer Products and the Environment: COT Statement on a toxicological evaluation of chemical analyses carried out as part of a pilot study for a breast milk archive, 2004, at https://cot.food.gov.uk/sites/default/files/cot/cotsuremilk.pdf
20) Schecter et al., Polybrominated Diphenyl Ether (PBDE) Levels in an Expanded Market Basket Survey of U.S. Food and Estimated PBDE Dietary Intake by Age and Sex Environ Health Perspect. 2006 October; 114(10): 1515–1520. Published online 2006 July 13. doi: 10.1289/ehp.9121, PMCID: PMC1626425)
20a) Page 5-55 of EPA: An exposure assessment of polybrominated diphenyl ethers. National Center for Environmental Assessment, Washington, DC; EPA/600/R-08/086F. online at http://www.epa.gov/ncea or directly at http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=210404 )
21) Near end of Section 5.6.2 ("Impacts to Infants from Consumption of Breast Milk"), p. 5-79, of An exposure assessment of polybrominated diphenyl ethers. National Center for Environmental Assessment, Washington, DC; EPA/600/R-08/086F. online at http://www.epa.gov/ncea or directly at http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=210404
21a) Abraham et al. Time course of PCDD/PCDF/PCB concentrations in breast-feeding mothers and their infants. Chemosphere 1998; as described in Massart et al., Human Breast Milk and Xenoestrogen Exposure: A Possible Impact on Human Health, Journal of Perinatology, 2005, at https://www.academia.edu/26062655/Human_Breast_Milk_and_Xenoestrogen_Exposure_A_Possible_Impact_on_Human_Health?email_work_card=view-paper
Also Patterson et al., Correlation between serum and adipose tissue levels of 2,3,7,8-tetrachlorodibenzo-p dioxin in 50 persons from Missouri. Arch Environ Contam Toxicol 1988, at https://link.springer.com/article/10.1007/BF01056017
22) GM Lehmann, JS LaKind, MH Davis, EP Hines, SA Marchitti, C Alcala, and M Lorber: Environmental Chemicals in Breast Milk and Formula: Exposure and Risk Assessment Implications, Env. Health Perspect, 2018, at https://ehp.niehs.nih.gov/doi/full/10.1289/ehp1953 , in "Example 3…" section near end. The statement quoted is found at the bottom of this screenshot from that document:
Above page found at https://ehp.niehs.nih.gov/doi/full/10.1289/ehp1953
23) Patel et al., Cohort Profile: The Promotion of Breastfeeding Intervention Trial (PROBIT), Int J Epidemiol. 2014 Jun; at https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4052126/
31) American Academy of Pediatrics: Pediatric Environmental Health, 3rd Edition, 2012, p. 200; available at AAP EBooks, at https://ebooks.aappublications.org
31a) NTP Technical Report on the Toxicology and Carcinogenesis Studies of Perfluorooctanoic Acid (CASRN 335-67-1) Administered in Feed to Sprague Dawley (Hsd:Sprague Dawley® SD®) Rats, May 2020, at https://ntp.niehs.nih.gov/ntp/htdocs/lt_rpts/tr598_508.pdf
31b) Grandjean et al., Estimated Exposures to Perfluorinated Compounds in Infancy Predict Attenuated Vaccine Antibody Concentrations at Age 5-Years, J Immunotoxicol, 2018, at https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6190594
31c) Mogensen et al., “Breastfeeding as an Exposure Pathway for Perfluorinated Alkylates.” Environmental Science & Technology, 2017,
31d) Mondal et al., Breastfeeding: A Potential Excretion Route for Mothers and Implications for Infant Exposure to Perfluoroalkyl Acids, Environ Health Perspect, 2014, at https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3915259/
31e) Figure 3.b in Kingsley et al., Variability and predictors of serum perfluoroalkyl substance concentrations during pregnancy and early childhood, Environ Res, 2018, at https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6309672/#SD1
32) Infant Exposure to Dioxin-like Compounds in Breast Milk Lorber and Phillips Vol. 110., No. 6 June 2002, Environmental Health Perspectives at http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1240886/pdf/ehp0110-a00325.pdf, 242 pg TEQ/kg-day (body-weight basis) at initiation;
Formula: Results from what appears to be the most recent major survey of dioxins in infant formula are from the following: Government of Canada: Dioxins and Dioxin-Like Compounds in Selected Foods - April 1, 2012 to March 31, 2014, at https://www.inspection.gc.ca/food-safety-for-industry/chemical-residues-microbiology/food-safety-testing-bulletins/2019-04-17/dioxins-and-dioxin-like-compounds-in-food/eng/1551804364813/1551804419782 (average 0.029 pg TEQ/g fat, based on 120 predominantly imported samples, which should be compared with the EU's limit of 5.5 pg TEQ/g fat for milk and dairy products)
Compatible figures were found in Weijs PJ, et al., Dioxin and dioxin-like PCB exposure of non-breastfed Dutch infants, Chemosphere 2006 Aug;64(9):1521-5. Epub 2006 Jan 25 at www.ncbi.nlm.nih.gov/pubmed/16442144
33) WHO, Persistent Organic Pollutants: Impact on Child Health, p. 6, at http://whqlibdoc.who.int/publications/2010/9789241501101_eng.pdf
42a) Ch. 6 Risk Assessment (Overall Conclusions section) in Possible Health Effects of Exposure to Residential Electric And Magnetic Fields, 1997, U.S. National Academy of Sciences, National Academies Press, at https://www.ncbi.nlm.nih.gov/books/NBK232740/
Merriam-Webster at https://www.merriam-webster.com/dictionary/epidemiological?pronunciation&lang=en_us&dir=e&file=epidem07
51) The Surgeon General's Call to Action to Support Breastfeeding, 2011 (including page 33), U.S. Department of Health and Human Services, Washington, D.C., at https://www.ncbi.nlm.nih.gov/books/NBK52682/pdf/Bookshelf_NBK52682.pdf
52) Martin et al., Effects of promoting longer-term and exclusive breastfeeding on adolescent adiposity, blood pressure, and longitudinal growth trajectories: evidence from the PROBIT cluster-randomized trial, JAMA Pediatr, 2017, at https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5576545
53) Colen and Ramey, Is Breast Truly Best? Estimating the Effect of Breastfeeding on Long-term Child Wellbeing in the United States Using Sibling Comparisons, Soc.Sci.Med. 2014, at https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4077166/
None of the regression coefficients in this study reached statistical significance, but the true effects of breastfeeding may well be worse for the child than as found in this study; the authors pointed out that, despite the usefulness of the discordant sibling method of comparison, there could still be bias in favor of the breastfed infants: "All of the scenarios we can call to mind in which siblings are differently fed favor the breastfed sibling -- for example, sibling A is born full-term and breastfed while sibling B is born preterm and bottle-fed." (A significant positive relationship has been found between pre-term birth and bottle feeding.57) So the true effects of breastfeeding could well be even worse than as found in this high-quality study.
54) Jacobson et al., Effects of Exposure to PCBs and Related Compounds on Growth and Activity in Children, Neurotoxicology and Teratology, 1990, Vol. 12, pp. 319-326, at http://ac.els-cdn.com/089203629090050M/1-s2.0-089203629090050M-main.pdf?_tid=5f238b74-1d4f-11e7-8a29-00000aacb35d&acdnat=1491761419_84450b69c691bfdc5ffd9534957f7940
55) Kostyniak et al., Relation of Lake Ontario Fish Consumption, Lifetime Lactation, and Parity to Breast Milk Polychlorobiphenyl and Pesticide Concentrations, Environmental Research, 1999, at http://www.sciencedirect.com/science/article/pii/S0013935198939391. Great Lakes fish-eaters have been found to have PCB levels about 35% higher than non-fish-eaters for some of the various forms of PCBs.
56) Koja et al., Changes of gross behavior
with polychlorinated biphenyls (PCB) in
immature rats [in Japanese]. Kagoshima
Daigaka Igaka Zasshi. 1978 Cited in Tilson et al. (below)
Koja et al., Effects of polychlorinated
biphenyls (PCB) on the gross
behavior of immature rats and the influence
of drugs upon them [in Japanese].
Kagoshima Daigaka Igaka Zasshi 1979;31:
315-319. Cited in Tilson et al. (below)
Tilson HA et al., Polychlorinated biphenyls and the developing nervous system: cross-species comparisons.
Neurotoxicol Teratol. 1990; at https://pubmed.ncbi.nlm.nih.gov/2115098/
Also Johansen et al., Postnatal exposure to PCB 153 and PCB 180, but not to PCB 52, produces changes in activity level and stimulus control in outbred male Wistar Kyoto rats, Behavioral and Brain Functions, BioMed Central Ltd. 2011, at https://behavioralandbrainfunctions.biomedcentral.com/articles/10.1186/1744-9081-7-18 One of the experiments was by a team of researchers who said that the concentrations of the toxins found in their test animals' brains were "about the same order of magnitude as observed in infants less than 1 year old."
56u) Perkins et al., Polychlorinated Biphenyls and links to Cardiovascular Disease, Environ Sci Pollut Res Int, 2016, at https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4609220/#:~:text=Chlorinated%20organic%20pollutants%2C%20such%20as,factors%20for%20developing%20cardiovascular%20disease.
56v) NIH: Physical Activity and Cancer, at https://www.cancer.gov/about-cancer/causes-prevention/risk/obesity/physical-activity-fact-sheet
56w) Vineis and Wild, Global cancer patterns: causes and prevention, The Lancet, 2014, at https://www.antsz.hu/data/cms62790/Lancet.pdf
56x) Ombra et al., Dietary compounds and cutaneous malignant melanoma: recent advances from a biological perspective, Nutrition and Metabolism, 2019, at https://link.springer.com/article/10.1186/s12986-019-0365-4
57) Chiang et al., Receipt of Breast Milk by Gestational Age - United States, 2017, MMWR Morb Mortal Wkly Rep. 2019 Jun, at https://www.ncbi.nlm.nih.gov/pubmed/31170123?deliveryName=USCDC_1296-DM10201
58) Doll R, Peto R. The causes of cancer: quantitative estimates of avoidable risks of cancer in the United States today. J Natl Cancer Inst. 1981
59) WC Willett (Editorial): Diet and Cancer: An Evolving Picture, JAMA, 2005
60) Schottenfeld et al., Current Perspective on the Global and United States Cancer Burden Attributable to Lifestyle and Environmental Risk Factors, Annual Review of Public Health, 2013, at https://www.annualreviews.org/doi/full/10.1146/annurev-publhealth-031912-114350
Also WC Willett (Editorial): Diet and Cancer: An Evolving Picture, JAMA, 2005
60a) M. White, The Causes of Cancer: What Has Happened Since Doll and Peto's Landmark Paper?, Epidemiology, 2008, at https://journals.lww.com/epidem/Fulltext/2008/01000/The_Causes_of_Cancer__What_Has_Happened_Since_Doll.75.aspx
60b) William J. Blot, Robert E. Tarone, Doll and Peto’s Quantitative Estimates of Cancer Risks: Holding Generally True for 35 Years, JNCI: Journal of the National Cancer Institute, 2015, at https://academic.oup.com/jnci/article/107/4/djv044/894954
60c) Song and Giovannucci, RE: Doll and Peto’s Quantitative Estimates of Cancer Risks: Holding Generally True for 35 Years, Journal of the National Cancer Institute, 2015, at https://academic.oup.com/jnci/article/107/10/djv240/987225
60f) Breastfeeding and the Use of Human Milk, AAP Section on Breastfeeding, Pediatrics, 2012, at https://pediatrics.aappublications.org/content/129/3/e827.full#content-block
60h) American Academy of Family Physicians: Breastfeeding (Policy Statement) at http://www.aafp.org/about/policies/all/breastfeeding.html
60n) Greiner, Exclusive breastfeeding: measurement and indicators, International Breastfeeding Journal, 2014, at https://internationalbreastfeedingjournal.biomedcentral.com/articles/10.1186/1746-4358-9-18#Sec1 "Internationally, most surveys use a point-in-time or current status measurement (usually 24-hour recall)…."
Fewtrell, Promoting and Protecting Breast-feeding: The Importance of Good Quality Data, JPGN, 2019, at https://journals.lww.com/jpgn/Fulltext/2019/03000/Promoting_and_Protecting_Breast_feeding__The.5.aspx
"The WHO indicator … infants aged 0 to 5 months who are fed exclusively with breast milk, based on the diet during the 24 hours before the survey (to avoid recall bias).
"…many HIC (high income countries) record the proportion of infants who are still exclusively breast-fed at around 6 months. … asked about infant feeding during the last 24 hours."
UK National Health Service: Infant Feeding Survey 2005, Appendix I: Survey Methods, page 383, at https://files.digital.nhs.uk/publicationimport/pub00xxx/pub00619/infa-feed-serv-2005-apx.pdf
U.S. CDC: Methods: Breastfeeding Rates, National Immunization Survey (NIS), at https://www.cdc.gov/breastfeeding/data/nis_data/methods.html
61) Cancer Research UK: Cancer Incidence Statistics: Cancer incidence for all cancers combined, at https://www.cancerresearchuk.org/health-professional/cancer-statistics/incidence#heading-Zero