|List of chapters|
Our chemical fears
P. 218 left and figure 117: FLAW
"Here we see only a slight increase in age-adjusted cancer rates . . ". Flaw: Lomborg writes that in the grey curve for "Age-adjusted cancer death rate", age has been adjusted to standard world population. Now, if we check the trend over time in USA for each age class (data from www-depdb.iarc.fr/who), we see that the mortality is increasing for age classes at 55 and above, but decreasing for lower age classes. As by far the greater amount of cancer deaths occur in people older than 55 years, the overall trend is that the mortality from cancer is increasing, even if we adjust for changes in the age composition of the US population. Lomborg´s "age adjustment", however, is a calculation of what the mortality would be if the age distribution were the same as in the whole world, i.e. with many more young people and fewer old people than in USA. As the trend is positive for the young age classes, and the young age classes are given greater weight, this tranformation will make the trend become more positive than it actually is. The transformation might be justified if Lomborg compared USA with the rest of the world, but he does not. Thus, this unmotivated transformation makes the situation look more favourbale than it is. This seems to be a case of deliberate misleading.
P. 221 right: BIAS
" . . regular screening for breast tumors . . has been instituted in the US, the UK and Canada . . ". Flaw: In the whole paragraph on mortality from cancer, Lomborg speaks only of USA, as if this were the whole world. Only in the quoted sentence he mentions two other countries. However, if we check the trends in mortality from breast cancer worldwide (data from www-depdb.iarc.fr/who), we see that the trend for slight reductions in mortality in later years is evident only in USA, Canada and UK (i.e. just the countries mentioned by Lomborg) and in a few others where regular screening is also implemented (e.g. Denmark, Sweden). In most other countries, the trend is towards increased mortality. This is true in most of Europe (e.g. Germany, France, Italy and Spain) and especially in east-central Europe and Russia and Ukraine. It is also true in most of Latin America, and in Japan. Altogether, the dominating trend in the world is for an increasing mortality. Allegedly, Lomborg wants to show us the global pattern. But by mentioning only countries where the trend is favourable, and omitting the majority of countries where it is unfavourable, his presentation is biased.
P. 222 right: ERROR OF OMISSION
"Cancer death rates are declining . . " Error: On p. 8 Lomborg writes: "I shall always attempt to present the most comprehensive figures in order to describe the development of the entire world or the relevant regions. What we need is global trends." So, by his own standards, he should write about cancer mortality in the whole world, not just USA. Actually, he has seen relevant figures. In several chapters, he refers to a book chapter by Murray and Lopez (1996) where they list mortality causes for the whole world in 1990 and - projected - in 2000, 2010 and 2020. It is clear from his legend to figure 8 (p. 26) that he has utilized this source of information, namely for infectious diseases where the projected trend is downward. In the very same tables, there are also data for cancer deaths worldwide, and here the projected trend is steadily increasing, in terms of absolute numbers as well as in terms of age-adjusted death rates. The basis for such a pessimistic prognosis is that when figures for cancer deaths and national income per capita for 47 countries are compared, there is a marked trend for cancer deaths to increase with increasing income (contrary to Lomborg´s point that the richer we get, the more problems we can solve). As the average income is expected to rise in the future, the cancer rate is also expected to rise. So when Lomborg has global figures for a disease type that will become less frequent, and for another disease type that will become more frequent, he omits the figures in the latter case, and speaks of the situation for only one nation, USA. This bias is so crucial that it is counted here as an error.
P. 226 bottom left: GROUNDLESS DEROGATION
"The European Environment Agency named pesticides in drinking water one of the `dominant health issues in Europe´ . . ". Flaw: This formulation seems due to malicious intent. The EEA report has a heading titled ´Dominant health issues in Europe´. Under this heading, there is an extensive treatment of pesticide residues in food. For completeness, this is followed by a short paragraph on pesticides in drinking water, from which it is evident that these constitute a minor problem, relative to pesticides in food. But because this paragraph is placed in the chapter with the above title, Lomborg feels that he can write as he does. His text on this point is grossly and deliberately misleading.
P. 232-233: FLAW
"Chemically, there seems to be no basis for distinguishing between natural and synthetic pesticides." Flaw: This is not true. Most of the most problematic synthetic pesticides are halogenated, whereas natural pesticides are not. In addition, there is a biological difference: Many synthetic pesticides are not readily decomposable by biological agents, e.g. bacteria, whereas the natural ones are so.
P. 233 top left: ERROR
"Arsenic has been used as a weed-killer . . " Error: What has been used is not arsenic as such (the element), but arsenic trioxide. This is not a mineral. It has not been used a a weed-killer, but as a wood preservative.
P. 233 top left: FLAW
"Aflatoxin is the most carcinogenic pesticide known . ." Flaw: Aflatoxin is not usually designated a pesticide, and even in the special context here it seems unjustified to use the word pesticide.
P. 233 left: FLAW
"Typically, natural pesticides make up 5-10 percent of a plant´s dry weight." Flaw. This is not correct. Pesticides, by definition, are compounds that may kill your enemies. Other compounds, repellants and defense substances, just keep your enemies away. Most of the alleged 5-10 percent are of this type - e.g. tannins and terpenes. What Lomborg does here, is to equal our consumption of tannins e.g. in tea with consumption of pesticides like lindane. This is misleading and unjustified.
P. 237 right: FLAW
" . . synthetic estrogens are not, like natural estrogen, blocked by the body´s other proteins" . . "Many plants also contain natural estrogens . . . " . . "our overall intake of synthetic estrogens calculated in estrogen equivalents is more than 40 million times lower than the average intake of natural plant estrogens." Flaw: It is completely unclear from the text whether natural plant estrogens are among the substances that are blocked by the type of proteins called SHBGs. Lomborg knows that this seems to be so for many of them, but forgets to mention this. He knows it, because Sharpe & Skakkebæk (1993) bring on their p. 1393 a table which is an overview of possible routes of human exposure to estrogens. Lomborg has quoted this table (on his p. 239), but has left out those parts which he does not like. Among these are the sentence: "Paradoxically, plant oestrogens may reduce exposure to endogenous oestrogens by increasing production of SHBG". The same is explained in their general text. Now, when we come to the bottom of p. 237, we are told that the exposure to natural plant estrogens is 40 mio. times larger than the exposure to "synthetic estrogens", but now all is forgotten about the point that the former activate blocking proteins, whereas the latter are not blocked by proteins.
P. 237 right: FLAW
"Many plants also contain natural estrogens for the same reason . . " Flaw. In the right column of page 237, Lomborg uses a lot of space to describe the negative effects of plant estrogens, but does not with one word mention the positive effects. For instance, isoflavones such as genistein bind heavily to mammalian/human estrogen receptors. The effect of this may be positive as well as negative. By blocking the receptor, they may reduce the impact of other estrogens and thus prevent the action of cancerogenic substances. On genistein in red clover, Lomborg´s source (Toppari et al. 1996) writes that depending on the dose they are either estogenic or antiestrogenic. In addition, isoflavones do not damage DNA and thus are not cancerogenic. Therefore, they should not be equalized with cancerogenic substances like Lomborg does. Lomborg writes that soy protein has an extremely high estrogen content. This is true. But this soy estrogen is probably beneficial, i.e. it protects against cancer. The high intake of soy in east Asian countries is believed to be one of the explanations why the incidence of breast cancer is much lower in Asian cultures than in the West. From Lomborg´s text, the reader may get the false impression that when the intake of natural estrogens is much larger than the intake of synthetic estrogens, the latter can impossibly be a source of cancer, because then the former should have caused much more cancer. In reality, however, Lomborg is equalizing compounds that prevent cancer with compounds that promote cancer. Thereby, his text is misleading.
P. 238 right, note 1858: (GROUNDLESS DEROGATION)
Lomborg exposes how Greenpeace exaggerated the problem of declining semen quality. Flaw. Lomborg does not mention in the main text that the scientists that discovered the phenomenon, publicly wrote against Greenpeace´s interpretation. It is not standard for scientists to write such protests, but Skakkebæk did that in this case, which testifies to his will to avoid misinformation. As this relevant point is not mentioned in the main text, most readers will not notice. However, on p. 239 right, we read in the main text: "Surprisingly, Skakkebæk has never publicly commented . . " on a flawed scientific paper which supports Lomborg´s opinion. So, when Skakkebæk does what Lomborg would want him to, this is concealed to most readers. When he does not, his omission is emphasized in the main text. With such a bias, Lomborg manages to give the impression that Skakkebæk is not trustworthy. This may appear to be a small detail to the English reader, but it is a part of the story that Skakkebæk actually yielded crucial help to Lomborg in July 1998. Without his help, Lomborg would simply not have been able to write the section on semen quality. But just afterwards, as a part of the marketing of Lomborg´s first book, he used the whole of a large newspaper article in 1998 (Politiken, 21./9.) to criticize Skakkebæk and cast unjustified aspersions on his motives. That is, Lomborg really has an intention to slate Skakkebæk, wherefore we know that the bias is deliberate.
P. 238 right, note 1860: ERROR OF REFERENCE
Error: The reference given for a rise in testicular cancer (Toppari et al. 1997) is not in the bibliography; if it should have been Toppari et al. 1996, which is in the reference list, then the page indicated does not fit.
P. 238 right, note 1860: FLAW
In the main text, Lomborg only very briefly refers to a possible link between the rising incidence of testicular cancer, and the trends in semen quality. Flaw: Lomborg has to some extent belittled the rise in testicular cancer; he speaks of a doubling since 1959, but the reference which he has used repeatedly (Toppari et al. 1996), shows an increase of 3-4 times from 1943 to 1988 (and it has continued since then). As Lomborg tries to prove that the semen quality has not declined at all, he has an interest in downplaying the rise in a related disease. Therefore, there is a suspicion that this slight downplaying might be deliberate.
P. 238 right, note 1860: FLAW IN REFERENCE
Lomborg´s bibliography includes a paper of Carlsen et al. (1995): "Declining semen quality and increasing incidence of testicular cancer; is there a common cause." Strangely, however, it seems that he does not refer to this paper anywhere in his notes. One would have expected a reference in note 1860, but that is not so. The omission does not hamper Lomborg´s argumentation, so this seems just to be a non-intended lapse.
P. 238 right, note 1860: FLAW
The text on top of p. 404 reads: "A link between testicular cancer and sperm quality is possible though not obvious. . ". Flaw: The same formulation is found in the Danish version from 1998, and at that time, it was to some extent a fair summary of existing knowledge. However, in 1999, it was pointed out to Lomborg that now a paper had appeared which demonstrated a significant relation between the two phenomena (Møller & Skakkebæk (1999): British medical Journal 318: 559-562). This paper also appeared in a Danish version in the Danish weekly journal for doctors. The authors conclude that: "The association between male subfertility and subsequent risk of testicular cancer is strong and consistent with the hypothesis of a common aetiology." So, from 1999 onwards, Lomborg´s formulation is no longer valid. But although he has up-dated note 1860 in other respects, on this point he deliberately keeps his out-dated text. The reference Olsen et al. 1996 is no strong support of Lomborg´s claim.
P. 238 right, note 1860: ERROR OF BIAS
"Moreover, the increase in incidence could have many other explanantions, e.g. smoking . . " Error: It appears that since 1998, Lomborg has tried to find papers that describe other possible links with testicular cancer, or downplay the established link with semen quality. Thus, he must have made a literature search of the subject. However, if one searches e.g. in medline for new literature 1998-2001 on testicular cancer, one will find a series of papers corroborating a link to semen quality. Besides the papers by Møller & Skakkebæk (1999) referred to above, these are:
Skakkebæk et al (1998): APMIS 106: 3-12. - Møller (1998): APMIS 106: 232-239. - Petersen et al. (1998): Seminars in oncology 25: 224-233. - Petersen et al. (1999): J. clinical oncology 17: 173-179. - Rørth et al. (2000): Scandinavian j. urology nephrology suppl. 205: 172. - Jacobsen et al. (2000): Brit. med. j. 321: 789-792. - Instead of citing these, he cites a paper by Sonneveld et al. 1999 which allegedly shows that the incidence rate of testicular cancer is inflated because of earlier detection; however, this paper does not show this. Furthermore, he cites 4 papers that describe alternative causes. The two last-mentioned of these are indeed relevant, but the two first are not. The hypothesis of a connection with iron is possible, but mostly just speculative, and the paper on smoking is nonsense; it is contradicted by a thesis showing that there is no connection between testicular cancer and smoking (Møller (2000). Epidemiological studies of testicular germ cell cancer. Doctor´s thesis). Altogether, we can conclude that Lomborg has indeed made a literature search in order to update his text on this issue, but all the papers he has found that support Skakkebæks theory, have been omitted - even when they have been pointed out directly to him - whereas a number of papers which point in alternative directions have been cited, even though 3 out of thes 5 are irrelevant or dubious. We can thus say with much certainty that Lomborg has been deliberately biased in his choice of literature.
P. 238- 239: FLAW
"In the article they listed seven different ways in which we may have become exposed to more estrogens . . " Flaw: Sharpe & Skakkebæk brought in their paper an overview of "routes of human exposure to oestorgens that have changed in the past half-century". Lomborg reproduces this overview in a slightly biased way. Those points which he is interested to stress, i.e. the most natural sources, are presented asslightly more certain. Thus, in item one, he writes "seems to increase" instead of "may increase". In item two "increases" instead of "can increase". Concerning diet, he omits to mention that soya is actually the richest source of plant estrogens, and that the consumption of soya has increase enormously. When it comes to man-made oestrogens, the change goes in the opposite direction; he writes "perhaps we are becoming more exposed to synthetic estrogens", which is different from the original´s ". . may find their way into drinking water. . ". He omits two estrogen sources, which are the use of orally active anabolic steroids in livestock, and certain pesticides. Thus, out of 3 artificial sources, he mentions only one "synthetic estrogens", and he then goes on to conclude that these constitute only a subset of all the possible explanations." In conclusion, Lomborg´s presentation is deliberately biased.
P. 239 top left: (COMMENT)
". . but it was this story the media chose to circulate." Comment: First. The Danish research team has been careful not to postulate that synthetic estrogens are the cause of the declining semen quality. They only present this as a hypothesis. And it is not true that the media have only circulated this story. They have also dealt with the hypothesis that declining semen quality could be due to growth regulators (chemicals applied to shorten straws). It should be added that recently, a paper has appeared that implicates a role for DDT and PCB´s in the declining semen quality (Hauser et al. (2002): Environmental health perspectives 110: 229-233).
P. 239 left: ERROR OF BIAS
"The 1992 article has led to numerous critical responses and new studies . . . " Error: In this paragraph, Lomborg gives a confusing impression of flickering trends. The proper thing to do, of course, would be to describe the results of the performed meta-analyses which give an overview. There exists a new meta-analysis, published by Swan et al. in 2000 (Environmental health perspectives 108: 961-966), which includes 101 investigations from all over the world. It corroborates the negative trend, i.e. it goes against Lomborg´s point. He has not mentioned it. This, of course, could be because he has simply not up-dated that section of his book since 1998. But not so. He does mention an overview of trends in USA published by Saidi et al. (1999); the authors of this paper include Fisch and Goluboff, on whom Lomborg very much supports his text, and whose data point in Lomborg´s direction. He may just by accident have overlooked the paper by Swan that points in the opposite direction; but he has talked personally with Swan in 1998, so it seems strange that he should not know that more papers from her hand were under way. Furthermore, science has advanced. Scientists have now begun to analyse semen samples according to the men´s birth year, instead of the year that they delivered their semen sample. It turns out that thereby the trends become more marked, and in some cases a significant decline with time is obtained even when a regression on the basis of the year of ananlysis did not show a decline (Zorn et al. (1999) Int. j. andrology 22: 178-183). Also Danish data have been re-analyzed according to the year of birth, and thereby certain trends become more clear, especially the trend for a continued regular decline in semen quality (Bonde et al. (1998) Scand. j. work environ. health 24: 407-413). This result agrees with that damage inflicted to the reprooductive system in pre-natal life can cause irreversible cahnges, while effects on adults will normally be reversible. - In addition, it would be natural for Lomborg to make a literature search to see if evidence published after 1998 could tell something new. If one makes a search on Medline with the key words "decline" and "semen quality", one will find 9 new papers published in the period relevant for at update of Lomborg´s book (1999-March 2001). "Increase" and "semen quality" does not yield any. Of the 9 papers, 7 show a continued decline in semen quality, whereas 2 show a rise. Lomborg cites one of these, namely one paper from 2000 that shows a rise (his note 1903). So, Lomborg has done something to update his text. How he has made his search, we do not know; we only know the result which is that he just happens not to have found any papers out of that majority which corroborates the decline. Altogether, there is an error of bias, but we cannot know if that bias is deliberate.
P. 239 right: FLAW
"New York is one of the cities in the world to have the highest sperm count, at around 130 million/ml. . . " Flaw: The paper by Fisch & Goluboff (1996) presents on its p. 1045 six previous studies of sperm counts in New York. Arranged in a time order, they show the following averages: 121- 134 - 101 - 107 - 110 - 79. This suggests that there might have been a falling trend in the New York area, like the overall trend in the rest of the world. However, their own recent figure is 132, which completely disrupts this falling trend. So, the claim that there is no falling trend here, depends on just one investigation. On this basis, one cannot postulate that the sperm count in New York is about 130 during the whole period; a weighted average of all seven studies gives an average of 92, which is not especially high.
P. 239 right, notes 1872 and 1873: (COMMENT)
". . we know from a study just of New York that there has not been a decline in sperm count from 1972 to 1994. . " Comment: This New York investigation (Fisch et al. 1996) is the key point in Lomborg´s discussion of the subject. If the 400 men participating in just this investigation had not had a high sperm count, his whole argumentation would fall to the ground. Lomborg was made aware of this paper by Skakkebæk in July 1998. Skakkebæk warned him that the paper might be suspect. He had a concrete suspicion that the main author was influnced and controlled by a specific large chemical firm. However, it is usually not possible for a scientist to raise the burden of proof in such cases, and therefore Skakkebæk is not able to say publicly that something may be wrong with the paper. So, in his opinion on this paper, Skakkebæk is vulnerable. Lomborg has immediately utilized this situation to taunt Skakkebæk; he stresses with extra emphasis how very wrong it is of Skakkebæk to disregard Fisch, knowing very well that Skakkebæk is unable to defend his view publicly. From a professional point of view, it seems wrong to base the key point of one´s discussion on a paper which is subject to such a suspicion.
P. 239 right: ERROR
". . a fall can no longer be statistically detected." Error: It is not true that a significant decline can only be detected as long as the early data from New York are included. The fact is that if the analysis is made for Europe alone, there is still a significant decline. This is clear from the paper of Swan et al. (1997), which Lomborg refers to in note 1872. Here, he demonstrates a feeling of offence that Swan et al. have made a division between Europe and USA, instead of a division between New York and the rest of the world. His accentuation ("completely incomprehensible") appears to be an example of the well-known trick that if one´s arguments are weak, one can raise the voice instead. There is absolutely no reason why Swan et al. should not be allowed to analyze Europe separately, and when they do this, they actually find a decline much steeper than in USA. Lomborg knows this, and when he expresses himself otherwise, his text is deliberately misleading.
P. 239 right, note 1874: (COMMENT)
See the comments to note 1858 and to notes 1873 and 1874.
P. 239 bottom right: ERROR
"Since 1970 no change in sperm count can be demonstrated." Error: The first paper (Carlsen et al 1992) that described a continued decline in semen quality, fitted a single downwards sloping regression line to the data points. In the vivid discussion that followed, several scientists tried to disprove this trend by alternative ways of analyzing the data. Thus, Olsen et al. (1995), cited by Lomborg in note 1875, thought that instead of using a line, one should rather talk of a step-wise change, i.e. all data before 1970 fluctuate around a high level, without any trend, whereas all data after 1970 fluctuate around a lower level, again without no trend. Others suggested other models. To resolve the issue, Swan et al. (1997) fitted all the proposed models to the same set of data, to see which model explained most of the variation. Their conclusion was that the linear regression gave a considerably better fit than the step model, provided that geographical variation is taken into consideration, which it should indeed in Lomborg´s opinion. This means that the criticism of the 1992 paper is not warranted. In addition, papers that have appeared after 1997 corroborate that the sperm quality in many places continues to decline also in recent years. Therefore, Lomborg´s postulate, accentuated with italics, is not warranted. What is more, he knows that, because he has thoroughly read Swans paper, and also talked with her. Thus, his text is deliberately misleading.
P. 240 top left: FLAW
"Here it seems as though there was a slight fall in sperm count, especially from 1942 to 1970." Flaw: This is a deliberate understatement. What Lomborg refers to here, is the step model advanced by Olsen et al., according to which all decline is conceived to have happened before 1970. In this model, the stepwise decline is certainly quite large (at least 33 %) and certainly significant. Therefore, it is not a proper choice of wrods to say "seems" and "slight fall".
P. 240 left: FLAW
Lomborg criticizes Skakkebæk because he ". . claims that `to our knowledge there are no data to indicate a change in masturbation or coital frequency since the 1930s." Flaw: It is crucial to Lomborg´s argumentation that the decline in semen quality that happened before 1970 can be explained by shorter abstinence time. So, the flaw is that he stresses all information that supports this view, and neglects or downplays all information that goes in the opposite direction. Actually, a Ph.D. report made at Skakkebæk´s institute (A.G. Andersen (2000) PhD thesis at the Danish state hospital, dept. of growth and reproduction & The fertility Clinic) has tried to look at this problem, and has found no overall trend for shorter periods between ejaculations now than before. Some of the data on frequence of ejaculation in recent time are actually rather low, e.g. data for UK in 1990-91. A source which was available to Lomborg (cited in Swan 1997) is MacLeod & Gold (1952): Fertility and sterility 3: 297-315. They write that "the admitted `intercourse rate´ throughout married life until at least the age of 40 is almost invariably twice to three times weekly, with twice weekly being by far the most frequent reply . . ". This does not sound as if things in 1952 were much different from today. Lomborg supports himself i.a. on figures from Hunt, 1974 (published on Playboy Press!) according to which the frequency of coitus for married 30-year olds rose from 1.9 times a week to 3.0 times a week from the 1940s to the 1970s. However, on closer inspection of the figures, these are the median values. The average values show a rise only from 2.5 to 2.8, and that is less impressive. One may discuss if the right thing is to use average values or median values; but it is beyond discussion that if one uses median values in one data set (frequency of coitus), one must do the same with the other data set (sperm counts). But if we use median values for sperm counts, then the decline becomes stronger than with the average values. So, as the published data on sperm counts are averages, the most correct for abstinence times would also be to use the averages, and if so, then Lomborg´s point is not so evident any more.
P. 240 right: FLAW
"A Swedish survey showed that the period of abstinence . . ". Flaw: Lomborg writes much about the frequency of ejaculation in daily life, but only in the sentence referred to here, does he touch on what is actually relevant: the period of abstinence before delivery of a semen sample. He omits to mention that on this point, there is no clear evidence of a general change. Thus, a source which was available to Lomborg (cited in Swan 1997), MacLeod & Gold (1952): Fertility and sterility 3: 297-315, says that 20 % of the persons did not comply with the prescribed period of abstinence for 3 days, and only 22 % had not ejaculated for 8 days or more. Thus, 50 years ago, the situation was hardly different from today, when abstinence for 3 days is still prescribed. As to the overall pattern, Lomborg has spoken with Swan and read her paper from 1997. Here, she discusses if abstinence time is a confounder, and she answers yes, but with a sign opposite of what Lomborg says. In the newest investigations, older persons are over-represented, and old persons have on average longer periods of abstinence than younger persons.
P. 240 right: ERROR
" . . a fall of 47 million sperm cells/ml over a period of 50 years is thus the equivalent of the reduction in the abstinence period of 3.6 years." Error: It is not true that fall in sperm count can be explained by a shorter period of abstinence. The reference that abstinence time in Sweden has fallen by 3.1 days, is Bendvold et al. (1991): Archives of andrology 26: 189-194. Indeed, in this paper we find the indicated drop in abstinence period from 1966 to 1976, i.e. across the breakthrough of the sexual revolution. But the men in this investigation lived in infertile couples who wanted to have children, wherefore their frequency of intercourse is hardly representative. In other aspects, the paper actually contradicts Lomborg´s claim. It demonstrates only a slight difference in sperm count between those that were abstinent for more than 5 days, and those that were abstinent for 5 days or less. Moreover, the sperm count for the group with short period of abstinence was much lower in 1986 than in 1956. It seems that in 1956, 3-5 days were enough to fill up the store of sperm cells, whereas in 1986, it took more than 5 days to fill up the stores, i.e. the rate of sperm cell production had probably declined. Thus, when we have a paper that separates the factors of abstinence time and year of sample, then we get a corroboration that the decline in semen quality has happened independently of abstinence time. Did Lomborg know this ? He has talked with Swan about this paper in 1998, and got the correct reference from her (note 1889). But he has not used the paper after all. Maybe he forgot to read this paper that is crucial for his argumentation - that is gross negligence - or he read it and left it out - which is deliberate bias.
P. 240 right: FLAW OF OMISSION
Flaw: Lomborg only writes about the trend in sperm count (no. of cells per ml). But many scientists have also reported other quality parameters - percent of normal sperm cells, and their mobility. Many studies find a signicant decline in these parameters, which are independent of preceding abstinence. For instance, the Swedish paper by Bendvold et al. referred to in the above comment finds a marked drop in these parameters from 1956 to 1986. For a fair treatment of the whole subject, Lomborg should have mentioned this aspect, but he did not.
P. 241 top left: FLAW
" . . traditional (non-organic) greenhouse gardeners also had better quality sperm . . " Flaw: From this and the preceding text the reader will get the impressions that organic and non-organic farmers/gardeners both had equally good sperm. Only if one reads note 1897 will one see that greenhouse gardeners did, however, have a sperm count 20 percent inferior to that of the ecologists. Furthermore, Lomborg forgets to mention that the 20 % difference refers only to the winter period. In the summer period, the difference was even larger. To hide this crucial information in the notes, and omit parts of it, is deliberately misleading.
P. 241 top left: ERROR OF OMISSION
It was pointed out to Lomborg in 1999 that the first investigation of organic farmers was followed up by additional investigations. An English reference to these is Abell et al. (2000): Scand. j. work environ health 26: 492-500. This study goes into detail with the greenhouse gardeners. It was carried out in a period when the now banned organochlorine pesticides were still in use, so presumably these caused the impaired semen quality. Those who had worked in the greenhouses for more than 10 years had much lower sperm counts than those who had worked there for less than 5 years. Those whose work functions implied the least exposition to pesticides had twice the sperm counts of those that were most exposed. these and other details strongly suggested a negative effect of pesticides on sperm counts. But although Lomborg knew about this study, and although it is very relevant to his theme, he chose not to mention it. This is deliberately misleading.
P. 241 top left: FLAW
"Finally, in 1999, a large study of 171 traditional and 85 organic farmers settled the question . . " Flaw: The review of this 1999 study is not correct. There were measured 9 different sperm quality parameters, not 15. Eight of the nine showed the highest values in organic farmers. In one of these, the difference was very significant in favour of the organic farmers. As this was a parameter is correlated with the ability to become a father, it is certainly relevant. Thus, although significance was this time not obtained with sperm counts, a relevant significant difference was indeed found, which means that the theory of better sperm quality in organic farmers was supported. Instead, Lomborg focuses on the spurious results of correlations with pesticide exposition. He forgets to mention that exposition was not actually measured; the figures are calculated presumed exposition. And in any case, the level of pesticide load was in any case so low that no negative effects could be expected (in Denmark, most non-organic farmers use fairly moderate amounts of pesticides). It is strange that Lomborg, claiming to be a statistician, uses such a diffuse word as "indistinguishable". It would have been more relevant to talk about significant and non-significant differences. Thus, one of the sperm parameters showed a difference with p< 6 %. although this is not significant, a few more samples in each group might have made it so, and to use the word "indistinguishable" about such a situation is not justified. It gives the false impression that the values in the two groups were practically identical. Finally, to say that this study "finally settled the issue", i.e. disproved the advantage of organic farmers, is certainly deliberately misleading.
P. 241 left: (COMMENT)
" . . we have sex about twice as often as we did before. And that is precisely what we see." Comment: Lomborg´s source tells that the average frequency of intercourse has gone up from 2.5 to 2.8 times per week. This is not a doubling. The Swedish study that Lomborg also refers to, deals with abstinence time before delivery of samples, not with frequency of sex in daily life. So, this is the same flaw as explained for p. 240 left.
P. 241 left, note 1901: ERROR
Note 1901 has: "Looking at Europe separately, we see a decline which, however, is not statistically significant." Error: This is in direct conflict with Swan et al. (1997), according to which the decline in Europe is significant. As Lomborg has read that paper thoroughly, this error is deliberate.
P. 241 bottom left: ERROR OF OMISSION
"In the US, rates of infertility have remained constant . . In the UK . . fertility has actually increased . . ". Error: It is possible to find studies that could not find any decline in fertility, but this is not the general trend. On the contrary, the management of infertility problems has become an increasingly important part of health services in the industrialized countries since about 1980. Declining semen quality is probably an important contributing factor for this. A recent investigation of young Danish men showed that 40 % of those that complied with the period of abstinence had so low counts of normal sperm, that it could have negative consequences for their fertility (Andersen et al (2000): Human reproduction 15: 366-372). Such information seems very relevant and should have been included in Lomborg´s text. However, there is no information that Lomborg was aware of it, which means that the error is probably not deliberate.
P. 241 right: WRONG STATEMENT
" . . today we know for certain that the scary vision of the general, overriding reduction in sperm quality was mistanken. . " Error: A scary vision has actually come true today when 40 % of all young men in Denmark have a semen quality so low that their ability to become fathers may be impaired. And this trend is evident all over the globe, albeit possibly with some exceptions. To use the words"we know for certain" is downright wrong.
P. 242 left and note 1913: (COMMENT)
"According to Poul Bjerregaard, professor of eco-toxicology . . " Comment: This is a secondary source, namely a journalist´s summary of what Bjerregaard said to her. Translated into English, the text referred to by Lomborg is as follows: "Today, according to professor in eco-toxicology Poul Bjerregaard, scientists agree that the estrogen-like substances have no synergy effects. If there is any sort of interaction, it is purely additive." When this statement was made (November 1997), it had already been disproven, as may be seen in the paragraph on p. 242 bottom left.
P. 242 bottom left: WRONG STATEMENT
"So, even though all the facts currently suggest that estrogens have no cocktail effect . . " Error: This is not true. Although the McLachlan article was withdrawn, other articles from before 2001 which have not been withdrawn, document synergistic effects in compounds with estrogen effect. These are P. M. Vonier et al. (1996): Environ. health perspect. 104: 1318-1322, and J. M. Bergeron et al. (1999): Environ health perspect. 107(2): 93-97. A study of the negative effects of a chlorinated dioxin and a PCB compound on heme biosynthesis in rats demonstrated very strong synergism between the two substances, raising the effect up to 800 times over that of the controls A.P.J.M van Birgelen et al. (1996): Environ health perspect. 104(5): 550-557. Although the latter study did not deal with an estrogenic effect, the compounds involved may also act as synthetic estrogens.
General remarks on synthetic estrogens and breast cancer:
Lomborg´s text on this subject (pp. 242-244) does not focus on the most relevant issues. On the basis of Davis et al. (see below) we can state some points which should be stressed. Lomborg writes that the connection between pesticides and breast cancer is based on the idea that they can mimic estrogens. Considering that the body is exposed to billions of times more estrogen equivalents from human estrogen than from pesticides, it is ridiculous to consider this mimicing effect as dangerous. However, when human estrogen is decomposed in the body, it may turn into one out of two compounds, one of which increases the risk of cancer, whereas the other stimulates cell repair and inhibits cancer. Pesticides disturb the balance between the two routes of decomposition.
Another point is that the dose
is less important than the timing of exposure. The most risky
periods are when the cells destined to form breast tissue are
growing and dividing at the greatest rate. This seems to be in
the fetus before birth, and in the young girl just before
puberty, when the breasts grow, even though the cancer is not
recordable until many decades later. So when a woman of 60
years´ age develops breast cancer, the most elevant question is:
what was her exposition to pesticides 45+ years ago ? - Some
chemicals, like DDE (transformed DDT) remain for long in the body
tissue, e.g. in the breast fat. DDE has a half-life of about 5
years, which means that in 45 years, only about 0.2 % fo the
original DDE remains. Therefore, to measure DDE in breast tissue
once the woman, aged 60, develops cancer, is not relevant. Most
of the DDE mesured will have entered her body long after the
cancer was initiated. But why then, are such studies made ?
Because they are feasible ! Studies of that type have been
described as analogous to "looking under the nearest
lamppost for lost keys because that is where there is light"
(vom Saal et al., New Engl. J. Med. 388 (1998): 988). - Other
pesticides, like lindane, are removed faster from the body, and
measurements already a few years after exposition may be
irrelevant. Especially suspicious are studies where patients were
investigated only at advanced stages of cancer, when the disease
as such may have affected the concentration of pesticide
A recent study avoided the above
mentioned flaws. It analysed frozen blood samples taken from
Californian women during the years 1959-1967, when direct exposition to
DDT was near its peak. When DDT and its decomposition product, DDE, was
measured in frozen samples from women who much later developed breast
cancer, and in a control group of the same age, it turned out that
there was a clear difference in blood DDT levels between those with
cancer and the control group. This was clearest for the group of women
who were less than 14 years old in 1945, when the general public was
exposed to DDT for the first time. This agrees with what was said
above: the impact occurs mainly before puberty. For such women, the
risk of getting breast cancer before the age of 50 was 5 times higher
if the blood DDT level was high relative to when it was low. There was
no relationship with DDE, only with DDT. It is also stated in the paper
that the effect of DDT is a poisoning effect, possible by action on the
cell DNA, not an estrogen mimicking effect. Reference: B. A. Cohn et
al. (2007): Environmental health perspectives 115 (10): 1406-1414.
Another relevant factor is the phasing out of the use of the chemicals. In Europe, DDT and PCB were phase dout during the 1970s, whereas dieldrin and lindane where phase dout only decades later. Thus, if one wants to demonstrate an effect of DDT or PCB, body samples should have been taken before 1980. Later samples, e.g. from the early 1990s, might be relevant in relation to lindane or dieldrin. - Most studies cited by Lomborg violate these preconditions. But not all. Most importantly, a Danish study (Høyer et al. 1998) used blood samples that had been stored for 17 years before the diagnosis was made, i.e. breast cancer was reported long time after the blood samples were drawn. There exist a few other prospective studies, but in these, the samples were taken only a few years before evaluation of cancer incidence. Therefore, the Danish study is more relevant to the subject than are most other studies. However, its importance is downplayed by Lomborg and by some British and American expert groups that seem to be characterized by anti-environmentalist bias (COC 1999, NRC 1999).
Reference: D. L. Davis et al. (1998): Rethinking breast cancer risk and the environment: The case for the precautionary principle. Environmental health perspectives 106: 523- 529.
P. 242 - 243 : (COMMENT)Lomborg uses much text to present allegedly exaggerated claims about the link between DDT and breast cancer Comment: The study by Cohn et al. (2007)referred to above - which was not published yet when Lomborg wrote his book - demonstrates that there is indeed such a link. Thus, the claims about this link are not wrong.
P. 243 left: (COMMENT)"Nevertheless, the real issue of course is whether synthetic estrogens can be causing breast cancer." Comment: No, that is not the `real issue´. When DDT causes breast cancer, this is not because it is a synthetic estrogen. See the paper by Cohn et al. (2007) referred to above.
P. 243 bottom left: (COMMENT)
"Third, the incidence of breast cancer has been increasing while concentrations of DDT, DDE and PCB in the environment have fallen." Comment: In the light of the comments above, there is nothing strange about this. The present incidence rate of breast cancer may reflect pollution levels as long as maybe 50 years ago.
P. 243 right: BIAS
"Since then, seven large studies . . . " Flaw: One of the studies cited by Lomborg is Güttes et al. (1998). Their paper has a survey of 11 studies. Out of these, 6 demonstrate a relationship between at least one "synthetic estrogen" and breast cancer. Although Lomborg has consulted this paper, he did not report this. Instead, he refers to the report by COC (1999), which has a survey of 8 studies (4 of which are also treated by Güttes et al.); three of these demonstrate a relationship. The report by COC seems to have some bias; it downplays the significance of the findings, and avoids the general considerations explained above. When combining the two surveys, 8 out of a total of 15 studies found at least one relationship. Concerning DDT, 6 out of the 15 studies found elevated levels in cancer patients (and one found a reverse relationship). However, all that Lomborg reports, is the study with a reverse relationship, and the conclusion by COC that there is no relationship.
P. 243 bottom right: FLAW
"For dieldrin, only two studies had tested the connection . . " Flaw: One of these two studies is the Danish study by Høyer et al. referred to above in the general remarks on breast cancer. This study is superior to all others because it combines a prospective investigation over a long period with a large number of cases. Thus, relatively great emphasis should be put on just this study. It finds a relationship with dieldrin that is highly significant. This makes some sense, because in Denmark the use of DDT was never large, and it was banned already in 1969, whereas dieldrin was used somewhat more extensively. The study included 28 forms ("congeners")of PCBs, and found no relationship with any of them. With these 28 congeners, and different degradation products of e.g. DDT, the total number of compounds tested was 46. With a significance level of 5 % = 1/20, one should find a relationship, by chance, in 46/20 = about 2 of the compounds. Therefore, it has been postulated, e.g. by COC (1999), that the relationship with dieldrin might be a chance finding. This is not sensible, however. First, the amounts of the 28 PCB congeners would be expected to be correlated, and thus we do not deal with 46 independent variables. Rather, we deal with less than 20. A statistician should have been able to understand this. Secondly, the relationship with dieldrin was significant at a level much stronger than 5 %. Therefore, the claim that this could be just a chance finding, is not justified. This issue has been discussed between Lomborg and the authors of the Danish study in the media before 2001, so Lomborg is aware of the arguments pro et con. In spite of this, all he writes here is that it is plausible that the single, statistical find was a chance finding. To stress that dieldrin is not an issue, he even writes that it has no estrogenic effect, although the question here is a different one - whether it has a cancerogenic effect. When Lomborg only cites the biased conclusion of the COC report, and omits arguments in favour of the opposite interpretation, his text is deliberately biased.
P. 244 top left: FLAW
"Of the three studies that have examined beta-HCH and lindane . . " Flaw: Actually, a Finnish investigation from 1990 showed a tenfold higher risk for breast cancer in persons who had elevated levels of lindane residue (cited in Wolff et al. 1993 in Lomborg´s reference list). Although Lomborg has seen this evidence, as he has read Wolff et al. (note 1915), he gives the reader the impression that no such relationship has been found. When confronted with this, Lomborg defends himself as follows: "Of course, I merely refer to the findings of the British advisory committee on carcinogenicity of chemicals", i.e. COC(1999). However, there is no statement in Lomborg´s text on lindane that he restricts himself to those studies dealt with by COC. Instead, the reader gets the impression that there exist only three studies. So, Lomborg´s text here is deliberately misleading.
P. 244 left: WRONG STATEMENT
"We now have the data, and they supply no evidence . . ". Error: This is an overstatement of the preceding paragraphs. Lomborg speaks of "the data", which must mean all available data, and he postulates that they supply no evidence, which the reader must understand as no single piece of evidence. Actually, however, as explained in the comments on DDT, 8 out of 15 studies did find at least one piece of evidence for a relationship to breast cancer. Lomborg knows that, and when he nevertheless uses the cited formulation, he is deliberately misleading.