Magic Statistics

Between the 1950s and the 1990s, over twenty medical research studies concluded that induced abortion, especially before the first full-term pregnancy, is a statistically significant risk factor in later development of breast cancer. That conclusion was, however, received with skepticism by medical experts. Beginning in the late 1990s, ten more investigations using a different analytical approach found no relationship between induced abortion and breast cancer. This accounts for the National Cancer Institute’s 2003 conclusion that the earlier studies reached erroneous conclusions and that there was no proven relationship between induced abortion and breast cancer.

In a new review of the ten recent studies, Dr Joel Brind argues that they all incorporate serious methodological errors. Therefore, he concludes that the more recent studies "do not invalidate the large body of previously published studies that established induced abortion as a risk factor for breast cancer". Dr Brind is president of the Breast Cancer Prevention Institute and a professor in the Department of Natural Sciences, Baruch College of City University of New York. His review is published in the Winter 2005 issue of the Journal of American Physicians and Surgeons; the article is posted here in pdf format.

The crucial difference between the previous studies and the more recent ones is methodological. Different approaches were utilized to gather the data to be analysed. The earlier studies took a retrospective approach: The data were collected via surveys that asked women about their medical histories, focusing on abortion and breast cancer. Have they been diagnosed with breast cancer and, if so, when? Did they have an abortion and, if so, when? (Obviously, this is an oversimplification. The surveys would have gathered detailed medical history information but, for present purposes, that is what it boils down to.) The alleged problem with such studies is what is known as recall bias or reporting bias. When individuals are asked about events years or decades in the past, they can forget, or they can deliberately choose not to reveal information because they are embarrassed or ashamed or simply don't think it's anybody else's business.

In order for this potential source of bias to invalidate the abortion studies, the critics must allege that women who had developed breast cancer were more likely to disclose induced abortions than were women who did not develop breast cancer. That would account for the higher incidence of breast cancer among women who reported having had an abortion. But is this indeed the case? The degree of reporting bias is an empirical question. In theory, it is possible to gather medical history data from an independent source and investigate the extent, if any, of reporting bias. The critics of the earlier abortion studies, according to Dr Brind, did that for only one of the 21 studies he notes. In that one instance, however, the researchers overstated the extent of reporting bias, backtracked when called on their error, and were left with statistically insignificant estimates. Yet, says Dr Brind, the authors "continued to cling to reporting bias as an explanation–as have many others, even in the absence of any significant positive evidence. On the contrary, evidence of the lack of such reporting bias has been produced repeatedly . . ."

So, the earlier studies used retrospective data; the more recent studies, on the other hand, used prospective data. This type of study eliminates the possibility of reporting bias because of the way the data are gathered. One type of prospective study uses data from medical records. No one is asked to answer any questions about medical history; rather, researchers go directly to administrative data files maintained by (usually governmental) medical authorities. This certainly does not mean that prospective studies are free of data problems; they just have different ones.

The medical data to be studied must be selected very carefully. Abortions are performed on younger women, while breast cancer is typically a disease of older women. The data analysed must cover a sufficient number of years to include women who are young enough to have had induced abortions and old enough to have developed breast cancer. Allowance must also be made for the fact that some women in the study may yet develop breast cancer in the future. Also, the data base should cover only the period since legalisation of abortion in the respective jurisdiction, because use of medical records from before that time period will bias the results toward no association between induced abortion and breast cancer.

Not only that, administrative data files are subject to errors and omissions. In medical records, classification errors and perhaps falsification can occur. One might legitimately suspect that it would not be uncommon incorrectly to classify induced abortions as spontaneous miscarriages. Any of these problems, if not recognised and handled properly, could cause large errors in estimates, causing the study to be rendered invalid.

The potential problems of using prospective data are just as methodologically demanding as those involved in retrospective data, if not more so. The largest of the prospective studies mentioned by Dr Brind is the 1997 study by Melbye et al. that looked at medical records of all 1.5 million women born in Denmark between 1935 and 1978. These women had close to 400,000 abortions and over 10,000 cases of breast cancer. The study concluded that "[T]he risk of breast cancer in women with a history of induced abortion was not different from that in women without such a history . . ." Brind points out that Melbye’s study was based on the assumption that abortion was legalized in Denmark in 1973. This is incorrect, however: abortion was actually legalized in 1939. So, 60,000 of the oldest women in Melbye's study were misclassified as not having had an abortion, even though they had a legal abortion on record. Moreover, Melbye included breast cancers (the outcome) beginning in 1968, but abortions (the risk factor being investigated) only from 1973. This would also bias the results toward finding no association.

The other prospective studies all had fatal methodological flaws. Two studies had samples too small to yield statistically significant estimates. At least two studies employed samples that did not represent all women who have abortions: in one, the sample included only women had had a live birth; another eliminated women who had an abortion before their first full-term pregnancy. Some studies employed databases with extensive misclassification of induced abortions as spontaneous miscarriages. Other studies covered time periods too short to allow for later development of breast cancer.

In the nine years since we reviewed the abortion-breast cancer literature, the ten published studies based on prospective data were widely touted as resolving the controversy in favor of no abortion-breast cancer association. Yet, as is clear from the present review, none of them has provided credible evidence to back up the oft-repeated claim of no association.

These studies also misrepresent, or completely fail to mention, the earlier retrospective studies that found such an association.

Incidence of breast cancer among American women has been increasing. Dr Brind points out that

virtually all the increase in breast cancer incidence between 1986 and 1998 occurred in women under age 65 in 1998, i.e., in women under age 40 in 1973, the year induced abortion was legalized nationwide by the Roe v. Wade decision. It is not unreasonable, therefore, to attribute a substantial portion of the increase in breast cancer incidence since 1986 to induced abortion.

The best evidence available after decades of research concludes that "induced abortion is indeed a risk factor for breast cancer, despite the strong and pervasive bias in the recent literature in the direction of viewing abortion as safe for women." Genuine support for women'’s rights would entail advising women of all the known risks of abortion. Informed consent requires that a woman contemplating abortion be told that, if she chooses to have an abortion, her long-term risk of breast cancer is higher than if she does not.

via Life Site News.

James Bowman looks at the war between proponents of Darwinian evolution and intelligent design (ID), and he's not sure who to root for. On ID he's a skeptic, but he finds that he has a lot of sympathy for ID advocates because their enemies are simultaneously smug and morally obtuse. Why, for example, do anti-religious scientists assume that, once religion has been cast aside, everyone will just naturally want to be nice to each other?

When The New York Times did a surprisingly—though obviously not whole-heartedly—sympathetic series of articles on Intelligent Design in August, you couldn't help noticing this in the comments of some of the scientists quoted. Professor Steven Weinberg, a Nobel Prize-winning physicist from the University of Texas, said that I think one of the great historical contributions of science is to weaken the hold of religion. That's a good thing. Another Nobel laureate and co-discoverer of DNA, James Watson, says that "one of the greatest gifts science has brought to the world is continuing elimination of the supernatural." Why are such scientists enthusiastic cheerleaders for the pointlessness of the natural machine that so fascinates them, and for the non-existence of anything outside it which might give it meaning? Well, it's hard to believe, but I think it's because, brainiacs though they obviously are in every other way, theologically they are on the level of the late John Lennon. "Imagine there's no heaven," wrote the ex-Beatle,

It's easy if you try,
No hell below us,
Above us only sky,
Imagine all the people
living for today. . .

Imagine there's no countries,
It isn't hard to do,
Nothing to kill or die for,
No religion too,
Imagine all the people
living life in peace. . .

It's really the most basic logical mistake, um, imaginable. Because people have quite often cited religion as a reason for killing other people–and the other people for killing them–if you take religion away from them they won't kill each other anymore. Put so baldly, the proposition could only be believed by a child, but scientists very often are child-like–as, of course, Lennon was.

Sounds rather elementary to me. Yet "highly-educated scientists are suckers for atheistic promises of utopian pie in the godless sky".

This post presents another instance of the child-like faith that Christian morality will persist after Christianity itself has been "debunked" by science.

So says Peter Singer, Princeton bioethicist:

During the next 35 years, the traditional view of the sanctity of human life will collapse under pressure from scientific, technological, and demographic developments. By 2040, it may be that only a rump of hard-core, know-nothing religious fundamentalists will defend the view that every human life, from conception to death, is sacrosanct.

"Nevertheless, when the Son of Man comes, will he find faith on earth?"

via The Pearcey Report.

Climate change scientists now insist that we believe "six impossible things before breakfast", says Philip Stott, Emeritus Professor of Biogeography at the University of London. Earlier this week a group of scientists predicted that global warming is weakening the Gulf Stream, and forecast that, as a result, Britain's average temperature will drop 1C, resulting in a climate like Canada's. (Horrors!)

But another group of scientists only last October warned that global warming would drive Britain's average temperature up 2-3C.

Then, last June, yet another bunch of British researchers predicted temperature rises of 3-6C. Clearly, British scientists need to spend time talking to each other. Prof Stott concludes:

In reality, nobody has a fog what will happen. This is Virtualia, not the UK. During the last year, global warming has been predicted to lead to wetter winters, drier winters, another ice age, blazing-hot Mediterranean summers killing thousands, greater biodiversity and less biodiversity.

Hence the impossible things to believe from the Today programme before breakfast. But I’m with Alice: "There’s no use trying," she said. "One can’t believe impossible things."

Perhaps, as the Red Queen suggests, we need more practice believing impossible things. Climate change scientists are leading the way.