One media-blasted notion that most women now seem to take at face value is the importance of not drinking a drop of alcohol during pregnancy. The most ironic part of the propoganda is that those passing it along know it is a double-edged sword: they try to keep an impossible balance between convincing you that one beer will cause irreparable harm to your fetus, and assuring you that if you had some drinks before you were pregnant, you shouldn't acutally worry.

This contradiction might make some women wonder where the truth lies. While it is clear that excessive alcohol consumption can result in Fetal Alcohol Syndrome (FAS), a suite of birth defects mainly affecting the brain (learning and memory), but Wikipedia's definition of the condition is misleading in that it states: "It is unknown whether amount, frequency or timing of alcohol consumption during pregnancy causes a difference in amount of damage done to the fetus." A perusal of the current literature on the topic of alcohol and pregnancy makes it quite clear that generally there is a positive relationship between the amount of alcohol consumed during pregnancy and the likelihood and severity of defects. In addition, it is also known that the most important brain development occurs in the first trimester, and thus that is almost certainly the most critical period to avoid consumption of alcohol.

What has been unknown since the FAS-prevention media blitz started is whether or not there is a threshold level of alcohol consumption below which there is no damage to the child. Frankly, there doesn't seem to have been a lot of interest in this among medical professionals, because it is simpler just to tell women not to drink at all than it is to consider nuances that might cause women to think it is okay to drink, and then drink too much, whatever too much is. The one certainty is that if a woman does not drink at all, there is no danger of impairments to the fetus caused by alcohol.

There have been occasional papers, however, focusing on the moderate drinkers, in order to possibly establish a safety threshold for alcohol consumption. One research project in particular is interesting because it is an ongoing longitudinal study that began when 580 mothers involved joined the study at 4 months pregnant. The most recent paper I was able to access on the study has data from when the focal children were 14 years old (Wilford, J.A., Richardson, G.A., Leech, S.L., Day, N.L., 2004. Verbal and visuospatial learning and memory function in children with moderate prenatal alcohol exposure. Alcoholism: Clinical and Experimental Research. 28:497-507.) This study is a genuine attempt to uncover potentially subtle effects on children whose mothers had around three drinks or less a week during pregnancy.

The positive aspects of the paper is that 1) it separates drinking by trimester, which is developmentally relevant and 2) it compares continuous variables in a regression analysis - e.g. drinking amounts are not placed into arbitrary categories, rather the actual volume drunk is used for the data.
The problems with the paper are pretty much the same ones that plague so many medical papers. First, by necessity, drinking levels are self-reported. Given that the first assessment was at 4 months - after the first trimester - the accuracy of the data is not convincing. Not only was each woman required to remember amounts drunk months previously (questionable even for those not experiencing the undulating body chemistry of pregnancy) but at the point these women were pregnant there was already some media blitzing discouraging drinking during pregnancy. The authors do not address the problem of potential underreporting bias.

Mainly, though, the problems are statistical. The authors used a large suite of learning and memory tests to score the children. They then regressed every variable they could think of against the scores, because of the need to correct for all the other potential factors influencing the scores, for example: race, mother's IQ, child's age, child's current substance use, mother's current substance use, etc. in addition to mother's alcohol and other substance use over each trimester. This makes for an extremely large number of statistical tests performed, although the number is unreported and thus must be estimated. The "significant" results alone number 31. For the six learning and memory tests that had "significant" results, all included mother's use of alcohol in the first trimester (but not the second or third in any case). This leads to a long discussion of the mechanism by which moderate drinking has caused "learning and memory deficits" in these children, etc.

But assuming the learning and memory tests are valid evaluations, we need to take another look at what the statistics actually showed. (Table 2 with p-value results is reproduced below.) First of all, one of the most basic statistical rules, ignored by doctors in every case I have seen, is that there is a probability of being wrong about the result. It is generally accepted that if the statistics show 95% probability of an effect, it is a valid result. The problem arises when as part of a single study, a large number of statistical tests are run, as in this case. The flip side of being 95% sure is that out of every 20 statistical tests conducted, we can expect, on average, for one of them to show significance when in fact there is actually no effect. To control for this problem, many scientists employ a Bonferroni correction, which simply means dividing the base significance level (0.05) by the number of tests conducted - in this case, apparently something greater than 200, although the true number is never stated by the authors. The appropriately conservative significance level for this paper would then be 0.00025.

The highest significance level for maternal consumption of alcohol in the paper was at <0.001 for "word-pair learning," and three of the other four significant tests were at 0.05. Perhaps, though, since first trimester alcohol consumption cropped up as a relatively high probability effect for five different learning and memory indexes (out of a total of 12, I think, but again it is not clear), it could be considered a real effect. But what other effects were real? The only effect that nearly every single time passes the Bonferroni level of significance was mother's IQ, which was significant at the 0.00001 level 5 times out of 6. So, even if there is a barely measurable effect on learning and memory on the children whose mothers drank, the IQ of their mother is 5000 times more important in determining that score. Also, for a given index, equally or more important than whether the mother drank was the child's current use of alcohol or tobacco. Race was more significant for four indexes. In two of the indexes, the child's gender was more significant.

The take-home message boils down to this. A lot of factors affect your children's mental abilities. Many of these are out of your control. The first trimester of pregnancy is certainly the most sensitive period for brain development. For a lot of women, morning sickness during that time means the last thing they are interested in is alcohol. Although these results are about as sketchy as you can get, it is certainly rational for women to abstain during the first timester, just in case. It took a lot of fishing for these authors to find a result they could publish, and they never discussed it in terms of relative importance to the other factors. Ironically, the mothers who seem to worry the most about alcohol during pregnancy are the high-powered, yuppie types with the big IQs to match - who it seems have the least to worry about. After all, their mothers went to cocktail parties while pregnant, and they still got into Ivy League schools.

So next time you see a visibly pregnant woman drinking a glass of wine in public (are there any of them left who have the guts to do it?), don't insist that it is your civic duty to give her a dressing down, because she is past the first trimester anyway. The stress you cause her is probably worse for her fetus than the wine.

Labels: health, humans, statistics

About a month ago, another new paper on obesity (Christakis, N.A. and Fowler, J.H., 2007. The spread of obesity in a large social network over 32 years. New England Journal of Medicine 357:370-379) made headlines. The paper is based on an enormous data set that started being compiled in 1948, consisting of health information of thousands of residents of the town of Framingham, Massachussets. The original purpose of the study was to learn about causes of heart disease. Everyone in the study has received complete physicals every couple of years throughout their lives, and the data collection has continued on to the second and third generation of patients, which will presumably provide some information about the genetics of heart disease in addition to external causes. Christakis and Fowler use information about relatives and friends of the study subjects, included as part of the original data set, to assert that people whose friends get fat are more likely to get fat themselves.

The authors make the case that "[t]he spread of obesity in social networks appears to be a factor in the obesity epidemic." Although they never use the specific word "disease" in reference to obesity as some sensationalizing media outlets do, the thrust of the paper is that there is yet another cause of our obesity out there that is not our fault.

The advantage of using such a data set in this paper is that there is a lot of data about a lot of people over a long time period, and it is certainly understandable that scientists might conceive of other uses than the original purpose. The disadvantage is that the data set was not really designed to draw conclusions about obesity - and one of the major problems with the study is that the authors are promoting a cultural influence - based on interpersonal relationships - on weight gain using an extremely homogeneous sample, which does not represent a real cross-section of society. But, approached with an understanding of its assumptions and limits, Framingham-type data can indeed be useful for secondary studies.

The biggest limit of this analysis is its dependence on overlapping relationships among people. Although "social-network analysis" is not a technique with which I am highly familiar, it appears to consist mainly of high levels of pseudoreplication, which is a major problem for statistical analyses. Pseudoreplication is the use of data that are not independent, violating an important assumption upon which proper statistical analysis depends. There were 5124 focal subjects, and over 12,000 people total in the study, with an average of 7.5 social ties per person. The math on this clearly indicates that some people were analyzed as friends of more than one person. Thus, these data are not exactly independent. If the data were on people scattered about the country, so that each person's social network was independent of everybody else's, pseudoreplication would be avoided. With the data used as is, the statistical assumption of independence has been violated (although one can only conclude this in a roundabout manner; the author's use of jargon and limited statistical explanation makes their methods difficult to discover).

The most sensational assertion of the paper, that physical distance from one's friend does not affect the probability of becoming obese - and thus obesity of friends cannot be explained simply by them all having bad habits together - is undermined by the actual data, which are not nearly so conclusive. The authors broke physical distance into 6 rather absurd categories: 0 miles, 0.26, 1.5 miles, 3.4 miles, 9.3 miles, and 471 miles. Effectively, only the last group has true physical distance. Their conclusion is based on a nonstatistical difference, which may just mean that variation in the data is too large to detect a difference. In fact, the variation in their data is huge, with 95% confidence intervals (the statistical standard) often ranging over 50 or more percentage points. There is only confidence that having fat friends makes you fatter if the confidence interval does not overlap with a probability of zero. Looking at category 6 (471 miles) compared to the other groups in the figure below from the paper, four of the confidence intervals overlap with zero, as opposed to not more than 1-2 in the other distance categories, and upper confidence levels of probabilities of becoming obese are much lower than with shorter distances. The difference between category 6 and the others may not have been significant, but it is quite a stretch to conclude from this that distance from the friend does not make a difference in the probability of following him or her into obesity. (The six bars in each category represent data from six different health examinations over a person's life.)

For the authors, this paper was a no-brainer in two ways, though. If you can find a way to publish another reason why it isn't really someone's fault they are obese, by implying that the condition spreads from person to person like a disease, you've struck gold. In addition, because the conclusions are not that surprising at all (although they were spun as such by the NEJM media machine) it's easier not to pay attention to the statistical problems in the paper. But surely the authors' explanation that people thing it's more acceptable to be obese if their friends are is laughable to anyone. Do you know anyone who chose to be fat? "It really must be OK to let myself go if my friend has" just does not seem like a thought many people would have. But people who are friends will do things together, and if one of them can no longer do something physical, the other will end up hanging out watching TV with them, in a lowest-common-denominator effect. Because the assertion about physical distance not mattering is essentially bogus, a much simpler one is mutual lifestyle choice. It just wouldn't be very nice to dump a close friend because she got fat and couldn't ride bikes anymore.

Labels: health, humans, sociality

At the risk of sounding heartless, the death of Alex the Talking Parrot last week leaves some of us, who have worked in the same department with Dr. Irene Pepperberg, a bit relieved. It was certainly bordering on irritating when those of us studying animal behavior using less glamorous species were putting in the long weekends to raise our sample sizes in order to make our work acceptable to the reviewers of actual science journals, while Dr. Pepperberg cranked out dozens of papers in such publications as the Journal of Comparative Psychology and Language Sciences, and was recruited to various appointments in science and psychology departments at the University of Arizona, MIT and Harvard using data from a single, often disagreeable, bird.

Certainly Dr. Pepperberg played the media like a fiddle, making her attractive as a "researcher" because of all the publicity she drew to that one bird. She laments Alex's passing in many respects, surely, but partly because after over 25 years of training, Alex still had not had the chance to show the world what a genius he was:

Alex could pull together a few simple concepts. Show him a group of objects and he could tell you, "What color is wood and four-corner?" or, "What shape is paper and purple?" Dr. Pepperberg was hoping to train Alex to spin his own recursions, informing her that the nut was "in the blue cup that’s on the tray" or "in the yellow box on the chair."

"I wish we had gotten further," Dr. Pepperberg wrote in an e-mail message. "We were just beginning to get him to designate things like 'in' and 'on.' "

Fortunately, though, he did last long enough to have a human's entire career built around him. Perhaps the two other parrots Pepperberg is training will achieve greater heights of language skill than Alex did. Still, one cannot help but wonder why Dr. Pepperberg never seemed to last more than a few years at a given institution. Perhaps it was her uninterest in interacting with her colleagues, which seemed to indicate that she had nothing of value to learn from such an interaction. Perhaps when the novelty of her research wore off at a given institution, the realization finally dawned that her career's work has added little of value to our body of scientific knowledge.

Sample size is a critical issue in science. Data from one individual (or even three individuals) are marginally useful at best because there is so much variation among individuals. What if an alien came to earth and collected data about the linguistic abilities of humans based on conversations with only Franklin Roosevelt? What if its data were based on conversations with a high school drop out with an IQ of 60? (or President Bush? Sorry, couldn't help that one.) Its conclusions would be quite different in the two cases.

Another problem with teaching animals English in order to draw conclusions about how their brain works is that we are testing them in a context that has no evolutionary relevance for their species. Ecologists and evolutionary biologists were interested in Alex's data, given more study about African grey parrots social structure in their natural habitat - because "language" is all about communication with others of one's same species. Unfortunately, although Dr. Pepperberg gave lip service to exploring such research directions in the future when that is what her colleagues wanted to hear, she never seemed much interested in actually pursuing that avenue (and a glance through her long reference list does not indicate any publications devoted to the parrots in the wild).

So people can argue forever about Alex's true abilities, parsing his every word and knowing glance, but what biologist can really muster any interest in the conclusion of that discussion? It will tell us nothing about Comparative Psychology, because all we have is an inadequate sample size about an animal living in a cage in a lab, interacting mostly with humans. But perhaps the Language Scientists love all the hullabaloo for inspiring passionate discussions about whether and how any species other than humans can use language. Perhaps training Alex had some purpose after all, but it was not science.

Labels: behavior, brain, statistics