STATS ARTICLES 2007
What are the Risks of Light to Moderate Drinking During Pregnancy?
Rebecca Goldin Ph.D, December 12, 2007
Conflicting advice from health experts - so what do the studies really say?
Also see STATS examination of heavy alcohol consumption and fetal alcohol syndrome
Alcohol has long been recognized as dangerous for pregnant women. It is a known teratogen (something that can disturb the growth and development of an embryo or fetus). It causes birth defects and developmental problems for tens of thousands of American babies per year – and possibly millions worldwide. In 1989 the United States required alcoholic beverages to be labeled with the warning that alcohol can cause birth defects.
And yet, plenty of doctors – even OB-GYNs – are comfortable with their pregnant patients consuming modest amounts of alcohol regularly over the course of their pregnancies. In the UK,
“The Department of Health advises that pregnant women or women trying to conceive should avoid drinking alcohol. If they do choose to drink, to minimize the risk to the baby, they should not drink more than one to two units of alcohol once or twice a week and should not get drunk [a small glass of wine is two units of alcohol].”
In contrast, the U.S. Surgeon general warns that:
“Pregnant women and women who may become pregnant [should] abstain from alcohol consumption in order to eliminate the chance of giving birth to a baby with any of the harmful effects of the Fetal Alcohol Spectrum Disorders (FASD).”
And this recommendation replaces the recommendation from 1981 that women should “limit” their consumption. What gives?
Over fifteen years ago, Dr. Ernest Abel and Dr. Robert Sokol, then Dean of the School of Medicine at Wayne State published an article suggesting there was no strong evidence that light to moderate drinking had an impact on unborn babies.
What they found was that few studies were consistent about what “light” or “moderate” drinking meant. If a study is meant to examine the effects of consuming two drinks a day, but it includes some women who actually consumed five drinks per day, the statistics will be skewed. On the other hand, if the researchers include women who only drink one drink a day in with “moderate drinkers” the data may be skewed the other way – an effect could be present, but it will be hard to observe. In order for the scientists to really evaluate what is going on, they need to isolate a very narrow group of drinkers, those who are consistently drinking in the narrow band that is considered “moderate” drinking.
At the time, their literature review revealed a lot of poorly done studies with weak results; usually “moderate” drinkers were lumped in with “heavy” drinkers, making it difficult to assess whether emerging problems were those due to the heavier or more moderate drinking.
They also noted the very serious problem of how these studies are conducted; women report on their own alcohol consumption. This makes all the studies vulnerable to the possibility that pregnant women underreport their alcohol consumption. Given how widely publicized the problem of drinking while pregnant, are these women going to report everything they drink? Are those who drink heavily going to report that they drink moderately? Even without it being intentional, mothers who are in denial about how their behavior could affect their unborn children may well be convincing themselves they aren’t drinking all that much.
Public health advocates, doctors, and scientists have still not come to terms with the issue. Because the consequences of excessive drinking is so well-documented and sad, there is a strong undercurrent of “better not to risk it”. And it may well be that an abstinence-only public health plan has the greatest likelihood of getting to those who are greatest risk for having babies affected by alcohol, those who abuse it. But there are also those of us who enjoy a glass of wine here or there, pregnant or not. And we want to know: has research found evidence that pregnant women should not drink, ever?
In our literature search, we found some studies that suggest that drinking moderately (one to two drinks a day) can have a small effect on children. But most of the studies we found showed no effect. We found no studies at all that suggested a negative impact due to drinking three or fewer glasses of wine or beer per week.
The Impact on IQ Many Years Later
Heavy drinking is known to cause mental retardation, behavioral and learning problems. But what about light drinking? Is there a modest effect that can be seen with an IQ test?
As with many statistical studies, there are occasionally strange phenomena that emerge when many statistical tests are performed, and they can be hard to interpret. A study by Willford, Leech, and Day (Alcoholism: Clinical and Experimental Research, 2006) found some effects of moderate drinking on African Americans and not on whites. This study evaluated 10-year-old children’s IQ and compared those whose mothers drank moderately to those whose mothers abstained, and similarly for those whose mothers drank lightly to those whose mothers abstained.
The definition of a “moderate drinker” was more than 0.4 drinks per day, but less than 0.89 drinks a day – in other words, from three to six drinks per week. For this group, there seemed to be no impact on the Caucasian kids, but a small shift in IQ scores among the African American kids whose mothers were moderate drinkers in the first or second trimesters.
But there were some very weird conclusions found in this study too. For white children, for example, maternal consumption of cocaine at the time the kids were 10 was associated with a higher performance on overall scores, and its impact on abstract/visual reasoning for white kids was larger than the impact of second trimester moderate drinking for black kids!
The fact that the authors didn’t pursue the hypothesis that coke makes whites smarter suggests that perhaps the standards for confounding factors and statistical significance might have been too low. Though it should be noted that the effects from moderate drinking for black children were observed across all the cognitive categories, while the positive influence of current mother’s cocaine use only showed up for abstract/visual reasoning for white kids. This at least suggests that the coke benefits were a statistical anomaly, but the alcohol detriment may not have been.
In contrast, for light drinking, no effect was observed. The definition of a light drinker was more than none, but less than three drinks per week. There was no effect of light drinking found on any group.
One has to be very wary, however, of making strong conclusions from this study. There is really no satisfactory explanation of why moderate drinking during pregnancy would impact blacks more than whites. Among social or cultural influences that were not considered is the possibility that blacks in this study were more likely to underreport their drinking than whites, perhaps because the interviewers were white. This could mean that some of the moderate drinkers were in fact heavy drinkers. The other possibility is that there are underlying genetic correlates with race that make some people more vulnerable to the negative effects of alcohol. And of course there may be some other factor that explains these differences.
Unfortunately the study had other statistical limitations: While it only reported statistically significant results, the researchers should have shown their estimates for whites even if they didn't reach statistical significance as they may have suggested a trend. And because they were doing a “racially stratified” analysis, they should have reported the number of black vs. the number of whites in the study; statistical significance depends on these numbers, and the significance for blacks and not whites could be an indicator that there simply weren't enough whites to make the same conclusions as they did for blacks.
If the trend had been in the same direction, this would suggested that drinking moderately for whites is also a problem, even if it didn't reach statistical significance. On the other hand, the authors didn't adjust for the fact that they conducted so many statistical tests; while these tests are all correlated, but the fact that so many tests were conducted leaves open the possibility that some of their results were statistical anomalies, such as the results they found for cocaine.
Doesn’t anyone look at the literature as a whole?
The literature on this topic is so vast, that one has to resort to literature reviews, meta-analyses and the like. The advantage of doing this is that these studies tend to exclude studies of lower quality, and they try to find overall trends in many studies. For this reason, they are also less likely to observe an effect that is a statistical anomaly. They are also more resilient against accidental bias that comes from an unknown source – at least, unless the source is correlated with the drinking itself.
We found two major recent analyses that measured the impact of prenatal alcohol exposure. The first, Testa, Quigley and Eiden (Alcohol and Alcoholism, 2003), examined whether alcohol consumption has an impact on IQ, and if so, at what dose. They did a literature review of all materials from 1973 to December 2000, and included only studies that measured mother’s alcohol intake and the babies’ Mental Development Index (MDI), a test that measures babies’ progress at six months, 12-13 months, and at 18 months (different tests are used for different ages). For Testa et al., light drinking is considered more than none, and less than one drink per day. Moderate drinking is one to 1.99 drinks per day, and heavy drinking is two or more drinks per day. It’s worth noting that what is considered moderate in this study is considered light in the Willford et al. study.
These authors looked at kids at all three of these age ranges: for light drinking, there seemed to be no effect at six months on the MDI, no effect at 12-13 months after adjusting for covariates, and, strangely, a statistically significant positive effect for children 18-26 months. Let’s just assume that’s a statistical fluke (like cocaine use). Fact is, no negative effect found over a large sample and many kids.
They also did not find an effect for moderate drinkers – keep in mind, these moderate drinkers were classified as light drinkers in the Willford study. Heavy drinkers’ babies did have developmental issues which resulted in a reduced MDI – however, it disappeared by the time those kids were 18-26 months.
A separate systematic literature review was done by Henderson, Gray and Brocklehurst (BJOG, 2007) in which the authors looked for any impact of light to moderate drinking on miscarriage, stillbirth, intrauterine growth restriction, prematurity, birthweight, small for gestational age at birth and birth defects including fetal alcohol syndrome. As the authors say,
“At low–moderate levels of consumption, there were no consistently significant effects of alcohol on any of the outcomes considered. Many of the reported studies had methodological weaknesses.”
So, are pregnant women off the hook?
Before we have a hay-day, keep in mind that bad studies tell us nothing – the news is neither good nor bad, and cannot either be interpreted to ease our conscience about drinking. Herderson et al. conclude their article with caution:
“This systematic review found no convincing evidence of adverse effects of prenatal alcohol exposure at low–moderate levels of exposure. However, weaknesses in the evidence preclude the conclusion that drinking at these levels during pregnancy is safe.”
As far as we know, drinking lightly during pregnancy does not have adverse consequences. The systematic studies and meta-analyses we looked at suggest that drinking fewer than three small drinks per week is a safe level of consumption for pregnant women. This result seems robust in that the literature reviews considered hundreds of articles.
However, there are always other studies to try and, potentially, a well-designed study would in fact show that even light drinking can cause harm. So while no one can fault a woman who decides to have a glass of wine, it’s perhaps understandable that some women still don’t want to take the chance. One thing that’s clear is that women who have had a few drinks shouldn’t worry. With recent (probably anecdotal) reports that anxiety over drinking is fueling abortion, we recommend that women relax – have a drink or not – and stop worrying about light alcohol consumption.