Junkfood Science: It’s not nice to scare mothers: the latest miscarriage scare

January 22, 2008

It’s not nice to scare mothers: the latest miscarriage scare

Pregnancy should be a time of joy, as a new life is about to be brought into the world. But, sadly, it can also be an anxious time for expectant parents, who worry for the health of their unborn babies and fear that something will go wrong. Sound information can help to ensure each pregnancy has a happy ending. This month brought reassuring information, as well as the re-emergence of a 30-year old scare targeting pregnant women.

A study published in the current January issue of Epidemiology found that among 2,407 pregnant women, their coffee and caffeine consumption, at all the times evaluated — prior to becoming pregnant, the first four weeks after their last menstrual period, or during the first trimester of pregnancy — was unrelated to any risk of miscarriage. “There is little indication of possible harmful effects of caffeine on miscarriage risk within the range of coffee and caffeine consumed,” the researchers reported.

Led by David A. Savitz, Ph.D., a professor at Mount Sinai School of Medicine in New York, the study found that the amount of caffeine consumed by most pregnant women need not be a worry. All of the adjusted odds ratios (correlations) between coffee and miscarriage were insignificant, ranging 0.7 to 1.3. The women from Texas, North Carolina and Tennessee had represented a variety of races, education and income levels, ages and lifestyles. The median daily coffee intake among the women prior to and early in their pregnancy was 350 mg caffeine* and during the first trimester when they were interviewed was 200 mg per day.

A strength of this study is that two-thirds of the women had been interviewed prior to miscarrying, minimizing recall bias which can create false findings. The null findings held, however, when the researchers analyzed just the women who had miscarried after they had been interviewed.

What is recall bias? Recall bias is a major threat to the validity and credibility of studies that use self-reported data, explained Eman Hassan, MBBCh, MHSc, at the Department of Health Care and Epidemiology at the University of British Columbia. In an examination of recall bias in research, published in the Journal of Epidemiology, she explained that recall bias comes from the natural tendency of people to report and remember past events differently than they report at the time they are happening. Recall bias is most pronounced if people are asked to report things after a traumatic or significant event (such as cancer or a baby is born with birth defects). After a traumatic health problem occurs, people instinctively look back to think of all the possible past events and exposures that could have caused it to happen, and will tend to overreport those things that they believe might have been to blame, and disregard others, leading to inaccurate risk estimates. If an exposure or behavior is seen as socially undesirable, it heightens recall bias. Instead, if data is gathered objectively on a group of people at the beginning and then they are observed (prospectively) to see what happens, the findings will be less likely to suffer recall bias. That’s one reason the timing of data collection matters and retrospective studies are more problematic than prospective.

Caffeine use in pregnancy has been debated for decades. The March of Dimes has long reassured women that most studies have found moderate levels of caffeine (under 300 mg/day) to have no effect on fertility or risks of miscarriage. Extremely high levels of caffeine, of 500 mg a day or more, have been linked to miscarriage in some studies, while others have reported no increased risk, even with high caffeine intake. The lack of consensus has understandably left some women needlessly concerned, rather than reassured. The confusion, however, stems in part from weak and flawed studies jumbled in with stronger ones. “A recent study that examined 15 studies on caffeine and miscarriage found problems with methods used in many of these studies,” said the March of Dimes. “Most of the studies showed a link between caffeine consumption during pregnancy and miscarriage, at least at higher doses. However, the authors concluded that the evidence was not strong enough to prove that caffeine causes miscarriage.”

It is important to remember that all of the evidence linking caffeine to miscarriage is epidemiological — those observational studies that cull through databases on groups of people and use computer models to look for correlations. None are intervention trials to identify if caffeine could be a cause because it would be considered unethical to randomly assign one group of pregnant women to high caffeine consumption and another to none to see what the outcomes might be. As we know, however, epidemiological studies are most fraught with problems in misinterpreted statistics, errors and biases, and are most easily manipulated. That means we have to be especially on our toes to recognize when findings are reliable and worthy of our attention. Countless fears would disintegrate if everyone understood that correlations with relative risks less than 5 to even 10 — that’s 10-fold higher risks — are often not real or tenable, and are often found to be explained by confounding factors.

Even the design of epidemiological studies on caffeine and miscarriage, though, make some stronger and more credible than others.

Understanding the strengths and weaknesses of the studies

The paper that the March of Dime referenced is a large review of the evidence on caffeine consumption and spontaneous abortion published in the journal Epidemiology. Researchers, led by Lisa B. Signorello, ScD, at the International Epidemiology Institute in Rockville, Maryland, reviewed studies on people that had been published between 1966 and 2003. They identified 15 epidemiological studies, which they then closely examined for “methodologic problems that would generate biased findings.” They found that:

All of the studies reviewed suffer from important methodologic limitations that hinder both the interpretation of each study individually and the comparison of results across studies. Despite the fact that most epidemiologic studies have observed a positive association between maternal caffeine intake and the risk of spontaneous abortion, we conclude that the evidence must be considered to be equivocal, given the biases likely present and the fact that most of the potential biases would tend to overestimate any association.

This paper offered valuable points that can help women trying to make sense of research on this issue, so let’s take a closer look at the points they emphasized. The methodological problems they found in most of the studies done to date, “would create or inflate an association between caffeine and spontaneous abortion,” they said.

Briefly, their detailed report found that the cross-sectional studies, done on more than 44,000 women who had been asked about past pregnancies, reported untenable relative risks even among women with high coffee intakes during the first trimester. But, the authors said, these studies are of limited value in evaluating causation because of the problem with recall bias.

The case-controlled studies that had been conducted in six countries all suffered from selection bias. Some used selection criteria for the control groups that skewed them to women with lower caffeine intakes, defined control groups as only women who gave birth to healthy babies, were admitted for normal deliveries, or had been getting prenatal care where they’d been counseled to avoid caffeine. The selection biases “could spuriously overestimate an association between caffeine intake and spontaneous abortion,” they found. These studies also suffered from recall bias, which affected the reporting of caffeine intakes made after miscarriages.

The researchers reviewed five large cohort studies that had been conducted in the United States. The largest, conducted on 5,000 California women, had found “no convincing evidence that caffeine intake during pregnancy increased the risk of spontaneous abortion.” Several smaller ones reported untenable and modest increased relative risks associated with women consuming higher amounts of caffeine, but the rates of spontaneous abortions varied extremely and reflected the different times during gestation when the women were enrolled in the studies, making conclusions unreliable.

Timing is everything. Many miscarriages occur before women realize they are pregnant. The authors explain that women change their caffeine intake considerably during pregnancy, consciously and unconsciously. Pregnant women generally decrease intake most remarkably during the first trimester and intakes remain lower for most women or fluctuate throughout pregnancy “in response to pregnancy symptoms, such as nausea, indigestion, food aversions, or increased olfactory sensitivity.”

Pregnancy symptoms, like nausea and aversions to tastes or smells, is the most significant confounding factor in these studies, they emphasized. “Nausea is more frequent and/or severe early in pregnancies that are eventually carried to term,” compared to those that miscarry. In other words, “women with nonviable pregnancies may continue to drink coffee unabated because pregnancy symptoms do not discourage it," they said. "As a result, their higher caffeine intake could be interpreted as causally related to the spontaneous abortion, when caffeine in fact was simply a marker for nonviability.” Yet most studies don’t control for pregnancy symptoms, even when they say they do. [If you’re a lucky woman not suffering from morning sickness, please don’t worry, though, that this means you’ll lose your baby.]

Women will also tend to increase their caffeine intake after the fetus dies and when pregnancy symptoms abate, the researchers said. But spontaneous abortion (and clinical signs of the miscarriage) can occur weeks after fetal demise, which can also skew correlations of higher caffeine consumption seen during that interim as being causation.

Bottom line, researchers cannot simply take a single measurement at one point in time, as most studies have done, they said.

The soundest methodology, according to these epidemiology experts, was in a study by researchers at the National Institutes of Health that assessed caffeine intake 5 weeks after the start of the last menstrual period, and then again at weeks 6, 8, 10, and 12, to arrive at a more accurate estimate of first-trimester intake. This study on 431 women had also enrolled the women very early after conception before miscarrying, with 76% enrolled before conception, eliminating recall bias. This study, like the similar study reported at the opening of this article, found no tenable risk of miscarriage associate with caffeine intake during the first trimester and “a test for trend indicated no significant increase in risk with increasing caffeine intake.”

Kaiser Permanente study

The importance of nausea as a confounding factor was also emphasized in a large study of 5,144 pregnant women that had been conducted by scientists at the State Department of Health, Kaiser Permanente Division of Research and University of California, San Francisco, in 1997. They found “no significant increased risk for spontaneous abortion, or miscarriage, associated with caffeine consumption.” Even among women considered heavy caffeine consumers (300 mg/day) relative risks associated with miscarriage was only an untenable 1.3.

But decaf coffee was associated with higher risks, too! (Relative risks were 2.4). The correlation, explained UCSF pharmacologist and caffeine metabolism researcher, is more likely an “epidemiological phenomenon.” The researchers said: “We suspect that the apparent risk associated with heavy decaffeinated coffee intake resulted from women with nonviable pregnancies experiencing fewer symptoms of pregnancy and consequently consuming more decaffeinated coffee.” When a women stops feeling morning sickness (a potential sign that the pregnancy is in trouble), she may be more likely to drink more coffee, regardless of its caffeine intake. But the researchers could find no link with any chemicals in decaffeinated coffee to miscarriage risks, either.

What happened yesterday?

Yesterday, precisely on the same day, we witnessed more than 500 media outlets all reporting the exact same sensational story: “Any Caffeine Could Increase Miscarriage Risk,” reported CBN News. “Caffeine causing miscarriages,” headlined KSNT in San Francisco. “Best evidence yet of caffeine-miscarriage lin,” reported the SF Chronicle. Plus 500 more stories virtually just like these... all taken from the same press release.

We witnessed, once again, marketing, not science. The public understandably believed that this new study must, indeed, be important since almost no medical professionals were heard contradicting the news reports.

But, there is a reason why:

Because the study had not yet been published in the medical journal or been made available to medical professionals. Anyone who might critique the study hadn’t seen it. It was released to the media before doctors and medical professionals with paid subscriptions had even had a chance to read it!

This is marketing. It is using the media to plant the spin and a fear. It is well known that after people have been frightened and a myth has saturated the media and made its way around the world that no amount of discussions after the fact will assuage fears, sound convincing, or probably even be reported! It will be old news.

I’m not even going to get into the funding source of the study, the California Public Health Foundation, and the lead researcher’s long-time interest in looking for environmental causes, such as electromagnetic field exposures, for miscarriages and other health risks. Whatever the motives for overstating the findings of this study are irrelevant.

The latest Kaiser study in the news

Let’s go right to the study and see what it actually found. You now have enough background to see for yourself how it epitomized the weaknesses and flaws that had been described by Signorello and colleagues at the International Epidemiology Institute.

The Kaiser Permanente researchers compiled all female healthplan members in the San Francisco area who had had positive pregnancy tests done at the Kaiser laboratory between October 1996 and 1998. [Your lab results are not confidential.] The researchers wrote to the 2,729 women and got fewer than half of them (1,063) to come in for an interview. The women were interviewed at an median gestational age of 71 days and were asked about their beverage consumptions (the researchers estimated servings sizes and caffeine intakes), exposures to magnetic fields, smoking, alcohol, Jacuzzi use, and pregnancy symptoms. Most of the women had already had their miscarriages by the time they were interviewed. The Kaiser researchers used patient medical records to determine the pregnancy outcomes at 20 weeks and identify the miscarriages. There were 264 miscarriages among the women who participated in the study, a higher percentage of miscarriages than in the general population.

You can already see a number of the problems: selection bias; recall bias; ambiguous caffeine intakes estimated from an interview; reporting only on 39% of the pregnancies, more likely those with bad outcomes; and caffeine consumptions estimated from a single point in time. These techniques make it easy to create a scare, but don’t make a sound way to answer a scientific question.

The lead author and the director of Kaiser’s women’s health were quoted as saying that their study offered proof that women should limit caffeine and that even low levels of caffeine were dangerous and doubled** the risk for miscarriage. This is disingenuous, however, because they lumped all consumptions of 200 mg/day and higher together and created an artificially low threshold — making it impossible to know where any risks may actually have increased.

But the women consuming 200 mg or more caffeine were also notably older (37.2% were 35+), more likely to have had previous miscarriages (30.49% had had at least one previous miscarriage), less likely to have experienced vomiting since their last menstrual period (72% compared to only 59% among those consuming no caffeine), and more likely to have been interviewed after their miscarriages. Nausea (RR=2.02) and previous miscarriage (RR=2.33) were associated with miscarriages. The researchers said their computer model (somehow) controlled for these confounding factors and found the same results. They concluded:

We provided new evidence that the observed association was not likely the result of confounding by the pregnancy related symptoms of nausea, vomiting, and aversion to caffeine consumption. Therefore, it may be prudent to stop or reduce caffeine intake during pregnancy.

A young mother was on the news last evening, saying that she now feared that the cup of coffee she had allowed herself was why she had miscarried her baby.

Miscarriage myths and fears

According to the National Institutes of Health, miscarriage is the spontaneous loss of a fetus before the 20th week of pregnancy. Most occur when there is a genetic problem in the unborn baby and these are usually unrelated to the mother. Embryonic causes account for 80-90% of miscarriages, with genetic abnormalities within the embryo the most common cause. Other possible causes for miscarriage can include various physical problems (structural abnormalities), older maternal age, infections and serious diseases (untreated thyroid disease or diabetes, or renal disease). This is one reason early prenatal care is important.

The most tragic thing about yesterday’s media deluge was the number of women frightened that any amount of coffee or the smallest deviation from perfection in their behavior, could cause them to miscarry. It is easy to scare a pregnant woman, as it’s one of the most vulnerable times in their lives, and to make them worry that their behavior could cause them to lose their babies. But it’s also cruel and unsound.

Last month’s issue of the Journal Archives of Womens Mental Health examined how myths and fears about miscarriages persist, despite the prevalence of medical information to the contrary. Dr. Jonathan Schaffir, a clinical assistant professor of obstetrics and gynecology at Ohio State University said:

Most miscarriages result from genetic or chromosomal abnormalities in the fetus, or from medical complications relating to hormonal imbalances or problems with the uterus or placenta. Most of these things are beyond anyone's control and can happen to anyone.

Tragically, women without an understanding of scientific thinking are more likely to believe that pregnancy problems are a mother’s fault, he said. He called for a need to educate women to prevent them from feeling any guilt in association with their pregnancies. “Health care providers can reassure patients that these ‘old wives' tales' should not contribute to any feelings of personal responsibility.”

© 2008 Sandy Szwarc

* Caffeine levels in foods and beverages vary widely, but coffee generally contains the highest amounts. According to the U.S. Dept. of Agriculture, an 8-ounce cup of coffee has an average of 137 mg of caffeine, brewed tea 76 mg, and a 12-ounce can of soda 37 mg. Chocolate has negligible amounts.

** Specifically, they reported an association with miscarriage at this consumption, with a relative risk of 2.23 (95% confidence level 1.34-3.69).

Bookmark and Share