Nausea and vomiting of pregnancy (NVP) is a ubiquitous complex of symptoms worldwide. Several studies from diverse populations report varying, but high, levels of NVP.1 Much of the variability in the reported incidence of NVP reflects problems and/or peculiarities with the procedures for obtaining data about pregnancy-related events. In the United States, nearly 90% of pregnant women display some symptom of nausea and/or vomiting of pregnancy. The etiology and epidemiology of this complex of symptoms is only vaguely understood.
We conducted a study of 414 women in Albany, New York, with singleton pregnancies in order to identify patterns of nausea and/or vomiting of pregnancy and to follow these patients until delivery.2 These women were well nourished, and their pregnancies were ascertained by the 12th week after their last menstrual period (LMP). Cases were excluded from analysis due to twin births, induced abortions before the 13th week of pregnancy, spontaneous abortions before 12th week, and loss of participants or refusal to participate in follow-up interviews. The initial study interview was during the 12th week of pregnancy. Information about the 12th week of pregnancy, and about the month before the last menstrual period, was obtained during the initial interview. Subsequent interviews were conducted by telephone to cover the 16th, 20th, 30th and 38th weeks of pregnancy, and then again after delivery. Information collected during these interviews included the date of onset of symptoms of nausea and vomiting, the time of day that symptoms were usually experienced, and the date each symptom stopped. The study participants kept a log of these events. Known or suspected non-pregnancy-related nausea and vomiting, e.g., urinary tract infections or diarrhoea, were excluded from the analysis.
Data on food consumption were also collected, based on 24-hour recall recording the specific times and quantities of consumption of all foods, beverages and nutrient supplements. The information was converted to constituent nutrients with the use of a computer program developed for the Ten-State Nutrition Survey.3
The final study cohort of 414 women (of which 400 were white) was skewed toward the upper end of the scale of socioeconomic status, despite our efforts to obtain a representative sample of pregnant women in the area. Women of lower socioeconomic status did not seek prenatal care as early, thus not meeting one of our enrollment criteria, and dropped out at a much higher rate, probably due to the need for considerable participant motivation for the several interviews during the course of the pregnancy.
Nausea and/or vomiting of pregnancy was reported by 89.4% of study patients; 10.6% had no symptoms; 32.9% had nausea without vomiting; 2.9% had vomiting without nausea; and 53.6% had nausea and vomiting. By the eighth week of pregnancy, nearly 80% of the women in the study had developed nausea, and 70% were still actively experiencing nausea. Of those women who had nausea, 30% had stopped having symptoms by the 12th week, and 50% had stopped having symptoms by the 15th week. Even so, 25% of those ever experiencing nausea were still having symptoms by the 20th week of pregnancy. Of those women experiencing vomiting of pregnancy, 50% had stopped having symptoms by the 15th week. Nausea persisted longer among women whose nausea was accompanied by vomiting than among women with nausea alone. In our study, patients were stratified according to symptoms of no NVP, nausea only, and vomiting with or without nausea. It was found that a significantly higher proportion of women with no NVP experienced foetal deaths compared with women in both groups having symptoms (p < 0.0015). Thirty-one women experienced foetal deaths: 18.2% of the 44 women with no symptoms, 9.6% of the 136 women with nausea only, and 4.3% of the 234 women with vomiting. Many studies have demonstrated a positive association between presence of early NVP and favourable pregnancy outcome. Successful uterine implantation of the embryo requires increased levels of progesterone, which, in turn, is thought to trigger NVP.
Women who experienced no NVP tended to have a lower pre-pregnancy weight. The pattern of maternal weight gain during the course of pregnancy was almost identical for women who experienced no NVP and women who experienced only nausea. Those who had vomiting of pregnancy tended to display reduced weight gain compared to women with no NVP, or women with only nausea, through about the 25th week of pregnancy, by which time the rate of weight gain of all three groups became identical. Until the 12th week of pregnancy, women experiencing vomiting of pregnancy actually lost weight compared to their pre-pregnancy weight. However, women who experienced vomiting of pregnancy gained weight faster than the other two groups after the 12th week of pregnancy until no difference existed by the 25th week.
We also found that infant birth weight varied with NVP: women with no NVP had a larger proportion of infants born with low birth weight, which is probably a function of shortened gestation. An increased intake of niacin early in our study population was associated with decreased infant birth weight. Furthermore, there was a negative correlation between intake of niacin during week 12 and the total days of nausea (p < 0.01). Increased niacin intake is usually accounted for by increased protein intake. The negative relationship between increased niacin intake and infant birth weight appears mostly to be accounted for by shortened gestation length. Increased protein intake and the related faster maturation of the foetus resulted in earlier delivery. Lower niacin intake was reported by women experiencing more days of NVP, who also delivered heavier infants. This may be because lower protein intake resulted in slower foetal maturation so that the pregnancy progressed to full term. By itself, within normal limits, lower infant birth weight is not a good measure of pregnancy outcome, especially as it relates to foetal/infant risk factors. If lower infant birth weight is a function of immaturity at birth (shorter gestation), then, all else being equal, the infant may be at greater risk. However, if foetal development has been accelerated by increased maternal protein intake during pregnancy, then the infant may not be at greater risk, i.e., somewhat earlier delivery of a mature infant results in lower infant birth weight. In addition, infants who are small for gestational age (SFGA) born to women who smoke during pregnancy tend to grow faster during the first two years of life than their non-SFGA counterparts born to non-smokers, resulting in no real difference in size at two years of age. Such an increased rate of postnatal growth tends also to occur in infants who are SFGA for some other reasons.
Maternal dietary aversions during pregnancy tend to be closely related to NVP, with up to 85% of women in some studies in the United States reporting at least one dietary aversion during pregnancy.4 Dietary cravings and aversions have usually been viewed as a related complex of symptoms, where cravings operate to increase the consumption of selected food items and aversions operate to decrease intake of other specific foods. However, it has recently become clear that cravings and aversions should be separated into two different complexes.5 Cravings typically occur for foods which pregnant women would normally eat (e.g., foods high in sugar), whereas aversions usually develop due to physiologic responses to exposure to specific food items (e.g., coffee, alcoholic beverages, meats) and non-food items (e.g., cigarette smoke) - usually through the mechanism of nausea and/or vomiting. The direct trigger seems to be a function of smell/taste sensitivity. Although associations between cravings and measures of foetal outcome have been difficult to measure, a number of associations between dietary aversions and measures of foetal outcome have been noted. In the Albany study, a significant positive association (p < 0.001) existed between the occurrence of any aversion and increased foetal growth index (FGI). Foetal growth index is a measure of mean size for gestation length adjusted for population means in the specific geographic area. Several positive associations (p < 0.05) were noted between increased FGI and aversions to meat consumption, implying that decreased consumption of meat produced increased FGI. The mechanism seems to be that somewhat decreased protein intake results in slightly slower foetal maturation, resulting in longer gestation time, the evidence of which is then increased infant birth weight.
Significant negative associations were observed between infant birth weight and coffee consumption (p < 0.05) and birth weight and total caffeine consumption (p < 0.01), when using relative odds ratios. Maternal smoking was negatively associated with all metric measures of foetal outcome, including infant birth weight (p < 0.001). Smoking and alcohol intake did not confound the effects of coffee consumption on infant birth weight. Using regression analysis of infant birth weight on exposure variables (caffeine, alcohol and maternal smoking) produced similar results. Maternal smoking, coffee consumption and the intake of distilled spirits accounted for about 4.5% of the variance in birth weight, which is a significant proportion of the total variance (p < 0.001). Therefore, all three independent variables were negatively correlated with infant birth weight.
The largest numbers of aversions reported were to cigarette smoke, alcoholic beverages and coffee. Since the action of these aversions is to decrease the consumption of items for which aversions develop, the potential foetoprotective aspect of such aversions is apparent. Furthermore, a common expression of these aversions is through the mechanism of NVP, thus the potential effect of NVP on exposure to embryotoxins could be large. Observation of such relationships has led to the development of several explanations for the origin of NVP as being foetoprotective in nature. The explanations hypothesize that natural selection has provided for these behaviours, i.e., NVP as a mechanism for protection of the foetus during pregnancy from harmful substances in the maternal diet. Whereas I also favoured this hypothesis when first studying NVP, over the years my zeal for this explanation has waned. It is clear that NVP does reduce foetal exposure to certain toxins in the maternal diet. It is not so clear, however, that selection has operated at the level of reducing embryotoxins in the diet versus selecting for or against other maternal features, and so just coincidentally reducing foetal exposure to some toxins in the maternal diet.
Numerous changes occur in the maternal system during pregnancy, many of which are intimately interrelated. One problem with establishing the immediate cause of NVP is that, although correlations are relatively easy to demonstrate, causal connections are much more difficult, owing to the many changes taking place. A further problem in establishing causal relationships relates to the confusion of different levels of explanation. An explanation of the origin of NVP at the proximate level (hormonal changes) may be viewed as being in opposition to an explanation at the level of evolutionary significance or purpose (often called the ultimate level of explanation), when in fact they explain the same event. This problem is especially true in the study of NVP, since researchers come from many different areas of academic and/or research interests, bringing to the problem their own baggage in the form of agenda and particular viewpoints. Such a situation is not necessarily bad, and even makes NVP a far more interesting topic to study, but it does present specific problems related to understanding varying categories of explanation. For example, as a biological anthropologist, when speaking of the etiology of NVP, I am much more interested in the evolutionary development of NVP than in understanding the specific hormonal underpinnings. I am also interested in the pattern of the expression of NVP across varied human populations and whether something similar to NVP also exists among our closest relatives, the great apes. Or is it that the increase in protein in our diet (through scavenging and/or hunting) that is thought to have accompanied our ancestors' movement onto the savannas of Africa about four million years ago, was the event that triggered the development of NVP in humans? Any discussion of the etiology of NVP that flows from such interests will most likely be quite different from that of someone interested clinically in the well-being of women during pregnancy and methods for "controlling" NVP for maternal comfort.
In addition, we are often driven to explain the origin of NVP by placing it in an adaptive context. This penchant for defining behaviours and/or physical traits in the context of a selectionist universe fails to take into account that many biological features are the result of stochastic processes. Furthermore, although selection may be operating, it may not be operating in the manner that we suspect to be the case. For example, as mentioned earlier, although NVP may reduce foetal exposure to certain toxins, selection may actually be promoting earlier foetal maturation through the mechanism of generally increasing protein intake in the diet, and NVP may be a simple by-product of this selection.
Back to home