Morning Sickness: a Mechanism for Protecting Mother and Embryo Author(S): Samuel M

Morning Sickness: A Mechanism for Protecting Mother and Embryo Author(s): Samuel M. Flaxman and Paul W. Sherman Source: The Quarterly Review of Biology, Vol. 75, No. 2 (Jun., 2000), pp. 113-148 Published by: The University of Chicago Press Stable URL: http://www.jstor.org/stable/2664252 . Accessed: 07/12/2013 23:06

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to The Quarterly Review of Biology.

http://www.jstor.org

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions VOLUME 75, No. 2 JUNE 2000 THE QUARTERLY REVIEW of B JOLOGY

MORNING SICKNESS: A MECHANISM FOR PROTECTING MOTHER AND EMBRYO

SAMUEL M.FLAXMAN Departmentof Neurobiology and Behavior,Cornell University Ithaca,New York14853 USA E-MAIL: [email protected]

PAUL W. SHERMAN Departmentof Neurobiology and Behavior, Cornell University Ithaca, New York 14853 USA E-MAIL: PWS6WCORNELL.EDU

ABSTRACT Approximatelytwo-thirds ofwomen experience nausea orvomiting during thefirst trimester of pregnancy.These symptoms are commonlyknown as morningsickness. Hook (1976) and Profet (1988) htypothesizedthat morning sickness protects the emzbryo by causing pregnant women to physicallyexpel and subsequentlyavoidfoods that contain teratogenic and abortifacientchemi- cals,especially toxic chemicals in strong-tastingvegetables, caffeinated beverages and alcohol.We examinedthis hypothesis by comprehensively reviewing the relevant medical, psychological and anthropologicalliterature. In its support,(i) symptomspeak whenembryonic organogenesis is mostsusceptible to chemical disruption (weeks 6-18), (ii) womenwho experience morningsickness aresignificantly less likely to miscarry than women who do not(9 of9 studies),(iii) womenwho vomitsufferfewer miscarriages than those who experience nausea alone,and (iv) manypregnant womenhave aversionsto alcoholicand nonalcoholic(mostly caffeinated) beverages and strong- tastingvegetables, especially during the first trimester. Surprisingly, however, the greatest aversionsare tomeats, fish, poultry, and eggs.A cross-culturalanalysis using the Human Relations AreaFilesrevealed 20 traditionalsocieties in whichmorningsickness has beenobserved and seven in whichit has neverbeen observed. The latter were significantly less likely to have animalproducts as dietarystaples and significantlymore likely to have onlyplants (primarilycorn) as staples thanthe 20 societiesin whichmorning sickness occurred. Animal products may be dangerous to pregnantwomen and theirembryos because they often contain parasites and pathogens,especially whenstored at roomtemperatures in warmclimates. Avoiding foodborne microorganisms is

The QuarterlyReview of Biology, June 2000, Vol. 75, No. 2 Copyright? 2000 by The Universityof Chicago. All rightsreserved. 0033-5770/2000/7502-0001$02.00

113

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions 114 THE QUARTERLY REVIEW OF BIOLOGY VOLUME 75

particularlyimportant to pregnant zwomzen' because they are immunosulppressed,presumably to reducethe chances of rejecting tissues of their own offspring (Haig 1993). As a result,pregnzant womenzare morevulner-able to serious,often deadly infections. We hytpothesizethat morning sicknesscauses womento avoidfoods thatmight be dangerousto themselvesor theirembryos, especiallyfoods that, prior to widespread refrigeration, were likely to be heavily laden zwith microor- ganismsanzd their toxins. The alternative hypotheses that morning sickness is (i) an epiplhenome- non ofmother-offspringgenetic conflict or hormones associated with viable pregnzancies, or (ii) an indicatorto potential sexual partners and kin thatthe woman is pregnant,resulting in reduced sexual behaviorand increasednepotistic aicl, werenot well supported. Available data are most consistentzvith the hypothesis that morning sickness serves an adaptive,propizylactIcftnction.

INTRODUCTION Whydoes NVP occur? This question can be M ORNING SICKNESS is the common addressed frommultiple perspectives or "levels termfor the nausea and vomitingthat of analysis"(Mayr 1961; Tinbergen 1963; Sher- most women experience during pregnancy. man 1988; Alcockand Sherman 1994). Answers These symptomshave been recognized for to questions about the proximatemechanisms thousands of years. Indeed, the earliest re- thatunderlie NVP (i.e., how the symptomsare corded description of vomitingin pregnancy brought about) and the functional signifi- dates fromabout 2000 BC (Uarnfelt-Samsioeet cance of NVP (whythe symptomsoccur) are al. 1983). complementary,not mutuallyexclusive. Full "Morning sickness" is actually a complete understanding requires explanations at both misnomer, however; symptomsgenerally oc- proximateand ultimatelevels of analysis.Until cur throughoutthe day, notjust in the morn- recently,however, most research on NVP fo- ing (Vellacott et al. 1988; Dilorio et al. 1992; cused on elucidating its proximate causes in Whitehead et al. 1992), and the term "sick- order to discoverways to amelioratethe annoy- ness" implies pathology,even though healthy ing and sometimes debilitatingsymptoms. women experience the symptomsand bear The neuroendocrine mechanisms underly- healthybabies (Nesse and Williams1994). For ing NVP have been well studied (reviewed by these reasons, other, more accurate terms Andrews and Whitehead 1990). Neural con- have been used to describe morningsickness, trol of nausea and vomitingis coordinated by including: nuclei in the brainstem. Symptomsare trig- (i) Emesis gravidarum. Jarnfelt-Samsioe gered by inputs fromtwo major pathways:the (1987:422) definesthis as "nausea alone area postrema (the so-called "chemoreceptor or the combinationof nausea, retching, triggerzone") and the gastrointestinalaffer- and occasional vomitingin early preg- ents. Both pathwaysare involvedin the body's nancy." response to ingested toxins. The area post- (ii) Pregnancy sickness. Profet (1992:327) calls this"a collectionof symptoms-food rema also functionsin the acquisition of con- aversions,nausea, and vomiting-one or ditioned taste aversions and control of food all of which occur in women duringthe intake. Although hormonal changes in early firsttrimester of pregnancy." pregnancymediate NVP, no consistentdiffer- (iii) Nausea and vomiting in pregnancy ences in levels of estrogens,progestagens, an- (NVP). Accordingto Fail-weather(1968: drogens,cortisol or human chorionic gonado- 135, quotingthe 1956 AmericanCouncil tropinhave been demonstratedbetween women on Pharmacyand Chemistry),this is nausea or vomitingthat is "commonlyob- who do and do not experience symptoms.This servedduring the first14 or 16 weeks [of findingindicates that the hormonal changes pregnancy]and characterizedby some function indirectlyby activating the neural disturbancein appetite and reaction to pathways,rather than being the emetic agents food in a fairlylarge percentage of cases," themselves(Walsh et al. 1996). but thatis not associated withdisturbed At the functionallevel, two questions arise. nutrition. First, why should hormonal changes make Of these terms,we favorNVP because it is de- pregnant women more likely to experience scriptive,objective, and acknowledged within nausea and vomiting?Second, is NVP adaptive the medical community. or pathological? These issues were raised 60

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions JUNE 2000 MORNING SICKNESS 115 years ago, when Irving (1940:719) reported trated phytochemicals in food preparation that "only six of our 225 cases [of pernicious (spices). Many spices have powerful antimi- vomiting] miscarried-a frequency of only crobial properties (e.g., Walker 1994; Hirasa one in 37.5 cases, as contrastedwith the usual and Takemasa 1998), and Billing and Sher- expectancy in pregnancyof one in everyfive man (1998) hypothesized that we use these or six." Although Irvingwas actuallyreferring naturalpharmaceuticals to reduce foodborne to an extremeand debilitatinglevel of nausea illnesses and food poisoning by inhibitingmi- and vomiting known as hyperemesis gravi- crobial and fungal growthand toxin produc- darum, his surprisingresults suggested an as- tion. These beneficial effectsmay account for sociation betweenNVP symptomsand positive the ubiquityof spice use, especiallyin hot cli- pregnancyoutcomes. mateswhere unrefrigeratedfoods spoil quickly Thirty-sixyears later, Hook (1976:182) pro- (Sherman and Billing 1999). However, as posed that"NVP (and other symptomsaffect- Gerber et al. (1999) recentlypointed out, too ing diet) in earlypregnancy evolved as a spec- much of a good thingcan be harmful.Spices trum of response to environmental factors are beneficial in the tinyquantities typically which are selectivelytoxic (or beneficial) to used in cooking, but in large doses manyphy- the fetus."Subsequently, Hook (1978, 1980) tochemicals can have deleterious effectsas al- explored the possible link between NVP and lergens, mutagens, carcinogens, teratogens maternalingestion of alcohol and caffeinated and abortifacients(Schardein 1985:679-699; beverages,as well as the use of tobacco. Follow- Ames et al. 1990a,b; Johns 1990; Shepard ing Hook's lead, Profet (1992:328) suggested 1992; Beier and Nigg 1994). For example, that "[t]he food aversions,nausea, and vom- small amounts of chili peppers can yield anti- itingof pregnancysickness evolved during the microbial and therapeutic effects,but inges- course of human evolution to protectthe em- tion of large amounts of capsaicin has been biyo against maternal ingestion of the wide associated withulceration, necrosis, and carci- arrayof teratogens(toxins thatcause birthde- nogenesis (Surh and Lee 1996). Likewise, fects) and abortifacients(toxins that induce large quantities of caffeinehave been associ- abortions) abundant in naturalfoods ... preg- ated with spontaneous abortions (Klebanoff nancy sickness represents a lowering of the et al. 1999). The embryoprotection hypothe- usual human thresholdof tolerance to toxins sis proposes that NVP functionsto shield the in order to compensate for the extreme vul- differentiatingembryo from these potentially nerabilityof the embryo to toxins during or- toxic chemicals. ganogenesis." Hook (1976, 1978, 1980) focused on the Hook (1976, 1978, 1980) and Profet (1988, harmful effectsof alcohol, caffeine and to- 1992, 1995) emphasized that the benefitsof bacco-substances that are often used nausea and vomitingwere the expulsion of because of the secondaiy compounds they dangerous foodborne chemicals and the sub- contain-whereas Profet(1992) presented an sequent avoidance of these chemicals via extensive list of common foods that contain learned aversions to the foods that triggered potentiallyembryotoxic chemicals. According illness.This "embryoprotection" hypothesis is to Profet: based on the followinglogic. The chemicals If first-trimesterpregnant women develop thatgive manyplants theirdistinctive aromas foodaversions because of recalibrated thresh- and flavors evolved to counter the plant's olds for detectingand toleratingtoxicity, biotic enemies, such as herbivorous insects then one would predictthat the following and vertebrates,fungi, pathogens, and para- foods and beverageswould elicitaversions: sites (Ehrlich and Raven 1964; Koul 1993). (a) bitter or pungent foods-indicating concentrationsof plant toxins-such These substances are known as phytochemi- high as coffee,tea, vegetables,spices and herbs, because cals or secondary compounds they and the more pungentor bitterof the alco- generallyare not essential to the plant's basic holic beverages;(b) foods thatemit burnt or primarymetabolism. Humans commonly or friedodors-indicating thecreation dur- ingest phytochemicalsthat occur naturallyin ing cooking of mutagens-such as barbe- vegetables, and also selectivelyuse concen- cued, roasted,or friedfoods (as opposed to

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions 116 THE QUARTERLY REVIEW OF BIOLOGY VOLUME 75

boiled foods); and (c) foodsthat emit smells and Reeve 1997). We have comprehensively suggestiveof spoilage-indicating parasiti- reviewed the relevantmedical, psychological zationby toxin-producing bacteria-such as and anthropological literature,and gathered animal products that are not extremely informationto testfive critical predictions of fresh... On theother hand, one would pre- the embryoprotection hypothesis: dict thatthe best-toleratedfoods would be thosethat have ratherbland odorsand tastes 1. NVP should be associatedwith positive preg- and thatdo notspoil easily, such as processed nancyoutcomes. breads,cereals, and grains(1992:346). 2. Foods thattrigger NVP should contain ter- Although the embryoprotection hypothesis atogens, mutagens and abortifacients. had not been testedrigorously, Profet (1995) 3. NVP should be more common when the published a semipopular book based on the embryois mostsensitive to toxic chemicals. subject, which included specific dietaryrec- 4. Foods containing toxins should be most ommendationsfor pregnant women. The book, aversiveto women when embryonicorgan- and a subsequent ScientificAmerican article ogenesis is most sensitiveto disruptionby about it (Holloway 1996), elicited harsh criti- exogenous chemicals. cism fromBrown et al. (1997), who stated that 5. The frequencyof NVP should depend on "claims made in the popular press about food the diet of a population: symptomsshould and health relationshipsshould be evaluated be uncommon in populations where staple by the media as fictionunless supported by foods rarelycontain substances that could scientificresearch" (p 179). In contrast,Brown damage embryos. et al. recorded the consumption of vegetables Our results confirmed that NVP is associ- their among 549 Minnesota women during ated withbeneficial effects, but notjust on the firsteight weeks of pregnancy:they found that embryo.The data suggesta more comprehen- and NVP consumed the women with without sive and specifichypothesis, namely, that NVP same numbersof servingsper week of Profet's protects the embryofrom teratogenicphyto- (1995:150-151) proscribed vegetables, and chemicalsand shieldsboth the motherand her no between the number of serv- relationship developingembryo from foodborne pathogens ings consumed per week and "adverse preg- and their associated toxins. This "maternal nancy outcomes" (miscarriages,fetal deaths and embryoprotection hypothesis"is consis- or congenital anomalies). However, Brown et tentwith observed variations in food aversions al. did not reportwhether the women who ex- and cravings,and variationsin the occurrence perienced NVP vomited or were nauseated of NVP among individualsand cultures. after consuming the proscribed vegetables, whetherthey avoided those vegetables during METHODS the restof theirpregnancies, or whethertheir pregnancy outcomes differeddepending on DEFINITIONS the occurrence of NVP or food aversions. In To avoid ambiguity,we brieflyexplain our addition, many of the vegetables that Brown use of some keyterminology. "NVP" is nausea et al. inquired about were consumed rarely, alone, or in combination withvomiting, that so there may have been littleopportunity for begins before the 20th week of gestationand them to affectembryos, or for maternal aver- is not associated with diseases, infectionsor sion learning to occur. These omissions are allergies. NVP encompasses a continuum of surprisingbecause Brown (1983:59) herself symptoms,from heartburn and mild nausea had previouslyreported thatstrongly-flavored to frequentvomiting. An extremelevel of nau- vegetablesand coffeefrequently "bring on the sea and vomitingknown as "hyperemesisgrav- feelingof nausea" in pregnancy,and thattaste idarum" occurs in less than 1% of pregnancies aversionscause pregnantwomen to eat less of (Tsang et al. 1996); vomitingis so severe that the offendingfoods (Brown and Toma 1986). patientsare unable to performdaily activities, In viewof these conflictingclaims, and their and even have difficultysleeping. According importancefor understanding women's health, to Fairweather(1968:136), hyperemesisgravi- we decided to evaluate the causes and repro- darum involvesvomiting "of such severityas ductive consequences of NVP (see Sherman to require the patient'sadmission to the hospi-

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions JUNE 2000 MORNING SICKNESS 117 tal," and, if untreated, it can result in dis- were stated in the original source or could be rupted nutrition,liver damage and ketosis calculated unambiguously. (Erick 1995). To focus on "normal" NV\P,we We sought informationon morning sick- excluded all known or likelystudies of hyper- ness-like symptomsin nonhuman mammals emesis gravidarum,based on diagnoses of the by searching textbooks of veterinaryphysiol- original authors and whether or not the sub- ogy and all 38 volumes (1959-98) of TheInter- jects required hospitalization. nationalZoo Yearbook.We also contactedveteri- The age of an embryo can be expressed as narians (i.e., at the NewYork State College of eitherits "postconception age" (i.e., itsage rel- VeterinaryMedicine) and other animal re- ative to the date of conception) or its "post- searchers who had published information menstrual age" (its age relative to the moth- about appetite during pregnancyin primates, er's last menstrualperiod). We use the latter, swine, sheep, cats, dogs, rats, rabbits,horses unless otherwiseindicated. Assumingthat fer- and goats. tilizationoccurs in the middle of a woman's Lastly,we used the Human Relations Area menstrual cycle, an embryo's postmenstrual Files (HRAF) to investigatethe distributionof age is about two weeks greater than its post- NVP in traditionalsocieties (Murdock 1981). conception age (Moore and Persaud 1998). A Previously,Minturn and Weiher (1984) re- developing human is an "embiyo" fromcon- ported thatinformation on morningsickness ception to eight weeks of age, and a "fetus" was available for 30 societies listed in the from the ninth week after conception until HRAF. We reviewed the printed HRAF on birth (Moore and Persaud 1998). Our analy- these cultures,as well as the new filesavailable ses focused on the firsttrimester (13 weeks) on CD-ROM (Installments43-45). For each of pregnancy,which straddles the definitional culture,we recorded all relevantinformation embryo-fetustransition. To avoid confusion, about pregnancysymptoms ("Pregnancy," cat- we simplyuse "embryo"throughout. egory 843) and dietary habits ("Diet," cate- "Miscarriage"is embryonicmortality during gory262). the first20 postmenstrualweeks, and "stillbirth"is mortalityfrom postmenstrual week 20 QUANTIFICATION OF AVERSIONS to parturition(Anderson et al. 1994). "Fetal AND CRAVINGS death" is the totalmortality attributed to both To quantify women's food cravings and miscarriagesand stillbirths."Preterm birth" is aversions during pregnancy,we follow Dick- parturition before the 37th postmenstrual ens and Trethowan (1971) in defininga "preg- week, and "low birthweight" refersto an in- nancy-relatedfood aversion"as a stronglyneg- fant that weighs < 2.5 kg (5.5 lb) at birth. ativeresponse to a particularfood or beverage "Neonatal mortality"is death of an infant thatwas not disliked prior to pregnancy,and withinone month afterbirth. a "pregnancy-relatedfood craving"as a strong urge to consume a food or beverage forwhich SEARCH TECHNIQUES there had not been an intense desire prior First,we gathered informationon NV\Pand to pregnancy. pregnancy-relatedfood aversions and crav- There is a large literature on pregnancy- ings using online databases (e.g., AGRICOLA, related food aversionsand cravings.To be in- BIOSIS, CAB Abstractsand MEDLINE). To lo- cluded in our analyses,studies had to present cate even earlier sources we also searched Bio- data specificallyon foods thatpregnant women logicalAbstracts from the firstvolume (1926) craved or found aversive (i.e., "endogenous" forward.We retrieved original sources, and responses to foods: Hook 1976), rather than used their referencelists to locate additional foods simply listed as "preferences" or "ta- sources. We also searched the ScienceCitation boos." This was done to exclude dietaryhabits Index for recent references to key sources. that were based on the advice of others, or When numerical data were extractedfor anal- what women thoughtthey should say theyate ysis,our only criterionfor inclusion was that whilepregnant. Studies that were excluded un- sample sizes (e.g., number of women in a der this criterion include Bartholomew and study, number of women exhibiting NVP) Poston (1970), Darwish and Amine (1982),

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions 118 THE QUARTERLY REVIEW OF BIOLOGY VOLUME 75

and Spielmann (1989). Wijewardene et al. greasy,and fried,""salty," "savory," "dressings (1994) was excluded because the sample sizes and condiments," and "other." We also ex- were unclear. cluded "nuts and peanut butter"and "water There are no standardized categories used and ice" because onlya fewstudies listed foods byresearchers to classifydietary aversions and in these categories and the number of aver- cravings. Some authors (e.g., Tierson et al. sions or cravingswas minuscule (inclusion of 1985) have presented data on women's re- these categories would not have altered our sponses to specificfood items,such as "olives," conclusions). "potatoes," and "liver,"whereas others (e.g., For each of the nine food categories that Dickens and Trethowan 1971) have tabulated we studied, the total numbers of aversionsor responsesto broaderfood categorieslike "meat, cravingsfrom all studieswere summed. These fish,eggs," "vegetables," and "fruit."These dif- totalsare not necessarilythe same as the total ferencesobviously could have affectedthe way number of women with aversions to or crav- in which individual responses were scored, ingsfor each categorybecause a woman could making it difficultto synthesizedata fromdif- express more than one aversion or cravingin ferentstudies. For example, awoman who had the same category.To minimize pseudorepli- aversions to "liver" and "chicken" could be cation, we calculated the average number of counted as having two differentaversions, or aversionsand cravingsper woman to each food onlyone aversionto "meats."To take thisproh categoryby dividingthe total number of aver- lem into account, we generated a list of food sions or cravingsto that categoryby the total categories that was broad enough to accom- numberofwomen sampled. As an example,sup- modate the variationfound in original refer- pose one studyof 20 women reportedthat two ences,but narrowenough to be usefulfor analy- had aversionsto "broccoli" and fivehad aver- sis. Nine categories emerged: (1) meats, fish, sions to "spinach," and a second studyof 20 poultry,and eggs, (2) vegetables,(3) grainsand womenreported thatfive had aversionsto "vege- starches,(4) dairyand ice cream, (5) fruitsand tables." In our analyses, these hypothetical firuitjuices,(6) alcoholic-beverages,(7) nonalco- studieswould account for12 aversionsto "veg- holic (caffeinated)beverages (whichincluded etables"and 0.3 aversionsper woman (12/40). onlycoffee, tea and soda), (8) sweets,desserts, There are several weaknesses in the avail- and chocolate, and (9) ethnic, strong and able informationabout cravingsand aversions. spicy foods. Categories 1-7 are self-explana- First,data were based on questionnaires in- tory. Regarding category (8), although we stead of on observationsof women's responses would have liked to consider chocolate sepa- to foods during pregnancy;it is possible that ratelybecause it contains many phytochemi- some women may have responded according cals (Matissek 1997), mostauthors did not dis- to what theythought they were "supposed" to tinguishsweets or dessertsthat did and did not crave or find aversive,rather than what they contain chocolate. Category (9) is obviously actuallyexperienced (see Sherman and Reeve ambiguous, but we retained it because of its 1997). Second, the timingof questionnairead- frequentoccurrence in numerous studiesand ministrationwas not standardized. Some au- its potential importance to testing Profet's thors (e.g., Fairburn et al. 1992) interviewed (1992) hypothesis,since strongand spicyfla- women once or more during gestation, and vors are often derived from phytochemicals others (e.g., Hook 1978) conducted only ret- (Billing and Sherman 1998). Tobacco and rospective(postpartum) interviews;women in smoke are clearlyimportant to Hook's (1976) the latter group may not have recalled their hypothesis,but since theyare not foods or bev- reactions to various kinds of foods as clearly erages, theywere considered separately.We as women in the formergroup. Third, most excluded eight categories of aversions and studies reported information on aversions cravingsbecause the ingredientsof these foods and cravingsthat occurred at any time during were either uncertain, or so heterogeneous pregnancy;only Maclntyre (1983) and Rodin thatit was impossible to inferexactly what was and Radke-Sharpe (1991) separated theirdata being craved or avoided. These categoriesare by trimester(we analyzed data fromthese two "soups," "casseroles,""chicken pot pie," "fatty, studiesseparately). Finally, some authors (e.g.,

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions JUNE 2000 MORNING SICKNESS 119

Fairburnand Welch 1990) onlyreported aver- in the United States (range: 27-89%, N = 22 sions or cravingsthat were "frequently"men- groups, 35,387 pregnancies), and 62% + 10 tioned by interviewees,so representation of in all other countriesin our sample combined some rarely-mentionedfood categoriesmight (range: 11-100%, N = 33 groups,38,283 preg- be artificiallylow. Although all these factors nancies; referencesin the legend to Figure 1). mayaffect the qualityof the originaldata, they Binomial proportion tests indicated signifi- do not systematicallybias our analyses for or cant differencesin the incidence of NV\Pbe- against any particular hypothesis about the tween the USA, UK, and all other countries adaptive significanceof NVTP,especially since combined (all pairwiseP < 0.01). In viewof the feworiginal investigatorsindicated awareness differencesamong countries,we reanalyzedthe of any adaptive hypotheses. frequencydata, thistime eliminatingpseudo- replication byusing only the mean frequency STATISTICAL ANALYSES for each country.For the 16 countries in our In general, we report results of statistical sample, the mean proportionof women expe- testsapplied by original authors. We also use riencingNVP was 62% ? 13, the median pro- G-tests incorporating Williams' correction portion was 64%, and the range was 35-84%. (Sokal and Rohlf 1995:698) to examine differ- Timingof NVP ences in (i) pregnancy outcomes between women who did and did not experience NVP Circadian and (ii) diet between cultures withand with- During weeks 6 to 13 of gestation, NVP out NVP, and binomial proportion tests to symptomsoccur with equal probabilitythrough- evaluate differencesin (iii) cravingsand aver- out normal waking hours: 40-45% of women sions and (iv) the incidence of NVP by geo- experienced nausea at least once during each graphic region. 4-hrperiod from0700 to 2300 (Figure 2). The probabilityof experiencing nausea dropped RESULTS to less than 10% during the two 4-hrperiods from 2300 to 0700. Differencesin the occur- CHARACTERIZING NAUSEA AND rence of symptomsbetween waking and sleep- VOMITING IN PREGNANCY ing hours were highlysignificant (P < 0.001), Frequencyof Occurrence but there were no significant differences There were 56 studies thatreported the fre- among the four time periods during normal quency of NVP in 75 groups of women from wakinghours. 16 countriesworldwide, with a total of 79,146 pregnancies of 64,876 individuals.The num- Weekly ber of groups exceeds the number of studies Tierson et al. (1986) and Vellacott et al. because some studiesincluded more than one (1988) documented the weeklyoccurrence of group of women. Groups ranged in size from NVP symptoms,based respectivelyon inter- 9 to 11,481, witha median of 151 and a mean viewsof 414 women in Albany,NewYork (89% of 865 ? 2,089 (SD). Among all the groups, of whom experienced NVP) and 500 women the mean proportion of women who experi- in London (76% of whom experienced NVP). enced NVP was 66% ? 16, the median propor- Both studies reported that NVP occurs pri- tion was 68%, and the range was 11-100% marilyin the firsthalf of pregnancy (Figure (Figure 1). 3a,b). Tierson et al.'s studyis especially reli- Frequencies of NVP differedamong coun- able because it excluded episodes of nausea tries.The highestmean frequencyfor a single or vomitingthat might have resulted fromill- countrywas 84% inJapan (1 group, 132 preg- nesses, infections,or "flu-like"symptoms. Be- nancies: Mori et al. 1988), and the lowestwas ginningin thefourth postmenstrual week (i.e., 35 % in India (1 group, 2,500 pregnancies:An- when the embryowas about two weeks old), anth and Rao 1993). In the United Kingdom, the fractionof pregnantwomen experiencing 75% ? 12 of women experienced symptoms nausea or vomitingrose rapidly.By the sixth (range: 53-94%, N = 20 groups, 5,746 preg- week, more than halfthe women experienced nancies), as compared to 64% ? 14 of women symptoms.The frequencyof NVP peaked dur-

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions 120 THE QUARTERLY REVIEW OF BIOLOGY VOLUME 75 0.32

O 0.24

0 0.16

O,, 0.08

v~~~~~~~~~~~~~~~~~~ z I I 0.0 0.2 0.4 0.6 0.8 1.0 Proportionof Womenwith NVP

FIGURE 1. FREQUENCY OF NVP AMONG 75 GROUPS OF WOMEN REPORTED IN 56 STUDIES. The mean frequencyof NVP was 0.66, and the average group size was 865 women (N = 79,146 pregnancies from64,876 women). The studieswere done in Australia (N = 116 pregnantwomen: Biggs 1975; N = 429: Krickeret al. 1986; N = 100: Abraham et al. 1994), Amsterdam (N = 396: Paarlberg et al. 1996), Canada (N = 20: Drake et al. 1988; N = 732: McBride et al. 1991), Greece (N = 102: latrakiset al. 1988), Hong Kong (N = 1,453: Chin 1989), India (N = 2,500: Ananth and Rao 1993), Israel (N = 100: Medalie 1957), Japan (N = 132: Mori et al. 1988), Kenya (N = 68: Pike 1997), Nigeria (N = 400: Jinadu and Daramola 1990), Norway (N = 22,241 pregnancies from8,675 women: Corey et al. 1992), South Africa (N = 1,771: Walker et al. 1985), Sri Lanka (N = 1,000: Wijewardene et al. 1994), Sweden (N = 152: Uddenberg et al. 1971; N = 5,377: Kullander and Kallen 1976; N = 948 pregnancies from 244 women: Jarnfelt-Samsioe et al. 1983; N = 102:Jarnfelt-Samsioe et al. 1985), the United Kingdom (N =100: Diggory and Tomkinson 1962; N = 105: Wolkind and Zajicek 1978; N = 42: Baylis et al. 1983; N = 50: Maclntyre 1983; N = 86: Fitzgerald 1984; N = 116: Masson et al. 1985; N = 212: Davies et al. 1986; N = 342: Evans et al. 1986; N = 86: Fairburn 1986; N = 242: Stewartet al. 1988; N = 500: Vellacott et al. 1988; N = 50: Fairburn and Welch 1990; N = 100: Fairburn et al. 1992; N = 1,000: Whitehead et al. 1992; N = 363: Gadsby 1994; N = 1513: Meyer et al. 1994; N = 569: Robinson et al. 1996), and the United States (N = 256: Speert and Guttmacher 1954; N = 3,853: Yerushalmy and Milkovich 1965; N = 7,027: Brandes 1967; N = 11,481: Milkovich and van den Berg 1976; N = 210: Little and Hook 1979; N = 78: Dilorio 1985; N = 180: Schuster et al. 1985; N = 70: Fawcett and York 1986; N = 7,767: Pettiti 1986; N = 414: Tierson et al. 1986; N = 55:Jenkins and Shelton 1989; N = 903: Weigel and Weigel 1989a; N = 270: Werler et al. 1989; N = 1,908: Fenster et al. 1991; N = 80: Rodin and Radke-Sharpe 1991; N = 126: O'Brien and Zhou 1995; N = 160: Snell 1996; N = 549: Brown et al. 1997). One additional study was unpublished (Abney 1986, cited in Dilorio 1988, location not stated, N = 144).

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions JUNE 2000 MORNING SICKNESS 121

WakingHours SleepingHours correspondence between the sensitiveperiods 0.5- in embryonicorganogenesis (Figure 3c) and the peak occurrence of NVP (Figures 3a,b). .3 0.4- Heritabilityof NVP 0.3- Women were significantlymore likelyto experience NVP if their mother (Whitehead et 0.2- 4, al. 1992; Gadsby et al. 1997) or sisters(Vella- cott et al. 1988; Corey et al. 1992) had also * 0.1 experienced the symptoms.These studies did 0 not include formalheritability estimates, and we were unable to calculate heritabilityfrom 6 0700- 1100- 1500- 1900- 2300- 0300- 1100 1500 1900 2300 0300 0700 the data presented. Corey et al. (1992) did re- Time Periods port highlysignificant (P < 0.001) tetrachoric correlationsin the likelihood of experiencing FIGURE 2. TIMES DURING THE DAY WHEN NVP between monozygotic twin sisterswho HEALTHY, PREGNANT WOMEN were raised together (54% ? 5 [SE], N = 830 EXPERIENCE NVP. pairs) and dizygotic twin sisters who were The bars represent the proportion of women in raised together (24% + 5, N = 902 pairs). a studywho experienced nausea orvomiting at least These resultsimply a genetic component to once during each 4-hour time period. Data (replot- experiencing NVP. ted) and statistical analyses are from Dilorio et al. (1992), based on diaries of symptoms kept by 19 NVP AND PREGNANCY OUTCOMES women for at least one week during weeks 6-13 of gestation. *** indicates P < 0.001. Miscarriages Information on the relationship between NVP and spontaneous abortions during the ing weeks 9 to 14, when 60-70% of the women first20 weeks was summarized by Weigel and experienced nausea and 30-40% vomited. Weigel (1989b). Their meta-analysisincluded Thereafter,the frequencydeclined gradually, seven studies based on 18,464 pregnancies of and by the 20th week only 20-30% of the 17,760 women. In everystudy, women who ex- women experienced nausea and 10-20% con- perienced NVP were significantlyless likelyto tinued to vomit.Some women (less than 10% ) miscarrythan women who did not experience experienced symptomsof NVP rightup to the NVP (Figure 4). time of birth. The validityof any meta-analysismay be challenged on the grounds that only positive Relativeto Embryonic Organogenesis results are published. Weigel and Weigel's Embryonic tissues are most susceptible to (1989b) studyis unlikelyto sufferfrom this damage fromteratogens during certain,well- problem, however,because there is no reason defined "criticalperiods" when cell division, to suppose thatcontrary results would be sup- cell differentiationand morphogenesis of a pressed. We located two additional relevant number of organ systemspeak simultaneously studies (Petitti1986; Fenster et al. 1991), and (Figure 3c). Teratogens rarelycause congeni- both of them reported thatwomen who expe- tal anomalies during the firstfour postmen- rienced NVP were significantlyless likely to strualweeks (Carlson 1994). Periods ofvulner- miscarrythan women who did not experience abilityfor various organ systemsbegin at about NVP, in agreement with Weigel and Weigel week five,when the developing central ner- (1989b). vous systemand heart become criticallysensi- There is one way the relationshipbetween tive. The peak of organogenesis and embry- NVP and lowermiscarriage rates could be mis- onic susceptibilityto teratogensoccurs during leading. In most of the studies analyzed by weeks 6 to 12. The embryo's central nervous Weigel and Weigel (1989b), women were en- systemcontinues to be sensitivethrough week rolled soon afterconception. If thosewho mis- 18. There is an obvious-indeed striking- carried usually did so in the firstfew weeks,

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions 122 THE QUARTERLY REVIEW OF BIOLOGY VOLUME 75

theymay have lost their embryobefore NVP Weigel (1989a) and Tierson et al. (1986), re- occurred (see Figures 3a,b). Thus early mis- spectively.In each case, the lowestfrequencies carriage would have avoided NVP. This possi- of both outcomes were among women who bility can only be ruled out for one study vomited during pregnancy, the next lowest (Brandes 1967: #4in Figure 4), in which most frequencies were among women who experi- women were enrolled aftertheir 8th postmen- enced only nausea, and the highest frequen- strualweek (i.e., near the peak of symptoms). cies were among women that did not experi- We located two additional studies that con- ence any symptoms(Figure 6). trolled for the timingof miscarriages.Kleba- noff et al. (1985) and Tierson et al. (1986) OtherPregnancy Outcomes only considered pregnancies thatlasted ' 14 NVP does not affectthe frequencyof still- weeks and ' 12 weeks, respectively.In both births(Yerushalmy and Milkovich1965; Bran- cases, fetal death rates (i.e., miscarriagesplus des 1967; Chin 1989). This implies that the stillbirths) were significantlylower among negativerelationships between NVP and fetal women who experienced NVP than among deaths (Figures 5 and 6b) are due primarily women who did not (Figure 5). These results to the association between NVP and reduced imply that NVP reduces miscarriages,rather frequencies of miscarriages. than vice versa. There were no consistent associations be- Cohen (1997) claimed thatan unpublished tween NVP and four other pregnancy out- studyin Bangladesh byK O'Connor et al. indi- comes: cated that the association between NVP and 1. Preterm birth. Brandes (1967), Kleba- reduced miscarriage rate disappeared when noffetal. (1985), and Ananth and Rao (1993) the effectof maternal age was controlled. Ac- reported that women who experienced NVP to cording Cohen, maternalage was positively were less likelyto give birthprematurely than correlated with miscarriages and negatively women who did not experience NVP, but Chin correlatedwith NVP symptoms.Neither Med- (1989), Jarnfelt-Samsioeet al. (1983, 1985), alie (1957: #2in Figure4) norJarnfelt-Samsioe Tierson et al. (1986), and Weigel and Weigel et al. (1983: #6in Figure 4) found any associa- (1989a) found no significantdifferences in tion between NVP and maternalage, however, frequenciesof preterm births between women althoughboth reportedhighly significant asso- who did and did not experience NVP. ciationsbetween NVP and reduced chances of 2. Low birthweight. Brandes (1967), Little miscarriage.Klebanoff et al. (1985) andJinadu (1980), and Tierson et al. (1986) found that and Daramola (1990) did find a higher inci- womenwho did not experienceNVP weremore dence of NVP among younger women, but likelyto givebirth to offspringweighing < 2.5 Vellacott et al. (1988), Chin (1989), Paarlberg kg (5.5 lb) than women who experienced et al. (1996), and Gadsby et al. (1997) found NVP, but no significantassociation between no relationshipbetween maternal age and NVP. NVP and low birthweight was observedbyJarn- Relationships between levels of NVP symp- felt-Samsioeet al. (1985), Klebanoff et al. toms and the likelihood of miscarriagesand (1985), Weigel and Weigel (1989a), Ananth fetal deaths were investigatedby Weigel and and Rao (1993), and Gadsby et al. (1997).

FIGURE 3. TIME COURSE OF NVP AND ITS RELATIONSHIP TO CRITICAL PERIODS IN EMBRYONIC ORGANOGENESIS. (a) Time course of nausea and vomiting (dashed line: nausea with or without vomiting; solid line: vomiting); reprinted (scanned) fromTierson et al. (1986), with permission from Mosby, Inc. (b) Time course of nausea, withor withoutvomiting, modified fromVellacott et al. (1988), withpermission from Elsevier Science. (c) Critical time periods when various developing structuresare most susceptible to disruptionby teratogens, based on informationin Moore and Persaud (1998:156). CNS = centralnervous system, H = heart, UL = upper limbs, Ey = eyes, LL = lower limbs, T = teeth, P = palate, EG = external genitalia, E = ears.

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions JUNE 2000 MORNING SICKNESS 123

0.6 4 < Nausea (a)

0.4-

?0.61 -N (b)

0.4 - Nausea or 0 j ,Vomiting ; 0.2-

CNS (c) H - ~UL-

; BEy ~ LL - T m CQ p

1B_

0 5 10 15 20 25 30 35 40 45 PostmenstrualWeek of Pregnancy

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions 124 THE QUARTERLY REVIEW OF BIOLOGY VOLUME 75

0.30 0.20 FETAL DEATHS MISCARRIAGES * 0.25 -

0.20 [n NoNV f 0.15- 0.15. " ~ ~~|no NVP|

0.10 ** o 0.10- 0.05

0.00 ; ~~~** 1 2 3 4 5 6 7 A, 0.05- StudyNumber

FIGURE 4. ASSOCIATION OF NVP WITH 0.00 I REDUCED MISCARRIAGE RATES (FETAL DEATH 1 2 AT < 20 WEEKS). StudyNumber Data are fromseven studies (N = 18,464preg- nanciesfrom 17,760 women). Statistical results are FIGURE 5. ASSocIATION BETWEEN NVP fromWeigel and Weigel's (1989b) meta-analysis, AND REDUCED FETAL DEATH with** P < 0.01 and *** P < 0.001. Studynumbers (MISCARRIAGE PLUS correspondto the following(with odds ratiosand STILLBIRTH) RATES. 95% confidenceintervals calculated by Weigel and Study 1 (Klebanoff et al. 1985) included 9,098 Weigel[1989b]): 1. Speertand Guttmacher1954, women whose pregnancies lasted ' 14 weeks; ** N = 256women, OR: 0.16,CI: 0.07-0.36;2. Medalie indicates P < 0.005. Study 2 (Tierson et al. 1986, 1957, N = 100, OR: 0.03, CI: 0.00-0.54; 3. Yeru- incorporating corrections of Tierson et al. 1989) shalmyand Milkovich1965, N = 3,853,OR: 0.34,CI: included 414 women whose pregnancieslasted - 12 0.26-0.45;4. Brandes1967, N = 7,027,OR: 0.41,CI: weeks; * indicates P < 0.05. Klebanoffet al. distin- 0.32-0.53;5. Kullanderand Kallen1976, N = 5,377, guished only between women who vomited during OR: 0.36, CI: 0.30-0.44; 6. Jarnfelt-Samsioeet al. pregnancy and those who did not (regardless of 1983,N = 948 pregnanciesfrom 244 women (analysis nausea). of outcomesbased on 911 pregnancies),OR: 0.41, CI: 0.24-0.70;7. Weigeland Weigel1989a, N = 903, OR: 0.25,CI: 0.12-0.52.All studies compared preg- Weigel (1989a) found no significantdiffer- nancyoutcomes for women who did and did not ences in frequencies of severe anomalies be- experienceNVP exceptMedalie (1957), who distin- tweenwomen who did and did not experience guishedbetween "none/mild" NVP versus "moder- NVP, Saxen (1975) and Golding et al. (1983) ate/severe"NVP. found no relationshipsbetween NVP and cleft lip or cleft palate, and Klebanoff and Mills (1986) found no relationship between NVP 3. Neonatal survival.Yerushalmy and Mil- and any birth defects. Kullander and Kalln kovich (1965) reported that babies born to (1976) reportedthat women who experienced women who had experienced NVP were signif- NVP were significantlymore likelyto bear chil- icantlymore likelyto survive2 1 month than dren withsevere anomalies than women who babies born towomen who had notexperienced did not experience NVP, and Kricker et al. NVP, but Brandes (1967) and Chin (1989) (1986) found that women who vomited dur- found no association between NVP and neo- ing pregnancy were significantlymore likely natal mortality. to bear childrenwith limb defectsthan women 4. Congenital anomalies. Yerushalmy and who did not vomit. Milkovich (1965) reported that women with FOOD AVERSIONS AND CRAVINGS NVP were less likelyto bear children with"severe" anomalies (e.g., disruptionsof major or- Frequency gan systems)than women who did not experi- We summarized the resultsof 20 studies of ence NVP. However, Milkovich and van den food aversions during pregnancy that in- Berg (1976), Petitti (1986), and Weigel and cluded informationfrom 5,432 women, and

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions JUNE 2000 MORNING SICKNESS 125

0.08 - MISCARRIAGES (a) 100 women, 62 experienced such aversions. Schwab and Axelson (1984) followedDickens 0.06 - and Trethowan's protocol and definitions, and obtained similarresults: of 60 women they 0.04- interviewed,37 (62%) experienced gestational aversions.A numberof additional studieshave t 0.02- reported connections between gastrointestinal distressduring pregnancyand avoidance Q 0.00 of the offendingfoods (e.g., Nobmann and ' 0.20 Adams 1970; Hook 1976, 1978; Ojofeitimiet FETAL DEATHS (b) al. 1982; Finley et al. 1985; Brown and Toma 1986; Al-Kanhal and Bani 1995). ', 0.15- Foods avoidedor cravedthroughout pregnancy , 0.10- Pregnantwomen most oftenreported aversions to "meat,fish, poultry, and eggs" (Figure 0.05- 7). Per capita aversionsto these animal products (0.28/woman) were nearlydouble those 0.00 of thesecond mostaversive food category,"non- NVP absent Nauseaonly Vomiting alcoholic beverages" (0.16 aversions/woman), and more than triplethe aversionsto "vegeta- Symptoms bles" (0.08/woman). Per capita aversionsto "alcoholic beverages"and "ethnic,strong and spicy FIGURE 6. ASSOCIATION OF NVP SYMPTOMS WITH MISCARRIAGES AND foods" were only 0.04/woman, and aversions FETAL DEATHS. to "dairyand ice cream,"and "sweets,desserts, Increased severityof NVP symptomsare signifi- and chocolate" were even less frequent.Aver- cantlyassociated withdecreased chances of (a) mis- sions to "grainsand starches"and "fruitand carriages(P < 0.001: Weigel and Weigel 1989a), and fruitjuice"were veryrare (< 0.02/woman). (b) fetaldeaths (P < 0.001: Tierson et al. 1986). Data In contrast,pregnant women most oftenre- wereextracted from the original papers and replotted. ported cravingsfor "fruit and fruitjuice"(0.20 cravings/woman) and "sweets,desserts, and chocolate" (0.17/woman). Per capita cravings were slightlyless common for "dairyand ice studies of food of women 21 cravings 6,239 cream" (0.12/woman), "meat, fish, poultry, (see Table 1). The majorityof women (65% + and eggs" (0.12/woman), and "grains and at least aversion 15) experienced one during starches"(0.08/woman). Pregnantwomen sel- ? 15 pregnancy;similarly, 67% experienced dom craved "vegetables"(0.06/woman), "eth- at least one craving. Corresponding median nic, strong and spicy foods" (0.04/woman), proportionswere 66% for aversionsand 68% and "nonalcoholic beverages" (0.03/woman). for cravings.Cravings were thus slightlymore Very few pregnant women craved alcoholic common than aversions (P < 0.05). beverages (< 0.01/woman). Interestingly,patterns of gestational crav- Ontogeny ings and aversions were virtuallymirror im- Food aversionsexpressed duringpregnancy ages (Figure 7). Per capita aversionswere sig- were apparentlynovel developments in wom- nificantlygreater than cravingsfor "meat, fish, en's attitudestoward particularfoods, rather poultry,and eggs," "nonalcoholic beverages," than exaggerationsof preexistingdislikes. For "vegetables,"and "alcoholic beverages" (P < example, Fairburnet al. (1992) found that80 0.001 forall), whereasper capita cravingswere of 100 women theyinterviewed experienced significantlygreater than aversions for "fruit gestationalfood aversions,and all 80 reported and fruitjuice,""grains and starches,""sweets, that theiraversions began during pregnancy. desserts,and chocolate," and "dairy and ice In an earlier study,Dickens and Trethowan cream" (P < 0.001 for all). Only for "ethnic, (1971) consideredas gestationalaversions only strong and spicy foods" were cravings and those that began during pregnancy.Among aversionsequally common (Figure 7, middle).

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions 126 THE QUARTERLY REVIEW OF BIOLOGY VOLUME 75

bp ob ob on on on on on on o- on on on o o o o < * v) *) s ~ s-< -< s4 s- sH abonXH a; X

>= Q AX Q A#Q@Q@@ @n Q A I2 * rrrrrr <> : : :::: : ::

- E n sn- =^ : tn

.to~~~~~~~~~~~Ec

>~~~~~b n

> i t ' 3 ' 2 S 4t - - G > F O t i ' i D E . X

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions JUNE 2000 MORNING SICKNESS 127

j 0.30 o Aversions

H Cravings

; 0.20-

0.10 0

o NS ***

< 0.00 Meat N-A Veg Alc ESS D S G&S F Food Category

FIGURE 7. FOOD AVERSIONS AND CRAVINGS OF PREGNANT WOMEN (ALL TRIMESTERS). Numbers on the verticalaxis representthe average number of aversionsor cravingseach woman in the sample had to each food category.These numbers were calculated by dividing the total number of aversions or cravingsto each categoryby the total number of women in the sample. Aversionswere based on 20 studies thatincluded 5,432 women, and cravingswere based on 21 studies thatincluded 6,239 women (see Table 1). Here *** indicates P < 0.001, and NS indicates no significantdifferences. Food category abbreviations:"Meat" is meats,fish, poultry, and eggs, "N-A"is nonalcoholic beverages,"Veg" is vegetables, "Alc" is alcoholic beverages,"ESS" is ethnic,strong and spicyfoods, "D" is dairyand ice cream, "S" is sweets, desserts,and chocolate, "G&S" is grains and starches,and "F" is fruitsand fruitjuices.

Foodsavoided in each trimester of seven food categories, namely "meat, fish, Per capita aversions to all food categories poultry,and eggs," "vegetables," and "non- were highestearly in pregnancy,and declined alcoholic beverages." Declines in per capita dramaticallythereafter (Figure 8). Aversions aversions between the successive trimesters were significantlymore frequent in the first thuswere mostconsistent for these threefood trimesterthan in the second trimesterfor six categories. Interestingly,cravings for animal of seven food categories (all P < 0.05); aver- products and vegetables also declined signifi- sions to the odd category ("vegetables") also cantly between the firstand third trimesters were lower in the second trimester,but the (data not shown). decline was only marginallysignificant (P = Rodin and Radke-Sharpe(1991) monitored 0.06). Aversions were significantlymore fre- dietarypreferences of women in the firsttri- quent in the firsttrimester than in the third mesterand, simultaneously,nonpregnant "con- trimesterfor six of seven food categories(all P < trol"women. The pregnantwomen exhibited 0.05); aversionsto theodd category("grains and more aversionsper capita than controls to all starches") also were lower,but the decline was seven food categories (Figure 9). Six of these only marginallysignificant (P = 0.07). Aver- seven comparisonswere highlysignificant (P < sions were significantlygreater in the second 0.01), and theseventh ("grains") was marginally trimesterthan in the thirdtrimester for three significant(P = 0.07). In general, food aver-

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions 128 THE QUARTERLY R,VIEW OF BIOLOGY VOLUME 75

0.60 1-

nn U lst1Trimester 0.50 * 2ndTrimester 0 ED3rd Trimester 0 0.40- L

Os 0.30 nn n - - ~~~~~~~~~~~~NSF 0.20 i n nnNi f N nfl* NSn 0.10 0

0.00 Meat N-A Veg G&S S D F Food Category

FIGURE 8. CHANGES IN FOOD AVERSIONS DURING THE THREE TRIMESTERS OF PREGNANCY. Based on interviewsby Maclntyre (1983, N = 50 women) and Rodin and Radke-Sharpe (1991, N = 80 women). Food categoryabbreviations and calculationsof per capita numbersof aversionsare as in Figure 7. Significantdifferences are indicated as follows:+ 0.05 < P < 0.1, * P < 0.05, ** P < 0.01, *** P < 0.001, NS = not significant.

sions were rare (< 0.10/woman) among con- and Weiher. Among these 27 societies (Table trols."Meat, fish,poultry, and eggs" were the 2), morningsickness was observedin 20 (74%). mostcommon targetsof aversionsin nonpreg- For the otherseven, original investigators made nant women, in agreement with previous re- specific statementsto the effectthat theydid ports (e.g., Midkiffand Bernstein 1985; De not observe any symptomsof morning sick- Silva and Rachman 1987; Mattes 1991). How- ness. These seven cultures are not clustered ever, aversions to these animal productswere geographically:they are located on fiveconti- more than twiceas frequentamong pregnant nents (Table 2). Of course, we cannot rule out women (0.78/woman) as among controls the possibilitythat morning sickness actually (0.30/woman). occurred in these societiesbut was undetected. However, HRAF sources gave us no reason to NVP IN TRADITIONAL SOCIETIES suppose thatobservations of these seven socie- Minturnand Weiher (1984) claimed thatin- ties were any less complete or accurate than formation on the occurrence of "morning observationsof the 20 societies in whichmorn- sickness" was available in the Human Rela- ing sickness was recorded. Therefore, we ac- tionsArea Files for30 traditionalsocieties. We cepted the HRAF data at face value. carefullychecked the HRAF fordata on these For the 27 societies in our sample, we tabu- 30 societies, but were able to locate relevant lated (Table 2) all foods listedas "staples" (i.e., informationfor only 26. We also located data foods eaten everyday). Meat was a staple in 14 on one societythat was not cited by Minturn societies (52%), milk in two (7%), corn in 10

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions JUNE 2000 MORNING SICKNESS 129

0.80 *** 1StI Trimester

0.60 Controls

p 0.40

0.20 **

0.00 Meat N-A Veg G S D F Food Category

FIGURE 9. FOOD AVERSIONS OF FIRST TRIMESTER PREGNANT WOMEN COMPARED TO AVERSIONS OF NONPREGNANT WOMEN. Data are from Rodin and Radke-Sharpe (1991 ), based on concurrent surveys of 80 women who were in theirfirst trimester of pregnancyand 80 nonpregnant "control"women. Food categoryabbreviations and calculation of per capita numbers of aversions are as in Figure 7; statistical results are indicated as in Figure 8.

(37%), ricein 6 (22%), and otherplants (mostly some lab colonies, thischange in appetite is a tubers) in 19 (70%). Societies in which morn- reliable indicator of pregnancy (Lewis et al. ing sickness was not observed were equally 1987; D M Bebiak, personal communication). likelyto have plants as dietarystaples as socie- Captiverhesus macaques (Macaca mulatta)also ties in which morning sickness was observed often exhibit a decrease in appetite during (Table 3). Compared to societies in which weeks 3 to 5 of their23-week gestation (Czaja morningsickness was observed,however, soci- 1975). Their appetite loss is accompanied by eties in which it was not observed were signifi- hormonalchanges thatparallel those occurring candy more likelyto have (i) onlyplants as sta- in women during the firsttrimester of preg- ples, (ii) corn as a staple, and (iii) corn as the nancy. Finally,for captive chimpanzees (Pan onlystaple, and theywere (iv) significantlyless troglodytes),Keeling and Roberts (1972:145) likelyto have meat as a staple, and (v) slightly noted that among the "subtle factors" that less likelyto have rice as a staple (Table 3). mightstrengthen a questionable diagnosis of NVP IN OTHER MAMMALS pregnancy, "[t]he pregnant female may ini- We searched widelyfor information on NVP tiallyexperience morningsickness and irregu- in nonhuman mammals. We found only sug- laritiesin appetite." The authors provided no gestiveevidence for three species. Female do- furtherinformation or references,and we were mestic dogs (Canisfamiliaris)typically exhibit unable to locate any other mention of morn- a sharp drop in food consumption during ing sickness in wild or captive chimpanzees weeks 3 to 5 of their 9-week gestation (Lewis (e.g., in Boume 1970-73; Goodall 1986; Wrang- et al. 1987; Bebiak 1988; Jackson 1995). In ham et al. 1994).

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions 130 THE QUARTERLY REVIEW OF BIOLOGY VOLUME 75

TABLE 2 Culturesfor whichinformation on NIP zvasavailable in Human RelationsArea Files Culture Location Staples (eaten everyday) NVP? Meat Milk Corn Rice Other Plant'

Alor Oceania X Yes Aranda Australia X X Yes Bhil Asia X X No Burmese Asia X X X Yes Cuna South America X X Yes Garo Asia X X X Yes Goajiro South America X X X Yes Gond Asia X X Yes Hottentot Africa X X X Yes Ifugao Oceania X X X Yes Kaska North America X Yes Mbundu Africa X No Okinawa Asia X X X Yes Omaha2 North America X No Papago North America X No Pukapuka Oceania X X Yes Siriono South America X X No Tallensi Africa X X Yes Tarahumara North America X No Terena South America X Yes Tonga Oceania X X Yes Toradja Oceania X X Yes Trobriand Oceania X Yes Truk Oceania X X Yes Wogeo Oceania X Yes Woleai Oceania X No Zulu Africa X X X Yes ' The more common "other plant" staples were taro, plantains, millet, breadfruit,sweet potatoes, cassava, and other tuberous plants. 2 In addition to HRAF, Fletcher and La Flesche (1972) was also used for informationabout the Omaha.

DISCUSSION The embryoprotection hypothesis also pre- Hook (1976, 1978, 1980) and Profet (1988, dictsthat women who experience NVP should 1992, 1995) hypothesized that NVP serves a have positive pregnancy outcomes more fre- useful function-protecting the embryo-by quently than women who do not experience causing pregnant women to physicallyexpel NVP. In support of thisprediction, in all nine and subsequentlyavoid foods thatcontain ter- ofnine studies,women who experienced "nor- atogenic, mutagenic,and abortifacientchemi- mal" NVP (i.e., excluding hyperemesisgravi- cals. This hypothesispredicts that NVP should darum) were significantlyless likely to mis- coincide withthe greatestsusceptibility of the carrythan women who did not (Figure 4). In embryoto developmental disruption.Consis- two additional studies, fetal deaths (miscar- tentwith this hypothesis, NVP symptomspeak riages plus stillbirths)were significantlyless during the firsttrimester, which is a critical likelyto occur in women who experiencedNVP period in organogenesis (Figure 3). During (Figure 5), and pregnantwomen who vomited gestational weeks 6 to 14, the embryo is es- were less likelyto suffermiscarriages and fetal pecially sensitiveto disruptionby exogenous deaths thanwomen who feltnauseated but did chemicals because cell division,cell differenti- not vomit (Figure 6). The latterresult may im- ation, and morphogenesis of multiple organ plicate physical expulsion of certain foods in systemsare occurringsimultaneously. enhancing pregnancyoutcomes.

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions JUNE 2000 MORNING SICKNESS 131

TABLE 3 Chiaracteristicsof diets in traditionalsocieties in relationto thepresence or absenceof NVP Dietary Number of societies Number of societies G Significance characteristic with NVP (N = 20) without NVP (N = 7) statistic level [fraction (proportion)] [fraction (proportion)] Plantsare staples 18/20 (0.90) 7/7 (1.0) 0.928 Notsignificant Onlyplants are staples 7/20 (0.35) 6/7 (0.86) 5.337 P < 0.05 Meatis a staple 12/20 (0.60) 1/7 (0.14) 4.388 P< 0.05 Cornis a staple 4/20 (0.20) 6/7 (0.86) 9.063 P < 0.005 Corn is the onlystaple 2/20 (0.10) 5/7 (0.71) 8.586 P< 0.005 Rice is a staple 6/20 (0.30) 0/7 (0.0) 3.709 0.05 < P< 0.1

In contrastto the strongand consistentneg- ing these chemicals do not have such geno- ative associations between NVP and miscar- toxic effects. riage-a typicalconsequence of developmental In contrast,however, Brown et al. (1997: disruptionsthat occur earlyin embryonicor- 181) statedthat " [s] peculation offeredin a re- ganogenesis-NVP was not reliablyassociated cent popular-press book [Profet 1995] that withpregnancy outcomes in later fetal devel- links intake of certain vegetables and other opment or postpartum.Whether women ex- foods to nausea and vomitingof early preg- perienced NVP or not,they were equally likely nancy and to adverse pregnancyoutcomes is to sufferstillbirths and neonatal mortality. not supported by thisstudy. We conclude that The frequencyof NVP (Figure 1), its timing intake of pungent and bittervegetables and relativeto criticalperiods in embryonicorgano- other proscribed foods is not hazardous to genesis (Figure 3), and the positivepregnancy pregnantwomen." NeitherProfet (1988, 1992, outcomes associated with"normal" NVP (Fig- 1995) nor Brownetal. (1997) quantifiedwhat ures 4-6)-especially when juxtaposed with foods triggeredNVP or resultedin pregnancy- the absence of negative outcomes-together related aversions.We did so. Among the food imply that NVP indeed serves a protective categorieswe recognized (Figures 7-9), "non- function. alcoholic (caffeinated) beverages," "vegeta- What exactlydoes NVP protectthe embryo bles," and "ethnic,strong and spicyfoods" all from?Hook (1976, 1978) focused on the ter- include items that contain potentiallyterato- atogens in alcohol, caffeinatedbeverages and genic phytochemicals. If we therefore com- tobacco. Profet(1992, 1995) presenteda more bine theminto a single super-category(Figure comprehensive list of potentiallydangerous 10), theyindeed account for a large number foods,and recommendedthat pregnantwomen of per capita aversions (0.29/woman)-as "protecttheir baby-to-be" by avoiding vegeta- manyas to meat products.Also consistentwith bles thatcontain teratogenicphytochemicals. Profet'sversion of the embryoprotection hy- Consistent with Profet's hypothesis,in large pothesis,pregnant women reported (i) signif- quantities phytochemicalsin many common icantlymore aversionsthan cravingsto "vege- vegetables and caffeinatedbeverages are mu- tables" and "nonalcoholic beverages" (Figure tagens, teratogens, abortifacients,and aller- 7), (ii) significantlymore aversionsto "vegeta- gens (e.g., Pieters 1982; Nagao et al. 1986; bles" and "nonalcoholic beverages"in the first Ames et al. 1990a,b; Friedman et al. 1991; trimesterof pregnancythan in the second or Shepard 1992; Beier and Nigg 1994; Klebanoff third trimesters(Figure 8), and (iii) signifi- et al. 1999). For example, crudejuice extracts candymore aversionsto "vegetables"and "non- from plants in the genus Brassica (e.g., cab- alcoholic beverages" than did nonpregnant bage, Brussels sprouts) contain isothiocya- women (Figure 9). nates and other breakdown products of glu- Hook's (1976) version of the embryo pro- cosinolates that can induce chromosomal tection hypothesisalso correctlypredicted a aberrations in mammalian cells (Kassie et al. subsetof pregnancy-related aversions. "Nonal- 1996). Extractsfrom vegetables not contain- coholic (caffeinated) beverages" was consis-

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions 132 THE QUARTERLY REVIEW OF BIOLOGY VOLUME 75

NS 0.3 F 1

0O.2-

~0.1

Meat N-A, Alc D S G&S F Veg, ESS Food Category

FIGURE 10. FOOD AVERSIONS OF PREGNANT WOMEN (ALL TRIMESTERS). This figurereplots the data in Figure 7, combining the three categories that include foods containing phytochemicals (e.g., "nonalcoholic beverages," "vegetables,"and "ethnic,strong and spicyfoods"). Per capita frequencies of aversions to meats and these phytochemical-containingfoods do not differsignifi- cantly,but each of these twocategories is a targetof aversionssignificantly more frequentlythan any other category (*** indicates P < 0.001).

tentlythe second (Figure 7) or third (Figures cies) and Stewartet al. (1988, N = 242) re- 8, 9) most aversivefood category,and aver- ported that per capita aversions to alcohol sions to "nonalcoholic beverages"were signifi- were almost as common as aversionsto "meat, cantly higher in the firsttrimester of preg- fish,poultry, and eggs" and "nonalcoholic bev- nancy than in the second or thirdtrimesters erages," and significantlymore common than (Figure 8). Contraryto Hook, however,aver- aversionsto "vegetables."In contrast,Tierson sions to alcoholic beverages were much less et al. (1985, N = 400) and Finleyet al. (1985, common (Figure 7). This is surprising,be- N = 60) found that gestational aversions to cause alcohol is a well-knownteratogen (Kauf- alcoholic beverageswere significantlyless com- man 1997). It is possible thatsome researchers mon than aversionsto "meat,fish, poultry, and did not consider alcohol to be a "food," and eggs" or "nonalcoholic beverages,"and about thereforedid not question women about its as frequent as aversions to "vegetables." Fur- use. Additionally,some women mayhave been thermore,Dickens and Trethowan (1971, N = reluctantto reveal informationabout alcohol 100), Schwab and Axelson (1984, N = 60), consumption during pregnancy. and Alberti-Fidanzaet al. (1996, N = 86) re- Several investigatorsdid include "alcoholic ported that aversions to alcoholic beverages beverages" in their questionnaires, however. were relativelyinfrequent (< 0.1 aversionsper Among them,Hook (1978, N = 250 pregnan- woman), and Harries and Hughes (1958, N =

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions JUNE 2000 MORNING SICKNESS 133

509), Baylis et al. (1983, N = 42), and Knox the association between animal products, (1993, N = 100) found thataversions to alco- food aversionsand NVP. First,digestive break- holic beverages were extremelyrare (< 0.05 down of animal products could conceivably aversionsper woman). Most reportsthus offer create or release teratogenicor abortifacient littlesupport for Hook's hypothesisas applied substances. Categories of compounds associ- to alcoholic beverages. ated with meat digestion include dipeptides, Also apparently contraryto Hook (1976), tripeptides,triglycerides, phospholipids, ster- only two studies reported aversions to ciga- ols, amino acids, uric acid, and the fat-soluble rettesmoke (Tierson et al. 1985; Fairburn et vitaminsA, D, E and K (Davenport 1982:205- al. 1992). However, the relationshipbetween 210;Johnson1997). Among these compounds, NVP and tobacco use is complicated. Among only vitamin A (retinoic acid) is known to women who smoked before conceiving,those cause birthdefects (Schardein 1985; Keen et who experienced NVP were more likelyto quit al. 1993), and thisoccurs onlyif it is ingestedin than those who did not experience NVP (Wol- pharmacological doses daily (that is, > 15,000 kind and Zajicek 1978; Meyer et al. 1994; but IU) over several months of gestation (Hath- see Paarlberg et al. 1996). This suggests that cock et al. 1990; Rothman et al. 1995). Thus, NVP contributed to an aversion to smoking. constituentsof meats in a normal diet are un- If so, it provides a novel explanation for the likelyto endanger a developing embryo. frequentlyreported, but heretoforeparadoxi- The second possibilityis that certain culi- cal, association between smoking and the ab- narypractices (e.g., frying,broiling and smok- sence of NVP (e.g., Wolkind and Zajicek 1978; ing) create cooking mutagens (Profet 1992) Little and Hook 1979; Vellacott et al. 1988; and introduce phytochemicals,such as those Meyer et al. 1994; Gadsby et al. 1997; but see in spices (Billing and Sherman 1998). Before Palmer 1973; Klebanoffet al. 1985; Paarlberg the widespreadavailability of refrigeration(i.e., et al. 1996). Alternatively,it is possible that before the 20th century), these practices- chemicals in cigarette smoke reduce NVP along with heavy salting-were the primary symptomsby interfering with their underlying means of preservinganimal products.By avoid- neuroendocrine mechanisms (e.g., Bernstein ing roasted,burned, smoked or heavilyspiced et al. 1989), or by damaging placental cells meats, pregnant women could minimize the that may be responsible for triggeringNVP exposure of their embryoto toxic chemicals. (e.g., Langhan's cells: Sastryet al. 1989; Arn- But exposure to bacteria or fungi and their holdt et al. 1991). Placental cell damage also toxins is a serious potential cost of eating could explain whymothers who smoke are less meats thatare not thoroughlycooked, spiced likely than nonsmokers to experience pre- or salted (Sherman and Billing 1999). eclampsia (Klonoff-Cohenet al. 1993; Cnat- This suggeststhe thirdand most likelyrea- tinguiset al. 1997). son whypregnant women benefitfrom avoid- Undoubtedly our most intriguingfinding is ing animal products: to minimizeexposure to thatpregnant women consistentlyfind "meats, foodborne illnesses and food poisoning. Raw fish,poultry, and eggs" to be more aversive meats and meat dishes that are prepared in than any other single food category (Figures advance and stored at ambient temperatures 7-9). This was not predicted by Hook (1976) formore than a fewhours, especiallyin tropi- or Profet (1988). Interestingly,Hook (1978: cal climates, typicallyshow massive increases 1360) reported resultssimilar to ours, but he in bacterial and fungal content (Bryan et al. termed them "unexpected" in lightof his em- 1979; Michanie et al. 1988; Hobbs and Roberts phasis on alcohol, tobacco and caffeine.Pro- 1993). Meats are more dangerous thanvegeta- fet (1992:346) mentioned the potential dan- bles because meats spoil faster (an animal's gers of spoiled or burned meats, but her immune systemceases functioningat death), argumentsfocused primarilyon plant toxins, and thusmeats are more oftenassociated with and no animal productswere included in her foodborneillnesses (Bryan 1988; Roberts1990; list of "worstfoods to eat during the firsttri- Todd 1994, 1996; Sockett 1995). Vegetables mester" (1995:151). and plant seeds are colonized less rapidly, There are three possible explanations for partlybecause cellulose and ligninare not eas-

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions 134 THE QUARTERLY REVIEW OF BIOLOGY VOLUME 75

ilybroken down by aerobic bacteria, and also phoid fever,-smallpox, coccidioidomycosis, because of the antimicrobial and antifungal malaria (P. falciparum),and influenza (Brabin properties of the phytochemicalsthat many 1985). Second, viral,fungal and protozoan in- plants contain, including protease inhibitors, fectionsare generallymore severeduring preg- chitinases, glucanases, phenolics, and ribo- nancythan similar infectionsexperienced by some-inactivatingproteins (Vigers et al. 1991; nonpregnant women (Gennaro and Fehder Darnettyet al. 1993). 1996), including influenza,cholera, measles, There is a special reason why pregnant varicella,diphtheria, scarlet fever, gonorrhea, women maybenefit from avoiding meat prod- giardiasis, and babesiosis (Brabin 1985). As ucts: their immune system is suppressed one example, Toxoplasmagondii-a common, (Formby 1995; Gennaro and Fehder 1996; foodborne protozoan parasite,often acquired Matthiesen et al. 1996). Although a woman's by handling or eating raw or undercooked humoral immune response is unaltered or meat (Kapperud et al. 1996)-rarely causes possibly strengthened during gestation, her toxoplasmosis in immunocompetent individ- cell-mediated immune response is weakened uals (Smith 1997). Toxoplasmosis is a serious (Bisset et al. 1990; Formby1995; Sabahi et al. riskfactor for individuals with compromised 1995; Blumberg and Heal 1996; Lim et al. cell-mediatedimmunity, however, such as preg- 1996). Numerous investigatorshave reported nantwomen (Smith 1997). thatthe activitiesof natural killercells are de- Maternal immunosuppression also creates pressed during pregnancy (e.g., Barrettet al. risksfor the embryo. Miscarriages and birth 1982; Gonik et al. 1987; Salm6ron et al. 1991; defectscan resultif, in the firsttrimester, the Haig 1993; Beer et al. 1996; Gennaro and woman contractsa febrileillness (Shaw et al. Fehder 1996; Matthiesen et al. 1996; but see 1998), includingviruses (Dickinson and Gonik Hidaka et al. 1991; Opsahl et al. 1994). This 1990; Kurppa et al. 1991), respiratoryinfec- occurs in some nonhuman mammals as well tions(Kricker et al. 1986), and influenza(Sax6n (Chaouat and Menu 1997). Temporaryimmu- 1975; Lynberget al. 1994). Foodborne patho- nosuppression is essential for a successful gens can also directlyaffect the viabilityof the pregnancy;if the immune systemfunctioned embryo. For example, gestational toxoplas- normally,the mothermight reject her own off- mosis has been linked to congenital neurolog- spring,whose cellular phenotypesare foreign ical birth defects (Fichera and Roos 1997), because halfof theirgenotype is paternallyde- spontaneous abortions,neonatal diseases, and rived (Haig 1996). Indeed, women who do not ocular defects (Holliman 1995). Dickinson exhibitnormal decreases in levels of decidual (1994) has estimated that 3% of congenital killercell activitymore frequentlysuffer recur- deformitiesresult from gestational infections. rentspontaneous abortions (Chao et al. 1995; In viewof all these dangers,we maintainthat Beer et al. 1996; Christiansen1996). the embryoprotection hypothesis must be ex- Impairment of killer cell activityreduces a panded. We suggest that NVP protects both pregnantwoman's innate immunityto bacte- the mother and her embryo from infections ria, viruses,and tumor cells (i.e., her "firstre- by foodborne microorganismsand poisoning sponse"; Blumberg and Heal 1996), as well as fromtheir toxins, in addition to protectingthe her antibody-dependent,cell-mediated cyto- embryofrom teratogens and abortifacientsin toxicityresponse (i.e., killingof cells that are the mother's diet. This expanded hypothesis labeled withantibodies; Janeway and Travers would explain the high frequencies of aver- 1994) . As a result,immunosuppression during sions to "meats, fish,poultry, and eggs" (Fig- gestationcan have importantnegative conse- ure 7) in pregnant women, especially during quences (Formby1995). First,pregnant women the firsttrimester of pregnancy.Decreases in are at greaterrisk of seriousillnesses and death these aversions as pregnancyprogresses (Fig- fromfoodborne and waterborneenteric micro- ure 8) are consistentwith increasing maternal organismsthan nonpregnant women (Gerba and embryoprotection. In the firsttrimester, et al. 1996). Pregnant women have higher fa- embryonictissues are so vulnerable to disrup- talityrates than nonpregnant women from tion (Figure 3c) thateven fat-and protein-rich certain hepatitis strains, amoebic colitis, ty- foods are rejected. As pregnancy progresses,

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions JUNE 2000 MORNING SICKNESS 135 however,nutrient requirements increase and The maternal and embryo protection hy- fetal vulnerabilitydecreases, so the possible pothesis predicts that the frequency of NVP costs to the woman and fetusof ingestingani- should be affectedby the characteristicdiet of mal products are exceeded by the benefitsof a population. There were wide variations in consuming these nutrient-densefoods. frequenciesof NVP among countries (i.e., 35- It is puzzling thatwomen should ever crave 84% in 16 countries;Figure 1) and subpopula- animal products during the first trimester tions. In the United States alone, the range in (Figure 7). Perhaps in some populations these NVP frequencies was 27-89% in 22 samples. foods are the principal or onlysources of pro- The frequencyof NVP in the United Kingdom tein,fats, iron, folate and other vitamins,and wassignificantly higher than in theUnited States other kinds of nutrients.Adequate proteinin- and elsewhere.We triedto relate these differ- take is crucial to normal development (espe- ences to diet, i.e., by determining whether cially later in pregnancy); low protein diets study subjects in the United Kingdom con- have been linked to decreased placental and sumed more meats or more pungent vegeta- birthweights (Campbell et al. 1996; Godfrey bles than subjects in the United States or else- et al. 1996). Also,fresh meat thatis boiled thor- where. Appropriate dietaryinformation was oughly(so no teratogensare produced via cook- unavailable, however. ing), and to which no spices are added, would Dietary data were available in the Human pose littlethreat to a woman or her developing Relations Area Files for the 27 nonindustrial embryo.In fact,sterilized meat productsprob- cultures with informationon NVP. Minturn ably contain fewer dangerous compounds and Weiher (1984) reported that societies in than manyvegetables (Profet 1992, 1995). which NVP symptomswere not observed had We have argued thatNVP causes the expul- corn as a staple food significantlymore often sion of foods thatcontain microorganismsand than societies where NVP occurred. They also toxic chemicals,and also encourages pregnant reported thatsocieties with no symptomscon- women to eliminate the offendingfoods from sumed significantlymore fatsand green vege- theirdiet. Nausea and vomitingdo cause indi- tables, and that NVP was more prevalent in viduals to subsequentlydislike the foods that the Insular Pacific (91 % of the societies) than triggeredgastrointestinal distress (Milgram et in their overall sample (73%). Minturn and al. 1977; Pelchat and Rozin 1982; Rozin and Weiher found no associations between NVP of Vollmecke1986), and theadaptive value such and altitude, latitude, longitude, the type or food aversion learning is obvious (Letarte et intensityof agriculture,presence of food ta- al. 1997). Consistent with this hypothesized boos, importanceof women in the subsistence "prophylactic"function, multiple studies have economy, settlementpatterns, or the sizes of reporteda connectionbetween NVP and learn- communities. ing food aversions(Nobmann and Adams 1970; Our results (Tables 2 and 3) extend those Hook 1976,1978; Ojofeitimiet al. 1982; Finley of Minturnand found that et al. 1985; Al-Kanhaland Bani 1995), and be- Weiher (1984). We tween aversionsand reduced consumption of the seven societies in which NVP symptoms those foods (Tierson et al. 1985; Brown and were not observedwere significantlyless likely Toma 1986). For example, Rodin and Radke- to have meat as a dietarystaple, and signifi- Sharpe (1991:334) commented that"[f] or the cantlymore likelyto have onlyplants as dietary most part, the food groups that were more staples, than the 20 societies in which symp- commonlyavoided among pregnantthan non- toms occurred. Moreover, societies without pregnantwomen were stronglyrelated to the NVP were significantlymore likelyto have corn symptomsof NVP"; Finley et al. (1985:684) as a dietarystaple (and corn as the onlystaple) statedthat " [t]he mostcommon aversionswere than societies withNVP. Of course our analy- directed towardvegetables, usually onions or ses, as well as those of Minturn and Weiher membersof the Brassicafamily, strongly spiced (1984), may be questioned because societies or strong-smellingmixed dishes (usually of are not independent data points-owing to Chinese, Italian,or Mexican ethnicorigin), or possible recent common ancestryor diffusion 'greasy'foods. Usuallythe foods were reported of dietarypreferences. Independence is statis- aversivebecause theyprovoked nausea." ticallydesirable but, as discussedby Ember and

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions 136 T7HE QUARTERLY REVIEW OF BIOLOGY VOLUME 75

Otterbein (1991) and Mace and Pagel (1994), geophagy. Clay eating is widespread in tradi- independence of specificcultural practices is tionalsocieties. Many types of claycontain com- oftenimpossible to assess.Use ofcladistic meth- pounds that (i) coat the mucous membranes ods to inferindependence of cultureshas been of the digestivetract and protect them from suggested, but as pointed out by several re- damage by toxins,and (ii) bind and prevent spondents to Mace and Pagel (1994:557-564) absorption of ingested phytochemicals,mi- as well as Mace and Pagel themselves (1997: crobial toxins,and microorganisms(Vermeer 305), it maybe inappropriateto infercultural and Ferrell 1985; Phillips et al. 1990; Johns "phylogenies"by applying maximum parsimony and Duquette 1991a). Even acorns and wild techniques thatwere developed forinvestigat- potatoes can be made edible by preparing ing evolutionaryhistories. Hartung (1997:347) them withcertain clays (Johnsand Duquette argued thatvariations in independence among 1991b; Diamond 1999). cultures are "like noise in a signal . . . more If absence of NVP is associated withlack of likely to obscure true relationships than to dietarytoxins, rather than nutritionaldeficien- generatefalse ones," and Otterbein(1994:559) cies, then in culturesexhibiting NVP women stated he "certainlywould not cease using should be able to reduce symptomsthrough worldwide samples in comparative research practices that remove or detoxifydietary te- because of the alleged difficultiesthat arise ratogens. In support, Wiley and Katz (1998) fromthe nonindependence of cases." Follow- found thatwomen in manytraditional African ing his advice, we analysed (Table 3) all 27 societiespractice geophagy in earlypregnancy culturesfor which informationon NVP symp- to relieve NVP symptoms.They also reported toms were available in the HRAF. that geophagy was more prominent in non- There are twopossible explanations forthe dairyingsocieties, and that thiswas likelydue HRAF results.First, pregnant women in corn- to the higher levels of phytochemicalsin the based societies may rarelyeat foods that trig- diets of these societies. ger NVP and subsequent aversions. Domesti- Geophagy also has been shown to serve a cated corn is bland-tasting,and some strains protectivefunction in nonhumans. Chimpan- have minimal (Darnetty et al. 1993) or no zees in Uganda and Tanzania practice geoph- chemical resistanceto bacteriaand fungi(Guo agy, and Mahaney et al. (1996, 1997) found et al. 1998). This does not mean that corn is that the soils theyeat have a high content of alwayssafe, because some corn-dwellingspoil- metahalloysite,a partiallyhydrated clay min- age microorganismsproduce teratogens(e.g., eral with antidiarrheal properties like the Penicillium,Fusarium and Aspergillusspp.; Reddy pharmaceutical Kaopectate T. Mahaney et al. and Reddy 1993). However, fresh cultivated suggestedthat the chimpanzees were medicat- corn and unspoiled, dried corn contain mini- ing themselvesin response to gastrointestinal mal teratogens.The alternativehypothesis is distress.Geophagy also occurs in many other that corn-based diets are nutritionallydefi- herbivorousmammals and seed-eating birds; cient in some waythat specifically disrupts the in parrots at least, it reduces absorption of physiologicalmechanisms underlying NVP. For toxic phytochemicals into the bloodstream example, corn contains so littlelysine, trypto- (Gilardi et al. 1999). These comparativeobser- phan and niacin that malnutritionwould oc- vations support the hypothesisthat it was the cur in societies with corn-based diets, were it absence of toxins,rather than nutritionaldefi- not foralkali processing (Katz et al. 1974). Of ciency, that eliminated NVP in certain tradi- the corn-based societies in which NVP symp- tional societies (Table 2). toms have not been observed,information on A more decisive testof the alternativeexpla- food preparation was available only for two: nationsfor the HRAF data would involvecom- the Tarahumara treatcorn withalkali, but the paring miscarriage rates. The maternal and Papago do not (Katz et al. 1974). In the Tara- embryo protection hypothesis predicts that humara at least, nutritionaldeficiency is un- miscarriagerates of women in corn-basedsoci- likelyto account for the absence of morning eties (or women in anysociety who eat nothing sickness. but bland-tastingvegetables and do not ex- Additional evidence derivesfrom studies of hibit NVP) should be as low as miscarriage

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions JUNE 2000 MORNING SICKNESS 137 rates of women who exhibit NVP in societies (1997) noted that theirresults could indicate where meats and strong-tastingvegetables are eitherthat eliminating vomiting reduces anom- dietarystaples (e.g., Figure 4). In contrast,the alies by creating better metabolic conditions nutritionaldeficiency hypothesis predicts that forfetal development, or thatwomen who ex- miscarriagerates of women in corn-basedsoci- perienced NVP severe enough to seek drug eties (or bland vegetarians) should be as high reliefwere bearing the healthiestembryos. In as miscarriagerates of women who do not ex- addition, since women in these studies took hibitNVP in societieswhere meats and strong- antiemetics to alleviate symptoms,they may tastingvegetables are dietarystaples. Unfortu- have already developed aversions to poten- nately,HRAF sources gave no indications of tial pathogen- or teratogen-containingfoods, relative miscarriage rates in corn-based and therebyprotecting themselves and their em- meat-basedsocieties, and thereare no data on bryos by a judicious selection of diet before miscarriage rates for women who ate only NVP symptomswere suppressed. bland-tastingvegetables (e.g., corn, grains). Given the frequency and significance of Two other findingsare less consistentwith NVP in humans, we were surprisedto find so the maternaland embryoprotection hypothe- littleevidence of it among nonhuman mam- sis. First,although NVP was clearlyassociated mals. Indeed, one reason the embryoprotec- withreduced miscarriagerates, its occurrence tion hypothesishas not been adequately tested did not reduce the chances of low birthweight is thatno appropriate animal models are avail- or birthdefects. These negativeoutcomes can able. There are several possible reasons why resultfrom environmentalinsults that occur NVP is unknown outside of dogs, rhesus ma- in the firsttrimester (Smith et al. 1998), so caques, and chimpanzees. First,domesticated under the maternal and embryo protection and captivemammals (e.g., lab animals,house- hypothesis,NVP should have been associated hold pets, zoo specimens) often receive steri- with reduced frequencies of both outcomes. lized and bland diets that are relativelyfree Second, if NVP indeed serves a protective frommicroorganisms and toxic phytochemi- function,then artificiallyalleviating the symp- cals, so NVP mightseldom be triggered.Alter- toms should leave the mother and embryo natively,females that exhibited malaise (nausea more vulnerable.However, Kullander and Kal- and vomiting)during gestation may have been len (1976) reported that pregnant women artificiallyselected againstin captivebreeding who used antiemeticdrugs (mainlyantihista- programs. Second, NVP may be unnecessary mines) to alleviate NVP were less likelyto mis- for maternal and embryoprotection in mam- carrythan women who did not use antiemet- mals that have efficientphysiological mecha- ics. Many other investigatorshave studied the nisms for destroyingingested bacteria and effectsof antinauseant drugs on congenital fungi,and detoxifyingsecondaiy metabolites malformations.Seto et al. (1997) conducted of plants and microorganisms(e.g., pegastric a meta-analysisof such studies.They reported fermentationin brocket deer, Mazama spp.: that pregnant women who had taken anti- Bodmer 1991) . Alternatively,herbivorous mam- emetics in the firsttrimester were slightlybut mals may simplyavoid eating the most toxic significantlyless likely to bear children with plants during gestation,and pregnant carni- major malformationsthan women who had voresmay pass up carrionthat has been heavily not taken antiemetics. infestedwith especially virulent or toxicbacte- Although these findingsappear to contra- ria or fungi.Third, NVP mayactually be more dict our hypothesis,they may only indicate widespread, but it has gone unnoticed owing that women with the most severe symptoms to difficultiesof detection. Free-livingfemale (i.e., those least likelyto miscarryanyway: Fig- mammals would have to be observed pre-and ure 6) were the most likelyto use antiemetics. postconception,and detailed time budgets of Indeed, Kullander and KYllen (1976:105) ar- foragingbehavior and food preferenceswould gued that the correlation "is probably com- have to be assembled to reveal subtle changes pletelydue to less frequent and possiblyless in dietaryhabits. severe morning sickness complaints in preg- Two general hypothesesalternative to ma- nancies ending in miscarriage,"and Seto et al. ternal and embiyo protectionhave been pro-

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions 138 THE QUARTERLY REVIEW OF BIOLOGY VOLUME 75 posed to explain NVP. First,it may simplybe conflicthypothesis, manifestations should be a side effectof viable pregnancies. Stein and more pronounced in youngerwomen because Susser (1991:165) argued that "nausea is af- they should withhold more resources from fectedby the outcome [of pregnancy].It man- current offspringsince they have greater fu- ifestswith differingfrequencies according to ture reproductive opportunities than older whethera pregnancywill terminateeither as women. Conflictshould be especially evident a miscarriageor as a live birth."Haig (1993, in adolescent mothers because they them- 1996) agreed thatnausea and vomitingmight selvesare stillgrowing and would directlycom- be effectsrather than causes, either of some pete with the embryofor key resources (e.g., placental factorthat itself reduces the chances Wallace et al. 1996). Most studies, however, of miscarriage,such as human chorionic go- have found no relationship between mater- nadotropin (hCG), or of genetic conflictsof nal age and NVP (e.g., Medalie 1957;Jarnfelt- interestbetween a fetusand its mother. The Samsioe et al. 1983; Vellacott et al. 1988; Chin latter idea, which originated with Trivers 1989; Paarlberg et al. 1996; Gadsbyet al. 1997; (1974), is thatbecause the fetusis more related but see Klebanoffet al. 1985;Jinadu and Dara- to itselfthan to itsfuture siblings, especially if mola 1990), and in a study of 78 pregnant theyhave differentfathers, it is selected to take 14-19 year olds, Dilorio (1985) reported that more nutritionfrom the mother than is opti- the youngest women (ages 14-15) had the mal for her to give to any one offspring.Ac- lowest incidence of NVP (27% vs. 61% of cording to Haig (1993), the more healthyand 16-19 year olds). Finally,Haig's conflicthy- vigorous the fetus,the more capable it is of pothesis suggests that NVP symptomsshould competing effectivelyfor maternal resources, also be more pronounced later in pregnancy resultingin reduced chances of being miscar- because an older, larger fetus requires more ried but with more obvious indications of resources and may be more effectivein com- physiologicalconflict with its mother. peting forthem. NVP symptomspeak earlyin Five facts do not accord with these side- pregnancy,however, when the embryois tiny effecthypotheses. First, NVP is nota necessary (Figure 3). concomitant of a viable pregnancy,nor is it Rather than dismissingStein and Susser's onlyassociated withviable pregnancies. In our (1991) and Haig's (1993) hypotheses,we sug- sample of 5,235 pregnancies in whichNVP did gest that it may be more appropriate to view not occur (Figure 4), only535 (10%) resulted them as mechanistic complements to mater- in miscarriages,and in our sample of 13,192 nal and embryoprotection (i.e., hypothesesat pregnanciesin whichNVP did occur, 509 (4%) differentlevels of analysis). Under this inter- resulted in miscarriages. Second, no consis- pretation,reproductive hormones thataccom- tentdifferences in typesor levels of reproduc- pany viable pregnancies set the physiological tivehormones between emetic and nonemetic stage forthe expression of NVP by sensitizing viable pregnancies have been discovered. Al- the neural pathwaysthat triggernausea and though reproductivehormones are necessary vomiting (Andrews and Whitehead 1990). A for the expression of NVP, they are not, by likelycandidate is hCG, whose concentration themselves, sufficientto triggerNVP symp- in maternalblood peaks duringweeks 8 to 10 toms (Andrewsand Whitehead 1990). Third, of pregnancy and declines dramatically by there are seven societies in which NVP symp- weeks 14 to 18 (Riley 1959; Tulchinskyand toms have never been reported (Table 3), al- Hobel 1973; Kauppila et al. 1984), thusclosely thoughviable pregnanciesroutinely occurred. paralleling the time-courseof NVP (Figure 3). Complete absence of NVP obviouslyruns con- A rise in estradiol coincides withthe onset of traryto the side-effecthypotheses. Fourth, these NVP, but its concentration continues to in- hypothesesdo not predicteither (i) specificity crease afterNVP symptomshave diminished of food aversionsor (ii) patternsof change in (Busterand Simon 1989). Althoughthese hor- aversionsacross the course of pregnancy (Fig- mones (and others) are necessaryto support ures 7-9), whereas both phenomena are con- pregnancy,their association withnausea and sistentwith protecting the motherand embryo. vomiting,especially in response to specificdi- Fifth,under Haig's (1993, 1996) genetic etarytriggers (Figures 7-10), requires a func-

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions JUNE 2000 MORNING SICKNESS 139 tional analysis.This reemphasizes the comple- Haig (1993:509) statedthat "various hypothe- mentarityof hypotheses about fitnesseffects ses ... have been proposed to account fornau- and their underlyingmechanisms (e.g., Tin- sea duringpregnancy, but I am unable to come bergen 1963; Sherman 1988). to clear conclusions because the evidence re- The second alternativehypothesis was pro- mains equivocal." Table 4 summarizes infor- posed byDeutsch (1994:277), namelythat "preg- mationthat supports, contradicts, and is equivo- nancysickness and mastalgia have evolved to cal regarding the three major alternatives. reduce frequency of sexual intercourse in Available informationis most consistentwith earlypregnancy." The idea is that NVP alerts the hypothesisthat NVP protectsthe mother a woman's partner to her pregnancy,thereby and her embryo from dangerous substances discouraging behavior that might harm the in food. Protectionis provided byimmediately embryo. NVP also might signal a woman's eliminating the offendingfoods-especially partner and her kin of the impending need animal products and plant toxins-from the foradditional food and protection.There are three reasons to doubt this "communication" woman's body (vomiting), and learning to hypothesis,however. First, intercourse does avoid the nauseatingtastes subsequently. More not affectthe viabilityof a pregnancy,except thorough testingof this hypothesiscould be possiblyduring the final 4 to 6 weeks before accomplished by comparing NVP symptoms birth(Pritchard and MacDonald 1980:319), by and pregnancyoutcomes among women from whichtime NVP has typicallywaned (Figure3). the same societywho ate veiy differentdiets Second, the communication hypothesisdoes (e.g., meats and vegetablesvs. vegetables only not predict the absence of NVP in any society, vs. grainsand cereals only). Such studies must unless pregnant and lactatingwomen always be conducted before specificdietary practices livealone (i.e., thereis no one to communicate based on the maternaland embryoprotection with).This certainlydoes not occur in the seven hypothesisare adopted. societies in which morning sickness has not Nonetheless, our resultshave two practical been observed (Table 2). Most importantly, implications.First, there is no reason to believe NVP peaks after8 weeksof gestation,well after that alleviatingthe symptomsof normal NVP otherless costlyand uncomfortableindications (e.g., excluding hyperemesisgravidarum) will of pregnancyhave become apparent.The most improvethe outcome of a pregnancy.Indeed, obvious of these is cessation of menstruation, doing so could have the opposite effect,if it in- whichoccurs some 4 weeksearlier. Our perusal terfereswith the expulsion of potentiallydan- of HRAF filesrevealed thatthis is a universally gerous foods, or withlearning to avoid them. recognized indicatorof pregnancy.For exam- Second, there is no reason to believe thatdis- ple, in the Okinawa (Maretzki and Maretzki couraging pregnant women from avoiding 1963:456), the "[c]essation of menstrualperi- foods to which theydevelop aversionswill im- ods signals the startof pregnancy,"and in the prove theirpregnancy outcome. In fact,forc- Gond (Grigson 1949:263) "all women know ing women to eat aversivefoods could be detri- when they miss their firstmenstruation that mentalif it increasesthe embryo'sexposure to theyare probably startingto bear a child." In foodborneillnesses and teratogenicor aborti- the Alor (DuBois 1944:28-29), "thewife knows facientchemicals. Of course,experiencing NVP she is pregnant . . . when she ceases to have monthlyperiods," and a Tongan woman "uses does not guaranteethat a pregnancywill have a several signs as an indication of pregnancy. positiveoutcome, nor does a lack of NVP symp- She finds herselfwithout much appetite and tomsportend pregnancy failure. An overwhelm- feels physicallyweak. She is also abnormally ing majorityof women carrytheir pregnancies sensitiveto bad smells. Menstruationceases" to termand have healthybabies, whetherthey (Beaglehole and Beaglehole 1941:78). And in experience NVP or not (in developed coun- the Terena (Oberg 1949:38), "a woman be- tries,at least). Our keypoint is that "morning comes aware that she is pregnant when her sickness"apparently has serveda usefulprotec- menses cease and she has spells of nausea and tivefunction, and should thereforebe consid- is particularabout her food." ered and treatedas an adaptation.

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions 140 THE QUARTERLY REVIEW OF BIOLOGY VOLUME 75

16In ct

ct :i bC bn

od od Ct ct ll bio cd n :i ct

od od -.5 0 42) L-.

'cl ct ct bo b bio bID - U 0 0 q)

biD ct b.0 bio

4-J biD >, zl u .,-( n 11 bio ct 4_j 4-1 4-J bio (t 7 N 4-J . - a ct 4-J t_ 4-J >, - -L, b .0 zl (t ct En -_- n 4-J 4-J 4-J

4-1 4-J 4-J 4-J 4-1 Cld 7 4-1 ct ct ct 4-J

4-J Cld 4-J Cld C) 4-J

4-J 4_j 4-J 4-J Cld 7 (4--i b.0 4-j 4-J 4-J - 4-1 0 cld 4-J c) Cld bk cld bio (-) q) 4-J 4-J 4-J oo Z 1--i 4-1 'jl 4-J b,0

4-J :j ct Cld 'ZI 4-J Cld Z, q) Cld 4-J 4-J biD in. (:) 0 'o q) 4-J -(: 4-J bi) C.C) In 4-J c) Z.,0 4-J ct 4-J bf) 4-1 4-J 4-1 C4-4 :j - u 4-1 > 0 - - Z, 4-1 4-J U 4-J Z, Cld .5 Cld biD cd Cld

C'd Ct 4-1 Cld ct 4-J Cl-d Z, ct Z, q) 4-J bio > m 4 J 4-J bio Cld u p- ,, : C4--, Z, c) 4-J 4-J

'4--l Z, Z, ct b.0 ct U to bl) b,0

b,0 J, bn 5 0

4-1 4-1

biD Z, 4-1 Z, (4-4 4-1 4 4-1 Z, q) 1-1 Ct (-)

cd 4-1 z ct 4- Z, ct Z, z :! zl q) (-) bi) > a-4 4-1 Cld 4-1 Cld Cld u

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions JUNE 2000 MORNING SICKNESS 141

ACKNOWLEDGMENTS the manuscript,we thankM C B Andrade, T Eisner, For helpful suggestionsduring thisstudy we thank T A Gavin, D Haig, R EJohnston, C Kagarise Sher- B P Chang, D Chaudhuri,KA Merchant,H K Reeve, man,H KReeve, P WTurke,G C Williams,LAWolfen- L E Southard, S Szmuilowicz, and participants in barger,and an anonymous reviewer.During the re- the Animal Behavior Lunch Bunch at Cornell Uni- search, SMF was partiallysupported by a Howard versity.For assistance withreferences, we thank the Hughes Undergraduate Fellowship, the Honors Cornell Universityreference librarians at Mann Program in Cornell University'sDivision of Biologi- and Olin Libraries, particularlyK S Chiang, P M cal Sciences, and bya National Science Foundation Davis, M L Schlabach, and T P Turner.For statistical Graduate Research Fellowship. PWS was supported advice, especiallyon the binomial proportion tests, by grants from the National Science Foundation we thank G Casella of Cornell's Biometrics Unit. and the College of Agricultureand Life Sciences at For helpful comments on preliminaryversions of Cornell University.

REFERENCES

Abraham S, King W, Llewellyn-JonesD. 1994. Atti- BarrettD S, RayfieldL S, BrentL. 1982. Suppression tudes to body weight,weight gain and eating be- of natural cell-mediated cytotoxicityin man by havior in pregnancy.Journal of Psychlosomatic Ob- maternaland neonatal serum. Clinicaland Exper- stetfricsand Gynaecology15:189-195. imentalI7nnunology 47:742-748. Alberti-FidanzaA, Coli A M, Ottavi I, Panarelli W. Bartholomew MJ, Poston F E. 1970. Effectof food 1996. Nutritionalstudies on pregnantwomen in taboos on prenatal nutrition.Journal of Nutrition Umbria (Italy). Pages 11-22 in Nutritionin Preg- Education2:15-17. nancyand Growth,edited byM Porriniand P Wal- BaylisJM, Leeds A R, Challacombe D N. 1983. Per- ter. Basel: Karger. sistentnausea and food aversionsin pregnancy. Alberti-Fidanza A, Fidanza R. 1986. A nutrition ClinicalAllergy 13:263-269. studyinvolving a group of pregnant women in Beaglehole E, Beaglehole P. 1941. Pangai: Villagein Assisi, Italy. Part 1: Anthropometry,dietary in- Tonga.Wellington (New Zealand): The Polyne- take and nutritionknowledge, practices and atti- sian Society. tudes. InternationalJournalforVitamin and Nutfi- Bebiak D M. 1988. Nutritionand the reproductive tionResearch 56:373-380. roller coaster. Proceedingsof the Annual Meetingof AlcockJ,Sherman P W. 1994. The utilityof the prox- theSocietyfor Theriogenology 1988:167-173. imate-ultimatedichotomy in ethology. Ethology Beer A E, KwakJY H, RuizJ E. 1996. Immunophe- 96:58-62. notypic profiles of peripheral blood lympho- Al-Kanhal M A, Bani I A. 1995. Food habits during cytesin women withrecurrent pregnancy losses pregnancy among Saudi women. International and in infertilewomen with multiple failed in JournalforVitamin and NutritionResearch 65:206- vitro fertilizationcycles. AmericanJournal of Re- 210. productiveImmunology 35:376-382. Ames B N, ProfetM, Gold L S. 1990a. Dietarypesti- Beier R C, Nigg H N. 1994. Toxicology of naturally cides (99.99% all natural). Proceedingsof the Na- in in tionalAcademy of Sciences 87:7777-7781. occurring chemicals food. Pages 1-186 Ames B N, Profet M, Gold L S. 1990b. Nature's FoodborneDisease Handbook, Volume3: Diseases chemicals and syntheticchemicals: comparative Causedby Hazardous Substances, edited byYH Hui toxicology.Proceedings of the National Academyof et al. New York: Marcel Dekker. Sciences87:7782-7786. BernsteinL, Pike M C, Lobo R A, Depue R H, Ross Ananth C V, Rao P S S. 1993. Epidemiology of nau- R K, Henderson B E. 1989. Cigarettesmoking in sea and vomitingof pregnancyand itsrelation to pregnancyresults in marked decrease in mater- fetal outcome in a rural area. Journalof Tropical nal hCG and oestradiol levels. BritishJournal of Pediatrics39:313. Obstetricsand Gynaecology96:92-96. Anderson K N, Anderson L E, Glanze W D, editors. Biggs J S G. 1975. Vomiting in pregnancy: causes 1994. MIosby'sMedical, Nursing, and AlliedHealth and management. Drugs9:299-306. Dictionaiy.St. Louis (MO): Mosby-YearBook. BillingJ, Sherman P W. 1998. Antimicrobialfunc- AndrewsP, Whitehead S. 1990. Pregnancysickness. tions of spices: why some like it hot. Quarterly Newsin PhysiologicalScience 5:5-10. Reviezvof Biology 73:3-49. Arnholdt H, Meisel F, Fandrey K, Lohrs U. 1991. Bisset L R, Fiddes T M, GillettW R, Wilson P D, Proliferationof villous trophoblast of the hu- GriffinJ F T. 1990. Altered humoral immuno- man placenta in normal and abnormal pregnan- regulation during human pregnancy. American cies. VirchowvsArchiv B CellPathology 60:365-372. Journalof Reproductive Imnmunology 23:4-9.

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions 142 THE QUARTERLY REVIEW OF BIOLOGY VOLUME 75

Blumberg N, Heal J M. 1996. The transfusionim- vomiting-complicatedpregnancy. EuropeanJour- munomodulation theory: the Th 1/Th2 para- nal of Obstetricsand Gynecologyand Reproductive digm and an analogy withpregnancy as a unify- Biology33:215-219. ing mechanism. Seminarsin Hematology33:329- Christiansen 0 B. 1996. A freshlook at the causes 340. and treatmentsof recurrentmiscarriage, espe- Bodmer RE. 1991. Strategiesof seed dispersal and cially its immunological aspects. Human Repro- seed predation in Arnazonianungulates. Biotrop- ductionUpdate 2:271-293. ica 23:255-261. CnattingiusS, MillsJL, YuenJ, Eriksson0, Salonen Bourne G H, editor. 1970-73. The Chimpanzee,Vol- Ros H. 1997. The paradoxical effectof smoking umes1-6. Baltimore (MD): UniversityPark Press. in preeclamptic pregnancies: smoking reduces Brabin BJ. 1985. Epidemiology of infectionin preg- the incidence but increases the ratesof perinatal nancy. Reviewsof Infectious Diseases 7:579-603. mortality,abruptio placentae, and intrauterine BrandesJ M. 1967. First-trimesternausea and vom- growthrestriction. American Journal of Obstetrics itingas related to outcome of pregnancy. Obstet- and Gynecology177:156-161. ricsand Gynecology30:427-431. Cohen P. 1997. Morning sicknesslink 'misleading.' Brown J E. 1983. Nutritionfor YourPregnancy: The NewvScientist 153:11. UniversityofA/Iinnesota Guide. Minneapolis (MN): Corey L A, Berg K, Solaas M H, Nance W E. 1992. Universityof Minnesota Press. The epidemiology of pregnancy complications BrownJ E, Kahn E S, Hartman TJ. 1997. Profet, and outcome in a Nolrwegiantwin population. profits,and proof: do nausea and vomiting of Obstetricsand Gynecology80:989-994. earlypregnancy protect women from"harmful" Coronios-VargasM, Toma R B, Tuveson RV, Schutz vegetables? AmericanJournal of Obstetrics and Gy- I M. 1992. Cultural influences on food cravings necology176(1):179-181, part 1. and aversions during pregnancy.Ecology of Food BrownJ E, Toma R B. 1986. Taste changes during and Nutrition27:43-49. pregnancy. AmericanJournal of Clinical Nutrition Czaja J A. 1975. Food rejection by female rhesus 43:414-418. monkeys during the menstrual cycle and early F 1988. Risksof and Bryan L. practices,procedures, pregnancy.Plhysiology and Behavior14:579-587. processes that lead to outbreaks of foodborne DarnettyJF L, MuthukrishnanS, Swegle M, Vigers diseases. Journalof Food Protection51:663-673. AJ, SelitrennikoffC P. 1993. Variabilityin anti- BryanF L, Fanelli MJ,Riemann H. 1979. Salmonella fungalproteins in the grainsof maize, sorghum, infections.Pages 73-130in FoodborneInfections and and wheat. PlhysiologiaPlantarum 88:339-349. Intoxications,Second Edition, edited by H Rie- Darwish 0 A, Amine E K. 1982. Food habits during mann and F L Biyan. New York:Academic Press. pregnancyand lactation in Iraq. Food and Nutri- BusterJ E, Simon J A. 1989. Placental hormones, tionBulletin 4:14-16. hormonal preparation forand control of partu- Davenport H W. 1982. PlhysiologyoftheDigestive Tract. rition, and hormonal diagnosis of pregnancy. FifthEdition. Chicago: Year Book Medical Pub- Pages 2043-2073 in Endocrinology,Second Edi- lishers. tion, edited by L J DeGroot. Philadelphia: T Williams D R R, WhitakerR H. 1986. W. B. Saunders. Davies W, Campbell D M, Hall M H, Barker D J P, Cross J, Riskfactors for undescended testis.International Shiell A W, GodfreyK M. 1996. Diet in preg- Journalof Epidemiology 15:197-201. nancy and the offspring'sblood pressure 40 De Silva P, Rachman S. 1987. Human food aver- years later. BritishJournal of Obstetrics and Gynae- sions: nature and acquisition. BehaviourResearch and Tlherapy25:457-468. cology103:273-280. Carlson B M. 1994. Human Embryologyand Develop- DeutschJ A. 1994. Pregnancysickness as an adapta- mientalBiology.St. Louis (MO): Mosby-YearBook. tion to concealed ovulation. Rivista di Biologia Chao K-H,Yang Y-S,Ho H-N, Chen S-U, Chen H-F, 87:277-295. Dai H-J,Huang S-C, Gill TJ III. 1995. Decidual DiamondJ M. 1999. Dirtyeating for healthyliving. natural killer cytotoxicitydecreased in norrpal Nature400:120-121. pregnancy but not in anembryonic pregnancy Dickens G, Trethowan W H. 1971. Cravings and and recurrentspontaneous abortion. American aversions during pregnancy.Journal of Psychoso- Journalof Reproductive Immunology 34:274-280. maticResearch 15:259-268. Chaouat G, Menu E. 1997. Maternal T cell reactivity DickinsonJ E. 1994. Viral teratology.Pages 12-23 in in pregnancy? CurrentTopics in Microbiologyand ViralDiseases in Pregnancy,edited by B Gonik. Immunology222:103-126. New York: Springer-Verlag. Chin R K H. 1989. Antenatal complications and DickinsonJE, Gonik B. 1990. Teratogenicviral infec- perinatal outcome in patients with nausea and tions. ClinicalObstetrics and Gynecology33:242-252.

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions JUNE 2000 MORNING SICKNESS 143

Diggory P L C, Tomkinson J S. 1962. Nausea and FitzgeraldC M. 1984. Nausea and vomitingin preg- vomitingin pregnancy:a trialof meclozine dihy- nancy. BritishJournal of Medical Psychology57: drochloride withand withoutpyridoxine. Lancet 159-165. ii:370-372. Fletcher A C, La Flesche F. 1972. The Omaha Tribe. Dilorio C. 1985. Firsttrimester nausea in pregnant Lincoln (NE): Universityof Nebraska Press. teenagers: incidence, characteristics,interven- Formby B. 1995. Immunologic response in preg- tion. NursingResearch 34:372-374. nancy: its role in endocrine disorders of preg- Dilorio C. 1988. The management of nausea and nancy and influence on the course of maternal vomitingin pregnancy.Nurse Practitioner 13:23-28. autoimmune diseases. Endocrinologyand Metabo- Dilorio C, van Lier D, ManteuffelB. 1992. Patterns lismClinics of North America 24:187-205. of nausea during firsttrimester of pregnancy. Friedman M, RayburnJR, BantleJ A. 1991. Devel- ClinicalNursing Research 1:127-140. opmental toxicology of potato alkaloids in the Drake M L, VerhulstD, FawcettJ.1988. Physicaland frogembryo teratogenesis assay -Xenopus (Fetax). psychological symptomsexperienced by Cana- Food Chemistryand Toxicology29:537-547. dian women and their husbands during preg- R. 1994. sicknessand nancy and the postpartum.Journal of Advanced Gadsby Pregnancy symptoms. Nursing13:436-440. ProfessionalCare of Mother and Child4:16-17. DuBois C A. 1944. ThePeople of Alor: A Social-Psycho- Gadsby R, Barnie-Adshead A M, Jagger C. 1997. logi.calStudy of an East Indian Island. Minneapolis Pregnancynausea related to women's obstetric (MN): Universityof Minnesota Press. and personal histories. Gynecologicand Obstetric Edwards C H, McSwain H, Haire S. 1954. Odd di- Investigation43:108-111. etarypractices of women. Journalof the American Gennaro S, Fehder W P. 1996. Stress,immune func- DieteticAssociation 30:976-981. tion, and relationship to pregnancy outcome. Ehrlich P R, Raven P H. 1964. Butterfliesand plants: NursingClinics of North America 31:293-303. a studyin coevolution. Evolution18:586-608. Gerba C P, RoseJ B, Haas C N. 1996. Sensitivepopu- Ember M, Otterbein K F. 1991. Sampling in cross- lations: who is at the greatestrisk? International cultural research. BehaviorScience Research 25: Journalof Food Microbiology30:113-123. 217-233. Gerber L M, WilliamsG C, GraySJ. 1999. The nutri- Erick M. 1995. Hyperolfaction and hyperemesis ent-toxindosage continuum in human evolu- gravidarum:what is the relationship? Nutrition tion and modern health. QuarterlyReview of Biol- Reviews53:289-295. ogy74:273-289. Evans AJ,Li T C, Selby C,Jeffcoate 1986. Morn- WJ. GilardiJ D, DuffeyS S, Munn C A, Tell L A. 1999. ing sicknessand thyroidfunction. BritishJournal Biochemical functionsof geophagy in parrots: ofObstetrics and 93:520-522. Gynaecology detoxificationof toxins and cytoprotec- Fairburn C G. 1986. Nausea and vomitingin preg- dietary tiveeffects.Journal of 25:897-922. nancy. BritishJournal of Medical Psychology57: ChemlicalEcology 159-165. GodfreyK, Robinson S, Barker D J P, Osmond C, Fairburn C G, Stein A, Jones R. 1992. Eating habits Cox V. 1996. Maternal nutritionin earlyand late and eating disorders during pregnancy. Psycho- pregnancy in relation to placental and fetal somaticMedicine 54:665-672. growth.British MedicalJournal 312:410-414. Fairburn C G, Welch S L. 1990. The impact of preg- Golding J, Vivian S, Baldwin J A. 1983. Maternal nancy on eating habits and attitudes to shape anti-nauseantsand cleftsof lip and palate. Hu- and weight. InternationalJournalof Eating Disor- man Toxicology2:63-73. ders9:153-160. Gonik B, Loo L S, West S, Kohl S. 1987. Natural Fairweather D V 1. 1968. Nausea and vomitingin killer cell cytotoxicityand antibody-dependent pregnancy.American Journal of Obstetrics and Gy- cellular cytotoxicityto herpes simplex virus- necology102:135-175. infected cells in human pregnancy. American FawcettJ,York R. 1986. Spouses' physical and psy- Journalof Reproductive Immunology and Microbiol- chological symptomsduring pregnancyand the ogy13:23-36. postpartum.Nursing Research 35:144-148. Goodall J. 1986. The Chimnpanzeesof Gombe:Patterns Fenster L, Eskenazi B, Windham G C, Swan S H. ofBehavior. Cambridge (MA): Harvard Univer- 1991. Caffeine consumption during pregnancy sityPress. and spontaneousabortion. Epidemiology 2:168-174. Lon- Fichera M E, Roos D S. 1997. A plastid organelle as Grigson W V. 1949. TheMaria Gondsof Bastar. a drug targetin apicomplexan parasites. Nature don: Oxford UniversityPress. 390:407-409. Guo B Z, Brown R L, Lax A R, Cleveland T E, Russin FinleyD A, Dewey K G, Lonnerdal B, GrivettiL E. J S, WidstromN W. 1998. Protein profiles and 1985. Food choices of vegetariansand nonvege- antifungalactivities of kernal extractsfrom corn tarians during pregnancy and lactation.Journal genotypesresistant and susceptible to Aspergillus ofthe American Dietetic Association 85:678-685. flavus.Journal of Food Protection61:98-102.

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions 144 THE QUARTERLY REVIEW OF BIOLOGY VOLUME 75

Haig D. 1993. Genetic conflicts in human preg- pregnancy:a review.Obstetrical and (ynecological nancy. QuarterlyReview of Biology 68:495-532. Survey41 :422-427. Haig D. 1996. Altercation of generations: genetic Jarnfelt-SamsioeA, Eriksson B, Waldenstr6m J, conflictsof pregnancy.AmericanJournal of Repro- Samsioe G. 1985. Some new aspects on emesis ductiveImmunology 35:226-232. gravidarum.Gynecologic and ObstetricInvestigation HarriesJM, Hughes T F. 1958. Enumeration of the 19:174-186. "cravings"of some pregnantwomen. BritishMed- Jarnfelt-SamsioeA, Samsioe G, Velinder G M. 1983. icalJournalii:39-40. Nausea and vomitingin pregnancy-a contribu- HartungJ. 1997. If I had it to do over again. Pages tion to its epidemiology. Gynecologicand Obstetric 344-348 in Human Nature:A CriticalReader, ed- Investigation16:221-229. ited byL Betzig.Oxford: Oxford University Press. JenkinsM L, Shelton BJ. 1989. The effectivenessof HathcockJ N, Hattan D G,JenkinsMY, McDonald self-care actions in reducing "morning sick- J T, Sundaresan P R, WilkeningV L. 1990. Evalu- ness." Pages 267-272 in KeyAspects of Comfort: ation of vitaminA toxicity.American Journal of Managementof Pain, Fatigue,and Nausea, edited ClinicalNutrition 52:183-202. by S G Funk et al. New York: Springer. Hidaka Y, Amino N, IwataniY, Kaneda T, Mitsuda JinaduM K, Daramola S M. 1990. Emotional changes N, Morimoto Y, Tanizawa 0, Miyai K. 1991. in pregnancyand earlypuerperium among the Changes in natural killer cell activityin normal Yoruba women of Nigeria. InternationalJournal pregnant and postpartumwomen: increases in ofSocial Psychiatry36:93-98. the firsttrimester and postpartum period and JohnsT. 1990. WithBitter Herbs Thley ShallEat It: Clhemi- decrease in late pregnancy.Journal of Reproduc- calEcologyand theOrigins of Human Dietand Medi- tiveImmunology 20:73-83. cine.Tucson (AZ): Universityof Arizona Press. Hirasa K, Takemasa M. 1998. SpiceScience and Tech- JohnsT, Duquette M. 1991a. Detoxificationand min- nology.New York: Marcel Dekker. eral supplementationas functionsof geophagy. Hobbs B C, RobertsD. 1993. FoodPoisoning and Food AmericanJournal of Clinical Nutrition 53:448456. Hygiene.Sixth Edition. London: Edward Arnold. Johns T, Duquette M. 1991b. Traditional detoxifi- Holliman RE. 1995. Congenital toxoplasmosis:pre- cation of acorn bread with clay. Ecologyof Food vention, screening, and treatment.Journal of and Nutrition25:221-228. HospitalInfection 30 (supplement):179-190. Johnson L R. 1997. Digestion and absorption. Pages Holloway M. 1996. Profile: Margie Profet.Scientific 113-134 in GastrointestinalPhysiology, Fifth Edi- American274:40-41. tion, edited by L R Johnson. St. Louis (MO): Hook E B. 1976. Changes in tobacco smoking and Mosby-YearBook. ingestion of alcohol and caffeinatedbeverages Kapperud G, Jenum P A, StrayPedersen B, Melby during earlypregnancy: are these consequences, K K, Eskild A, Eng J. 1996. Risk factorsfor Tox- in part,of feto-protectivemechanisms diminish- oplasmagondii infection in pregnancy.Results of ing maternal exposure to embryotoxins?Pages a prospective case-control study in Norway. 173-183 in Bir-thDefects: Risks and Consequences, AmericanJournalof Epidemiology 144:405-412. edited byS Kellyet al. New York:Academic Press. Kassie F, ParzefallW, Musk S,Johnson I, Lamprecht Hook E B. 1978. Dietarycravings and aversionsdur- G, Sontag G, Knasmuller S. 1996. Genotoxic ef- ing pregnancy.AmericanJournal of Clinical Nutri- fectsof crudejuices fromBrassicavegetables and tion31:1355-1362. juices and extracts from phytopharmaceutical Hook E B. 1980. Influence of pregnancyon dietary preparationsand spices of cruciferousplants or- selection.InternationalJournal ofObesity 4:338-340. igin in bacterial and mammalian cells. Chemico- Human RelationsAreaFiles. Ann Arbor (Ml): Univer- BiologicalInteractions 102:1-16. sityMicrofilms International. Katz S H, Hediger M L, Valleroy L A. 1974. Tradi- latrakis G M, Sakellaropoulos G G, Kourkoubas tional maize processing techniques in the New A H, Kabounia S E. 1988. Vomiting and nausea World. Science184:765-773. in the first12 weeks of pregnancy.Psychotherapy Kaufmnan M H. 1997. The teratogenic effectsof al- and Psychosomatics49:22-24. cohol following exposure during pregnancy, IrvingF C. 1940. The treatmentof pernicious vom- and its influence on the chromosome constitu- iting of pregnancy. VirginiaMedical Monthly67: tion of the pre-ovulatoryegg. Alcoholand Alcohol- 717-724. ism32:113-128. JacksonP G G. 1995. Handbookof Veterinary Obstetrics. Kauppila A, Heikinheimo M, Lohela H, Ylikorkala Philadelphia: W. B. Saunders. 0. 1984. Human chorionic gonadotrophin and JanewayC A,Jr,Travers P. 1994. Immunobiology:The pregnancy-specificbeta-l-glycoprotein in pre- ImmuneSystem in Health and Disease. New York: dicting pregnancy outcome and in association Garland Publishing. with early pregnancy vomiting. Gynecologicand Jarnfelt-SamsioeA. 1987. Nausea and vomitingin ObstetricInvestigation 18:49-53.

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions JUNE 2000 MORNING SICKNESS 145

Keeling M E, RobertsJR. 1972. Breeding and repro- Little R E, Hook E B. 1979. Maternal alcohol and duction in chimpanzees. Pages 127-152 in The tobacco consumption and theirassociation with Chimpanzee,Volume 5, edited by G H Bourne. nausea and vomitingduring pregnancy. Acta Ob- Baltimore (MD): UniversityPark Press. stetriciaet GynecologicaScandinavica 58:15-17. Keen C L, Bendich A, Willhite C C, editors. 1993. LynbergM C, KhouryMJ,Cocian T. 1994. Maternal MaternalNutrition and PregnancyOutcome. New flu, fever,and the risk of neural tube defects: York: New York Academy of Sciences. a population-based case-controlstudy. American KlebanoffM A, Koslowe P A, KaslowR, Rhoads G G. Journalof Epidemiology 140:244-255. 1985. Epidemiology of vomiting in early preg- Mace R, Pagel M. 1994. The comparative method nancy. Obstetricsand Gynecology66:612-616. in anthropology.Current Anthropology 35:549-564. KlebanoffM A, Levine RJ, DerSimonian R, Clem- Mace R, Pagel M. 1997. Tips, branches, and nodes: ensJD, WilkinsD G. 1999. Maternal serum para- seeking adaptation through comparative stud- xanthine, a caffeinemetabolite, and the riskof ies. Pages 297-310 in Human Nature:A Critical spontaneous abortion. New England Journal of Reader,edited by L Betzig. Oxford: Oxford Uni- Medicine341:1639-1644. versityPress. KlebanoffM A, MillsJ L. 1986. Is vomitingduring Maclntyre S. 1983. The management of food in pregnancy teratogenic? BritishMedical Journal pregnancy. Pages 57-72 in The Sociologyof Food 292:724-726. and Eating,edited byA Murcott.Aldershot (UK): Klonoff-CohenH, Edelstein S, SavitzD. 1993. Ciga- Gower Publishing. rette smoking and preeclampsia. Obstetricsand Mahaney W C, Hancock R G V, AufreiterS, Huff- Gynecology81:541-544. man M A. 1996. Geochemistryand clayminerol- Knox BJ. 1993. Dietaryhabits, taste acuity and pref- ogyof termitemound soil and the role of geoph- erence for fat during human pregnancy [Doc- agy in chimpanzees of the Mahale Mountains, toral dissertation]. Belfast (Northern Ireland): Tanzania. Primates37:121-134. Queen's Universityof Belfast. Mahaney W C, Milner M W, Sanmugadas K, Han- Koul 0. 1993. Plant allelochemicals and insect con- cock R G V, AufreiterS, Wrangham R, Pier H W. trol: an antifeedant approach. Pages 51-88 in 1997. Analysisof geophagy soils in Kibale Forest, ChlemicalEcology of Phytophagous Insects, edited by Uganda. Primates38:159-176. T N Ananthakrishnanand A Raman. New Delhi Marcus R L. 1965. Cravingsfor food in pregnancy. (India): Oxford & IBH. ManchesterMedical Gazette44:16-17. KrickerA,ElliottJ, ForrestJ, McCredieJ. 1986. Con- Maretzki T W, Maretzki H. 1963. Taira: an Okina- genitallimb deficiency:maternal factors in preg- wan village. Pages 363-539 in Six Cultures:Studies nancy.Australia and NewZealandJournal of Obstet- of Child Rearing,edited by B B Whiting. New ricsand Gynaecology26:272-275. York:John Wileyand Sons. Kullander S, Kallen B. 1976. A prospectivestudy of Masson G M, AnthonyF, Chau E. 1985. Serum cho- drugs and pregnancy.II. Anti-emeticdrugs. Acta rionic gonadotrophin (hCG), schwangerschafts- ObstetriciaetGynecologica Scandinavica 55:105-111. protein 1 (SP1), progesterone and oestradiol Kurppa K, Holmberg P C, Kuosma E, Aro T, Saxen levels in patients with nausea and vomiting in L. 1991. Anencephaly and maternal common early pregnancy. BritishJournal of Obstetricsand cold. Teratology44:51-55. Gynaecology92:211-215. LandmanJ, HallJ S E. 1983. The dietaryhabits and MatissekR. 1997. Evaluation ofxanthine derivatives knowledge of folklore of pregnant Jamaican in chocolate: nutritionaland chemical aspects. women. Ecologyof Food Nutrition12:203-210. ZeitschniftfuerLebensmittel-Untersuchung und -For- LetarteA, Dube L, Troche V. 1997. Similaritiesand schungA205:175-184. differencesin affectiveand cognitiveorigins of Mattes R D. 1991. Learned food aversions:a family food likingsand dislikes.Appetite 28:115-129. study.Physiology and Behlavior50:499-504. Lewis L D, Morris M L, Jr,Hand M S. 1987. Small Matthiesen L, Berg G, Ernerudh J, Hakansson L. AnimalNutrition III Third Edition. Topeka (KS): 1996. Lymphocytesubsets and mitogen stimula- Mark MorrisAssociates. tion of blood lymphocytesin normal pregnancy. Lim KJ H, Odukoya 0 A, Li T C, Cooke I D. 1996. AmericanJournal of ReproductiveImmunology 35: Cytokinesand immuno-endocrinefactors in re- 70-79. currentmiscarriage. Human ReproductionUpdate Mayr E. 1961. Cause and effectin biology. Science 2:469-481. 134:1501-1506. Little R E. 1980. Maternal alcohol and tobacco use McBride M L, VanDenSteen N, Lamb C W, Gal- and nausea and vomitingduring pregnancy:re- lagher R P. 1991. Maternal and gestational fac- lation to infantbirthweight. Acta Obstetriciaet Gy- torsin cryptorchidism.Internationaljournal ofEp- necologicaScandinavica 59:495-497. idemiology20:964-970.

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions 146 THE QUARTERLY REVIEW OF BIOLOGY VOLUME 75

Medalie J H. 1957. Relationship between nausea Osman A K. 1985. Dietaiy practices and aversions and/or vomitingin early pregnancy and abor- duringpregnancy and lactationamong Sudanese tion. Lancet273:117-119. women. Journalof Tropical Pediatrics 31:16-20. Meyer L C, PeacockJ L, BlandJ M, Anderson H R. OtterbeinKF. 1994. Comment. CurrentAnthrolpology 1994. Symptomsand health problems in preg- 35:559-560. nancy: theirassociation with social factors,smok- Paarlberg K M, Vingerhoets A J J M, Passchier J, ing,alcohol, caffeineand attitudeto pregnancy. Heinen A GJJ,Dekker GA, van Geijn H P. 1996. Paediatricand PerinatalEpidemiology 8:145-155. Psychosocial factors as predictors of maternal Michanie S, Bryan F L, Fernandez N M, Vizcarra well-being and pregnancy-relatedcomplaints. M M, Taboada P D, Navarros 0, Alonso A B, Journalof Psychosomatic Obstetrics and Gynecology Santillan M L. 1988. Hazard analyses of foods 17:93-102. prepared byinhabitants along the PeruvianAm- Palmer R L. 1973. A psychosomaticstudy of vom- azon River.Journal of Food Protection51:293-302. iting of earlypregnancy. Journal of Psychosomatic MidkiffE E, BernsteinI L. 1985. Targets of learned Research17:303-308. food aversions in humans. Phlysiologyand Behav- Payton E, Crump E P, Horton C P. 1960. Growth ior34:839-841. and development. VII. Dietary habits of 571 MilgramN W, IKramesL, AllowayT M, editors. 1977. pregnant southern negro women. Journalof the Food AversionLearning. NewYork: Plenum Press. AmericanDietetic Association 37:129-136. MilkovichL, van den Berg BJ. 1976. An evaluationof Pelchat M L, Rozin P. 1982. The special role of nau- the teratogenicityof certain antinauseantdrugs. sea in the acquisition offood dislikesby humans. AmericanJournalof Obstetrics and Gynecology125: Appetite3:341-351. 244-248. PetittiD B. 1986. Nausea and MinturnL, WeiherAW. 1984. The influenceof diet pregnancyoutcome. on morningsickness: a cross-culturalstudy. Med- Birth13:223-226. Phillips T D, Clement B A, Kubena L R B. ical Anthropology8:71-75. F, Harvey Moore K L, Persaud T V N. 1998. The Developing 1990. Detection and detoxificationof aflatoxins: Human: ClinicallyOriented Embryology. Sixth Edi- prevention of aflatoxicosis and aflatoxin resi- tion. Philadelphia: W. B. Saunders. dues withhydrated sodium calcium aluminosili- Mori M, Amino N, Tamaki H, MiyaiK, Tanizawa 0. cate. Veterinaryand Human Toxicology32 (supple- 1988. Morning sicknessand thyroidfunction in ment) :15-19. normal pregnancy. Obstetricsand Gynecology72: PietersJJL. 1982. Nutritionalteratogens. A survey 355-359. of epidemiological literature.Pages 419-430 in Murdock G P. 1981. Atlas of WorldCultures. Pitts- Progressin Clinicaland BiologicalResearch: Epidemi- burgh (PA): Universityof PittsburghPress. ology,Early Detection and Therapy,and Environmen- Nesse R M, Williams G C. 1994. My We Get Sick: tal Factors,Volume 163B, edited by M Marois. TheNew Science of Darwinian Medicine. New York: NewYork: Alan R. Liss. Times Books. Pike I L. 1997. Pregnancy sickness and food aver- Nagao M, WakabayashiK, FujitaY, Tahira T, Ochiai sions during pregnancy for nomadic Turkana M, Sugimura T. 1986. Mutagenic compounds in women of Kenya. AmericanJournal of Physical An- soy sauce, chinese cabbage, coffee,and herbal thropology24 (supplement):186. teas. Pages 55-62 in GeneticToxicology of the Diet, PopeJ F, SkinnerJ D, Carruth B R. 1992. Cravings edited by I B Knudsen. New York: Alan R. Liss. and aversions of pregnant adolescents. Journal Nobmann E D, Adams S. 1970. Surveyof changes ofthe American Dietetic Association 92:1479-1482. in food habits during pregnancy. Public Health Posner L B, McCottryC M, Posner A C. 1957. Preg- Reports85:1121-1127. nancy cravingand pica. Obstetricsand Gynecology Oberg K. 1949. The Terenaand theCaduveo of South- 9:270-272. ernMato Grosso,Brazil. Washington (DC): Gov- PritchardJA, MacDonald P C. 1980. WilliamsObstet- ernmentPrinting Office. rics.16th Edition. O'Brien B, Zhou Q P. 1995. Variables related to NewYork: Appleton-Centuly- nausea and vomitingduring pregnancy. Birth 22: Crofts. 93-100. Profet M. 1988. The evolution of pregnancy sick- Ojofeitimi E 0, Elegbe I, Babafemi J. 1982. Diet ness as protection to the embryoagainst Pleisto- restrictionby pregnantwomen in Nigeria. Inter- cene teratogens.Evolutionary Theory 8:177-190. nationalJournalof Gynaecologyand Obstetrics20: ProfetM. 1992. Pregnancysickness as adaptation: a 99-103. deterrent to maternal ingestion of teratogens. Opsahl M, Hansen K, Klein T, Cunningham D. Pages 327-365 in TheAdapted Mind: Evolutionaiy 1994. Natural killer cell activityin earlyhuman Psychologyand theGeneration of Culture, edited by pregnancy. Gynecologicand ObstetricInvestigation J H Barkow et al. New York: Oxford University 37:226-228. Press.

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions JUNE 2000 MORNING SICKNESS 147

ProfetM. 1995. ProtectingYour Baby-to-Be: Preventing histamines:meta-analysis. AmericanJournal ofPer- BirthDefects in theFirstTrimester. New York: Addi- inatology14:119-124. son-WesleyPublishing. Shaw G M, Todoroff K, Velie E M, Lammer E J, Reddy RV, Reddy C S. 1993. Teratogenic potential 1998. Maternal illness, including fever, and of mycotoxins.Pages 270-309 in DietaryFactors medication use as risk factorsfor neural tube and BirthDefects, edited byR P Sharma. San Fran- defects. Teratology57:1-7. cisco (CA): Pacific Division AAAS. Shepard T H. 1992. Catalogof TeratogenicAgents. Sev- RileyG M. 1959. GynecologicEndocrinology.NewYork: enth Edition. Baltimore (MD): Johns Hopkins Paul B. Hoeber. UniversityPress. Roberts D. 1990. Trends in food poisoning. Food Sherman P W. 1988. The levels of analysis.Animal Scienceand TechlnologyToday 2:28-34. Behlaviour36:616-619. Robinson S, GodfreyK, Osmond C, Cox V, Barker Sherman P W, BillingJ. 1999. Darwinian gastron- D. 1996. Evaluation of a food frequency ques- omy:why we use spices. BioScience49:453-463. tionnaireused to assess nutrientintakes in preg- Sherman P W, Reeve H K. 1997. Forwardand back- nant women. EuropeanJournal of ClinicalNutri- ward: alternativeapproaches to studyinghuman tion50:302-308. social evolution. Pages 147-158 in Human Na- RodinJ, Radke-Sharpe N. 1991. Changes in appeti- ture:A CriticalReader, edited by L Betzig. New tivevariables as a functionof pregnancy. Pages York: Oxford UniversityPress. 325-340 in CGlemicalSenses. Appetite and Nutrition, Smith G C S, Smith M F S, McNay M B, FlemingJE. Volume 4, edited by M I Friedman et al. New 1998. First-trimestergrowth and the riskof low York: Marcel Dekker. birth weight. New England Journal of MIedicine RothmanKJ, Moore L L, SingerM R, NguyenU D T, 339:1817-1822. Mannino S, MilunskyA. 1995. Teratogenicityof SmithJ L. 1997. Long-termconsequences of food- high vitamin A intake. NewuEngland Journalof borne toxoplasmosis:effects on the unborn, the Medicine333:1369-1373. immunocompromised, the elderly,and the im- Rozin P, Vollmecke T A. 1986. Food likes and dis- munocompetent. Journal ofFood Protection60: likes. AntnualReview of Nutrition 6:433-456. 1595-1611. Sabahi F, Rola-PlesczcynskiM, O'Connell S, Fren- Snell L. 1996. Epidemiology and treatmentof nau- kel L D. 1995. Qualitativeand quantitativeanalysis sea and vomiting during pregnancy [Doctoral of T lymphocytesduring normal human preg- dissertation]. Buffalo (NY): State Universityof nancy. AmericanJournal of Reproductive Immunol- New York at Buffalo. ogy33:381-392. Sockett P N. 1995. The epidemiology and costs of Salmeron 0J, Vaquer S, Salmer6n I, Molt6 L, La- diseases ofpublic health significance,in relation pefia P, Manzano L, Alvarez de los Heros J I, to meat and meat products.Journal ofFood Safety Alvarez-Mon M. 1991. Pregnancy is associated 15:91-112. witha reduction in the patternof the cytotoxic Sokal R R, Rohlf F J. 1995. Biometiy:The Principles activitydisplayed by lymphokine-activatedkiller and Practiceof Statistics in BiologicalResearch. New cells. AmericanJournalof Reproductive Innmunology York: W. H. Freeman. 26:150-155. Speert H, GuttmacherA F. 1954. Frequencyand sig- SastryB V R, Horst M A, Naukam RJ. 1989. Maternal nificanceof bleeding in earlypregnancy.Journal tobacco smokingand changes in amino acid up- ofthe American Medical Association 155:712-715. takeby human placentalvilli: induction of uptake Spielmann KA. 1989. A review-dietary restrictions systems,gammaglutamyltranspeptidase and mem- on hunter-gathererwomen and the implications brane fluidity.Placenta 10:345-358. for fertilityand infantmortality. Human Ecology Saxen I. 1975. Epidemiology of cleftlip and palate. 17:321-345. BritishJournal of Preventiveand Social Medicine Stein Z, Susser M. 1991. Miscarriage,caffeine, and 29:103-110. the epiphenomena of pregnancy: the causal SchardeinJ L. 1985. ChemicallyInduced Birth Defects. model. Epidemiology2:163-167. NewYork: Marcel Dekker. StewartJ,Wheeler E, Schofield C. 1988. Regional Schuster K, Bailey L B, Dimperio D, Mahan C S.. differencesin Britishattitudes to diet in preg- 1985. Morning sickness and vitamin B6 status nancy: prioritiesand pragmatism.Ecology ofFood of pregnant women. Human Nutrition:Clinical and Nutrition20:211-229. Nutrition39C:75-79. Surh Y-J,Lee S S. 1996. Capsaicin in hot chili pep- Schwab E B, Axelson M L. 1984. Dietarychanges of per: carcinogen, co-carcinogen, or anticarcino- pregnant women: compulsions and modifica- gen? Food and ChemicalToxicology 34:313-316. tions. Ecologyof Food and Nutrition14:143-153. Tierson F D, Olsen C L, Hook E B. 1985. Influence Seto A, Einarson T, Koren G. 1997. Pregnancyout- of cravingsand aversions on diet in pregnancy. come followingfirst trimester exposure to anti- Ecologyof Food and Nutrition17:117-129.

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions 148 THE QUARTERLY REVIEW OF BIOLOGY VOLUME 75

Tierson F D, Olsen C L, Hook E B. 1986. Nausea Walker J R L. 1994. Antimicrobial compounds in and vomitingof pregnancyand association with food plants. Pages 181-204 in NaturalAntimicro- pregnancy outcome. AmericanJournal of Obstet- bial Systemsand Food Preseroation,edited by V M ricsand Gynecology155:1017-1022. Dillon and R G Board. Wallingford(UK): CAB Tierson F D, Olsen C L, Hook E B. 1989. Correction International. to Tierson et al. 1986. AmericanJournalof Obstet- WallaceJ M, Aitken R P, Cheyne M A. 1996. Nutri- ricsand Gynecology160:518-519. ent partitioning and fetal growth in rapidly Tinbergen N. 1963. On aims and methods of ethol- growingadolescent ewes.Journalof Reproduction ogy. ZeitschrzftftirTierpsychologe 20:410-433. and Fertility107:183-190. Todd E C D. 1994. Surveillance of foodborne dis- Walsh J W, Hasler W L, Nugent C E, Owyang C. ease. Pages 461-536 in FoodborneDisease Hand- 1996. Progesterone and estrogen are potential book:Diseases Caused byBacteria, Volume 1, edited mediators of gastric slow-wavedysrhythmias in byY H Hui et al. NewYork: Marcel Dekker. nausea of pregnancy.AmericanJournal of Physiol- Todd E C D. 1996. Worldwide surveillanceof food- ogy270:G506-G514. borne disease: the need to improve.JournalofFood Weigel M M, Weigel R M. 1989a. Nausea and vom- Protection59:82-92. iting of early pregnancy and pregnancy out- TriversR L. 1974. Parent-offspringconflict. Ameri- come. An epidemiological study.British Journal can Zoologist14:249-264. ofObstetrics and Gynaecology96:1304-1311. Tsang I S, Katz V L, Wells S D. 1996. Maternal and Weigel R M, Weigel M M. 1989b. Nausea and vom- fetaloutcomes in hyperemesisgravidarum. Inter- iting of early pregnancy and pregnancy out- nationalJournal of Gynecologyand Obstetrics55: come. A meta-analyticalreview. British Journal of 231-235. Obstetricsand Gynaecology96:1312-1318. TulchinskyD, Hobel CJ. 1973. Plasma human cho- Werler M M, Pober B R, Nelson K, Holmes L B. rionic gonadotropin, estrone, estradiol, estriol, 1989. Reporting accuracy among mothers of progesterone,and 17a-hydroxyprogesteronein malformed and nonmalformed infants.Ameri- human pregnancy.American Journal of Obstetrics canJournal of Epidemiology 129:415-421. and Gynecology117:884-893. Whitehead S A, AndrewsP L R, ChamberlainG V P. Uddenberg N, Nilsson A, Almgren P-E. 1971. Nau- 1992. Characterization of nausea and vomiting sea in pregnancy: psychological and psychoso- in earlypregnancy: a surveyof 1000 womenJour- maticaspects. Journal ofPsychosomaticResearch 15: nal ofObstetrics and Gynaecology12:364-369. 269-276. Wijewardene K, Fonseka P, Goonaratne C. 1994. Vellacott I D, Cooke EJ A,James C E. 1988. Nausea Dietary cravings and aversions during preg- and vomiting in early pregnancy. International nancy. IndianJournal of Public Health 38:95-98. Journalof Gynecology and Obstetrics27:57-62. WileyA S, Katz S H. 1998. Geophagy in pregnancy: VermeerD E, FerrellRE,Jr. 1985. Nigerian geopha- a test of a hypothesis. CurrentAnthropology 39: gical clay: a traditional antidiarrheal pharma- 532-545. ceutical. Science227:634-636. Wolkind S, Zajicek E. 1978. Psycho-socialcorrelates VigersAJ, RobertsW K, SelitrennikoffC P. 1991. A of nausea and vomitingin pregnancy.Journal of new familyof plant antifungalproteins. Molecu- PsychosomaticResearch 22:1-5. lar Plant-MicrobeInteractions 4:315-323. Wrangham R W, McGrew W C, DeWaal F B M, Walker A R P, Walker B F, Jones J, Verardi M, Heltne P G, editors. 1994. ChimpanzeeCultures. Walker C. 1985. Nausea and vomiting and di- Cambridge (MA): Harvard UniversityPress. etarycravings and aversions during pregnancy YerushalmyJ,Milkovich L. 1965. Evaluation of the in South Africanwomen. BritishJournalof Obstet- teratogeniceffect of meclizine in man. Awerican ricsand Gynaecology92:484-489. Journalof Obstetrics and Gynecology93:553-562.

This content downloaded from 152.14.136.96 on Sat, 7 Dec 2013 23:06:21 PM All use subject to JSTOR Terms and Conditions